! JhWit 19 if! : BIOLOGY INTERNATIONAL CHEMICAL SERIES H. P. TALBOT, PH. D., CONSULTING EDITOR APPLIED COLLOID CHEMISTRY GENERAL THEORY PUBLISHERS OF BOOKS F O R^, Coal Age * Electric Railway Journal Electrical World * Engineering News-Record American Machinist v Ingenieria Internacional Engineering 8 Mining Journal v Po we r Chemical & Metallurgical Engineering Electrical Merchandising iiiiiiiiiiiiiimiiiiiiiiiiiiiiimfiiiiinumiiininnniiiniiiiiiiiiiniiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii APPLIED COLLOID CHEMISTRY GENERAL THEORY BY WILDER D. BANCROFT // '/ PROFESSOR OF HHYSICAL CHEMISTRY AT 1 CORNELL, UNIVERSITY FIRST EDITION McGRAW-HILL BOOK COMPANY, INC. NEW YORK: 370 SEVENTH AVENUE LONDON: 648 BOUVERIE ST., E. C. 4 1921 COPYRIGHT, 1921, BY THE McGRAw-HiLL BOOK COMPANY, INC, MAPLK PRKSS T O K K PA PREFACE The earlier books on colloid chemistry presented the subject empirically because no other method was then possible. While we do not now know much about gelatinous precipitates and jellies, the theory of the rest of the subject is in fairly good shape and consequently I have written this book deductively. While my point of view differs considerably from Freundlich's, it is a pleasure to acknowledge my debt to him for what he did in his admirable book. This volume on general theory should be followed by at least one volume on each of the following subjects: silicate industries; paints and varnishes; plastics; fibers and dyeing; photochemistry and photography; petroleum industries; ore flotation and allied subjects; foods and beverages; soils and crops; biology and medi- cine. Allowing a minimum of three years to a volume, it seems very improbable that I shall write all of these single-handed. With the general theory of the subject once cleared up, there is no reason why the other volumes should not be written by people whose knowledge of the special subjects is much greater than my own. I am indebted to Messrs. Bray; Briggs, Clowes, Holmes, Johnston and Lamb for reading the manuscript and for critical comments on it. CORNELL UNIVERSITY, November 1, 1920. 432049 TABLE OF CONTENTS PAGE INTRODUCTION 1 CHAPTER I ADSORPTION OF GAS OR VAPOR BY SOLID ........ 3 Adsorption of Gas or Vapor by Solid 3 The Adsorption Isotherm 29 Chemical Reactions 40 CHAPTER II ADSORPTION OF VAPOR BY LIQUID, AND OF LIQUID AND SOLID BY SOLID AND LIQUID 65 Adsorption of Vapor by Liquid 65 Adsorption of Liquid by Solid 67 Adsorption of Solid by Solid v 78 Adsorption of Liquid by Liquid 81 Adsorption of Solid by Liquid 83 CHAPTER III ADSORPTION FROM SOLUTION 90 Adsorption from Solution by Solid 90 The Adsorption Isotherm 100 Abnormal Adsorption 106 Negative Adsorption 108 Reversibility of Equilibrium 110 Specificity of Adsorption Ill Adsorption of Several Solutes 113 Chemical Action . . . , . . ...,. . . . 120 Adsorption from Solution by Liquid 126 CHAPTER IV SURFACE TENSION BROWNIAN MOVEMENTS 129 Adsorption and Surface Tension 129 Brownian Movements . . . , . . ... . . . . . . . V . . 136 CHAPTER V COALESCENCE '. ............ 146 Coalescence of Liquids 146 Coalescence of Solids 147 Types of Precipitates . . ... . '. . . . 161 Theory of Peptization .... '.'..../.*'. 166 vii viii CONTENTS CHAPTER VI PAGE PREPARATION OF COLLOIDAL SOLUTIONS 172 Condensation Methods 172 Dispersion Methods 177 CHAPTER VII PROPERTIES OF COLLOIDAL SOLUTIONS 182 General Properties 182 Viscosity 190 Color. " 196 Electrical Properties 206 Stability of Colloidal Solutions 212 CHAPTER VIII JELLIES AND GELATINOUS PRECIPITATES 236 Structure 236 Vapor Pressure Relations 245 Diffusion 258 CHAPTER IX EMULSIONS AND FOAMS 260 Emulsions . . . 260 Foams 268 CHAPTER X NON- AQUEOUS COLLOIDAL SOLUTIONS 274 CHAPTER XI FOG AND SMOKE 276 Fog v 276 Smoke . 286 Precipitation of Solid ...... . ' 300 CHAPTER XII GASES AND SOLIDS IN SOLIDS ..,,.. 304 Gases in Solids. . . . . . . . . ,. .... . r . 304 Solids in Solids . \\ .. " 305 CHAPTER XIII THICKNESS OF SURFACE FILMS '. 314 AUTHOR INDEX 321 SUBJECT INDEX . 333 APPLIED COLLOID CHEMISTRY INTRODUCTION In 1861 Graham 1 pointed out that substances like potassium hydroxide, potassium sulphate, magnesium sulphate, sugar, and alcohol diffuse much more rapidly in water than hydrous silicic acid, hydrous alumina, starch, dextrin, the gums, albumin, tannin, gelatine, etc. Since these latter substances occur usually in a non-crystalline or even gelatinous form, Graham suggested calling them colloids from the French word for glue. Substances which diffused rapidly were to be called crystalloids because many of them crystallized readily. Graham believed that the distinction between a crystalloid and a colloid was fundamental and was due to some molecular condition. Though modern colloid chemistry begins with Graham, 2 his distinction between crystalloids and colloids has been dropped. A colloidal substance is not neces- sarily amorphous, for colloidal gold is certainly crystalline at times and possibly always. We now speak of a colloidal state instead of a colloidal substance, and we call any phase colloidal when it is sufficiently finely divided or dispersed, 3 without com- mitting ourselves definitely as to what degree of subdivision is necessary in any particular case. The subdivided phase is often called the internal or dispersed phase and the enveloping phase the external phase or the dispersing phase. In fog the rain drops are the dispersed phase and the air the dispersing phase, while the solid particles are the internal phase in the case of smoke. When a milky liquid is obtained by precipitating barium sulphate in the cold, the barium sulphate is the dispersed 1 Phil. Trans. 151, 183 (1861); Jour. Chem. Soc., 16, 216 (1862); 17, 318 (1864). 2 Interesting work was done by Selmi in 1844; but it had no appreciable effect on the development of scientific thought. 3 Cf. Wolfgang Ostwald: Grundriss der Kolloidchemie, 79 (1909). 1 CHEMISTRY phase aot-Mfe' Solution the dispersing phase. In emulsions of oil in water, the oil is the internal phase because it is present in drops. When benzene is emulsified to a solid jelly in a small amount of soap and water, we have a mobile liquid as dispersed phase in a viscous mass. Bread or Ivory soap might be con- sidered as a system with air as internal phase, and a fairly solid external phase. In a porous plate or a fine sponge the two phases are, or may be, continuous and it is therefore better to classify these as interlacing systems. True ruby glass consists of gold as dispersed phase in glass as a non-crystalline dispersing phase. With salt colored by exposure to cathode rays, we have metallic sodium as internal phase in an external phase of crystalline sodium chloride. Adopting the very flexible definition that a phase is called col- loidal when it is sufficiently finely divided, colloid chemistry is the chemistry of bubbles, drops, grains, filaments, and films, because in each of these cases at least one dimension of the phase is very small. This is not a truly scientific classification because a bubble has a film round it and a film may be considered as made up of coalescing drops or grains. It is a convenient classi- fication, however for many purposes. At first sight colloid chemistry may not seem to be an important branch of chemistry, either theoretically or technically; but this opinion changes when we consider that a knowledge of colloid chemistry is essential to anybody who really wishes to understand about : cement, bricks, pottery, porcelain, glass, enamels; oils, greases, soaps, candles; glue, starch, and adhesives; paints, varnishes, lacquers; rubber, celluloid, and other plastics; leather, paper, textiles; filaments, casts, pencils, and crayons; inks; roads, foundry cores, coke, asphalt; graphites, patines; zinc, phosphorus, sodium, and alumi- num; contact sulphuric acid, hardened oils, etc.; beer, ale, and wine; cream, butter, cheese, and caseine products; cooking, washing, dyeing, printing; ore flotation, water purification, sew- age disposal; smoke prevention; photography; wireless telegraphy; illuminants; comets; pharmacy; physiology. In other words, colloid chemistry is the chemistry of e very-day life. CHAPTER I ADSORPTION OF GAS OR VAPOR BY SOLID ADSORPTION OF GAS OR VAPOR BY SOLID In an extremely subdivided phase, the surface is very large relatively to the mass of the phase and it is therefore advisable to discuss the properties of surfaces, beginning with the behavior of solids in contact with gases or vapors. All solids tend to adsorb, or condense upon their surface, any gases or vapors with which they are in contact. The amount of adsorption for given conditions of pressure and temperature varies with the nature and physical state of the solid and with the nature of the gas; in other words, the adsorption is specific or selective. With the same 'solid and the same gas the amount of adsorption is greater the higher the pressure of the gas and the lower the temperature. Hunter 1 showed that many different gases and vapors are ad- sorbed by charcoal to a marked extent. While his experiments are probably not very accurate, and are not made at strictly comparable temperatures and pressures, they cover more ground than those of anybody else. Some of his data are given in Table I. In Table II are some data by Dewar 2 with cocoanut charcoal at low temperatures. The volumes of gas adsorbed have been calculated back to standard conditions ; in other words to the volumes which the adsorbed gases would occupy at and 760 mm pressure. Thus at 185 and 760 mm the actual vol- umes of gas adsorbed would be only 88/273, or approximately one-third, of the values given in Table II. This correction was not applied by Hunter in his work. iPhil. Mag. (4) 26, 364 (1863); Jour. Chem. Soc., 18, 285 (1865); 20, 160, (1867); 21, 186 (1868); 23, 73 (1870); 24, 76 (1871); 25, 649 (1872). 2 Proc. Roy. Soc., 74, 124 (1904). 3 APPLIED COLLOID CHEMISTRY TABLE I. ADSORPTION OF GASES AND VAPORS BY COCOANUT CHARCOAL Volumes gas adsorbed per volume charcoal. Temperature in Celsius degrees. Pressures in millimeters of mercury. Gas Vol. Temp. Press. Gas Vol. Temp. Press. Acetic acid Water Water 83 24 44 159 159 128 676 24 624 Acetaldehyde Acetaldehyde Ammonia 67 139 22 155 100 127 687 681 655 Benzene . . 59 130 646 Ammonia . 83 70 760 Ethyl alcohol Ethyl alcohol. . 84 111 160 127 663 644 Ammonia Ammonia 88 96 60 50 760 760 Ethyl alcohol 141 100 652 Ammonia 114 40 760 Ethyl acetate Carbon tetrachloride. . Carbon tetrachloride. . Methyl alcohol Methyl alcohol Methyl alcohol 72 4 8 61 127 151 21 154 155 100 159 129 100 159 659 695 636 684 663 652 660 Ammonia Ammonia Cyanogen Cyanogen Cyanogen Cyanogen Cyanogen 149 176 87 92 96 99 107 20 70 60 50 40 20 760 760 760 760 760 760 760 Chloroform Acetone ... . 30 68 100 157 651 671 Cyanogen Hydrochloric acid 114 60 100 760 668 Acetone 105 100 642 Nitrous oxide 64 100 661 Carbon bisulphide. . Carbon bisulphide .... Ethyl ether 91 117 54 158 100 160 659 671 684 Carbon dioxide Carbon dioxide .Nitrogen 17 71 15 127 3 687 760 760 Ethyl ether Ethyl ether 68 87 128 100 661 644 Hydrogen 4 3 760 TABLE II. ADSORPTION OP GASES BY COCOANUT CHARCOAL Volumes gas adsorbed per volume charcoal. Pressures not given; volumes corrected to and 760 mm. Temperature in Celsius degrees. Gas Vol. Temp. Gas Vol. Temp. Argon 12 o Nitrogen 15 Argon 175 185 Nitrogen . 155 -185 Oxygen 18 o Hydrogen 4 Oxygen 230 185 Hydrogen 135 -185 Carbon monoxide. 21 Helium 2 Carbon monoxide . . . 190 -185 Helium 15 -185 In Table III are given some more recent data by Hempel and Vater, 1 who used a special animal charcoal, which they mixed to 1 Zeit. Elektrochemie, 18, 724 (1912). ADSORPTION OF GAS OR VAPOR BY SOLID a stiff paste with ox-blood diluted tenfold with water and which they then charred at 600. TABLE III. ADSORPTION OP GASES BY SYNTHETIC CHARCOAL Volumes gas adsorbed per volume synthetic 600 charcoal. Volumes probably corrected to and 760 mm. + 20 -78 -185 + 20 H 2 7 3 19.5 284.7 H 2 S 213 N 2 21.0 107.4 632.2 C1 2 304.5 CO 26.8 139.4 697.0 NH 3 197.0 O 2 ....... 25 4 122 4 SO 2 337 8 + 20 -78 + 20 -78 Ar 12.6 92.6 CH 4 41.7 174.3 NO 103 6 231.3 C 2 H 4 139.2 360 7 N 2 O 109.4 330.1 C 2 H 6 119.1 275.5 CO 2 83.8 568.4 C 2 H 2 135.8 488.5 TABLE IV. BOILING-POINTS OF VARIOUS LIQUIDS Liquid B. Pt. Liquid B. Pt. Liquid B. Pt. He. -268 7 C 2 H 6 84 CH 3 CHO +20 8 H 2 252 6 HCl -82 9 (C 2 H 6 ) 2 +34 6 N 2 195 7 C 2 H 2 -82 4 1 CS 2 +46 2 CO -190 CO 2 -79 1 Br 2 + 58 6 Ar -186.1 H 2 S -60 2 CHC1 3 +61 2 2 CH 4 -182.8 -164 C1 2 NH 3 -33.7 -33 5 CH 3 OH CC1 4 +64.7 + 76 7 NO -153 (CN)a -21. C 2 H 5 OH +78 4 C 2 H 4 -105 SO 2 10 C 6 H 6 +80 2 N 2 O - 89.8 H 2 O + 100.0 For purposes of reference there are given in Table IV the boil- ing-points of some of the substances whose adsorption by carbon has been studied. As a first approximation it is often stated that a gas or vapor is adsorbed more readily the higher its boiling- point. Thus helium is taken up by charcoal much less than hy- drogen, and hydrogen again is adsorbed to a. much less extent 1 Sublimation point. 6 APPLIED COLLOID CHEMISTRY than nitrogen or oxygen. Carbon dioxide is adsorbed less readily than ammonia, so that these substances follow the em- pirical rule. Argon, however, is adsorbed less completely by charcoal than is nitrogen, while carbon monoxide is adsorbed to a greater extent at than either argon or oxygen, though according to the rule this ought not to be so. Nitrous oxide is adsorbed less strongly than ethylene, and nitric oxide more strongly than methane, which is not according to the boiling- points. Ethane, ethylene, and acetylene are adsorbed more at +20 than is carbon dioxide, though the last is the most readily condensible gas of the four. The difference between carbon di- oxide and hydrogen sulphide is in the right direction; but seems out of all proportion to the difference in boiling-points. Hydrogen sulphide is adsorbed more than ammonia, which is wrong; and chlorine is adsorbed much more strongly than ammonia, although the two boiling-points are practically identical. Cyanogen is ad- sorbed more than ammonia at 70 and less at 0. In the case of vapors there is no relation between boiling-point and adsorption. Going from higher to lower boiling-points, we have the order: water, benzene, ethyl alcohol, carbon tetrachloride, methyl alcohol, chloroform, ether, and aldehyde. The order from greater to lesser adsorption is methyl alcohol, ethyl alcohol, aldehyde, ether, benzene, water, chloroform, and carbon tetrachloride. The truth of the matter is that adsorption is specific and varies with the nature of the gas and of the adsorbing solid. One charcoal is not like another. In Table V are Hunter's results on charcoal TABLE V. ADSORPTION OF GASES BY CHARCOAL FROM DIFFERENT WOODS Volumes gas adsorbed per volume charcoal. Temperature = 0C; pressure = 760 mm. Wood Vol. NH Vol. C0 2 Vol. (CN). Wood Vol. NH6 Vol. C0 2 Vol. (CN) 2 Logwood Ebony. . . . 111 107 55 47 87 90 Jamaica logwood .... Sapan wood 69 70 33 32 Camwood 91 45 Beech 58 Green ebony 90 41 Rosewood . 51 Fustic (Cuba) .... Lignum vitse Boxwood 90 89 86 58 47 31 29 Wistaria sinensis .... Vegetable ivory Cocoanut shell 44 176 50 71 57 114 ADSORPTION OF GAS OR VAPOR BY SOLID 7 made from different woods. The cocoanut charcoal has the greatest adsorbing power of all and, of the others, logwood char- coal has the greatest adsorbing action on ammonia, fustic charcoal on carbon dioxide, and ebony charcoal on cyanogen. It is not the most porous charcoal in the ordinary sense which adsorbs the most gas, because Hunter says that the charcoal made from the shell of the cocoanut is very dense and brittle. The pores are quite invisible and, when the charcoal is broken, the fracture presents a semi-metallic lustre. It has been suggested by De Saussure 1 that charcoals with very many pores of very small diameter are the best adsorbers because they present the larg- est surfaces for condensation. This is not quite right because there are other factors besides the extent of surface, among them the arrangement of the surfaces. A plane surface will adsorb a given amount of a gas under standard conditions, but two plane surfaces placed close to each other will apparently adsorb a good deal more gas than the same surfaces would if each did not re-enforce the other and modify the concentration gradient. The adsorption depends not only on the surface but on the structure of the the adsorbing material. Narrow pores, down to a cer- tain limit, should adsorb more than spherical pores of large dia- meter and this actually happens. Table VI shows the change in the adsorption of air 2 when different charcoals are heated from -100 to +500. In a number of cases the parallelism between density and adsorption holds but by no means in all cases. While there may be an actual difference in the amount of sur- face, we are undoubtedly dealing with another factor, the pres- ence in the charcoal of substances which affect the adsorption. Charcoal is far from being pure carbon and the great improve- ments in charcoal as used in warfare have been along the line of cleaning the charcoal very thoroughly in addition to increasing the surface. While cocoanut charcoal is usually considered the best, the sour cherry charcoal in Table VI does better; and the synthetic charcoal of Hempel and Vater 3 was better than the best cocoanut charcoal which could then be obtained, though far inferior to the best cocoanut charcoal of today. The varying 1 Cf. DEWAR: Proc. Roy. Inst., 18, 180 (1905). 2 PIUTTI and MAGLIA: Gazz. chim. ital., 40 I, 569 (1910). 3 Zeit. Elektrochemie, 18, 724 (1912). 8 APPLIED COLLOID CHEMISTRY behavior of charcoal from cocoanut shell, logwood, fustic, and ebony is probably due to the presence of different impurities which affect different gases in different ways. TABLE VI. BEHAVIOR OF DIFFERENT CHARCOALS Apparent density cc gas per cc charcoal 100 to +550 Air Cli Prunus cerasus Cocus nucif era Phillyrea media Sour cherry 1 1.6393 1.4497 1.4970 1.4383 1.1901 1.3000 1.4192 1.3398 1.2802 1 . 2359 1.6274 1.3994 1.3132 1.3114 1.2218 280.1 251.8 222.4 212.7 199.5 186.3 138.5 134.5 134.1 133.0 131.5 129.5 118.6 111.1 99.9 312.7 334 . 2 260.0 Cocoanut 1 Jasmine box Silk oak White spruce Grevillea robusta Abies alba Castanea vesca Buxus sempervirens . . . Liospunos kaki Populus nigra European chestnut . . Common box Chinese persimmon . . Black poplar Prunus virginiana Ceratonia siliqua Prunus armeniaca Juglans regia Pinus rigida. . . Robinia pseudo-acacia . Choke-cherry Carob 1 Apricot 1 English walnut Pitch-pine Common locust Charcoal is used in most gas masks, because it is the best all- round adsorbent for toxic gases. 2 The removal of the toxic materials must be surprisingly complete. The adsorbent must be capable of reducing the concentration of the toxic gas from, say 1000 p. p.m. 3 to 1 p.p.m. or less within the 0.1 second that the air takes in passing through the canister. This is accomplished with a safe margin by the present gas mask materials. In fact it has been shown that charcoal will reduce a concentration of 7000 p.p.m. of chloropicrin, CC1 3 NO 2 , in a rapidly moving cur- rent of air to less than 0.5 p.p.m. in something under 0.3 seconds. Some use has been made in peace times of the fact that properly 1 Shells and kernels. 2 LAMB, WILSON and CHANEY: Jour. Ind. Eng. Chem., 11, 420 (1919). 3 Parts per million by volume. ADSORPTION OF GAS OR VAPOR BY SOLID 9 prepared charcoal will adsorb gases to a considerable extent. Melsens 1 placed wood charcoal, saturated with dry chlorine, in the longer branch of a bent sealed tube and dipped that end of the tube into boiling water, while the shorter branch was placed in a freezing mixture. A portion of the gas was volatilized and the pressure thus developed caused liquefied chlorine to appear in the cold portion of the tube. Sulphur dioxide, hydrogen sulphide, hydrobromic acid, and cyanogen can be liquefied in the same way; but not bromine, hydrochloric acid, carbon bi- sulphide, ether, or alcohol. In 1874 Tait and Dewar 2 described the use of charcoal as a means of getting a high vacuum. They placed a piece of char- coal in a tube having platinum electrodes, heated the tube to a red heat, exhausted it with a Sprengel pump and sealed it. If the tube was cool, no spark would pass when using an induction coil capable of giving a quarter of an inch spark. On heating the tube slightly some gas was given off from the charcoal and the spark passed. By making use of the increased adsorption at low temperatures quite surprising results Were obtained. 3 A bulb of 300 cc capacity, containing air at 15 and an initial pressure of 1.7 mm, was connected with another bulb holding five grams of charcoal. When the charcoal was cooled in liquid air, the pressure dropped to 0.00005 mm. A Crookes radiometer tube, filled with hydrogen at atmospheric pressure, had a charcoal bulb attached. On dipping the charcoal bulb into liquid air, no rotation took place even when the beam from an electric arc was focussed on the vanes. The motion became quite rapid however when the charcoal bulb was placed in liquid hydrogen because the adsorption of hydrogen by charcoal is sufficient at these very low temperatures, though not at the temperature of liquid air. The adsorption of helium by charcoal at -252 is not enough to make this experiment successful when the radio- meter tube is filled with helium. Since different gases are adsorbed in different amounts, the ratio of two gases in the charcoal will differ from the ratio in the gas phase and a separation will be possible to some extent. 1 Comptes rendiis, 77, 781 (1873); Jour. Chem. Soc., 27, 120 (1874). 2 Proc. Roy. Soc. Edin., 8, 348, 628 (1874). 3 DEWAR: Proc. Roy. Inst., 18, 437, 756 (1906). 10 APPLIED COLLOID CHEMISTRY Dewar 1 heated about 50 grams of charcoal and then saturated it at -185 in a current of pure dry air at atmospheric pressure, at 76 mm, and at 5 mm. The vessel containing the charcoal was taken out of the liquid air and the temperature allowed to rise to 15. The volumes of gas given off were 5.7, 4.8, and 4.5 liters respectively and the oxygen concentration varied irregu- larly between 56 and 58 percent. These experiments show the practical constancy of the mean composition of the gases ad- sorbed by the charcoal under the conditions of these experiments and they therefore show that wide changes in the pressure of the air current have but slight effect in altering the relative propor- tions of the adsorbed gases. In another experiment the vessel containing the saturated charcoal was placed in a vacuum vessel containing a little liquid air so that the temperature might rise slowly. This made it possible to collect and analyze separately the successive liters of gas as evolved. The data are given in Table VII. The mean composition of the six liters was 56 per- TABLE VII. DISTILLATION OF AIR ADSORBED BY CHARCOAL First liter Second liter Third liter 18.5 percent O 2 20. 6 percent O 2 53 percent O 2 Fourth liter Fifth liter Sixth liter 72.0 percent O 2 79.0 percent O 2 84 percent O 2 cent oxygen. A few experiments were also made with special mixtures of nitrogen and oxygen. When a gas mixture contain- ing 6.5 percent oxygen was used and the charcoal was heated rapidly to 15, five liters of gas were obtained, containing 23 per- cent oxygen. At higher temperatures the results seem to be quite different, though this may be due to the removal of oxygen by combustion. Skey 2 found that red-hot charcoal retains over three volumes of nitrogen which is given off when the cooled charcoal is immersed in water. Storey and Lewis 3 state that it is difficult to remove adsorbed nitrogen and carbon dioxide from coke even by heating 1 Proc. Roy. Soc., 74, 126 (1904); Proc. Roy. Inst., 18, 184 (1905). 2 Chem. News, 15, 15, 27 (1867). 3 Am. Chem. Jour., 4, 409 (1883). ADSORPTION OF GAS OR VAPOR BY SOLID 11 the containing flask to 200 and exhausting with a Sprengel pump. In these two cases the oxygen was probably burned com- pletely. It is not so easy, however, to account for Blumtritt's results. 1 He heated various forms of charcoal under mercury to 140 in a paraffin bath and analyzed the gases which were given off. With common wood-charcoal the gas was pure ni- trogen. After the same charcoal had been moistened and dried, the evolved gas contained 85.6 percent N 2 , 2.1 percent O 2 , 9.2 percent CO 2 , and 3.1 percent CO. The gas from poplar charcoal consisted of 83.5 percent N 2 , and 16.5 percent CO 2 , that from crushed alder charcoal had the composition 78.9 percent N 2 and 21.1 percent CO 2 , while charcoal made from ash yielded a gas consisting of 76.0 percent N 2 , 9.1 percent CO 2 , and 14.9 percent 2 . These variations, if real, probably depend on the nature and amount of the hydrocarbons left in each char- coal; but the experiments should be repeated. Craig 2 heated a Jena flask filled with lignite coke tp redness, closed the flask and analyzed the contents. The unadsorbed gas was nearly pure nitrogen and so was the adsorbed gas, which meant that the oxygen had oxidized some unspecified hydrocarbon without burning it to CO or C02. In Dewar's experiments the oxygen adsorbed at 185 was apparently given off at 15 without any appreciable contamina- tion with CO or CO 2 . Calvert 3 observed no reaction at ordinary temperatures, when carefully purified pieces of boxwood char- coal were heated to redness and, while still red-hot, were intro- duced into a cylinder standing over mercury and containing a measured volume of oxygen. Passing through the mercury undoubtedly cooled the charcoal a great deal before it came in contact with the oxygen. By special experiments Calvert satis- fied himself that no carbonic acid had been formed, or existed, in the cube of charcoal. Since the adsorbed oxygen oxidized ethyl alcohol to acetic acid, and ethylene to carbon dioxide and water, it is certain that there cannot have been a serious error in analysis, because such oxidations could not be caused by carbon monoxide or carbon dioxide. 1 Jour, prakt. Chem., 98, 418 (1886). 2 Chem. News, 90, 109 (1904). 3 Jour. Chem. Soc., 20, 293 (1867). 12 APPLIED COLLOID CHEMISTRY Quite different results were obtained by C. J. Baker 1 working with purified blood charcoal which was allowed to adsorb oxy- gen at 15 and was then pumped out. On heating the tube to 100 seven volumes of CO 2 were obtained when moist oxygen was used, while no gas was given off when dry oxygen was used. In a check experiment it was shown that no carbon dioxide was formed when water vapor and carbon were heated to 100 in a sealed tube for a week. The question arose whether dry oxygen was adsorbed or whether it was held more firmly than moist oxygen. Practically no gas was given off up to 350. At 450 a regular stream of gas was given off which consisted of 99 percent CO and 1 percent CO 2 in those cases where the drying had not been done so carefully. Intermediate values were found for intermediate degrees of drying. Rhead and Wheeler 2 do not confirm Baker's results though this may be because they did their experiments differently. Rhead and Wheeler treated purified carbon at some definite temperature, 200, 300, etc., with air, pumped out the air, and then heated the charcoal more or less rapidly to 1100 and analyzed the gas. With an initial temperature of 300 and rapid heating to 1100 the composition of the evolved gas was 11.3 percent CO 2 and 88.4 percent CO when dried air was used, and 15.4 percent CO 2 plus 84.4 percent CO when moist gas was used. When the initial temperature was 300, the gas coming off between 300 and 400 was composed of 57.0 percent CO 2 and 42.4 percent CO; between 400 and 500 the composition was 41.6 percent CO 2 and 58.0 percent CO, dropping to 0.8 percent C0 2 and 98.7 percent CO for the gas drawn off between 800 and 1100. Baker found 1 percent C0 2 and 99.0 percent CO at 450; but his initial temperature was 12. The two sets of experiments should be repeated so that we might know whether they are or are not contradictory. Neither Rhead and Wheeler nor Baker found any oxygen in the gas given off above 200. By determining the oxidizing action at different temperatures of charcoal containing adsorbed oxy- gen, it should be possible to determine approximately at what temperature all the oxygen was converted into CO and C0 2 . In spite of all the excellent work that has been done on the com- 1 Jour. Chem. Soc., 61, 249 (1887). 2 Ibid., 103, 461, 1210 (1913). ADSORPTION OF GAS OR VAPOR BY SOLID 13 bustion of carbon, there is still a great deal that we do not know. Hulett has found that there is a slow reaction between oxygen and charcoal at ordinary temperatures, apparently the formation of a non-volatile oxide. When one blows a slowly dying wood fire with the bellows, the fire will often start burning briskly again, in spite of the fact that the large influx of cooler air must tend to chill the embers and thus to put the fire out. Bacon 1 suggests that the dying embers are covered with an adsorbed film of carbon dioxide which is swept away to some extent by the draught from the bellows. When the fire is burning briskly, there is less carbon dioxide adsorbed at the higher temperature and the natural draught is sufficient to bring oxygen to the burning wood. Though this hypothesis has not been proved, it is rendered more probable by the experi- ments of Arndt and Schraube, 2 who found that even at a red heat carbon dioxide is adsorbed more strongly than carbon mon- oxide or nitrogen. Manville's fluptuations of the ignition point of charcoal 3 are probably due to hydrocarbons which had not been removed. Ramsay 4 has availed himself of Dewar's method of using cooled cocoanut charcoal as an adsorbent for gases to determine the amounts of neon and helium in the air. At 100 the oxygen, argon, and nitrogen are taken out practically quantitatively while the adsorption of neon and helium is negligible. At 185 practically all of the neon is adsorbed and none of the helium. From his observations, Ramsay concludes that the concentra- tion of neon in the air is about 0.00123 volume percent and that of helium about 0.00044 volume percent, while the amount of free hydrogen cannot be greater than one five-hundredth of the combined volumes of the neon and helium. At great elevations this changes completely, the gas with the lowest molecular weight concentrating in the higher regions. Alleman 5 gives some data for the estimated concentrations of hydrogen at different heights above sea-level, Table VIII. 1 Jour. Phys. Chem., 17, 768 (1913). 2 Nernst Festschrift, 46 (1912). 3 Jour. Chim. phys., 6, 297 (1907). *Proc. Roy. Soc., 76A, 111 (1905); 80A, 599 (1908). * Jour. Franklin Inst., 186, 161 (1918). 14 APPLIED COLLOID CHEMISTRY TABLE VIII. ESTIMATED DISTRIBUTION OF HYDROGEN IN THE ATMOSPHERE Height in kilometers Percent hydrogen Height in kilometers Percent hydrogen Height in kilometers Percent hydrogen 140 99.2 80 64.7 20 0.04 130 99.0 70 32.6 15 0.02 120 98.7 60 10.7 11 0.01 110 98.1 50 2.80 5 0.01 100 95.6 40 0.67 90 88.3 30 0.16 Dewar 1 has worked out a very effective lecture experiment, based on the selective adsorption by charcoal. A number of spectroscopic tubes, connected in series with a large U-tube con- taining charcoal are exhausted by cooling the charcoal until the electric discharge will barely pass. The charcoal tube is now placed in liquid air and a current of air is allowed to enter the system slowly, passing first through the charcoal tube which takes out the readily adsorbed gases such as oxygen, nitrogen, and argon, while helium, neon and hydrogen pass through. When the pressure in the first tube has risen sufficiently it begins to glow with the well-known rich orange hue of neon. In time the characteristic discharge of neon and helium appears in the other tubes. The sensitiveness of the neon tubes to induced electric oscillations from a coil of wire placed at right angles to the tube may be shown. Claude 2 points out that the remarkable luminosity of neon can be utilized for lighting purposes. One of the difficulties of the problem is the ease with which neon is masked by small quantities of certain other gases. It is not enough to introduce very pure neon into a tube with electrodes under suitable pressure in order to obtain an effectively luminous tube. This can be overcome by removing these other gases with cooled charcoal, when the beautiful orange luminescence of neon appears, and retains its brightness. The light is very rich in red rays and Claude considers it just the corrective required for the light of mercury vapor tubes. 1 Proc. Roy. Inst., 18, 444 (1906). 2 Mon. Sci., 74, 135; Jour. Franklin Inst., 171, 624 (1911). ADSORPTION OF GAS OR VAPOR BY SOLID 15 Krypton and xenon may be obtained from the air by means of cooled charcoal. 1 A current of air passes through a series of tubes immersed in liquid air for the purposes of purification, the last tube containing cotton wool in order to retain any dust of the solid, condensed impurities. This pure air is passed through a tube containing about 100 grams of charcoal for at least 24 hours. The charcoal tube is then removed and placed in solid carbonic acid, and the gas which is evolved is allowed to escape. The gas remaining in the charcoal at 78 is driven off by heat- ing and exhaustion, and all the carbon compounds and oxygen removed from it. The remaining gas, consisting of nitrogen, krypton, and xenon, is separated into its constituents by conden- sation and fractionation. Instead of passing a current of air over charcoal at 183, a few hundred grams of charcoal may be covered with liquid air and the latter may be allowed to evaporate in a silvered vacuum vessel. The gases remaining in the charcoal can then be separated as before. In this way spec- trum tubes of krypton and xenon may be readily prepared. Rutherford 2 found that cocoanut charcoal adsorbs radio-active emanations. A slow current of air charged with the emanations of radium, thorium, or actinium is deprived of some or all of its emanation in the passage through a tube filled with cocoanut charcoal, and for this purpose the charcoal need only be at ordinary temperature. The complete withdrawal of the eman- ation only takes place when the air is passed very slowly through the charcoal; if the speed is increased, the amount of unadsorbed emanation increases also. In another experiment Rutherford found that if a tube containing less than a gram of charcoal is connected to a vessel containing the emanation from several milligrams of radium bromide, the charcoal will adsorb the emanation in time. If some powdered willemite be mixed with the charcoal, the gradual adsorption is shown by the increas- ing brilliancy of phosphorescence of the willemite. It is not necessary to heat or exhaust the charcoal previously; but if this is done the emanation is adsorbed more rapidly. The charcoal retains the emanation at ordinary temperature but the greater 1 VALENTINER and SCHMIDT: Sitzungsber. Akad. Wiss. Berlin, 1905, 816; DEWAR: Proc. Roy. Inst., 18, 445 (1906). 2 Nature, 74, 634 (1906). 16 APPLIED COLLOID CHEMISTRY part of it is expelled by heating to a low red heat. Boyle 1 showed that cocoanut charcoal adsorbed thorium emanation the most and ordinary charcoal the least, while animal charcoal came in between the other two. The rate of adsorption increased very much when cocoanut charcoal was used at 75.5 and only slightly in the case of wood charcoal. At 211 cocoanut char- coal begins to give off carbon dioxide ; but it shows some power of adsorbing thorium emanation up almost to 300. It is not possible for the charcoal to become saturated with thorium emanation as may happen with other gases, because the thorium emanation adsorbed by the charcoal is decaying at the rate of half-value in 54 seconds. The general rule for mixtures of gases seems to be 2 that the more readily adsorbed gas displaces the other to some extent and is adsorbed to a greater relative amount than one would have predicted from experiments on the single gases. This cannot be universally true. Bergter 3 found that at pressures of 0.5 to 10 mm oxygen is adsorbed 30 to 40 times as strongly as nitrogen, and that at these pressures the presence of oxygen increases the amount of nitrogen adsorbed. With readily condensible gases, one may have one of the vapors helping to carry down the other. Cases of this sort have been studied by Hunter, 4 the most strik- ing instance being water and ammonia. It seems probable that the adsorbed water dissolves or adsorbs ammonia thus increas- ing the amount of ammonia apparently held by the charcoal. It has been shown by E. E. Reid that the presence of traces of benzene or ether in charcoal will increase the service time against chloropicrin. The data in Table IX are not tabulated as one would like to have them; but they do indicate that the presence of a second gas will cut down the adsorption of the first, 5 though such com- parisons should be made at equal partial pressures. Passing one gas over charcoal which has adsorbed another gas should theo- 1 Jour. Phys. Chem. 12, 284 (1908); Phil. Mag. (6) 17, 374 (1909). See also SATTERLY: Phil. Mag. (6) 16, 584 (1908); 20, 778 (1910); PORLEZZA and MORSI: Atti Accad. Lincei, 20 I, 932 (1911). 2 FREUNDLICH : Kapillarchemie, 99 (1909). 3 Drude's Ann. 37, 480 (1912). 4 Jour. Chem. Soc. 23, 73 (1870). 6 HEMPEL and VATER: Zeit. Elektrochemie, 18, 724 (1912). ADSORPTION OF GAS OR VAPOR BY SOLID 17 retically cause complete displacement of the adsorbed gas; but it is evident that the effectiveness will be much greater if one dis- places the slightly adsorbed gas by the readily adsorbed one rather than the other way round. It would be very interesting to determine an isotherm showing the relative amounts of two gases in the vapor phase and the charcoal phase when in equi- librium at constant pressure. TABLE IX. ADSORPTION OP MIXED GASES BY SYNTHETIC CHARCOAL Mixed gases with 0.4 g synthetic 600 charcoal. Could adsorb about 30 cc ethane. cc H 2 cc C 2 H 6 Percent C 2 H cc gas adsorbed Percent gas adsorbed Percent C 2 H 8 unadsorbed 95.0 5.0 5 3.7 3.7 1.5 45.0 5.0 10 3.8 7.6 2.5 20.0 5.0 20 3.8 15.2 5.0 11.7 5.0 30 4.4 26.4 7.5 6.1 5.0 45 4.6 41.4 9.5 cc H 2 cc CH 4 cc gas adsorbed cc H 2 cc C 2 H4 cc gas adsorbed 95.0 5.0 1.3 95.0 5.0 3.6 15.0 5.0 2.7 15.0 5.0 4.4 6.1 5.0 3.3 6.1 5.0 5.1 2.7 5.0 3.9 2.7 5.0 5.5 cc H 2 cc C 2 H 2 cc gas adsorbed cc N 2 cc C 2 H 4 cc gas adsorbed 95.0 5.0 3.6 95.0 5.0 6.2 15.0 5.0 4.7 30.0 5.0 9.6 6.1 5.0 5.2 12.2 5.0 10.7 2.7 5.0 5.7 5.4 5.0 11.6 A curious case has been cited by Mat win. 1 Charcoal will take carbon bisulphide and carbonyl sulphide out of illuminating gas, one kilogram of charcoal cutting the sulphur content of ten cubic meters of gas to 2.92 grams. Porous charcoals are the best, 1 Jour. Gasbeleuchtung, 52, 602 (1909). 2 18 APPLIED COLLOID CHEMISTRY such as pine and linden. Bone-black takes up almost no carbon bisulphide, and cocoanut charcoal is said to be even less effective. This is very remarkable, because cocoanut charcoal adsorbs carbon bisulphide strongly. We must either assume that the illuminating gas cuts down the adsorption of carbon bisulphide very much or that there is an error in the statement. If carbon bisulphide and illuminating gas were adsorbed in the same ratio in which they occur in the mixture, there would be no apparent purification even though the actual adsorption were very large. 1 This point calls for more study. The adsorption of gases by charcoal has been studied exten- sively, because charcoal adsorbs so many gases and adsorbs them so strongly. Experiments have also been made with many other solids silica, alumina, glass, wool, rubber, celluloid, meerschaum, metals, soils, etc. The behavior of glass is impor- tant because of its use in the laboratory, Miilfarth 2 states that the adsorption decreases in the order: NH 3 , SO 2 , N 2 O, C 2 H 2 . Kundt and Warburg 3 found that a small amount of water is retained obstinately by glass even when the pressure is reduced practically to zero. In order to remove this water the glass must be heated, in which case one is likely to cause changes in the glass, setting free water vapor and gases that were not ad- sorbed in any proper sense of the term. 4 On the other hand, the film of water which usually covers glass surfaces is due chiefly to alkali dissolved from the glass. 5 It is hard to tell whether the change in the vacuum of a Crookes' tube is due to adsorption of the gas by the glass or not. Swinton 6 found that the inner surfaces of these tubes are roughened perceptibly after being subjected to cathode ray bombardment. When these portions of the tube are heated strongly in a blow-pipe, they immediately become clouded, owing to the formation of minute spherical bubbles which average about 0.01 mm in diameter. The centers 1 Cf. LEIGHTON: Jour. Phys. Chem., 20, 32 (1916). 2 Drude's Ann., 3, 328 (1900). 3 Wied. Ann., 24, 327 (1884); cf. BTJKSEN: Pogg. Ann., 156, 201 (1875). 4 SHERWOOD: Jour. Am. Chem. Soc., 40, 1645; Phys. Rev. (2) 12, 448 (1918). 5 WARBURG and IHMORI: Wied. Ann., 27, 481 (1886); IHMORI: Ibid., 31, 1006 (1887;. 6 Proc. Roy. Soc., 79A, 134 (1907). ADSORPTION OF GAS OR VAPOR BY SOLID 19 of the bubbles are about 0.12 mm from the inner surface of the glass. The number of bubbles per square centimeter of glass was found to be about 625,000, or about 0.000133 cc of gas per square centimeter. The gas is mostly hydrogen and is given off when the glass is powdered. It is not known whether the bubbles are connected with the inner part of the tube by minute channels or not. If they are, this is a case of adsorption. If the bubbles are actually buried in the glass owing to the latter flowing round them, that is not adsorption. L. J. Briggs 1 showed that finely divided quartz adsorbs water vapor, though not to the same extent as amorphous silica. About 9 mg water was adsorbed at 30 by 50 g quartz having an estimated surface of 20,000 cm 2 when the vapor pressure was 26.1 mm, about 4.5 mg when the vapor pressure was 19.6 mm, and about 0.5 mg when the vapor pressure was 0.2 mm. Cohn- staedt 2 states that the gas given off by aluminum electrodes in a Crookes' tube is water vapor and not hydrogen, as usually be- lieved. He claims to prove that glass and aluminum surfaces will retain water even when heated to 500 or when dried over phosphorus pentoxide. It is the presence of a water film of varying thickness on a glass or other vessel which makes it necessary to keep the moisture constant if accurate weighings are to be made. Professor Morley once said that it was a mistake to believe that it is a difficult matter to dry a gas. That is a very simple thing to do; but the difficulty is to dry the vessel which holds the gas. Drucker and Ullmann 3 showed that the adsorption of a vapor by the glass containing vessel causes only a negligible error in vapor-density determinations with ether, alcohol, benzene, and chloroform; but the x error may amount to 2 percent with acetic acid at 80. At higher temperatures the adsorption decreases and the error is considerably less. The adsorption of ammonia by solid ammonium hydrosulphide introduces a serious error in the equilibrium relations for ammonia and hydrogen sulphide. 4 1 Jour. Phys. Chem., 9, 617 (1905). 2 Drude's Ann., 38, 223 (1912). 3 Zeit. phys. Chem., 74, 567 (1910). 4 MAGNUSSON: Jour. Phys. Chem., 11, 21 (1907). 20 APPLIED COLLOID CHEMISTRY While most solids, even though porous, do not adsorb gases as much as charcoal, yet Dewar 1 reports that dry aluminum oxide shows a remarkable power of adsorbing air at low temperatures, one gram condensing some 70 cc at atmospheric pressure. The adsorption by alumina decreases very, rapidly with decreasing pressure, and consequently alumina cannot be used as a means of obtaining high vacua. Meerschaum and silica behave some- what like alumina. On the other hand, high vacua can be obtained by means of copper powder. 2 Carbon dioxide 3 and water vapor 4 are both adsorbed strongly by alumina. Platinum black may take up 100 volumes of hydrogen and palladium nearly up to 3000 volumes of hydrogen, 5 but it is not certain whether this hydrogen is all adsorbed or is partly dissolved. In gas analysis palladium, heated to 100, is used to remove hydrogen. This is purely a question of rate because equilibrium is reached more rapidly at the higher temperatures. Paal and Hohenegger 6 report on the adsorption of acetylene by palladium. Platinum black will take up more than 800 volumes of oxygen; but it seems certain that some of the platinum has been oxidized. 7 Celluloid adsorbs carbon dioxide 8 and rubber apparently adsorbs 9 it and also other gases, though it is possible that we have true solution in the case of rubber. Wool 10 adsorbs both hydrochloric acid and ammonia gas. Rohland 11 reports that unsaturated com- pounds are adsorbed by clay and oxygenated compounds by the oxides of silicon, aluminum, iron and titanium. If the film of condensed air on a small particle remains of the same thickness irrespective of the diameter of the particle or if it does not decrease too rapidly with decreasing diameter, the ratio of air film to mass of particle will increase as the particle 1 Proc. Roy. Inst., 18, 435 (1906). 2 MERTON: Jour. Chem. See., 106, 645 (1905). "SCHEERMESSER: Centralblatt (3) 1, 165 (1870). 4 JOHNSON: Jour. Am. Chem. Soc., 34, 911 (1912). 6 PAAL and GERUM: Ber. deutsch. chem. Ges., 41, 805 (1908). 6 Ber. deutsch. chem. Ges., 43, 2684, 2692 (1910). 7 ENGLER and WOHLER: Zeit. anorg. Chem., 29 (1909). S LEFEBURB: Jour. Chem. Soc. 105, 328 (1914). 9 REYCHLER: Van Bemmelen Gedenkboek, 55 (.1910). 10 FREUNDLICH : Kapillarchemie, 106 (1909). 11 Zeit. anorg. Chem., 66, 108 (1909). ADSORPTION OF GAS OR VAPOR BY SOLID 21 grows smaller. Cushman and Coggeshall 1 found that a rock powder which would pass through a 200 mesh sieve surged like a liquid. When poured into a vessel, it only filled 46 percent of the space, while a coarser powder filled more. It is not sur- prising that the air-cushion around each particle makes the grains move one over another like a liquid. A similar phenome- non has been observed with ground phosphate rock. With a substance like carbon black, which adsorbs gases very markedly, as little as 5 percent of the apparent volume may be due to the carbon black 2 and a liter of carbon black may contain 2.5 liters of air. 3 When indigo is reduced to a very fine powder by means of a disintegrator 4 the single particles appear to be separated one from another by an envelope of air, so that the dry, solid powder occupies a space equal to that of a 20 percent paste of indigo, containing the same amount of indigo. Stoney 5 points out that the freedom of motion of finely divided solid particles becomes more marked when there is a temperature gradient. He cites the mobility imparted to very fine powders such as magne- sium carbonate or precipitated silica by heating them in a metal dish. When the dish is disturbed the powder glides about as if floating whenever the powder is able by radiation to maintain a lower temperature than the dish. Ives, Kingsbury and Karrer 6 state that the characteristics of thermocouples of varying diameter can be accounted for rather satisfactorily if it is assumed that they may be considered as cylindrical wires to which heat is conveyed through thin films of hot gas, of thickness independent of the temperature and size of the wire. The presence of a film of air or vapor round a solid particle is shown in a striking way by the fact that it is quite a simple matter 7 to pass suspended phosphorus pentoxide through several wash-bottles in spite of its tremendously hygroscopic nature. 1 Jour. Franklin Inst., 174, 672 (1912). 2 CABOT: Eighth Int. Congress Applied Chem. 12, 18 (1912). 3 SABIN: Technology of Paint and Varnish, 201 (1917). 4 Jour. Soc. Dyers and Colourists, 17, 294 (1901). 6 Phil. Mag. (5) 4, 443 (1877). 6 Jour. Franklin Inst., 186, 421 (1918). ? ENGLER and WILD: Ber. deutsch. chem. Ges., 29, 1929 (1896). 22 APPLIED COLLOID CHEMISTRY Tyndall 1 found it very difficult to get experimental tubes which were absolutely clean on the surface and which contained no suspended particles. Dust-motes from the external air passed through U-tubes containing potash and sulphuric acid and through flasks containing ethers or alcohols. He states that it requires long-continued action on the part of an acid, first to wet the motes and then to destroy them. 2 The difficulty in stopping smokes in gas warfare is that the particles do not come in contact with the materials in the mask. Schloesing 3 has put the thing generally by saying that it is much easier to take gas out of a gaseous mixture than to remove suspended liquid or solid particles. If the sheath of adsorbed air or gas is removed, wetting may take place. Since the sulphur trioxide from the contact process can be caught in concentrated acid and not satisfactorily in water, the concentrated acid must displace the gas film rapidly and fairly completely, while water does not; but we do not yet know the reason for this. One can- not agree with Knietsch 4 however when he says that it is a very remarkable and not satisfactorily explained fact that the sulphuric acid fog from the burner gases is more difficult to precipitate when it is cooled rapidly than when it is cooled slowly. Rapid cooling gives smaller particles and the air-cushion is larger relatively to the mass; consequently precipitation is more difficult. Schuster 5 studied the disruptive discharge through gases and found an apparent diminution of dielectric strength with diminu- tion of pressure. Since no sufficient decrease in inductive capac- ity could be observed in the mass of the gas, he assumed that layer in contact with the electrode there was a condensed surface of gas having a large inductive capacity. Schuster pointed out that some of the most puzzling facts of the disruptive dis-^ charge admit of explanation if we once admit the possibility of such a contact layer, diminishing in density with decreasing gas pressure. If this layer of adsorbed gas offers an increased re- sistance to the passage of an electrical discharge, it follows, from 1 Phil. Trans., 160, 337 (1870). 2 See also EHRENBERG: Die Bodenkolloide, 226 (1915). 3 Comptes rendus, 94, 1187 (1882). 4 Ber. deutsch., chem. Ges., 34, 4081 (1901). 6 Phil. Mag. (5) 29, 197 (1880). ADSORPTION OF GAS OR VAPOR BY SOLID 23 the Theorem of LeChatelier, that an electrical stress will tend to remove the film of adsorbed gas. This enables us to account for many apparently unrelated facts. Electrical waves will tend to remove active oxygen or active hydrogen from an electrode and will therefore cut down the over- voltage. 1 Superposing an alternating current on a direct current also decreases the over- voltage, making other reactions possible. 2 The experiments of Margules 3 and Ruer 4 on the dissolving of platinum find their explanation in the cutting down of the over-voltage. 5 With direct current there is oxidation to a higher and insoluble stage. With alternating current the over-voltage is decreased and little or none of the insoluble compound is formed. The electrolytic detector, the crystal detector, and the coherer, as used in wire- less telegraphy, seem to owe their action to the partial or com- plete elimination of an air film by means of electrical stress. The essential difference between the coherer and the detector is that coalescence takes place readily in the former case and not in the latter. It is interesting to note that Lenard 6 found that mercury wets platinum only when a current is flowing. At other times there is evidently an air film. It has been claimed by Allen 7 that the photo-electric fatigue of metals is due to a change in the surface film or gas or in the gas included in the metal (Hallwachs). It cannot be due to a chemical change such as oxidation; to a physical change such as roughening of the surface; to anelectrical change in the formation of an electrical double layer (Lenard) ; or to a disintegration of the metal due to the expulsion of electrons by light (Ramsay and Spencer). The difference in the results : Drude's Ann., 15, 193 (1904); BENNEWITZ: Zeit. phys. Chem., 72, 223 (1910). 2 ARCHIBALD and VON WARTENBERG: Zeit. Elektrochemie, 17, 812 (1911); REITLINGER: Ibid., 20, 261 (1914); GHOSH: Jour. Am. Chem. Soc., 37, 33 (1915); STEPANOFF: Chem. Abstracts, 10, 2431 (1916). 3 Wied. Ann., 66, 629; 66, 540 (1898). 4 Zeit. phys. Chem., 44, 81, Zeit. Elektrochemie, 9, 235 (1903); 11, 10, 661 (1905); HABER: Zeit anorg. Chem., 61, 365 (1906). 6 REITLINGER: Zeit. Elektrochemie, 20, 261 (1914). 6 Wied. Ann., 30, 212 (1887). 7 Phil. Mag. (6) 20, 564 (1910); cf. LENARD: Drude's Ann., 8, 196 (1902); T2.490 (1903). 24 APPLIED COLLOID CHEMISTRY obtained by Allen may be due to the fact that Allen worked at atmospheric pressure, while Robinson's measurements were made on electrodes at a high vacuum. 1 n cases of marked adsorption of gases it is interesting to determine what is the probable density of the adsorbed gas. Mitscherlich 2 calculated that, when carbon dioxide at atmos- pheric pressure and 12 is adsorbed by boxwood charcoal, the carbon dioxide occupies only one fifty-sixth of its original volume. Since this is a lesser volume than the same amount of carbon dioxide can occupy as a gas at this temperature, it is usually assumed that some of it has been liquefied. It is possible, how- ever, to consider the carbon dioxide as present in the form of a highly condensed gas which has not liquefied. It might be inter- esting to study the behavior of carbon dioxide near the critical point in presence of charcoal. The magnitude of the heat of adsorption is an argument in favor of the actual liquefaction of the gas. Dewar 3 has calculated the apparent density of some gases adsorbed by cocoanut charcoal at low temperatures. The data in Table X show that the densities are of the same order as those of the liquid gases and are greater in some cases. At higher temperatures the amount of adsorption would be smaller and the apparent condensation less. TABLE X. DENSITY OF GASES ADSORBED BY CHARCOAL Density in grams per cubic centimeter. Gas adsorbed Temperature of adsorp- tion Density of adsorbed gas Density of liquefied gas Carbon dioxide + 15 70 0.80 Oxygen -183 1 33 1 12 Nitrogen -193 1.00 0.84 Hydrogen -193 0.06 0.07 Hydrogen 210 08 Hydrogen Helium -252 -258 0.11 0.17 (0.122) 1 Phil. Mag. (6) 23, 255 (1912). 2 Sitzungsber. Akad. Wiss. Berlin, 1841, 376. 3 Proc. Roy. Inst., 18, 438 (1906). ADSORPTION OF GAS OR VAPOR BY SOLID 25 It is to be noticed that the vapor pressure of a film of adsorbed gas in an apparently liquefied state stands in no relation to the vapor pressure of a mass of the liquefied gas. The phase rule generalization that the vapor pressure is constant for the system, liquid and vapor at constant temperature, holds good explicitly only in case the disturbing effects due to gravity, distortion of the solid masses, and capillary tensions are eliminated. So soon as a liquid film becomes thin enough so that the liquid no longer has the properties of matter in mass, the vapor pressure need not remain constant. When we have an adsorbed liquid or liquefied gas, the vapor pressure will vary also with the nature of the adsorbing agent. The lowering of the vapor pressure is shown by the fact that oatmeal, 1 previously dried at high tem- perature, has been used instead of sulphuric acid to produce the freezing of water under the receiver of an air-pump. L. J. Briggs 2 showed that the weight of the water film condensed on the sur- face of quartz at 30 in an atmosphere within one percent of saturation is 26.6 mg per square meter as determined by drying afc 110. This corresponds to a calculated thickness of the water film of 26.6juju assuming the density of the film to be uni- form and equal to unity. Neither assumption is absolutely accurate. Since the amount of adsorption increases rapidly with falling temperature, adsorption must be accompanied by a marked evolution of heat. In Table XI the data not bracketted are by Favre. 3 All the molecular heats of adsorption are higher than the corresponding heats of liquefaction. The abnormally high values for hydrogen in palladium and platinum cannot be due to dissociation into monatomic hydrogen, because that would be accompanied by an adsorption of heat. They may be due in part to an oxidation of hydrogen. Masson 4 considers that the heat of adsorption of water vapor does not differ materially from the heat of liquefaction; but Dewar's data, 5 Table XII, confirm those of Favre, and we know also that there is always a 1 THOMSON: Phil. Mag. (4) 42, 448 (1871). 2 Jour. Phys. Chem., 9, 617 (1905). 3 Ann. Chim. Phys. (5) 1, 209 (1874). 4 Proc. Roy. Soc., 74, 249 (1904). S DEWAR: Proc. Roy. Inst., 18, 183 (1905). 26 APPLIED COLLOID CHEMISTRY heat effect when porous solids are wetted by liquids. 1 This is probably due to an actual increase in the density of the adsorbed liquid. TABLE XI. MOLECULAR HEATS OF ADSORPTION AND OP LIQUEFACTION OF GASES Adsorbing agent Gas Adsorption g cal. per mol. Liquefaction g cal. per mol. Platinum H 2 46,200 (240) Palladium H 2 18,000 (240) Charcoal NH 3 5 ? 900 8,500 (5,000) Charcoal CO 2 6,800 7,800 6,250 Charcoal N 2 O 7,100 7,700 4,400 Charcoal . . ... SO 2 10,00010,900 5,600 Charcoal HC1 9,200 10,200 (3,600) Charcoal. HBr 15,20015,800 (4,000) Charcoal . . . HI 21,000 23,000 (4,400) Melsens 2 found that when 25 cc of different liquids were added to 10 grams of freshly ignited wood charcoal there was a rise of temperature of several degrees with ether, alcohol, and TABLE XII. MOLECULAR HEATS OF LIQUEFACTION AND ADSORPTION OF GASES Charcoal as adsorbing agent at 185 (Dewar). Gas Adsorption g. cal. per mol. Liquefaction, g. cal. per mol. Hvdroffen 1600 238 Nitrogen 3686 1372 Argon 3636 Oxvsren 3744 1664 Carbon monoxide 3416 carbon bisulphide. When 97 grams of bromine were added to 11 grams of charcoal the rise of temperature was about 30. Since a sudden compression of ten atmospheres raises the tem- ^f. POUILLET: Ann. Chim. Phys. (2) 20, 141 (1822); CHAPPUIS: Wied. Ann. 19, 21 (1883); MARTINE: Phil. Mag. (5) 47, 329 (1889); 60, 618 (1900); (6) 15, 595 (1903); PARKS: Ibid. (6) 4, 240 (1902); 6, 517 (1903). 2 Ann. Chim. Phys. (5) 3, 522 (1874). ADSORPTION OF GAS OR VAPOR BY SOLID 27 perature of water only 1/77 of a degree he concluded that the rise of temperature of 1.16 when charcoal is wetted by water corresponds to a pressure of 893 atmospheres. Calculating in the same way he drew the conclusion that the force of adsorption is 3100 atmospheres for bromine, 13,090 for carbon bisulphide, 4620 for ether and 3080 for alcohol. Gaudechon 1 obtained the data given in Table XIII for different liquids, with wood charcoal. Since the heat of liquefaction varies with the temperature, it is probable that the heat of adsorption of gases and vapors does also; but this point seems not to have received any special attention. TABLE XIII. HEAT EVOLUTION ON WETTING WOOD CHARCOAL Gram calories per gram charcoal. g. cal. g. cal. Water 3.9 Acetone 3.6 Methyl alcohol 11.5 Chloroform . 2 3 Ethyl alcohol 6 9 Ether 1 2 Propyl alcohol. 5.6 Benzene 4.2 Amyl alcohol 3 7 CC1 4 1 5 Formic acid 12.0 CS 2 4.0 Acetic acid 6 0.4 Lamb ^and Coolidge 2 find that curves, which approximate closely to straight lines, are obtained when the heats of adsorp- tion of the vapors of some organic liquids by charcoal are plotted against the amounts adsorbed. The difference between the heat of adsorption and the heat of vaporization is called by them the net heat of adsorption and is practically constant for the different liquids studied, when referred to equal volumes of adsorbed liquid. Assuming that this net heat of adsorption is a heat of compression, the attracting force is estimated to be about 37,000 atmospheres when one cubic centimeter of liquid is adsorbed by ten grams of charcoal. The action of water vapor on cottdn has been studied by 1 Comptes rendus, 157, 209 (1913). 2 Jour. Am. Chem. Soc. 42, 1146 (1920). 28 APPLIED COLLOID CHEMISTRY Beadle 1 and by Masson. 2 If the bulb of a thermometer be covered with carefully dried cotton wool, and if the bulb be then dipped in a beaker or water at the same temperature, the thermom- eter will rise 8-12 or even more. The relatively large amount of air which is always entangled in the cotton wool does not escape during or after immersion, and the fibers present a glisten- ing appearance under water, which suggest that they are not wetted completely. Medicated or absorbent cotton differs from the ordinary, nearly pure, material in that the former has been freed from traces of natural cotton wax by treatment with solvents. When absorbent cotton is dipped into water, the rise in temperature is less than in the previous case even though more water is taken up by the cotton. In the case of the nearly pure cotton wool, the water evidently distills across the air space, while there is a certain amount of direct wetting with the absorbent cotton. There should therefore be a marked evolu- tion of heat if cotton is placed in saturated water vapor and the two cottons should behave exactly alike. Both these predictions were confirmed experimentally. Masson obtained similar re- sults when guncotton or glass wool was exposed to saturated water vapor or cotton to the vapor of absolute alcohol ; but the heat effects were much smaller in these instances owing to the lesser adsorption. L. B. Loeb 3 finds that the adsorption of methyl alcohol vapor by 0.41g platinum black causes a rise of temperature of about 0.36 when there is no oxidation. The effect of traces of impurity on the adsorption is shown in some experiments by Berliner. 4 His purified palladium foil adsorbed about 860 volumes of hydrogen. If allowed to stand in a vacuum for an hour, it then adsorbed practically no hydrogen. If left for an hour or two in the air, the same thing happened. The disturbing factor is apparently a trace of grease, because water did not wet the soiled palladium. Similar results were obtained with platinum. 1 Nature, 49, 457 U884); Chem. News, 71, 1 (1895); 73, 180 (1896). 2 Proc. Roy. Soc., 74, 230 (1904); MASSON and RICHARDS: Ibid., 78A, 412 (1906). 3 Proc. Nat. Acad. Sci., 6, 107 (1920). 4 Wied. Ann., 36, 803 (1888). ADSORPTION OF GAS OR VAPOR BY SOLID 29 THE ADSORPTION ISOTHERM So far we have only committed ourselves to the qualitative statement that the amount of adsorption increases with increase of pressure. It is now advisable to consider the quantitative side. Dewar 1 has made some measurements with hydrogen at varying pressures in charcoal at 185. The results are given in Table XIV. Dewar says in regard to them that the amount adsorbed is seen to increase with the pressure to 10 atmos- pheres, after which the adsorption in the pores of the charcoal seems to be independent of the pressure. At the temperature of liquid air this sample of charcoal (6.7 g) would not adsorb more than about one liter of hydrogen even when the pressure was raised from 10 to 25 atmospheres. It appears that Dewar considers the decrease from 156 cc to 138 cc as due to experi- mental error for he says distinctly that the adsorption had come to a limit. Since that time a number of careful investigations have been made on the quantitative variation of the adsorption with the pressure. 2 TABLE XIV. HYDROGEN ADSORBED BY CHARCOAL AT 185 Atm. cc/gram Atm. cc/gram 1 92 15 149 5 138 20 145 10 156 25 138 If we plot the amounts of adsorbed gas against the pressures, we get the so-called adsorption isotherm, a smooth curve concave to the pressure axis. Trouton 3 obtained a curve with the oppo- site flexure for the adsorption of water by flannel; but Masson 4 1 Proc. Roy. Inst., 18, 437 (1906). 2 TRAVERS: Proc. Roy. Soc., 78A, 9 (1906); DAVIS: Jour. Chem. Soc., 91, 1066 (1907); GEDDES: Drude's Ann., 29, 797, (1907); BOYLE: Jour. Phys. Chem., 12, 484 (1908); Phil. Mag. (6) 17, 377 (1909); McBAiN: Ibid 18, 916 (1909); Zeit. phys. Chem., 68, 471 (1909); HOMPRAY: Ibid., 74, 139, 687 (1910); TITOFP: Ibid., 74, 641 (1910); PIUTTI AND MAGLIA: Gazz. chim. ital., 40 I, 569 (1910). 3 Proc. Roy. Soc., 77A, 292 (1905). 4 Ibid., 78A, 9 (1906). 30 APPLIED COLLOID CHEMISTRY obtained normal curves for water and cotton, and Travers 1 obtained normal curves for wool and cotton with water vapor, so that Trouton's results are apparently due to experimental -7$ I A Z 5 4 c.c- J-fe ADSORBED F>ER GRAM CHARCOAL FIG. 1. error. Trouton 2 also obtained an abnormal type of curve for the adsorption of water vapor by glass; but it seems probable that alkali dissolved from the glass has been a disturbing factor. 1 Proc. Roy. Soc., 79A, 204 (1906). 2 Ibid., 79A, 383 (1906). ADSORPTION OF GAS OR VAPOR BY SOLID 31 If the gas is adsorbed but slightly, the curve may seem practi- cally a straight line over the ordinary pressure ranges. This is shown very clearly in the curves for hydrogen in charcoal Fig. 1 as determined by Tit off. 1 If the gas is adsorbed strongly, the isotherm bends around so as to run nearly parallel with the pressure axis, because then a marked increase in pressure causes FIG. 2. only a relatively slight increase in adsorption. These two ex- tremes can be realized very well with ammonia and charcoal 2 at 175 and -64, Fig. 2. In Table XV are given Titoff's data for the adsorption of ammonia by charcoal at and at 151.5. Under Ax/Ay is given the mean increase in adsorption with increase of pressure between each two consecutive observations. 1 Zeit. phys. Chem., 74, 641 (1910). 2 RICHARDSON: Jour. Am. Chem. Soc., 38, 1828 (1917). 32 APPLIED COLLOID CHEMISTRY At the values drop from 92 to 3, a ratio of over 30:1, while at 151.5, the drop is only from 51 to 11, a ratio of less than 5:1. If we plot the logarithms of the amounts of adsorbed gas TABLE XV. ADSORPTION OF AMMONIA BY CHARCOAL Temperature Temperature 151.5 Pressure Adsorption Ax/Ay Pressure Adsorption Ax/Ay in mm in cc cc/mm in mm in cc cc/mm 2.9 5.4 2.8 0.3 28.8 30.2 0.92 16.6 1.0 0.051 78.7 60.4 0.65 64.8 2.1 0.021 161.0 90.3 0.36 176.2 4.0 0.018 319.2 115.7 0.16 285.0 5.5 0.013 490.0 127.0 0.07 432.7 7.2 0.012 636.4 132.4 0.04 576.8 8.9 0.012 746.7 135.9 0.03 721.8 10.5 0.011 against the logarithms of the pressures, we get a curve which approximates a straight line, especially if the gas is not adsorbed too much. This shows that the data can be represented approxi- mately by an equation 1 of the form (x/m) n = kp, where x is the TABLE XVI. ADSORPTION OF CARBON DIOXIDE BY CHARCOAL Formula x ^"/P = 0.0602. x = cc CC>2 adsorbed per cc charcoal. p = pressure in mm Hg. Temp. 31. p x found x calc. P x found x calc. 41.5 1.7 1.7 453 6.5 6.4 120. 2.9 3.1 534 7.1 7.1 194. 4.0 4.0 602 7.6 7.6 276. 4.7 4.9 678 8.3 8.2 340. 5.4 5.5 698 8.6 8.3 405. 5.9 6.0 703 8.9 8.4 amount of gas adsorbed and m the amount of the adsorbing solid, while p is the pressure and k and n are constants to be determined experimentally for each temperature. The expo- 1 Chemists are in the habit of calling this an exponential formula which is not correct in the strict mathematical sense; but which is justified by usage. ADSORPTION OF GAS OR VAPOR BY SOLID 33 nential factor, n, is usually not an integer and its physical signifi- cance is unknown. 1 In Table XVI are given some data by Geddes 2 on the adsorption of C0 2 by charcoal. The formula enables us to calculate the adsorption with a moderate degree of accuracy. Of course data could have been picked out which would have looked better, but there seems no advantage in doing this. The formula cannot possibly represent the facts for pressures at which the curve runs nearly parallel to the pres- sure axis, and the experiments of Richardson 3 on the adsorption FIG. 3. of ammonia by charcoal makes it probable that the equation never describes the facts accurately over any wide range of pressures. For the present it should be considered as a service- able interpolation formula, though it may possibly belong in the same category as the simple gas law. The fact that we get a smooth curve when plotting pressure direct against adsorption shows that no definite compound is formed. In Fig. 3 the pressure is plotted along the vertical axis and the amount of gas condensed on or by unit mass is 1 For a slightly different formulation, see WILLIAMS: Trans. Faraday Soc., 10, 167 (1914); Proc. Roy. Soe., Edinburgh, 391, 48 (1919). 2 Drude's Ann., 29, 797 (1909). 3 Jour. Am. Chem. Soc., 38, 1828 (1917). 3 34 APPLIED COLLOID CHEMISTRY plotted along the horizontal axis. The typical adsorption iso- therm is represented by the curve OH, while OABCDE is the curve (not to scale) for the taking up of water by sodium sul- phate. Along OA the solid phase is anhydrous sodium sul- phate. Along AB the pressure remains constant because three phases are present: Na 2 SO 4 , Na 2 SO 4 .10H 2 0, and water vapor. Along BC the pressures change without any more water being taken up and consequently there are only two phases, water vapor and the compound Na 2 S0 4 .10H 2 O. The composition of the compound is given by F, the point at which the prolonga- tion of CB cuts the axis of abscissas. Along CD the phases are Na 2 SO 4 .10H 2 O, saturated solution and vapor, so the pressure remains constant, while unsaturated solution and vapor coexist along the curve DE. Many curves of this type have been de- termined by Browne 1 and others in his laboratory for the am- monia compounds. The curve OH is clearly not of this type and represents either a continuous series of solid solutions or an ad- sorption. The equilibrium is usually reached fairly rapidly, which is an argument against the existence of a solid solution though by no means a conclusive one. More convincing is the fact that the amount of adsorption by charcoal varies with the structure 2 and the extent of surface. The change produced by grinding charcoal 3 is only a slight one, because the surface of the unground charcoal is so enormous. Lagergren 4 estimates that one gram of charcoal has a surface of four square meters and is composed of about 1.4 billion particles. Lamb, Wilson and Chaney 5 consider that one cubic centimeter of active war charcoal has a surface of about 1000 square meters (density about 0.4). While the experiments on the taking up of gases by charcoal show fairly conclusively that surface phenomena are the im- portant ones, there is always the possibility that a solid solution may also be formed to some extent. McBain 6 considers that 1 See BROWNE and HOULEHAN: Jour. Am. Chem. Soc., 35, 649 (1913). 2 PIUTTI and MAGLIA: Gazz. chim. Hal., 40 I, 569 (1910). 3 CHAPPUis: Wied. Ann., 12, 164 (1881). Cf. also GURWITSCH: Zeit. Kolloidchemie, 11, 17 (1912). 4 See VAN BEMMELEN: Die Absorption, 410 (1910). 5 Jour. Ind. Eng. Chem., 11, 420 (1919). 6 Phil. Mag. (6) 18, 916; Zeit. phys. Chem., 68, 471 (1909). ADSORPTION OF GAS OR VAPOR BY SOLID 35 over 13 percent of the hydrogen taken up by cocoanut charcoal at the temperature of liquid air is present as a solid solution. 1 On the other hand Titoff 2 believes that the amount of hydrogen dissolved in the charcoal has been over-estimated by McBain. 3 The truth of the matter is that we have no definite criterion which will enable us to distinguish accurately in all cases between solid solution and adsorption. In the case of cellulose and of gelatine it is quite possible that some water is actually dissolved. Until we know more about the subject it is safer to treat any doubtful case as chiefly one of adsorption. On the other hand, the two special cases of metals and rubber with gases call for some special comment. Assuming the exist- ence of pores in platinum or palladium foil, one can account for the selective diffusion of hydrogen by postulating adsorption in the pores and subsequent diffusion. At high temperatures the adsorption would be much less, and it is difficult to explain the increased rate of selective diffusion. It is simpler to account for the diffusion of hydrogen through platinum by postulating that some of the hydrogen actually dissolves in the platinum. In that case the solubility should be independent of the physical struc- ture of the platinum. This seems to be true within narrow limits for platinum and within wide limits for palladium. It seems probable that platinum can dissolve several volumes of hydrogen at ordinary temperatures, and palladium nearly eight hundred volumes. 4 The bulk of the hundred volumes of hydrogen which are taken up by platinum black 5 must be adsorbed, and the same must be true of the nearly 3000 volumes taken up by colloidal platinum. 6 This is apparently the conclusion reached by Holt, Edgar, and Firth. 7 If platinum and palladium dissolve hydro- 1 Cf. TRAVERS: Proc. Roy. Soc., 78A, 9 (1906). 2 Zeit. ph-ys. Chem., 74, 641 (1910). 3 For other discussions, see MASSON: Proc. Roy. Soc., 74, 251 (1904); DAVIS: Jour. Chem. Soc., 91, 1066 (1907); GEDDES: Dmde's Ann., 29, 797 (1908); WOHLER, PLUDDEMANN and WOHLER: Zeit. phys. Chem., 62, 670 (1908); HOMFRAY: Ibid., 74, 139, 687 (1910). 4 Cf. BOSE: Zeit. phys. Chem., 34, 708 (1900). 5 RAMSAY, MOND and SHIELDS: Zeit. phys. Chem., 19, 40 (1896); 25, 657 (1898). 6 PAAL and GERUM: Ber. deutsch. chem. Ges., 41, 805 (1908). 7 Zeit. phys. Chem., 82, 513 (1913). 36 APPLIED COLLOID CHEMISTRY gen to some extent, the passage of carbon monoxide through heated iron cannot be attributed to adsorption. At present it is not possible to say whether an iron carbonyl plays any part in the diffusion or not. In the case of platinum and oxygen Ramsay, Mond, and Shields 1 consider that most of the oxygen taken up forms Pt (OH) 2 . They point out that Thomsen ob- tained an evolution of heat of 17,000 calories for the reaction Pt+O+H 2 O, while they find 17,600 calories for the so-called adsorption of sixteen grams of oxygen by platinum. In view of the relative behavior of rubber with water and with pyridine, it is hard to see how the permeability can be a question primarily of adsorption. 1 think that we must assume that carbon dioxide, pyridine, etc., actually dissolve in rubber. TABLE XVII. ADSORPTION OF CARBON DIOXIDE BY BLOOD CHARCOAL Formula x/p = 0.00024. x = millimols CC>2 per gram charcoal, p = partial pressure in mm Hg. Temperature 20. p X k p X k 613.1 0.1461 0.00024 337.7 0.0855 0.00025 611.1 0.1424 0.00023 325.5 0.0808 0.00025 605. 0.1320 0.00022 312.2 0.0706 0.00023 453.1 0.1014 0.00022 229. 0.0560 0.00024 437.2 0.1086 0.00025 221.2 0.0523 0.00023 435.2 0.1063 0.00024 176. 0.0452 0.00026 348.5 0.0844 0.00024 170.2 0.0451 0.00026 339. 0.088 0.00026 mean 0.00024 While the equation x/m = kp is a special case of the equation (x/m) n = kp, where n = 1, and therefore may hold even in cases of adsorption, it is also the simplest case of Henry's law for the solubility of gases in liquids. Reychler 2 found that the amount of carbon dioxide taken up by blood charcoal was pro- portional to the partial pressure of the carbon dioxide in the vapor phase. The data are given in Table XVII. It does not * Ibid., 26, 684 (1898).. 2 Van Bemmelen Gedenkboek, 55 (1910). ADSORPTION OF GAS OR VAPOR BY SOLID 37 seem probable that C0 2 dissolves in carbon, and I prefer to consider this a case of adsorption where the exponential factor happens to be unity. This view is confirmed by the fact that Geddes obtained very different relations for the adsorption of CO 2 by charcoal at 31 (Table XVI). Reychler also found that the amount of carbon dioxide taken up by rubber was proportional to the pressure over a narrow range of pressures. This may well be 'a case of solution of carbon dioxide in rubber. e concentration-pressure curve for hydrogen and palla- dium 1 has two ascending branches connected by a branch which is nearly but not quite, parallel to the concentration axis. It is not known to what extent the peculiarities of this curve are due to adsorption. While many people are willing to speak of adsorption and to let it go at that, attempts have been made to account for adsorp- tion either through the formation of known or unknown com- pounds. Rhead and Wheeler 2 consider that the fixation of oxygen by carbon involves the formation of a compound, C x O y , which breaks up into carbon dioxide and carbon monoxide when heated. Haber 3 points out that, according to Bragg, the ordi- nary solid crystalline salts are not systems in which one anion and one cation form a molecule which is separated from the next; / but that one anion is probably bound to all the surrounding \ cations. Unless the outer layer of a crystal differs fundamentally / from the inner portions, there must be forces at the surface of the } crystal similar to those acting between the outer layer and the next layer. If we call these forces chemical valences of some sort, it is possible that adsorption is due essentially to the satura- tion of these valences. These ideas have been developed by Langmuir 4 as a result of his experiments on electrically heated carbon and other filaments in presence of oxygen at pressures of 0.004-0.007 mm. Langmuir considers that any crystal or ( I HOITSEMA: Zeit. phys. Chem., 17, 1 (1895); RAMSAY, MOND, and SHIELDS: Ibid., 19, 25 (1896); HOLT, EDGAR and FIRTH: Jour. Chem. Soc., 82, 513 (1913). 2 Jour. Chem. Soc., 103, 462 (1913). 3 Jour. Soc. Chem. Ind., 33, 50; Zeit. Elektrochemie, 20, 521 (1914). Jour. Am. Chem. Soc., 37, 1139 (1915); 38, 1145, 2221 (1916). 38 APPLIED COLLOID CHEMISTRY any mass of liquid forms a giant molecule, and that an adsorbed gas forms a compound with the outer layer, the thickness of the reacting gas film being rarely over one or two molecules thick. Looking at the subject from a kinetic point of view he deduces an equation of the form where Q is the amount adsorbed at the pressure p, and a, b and c are constants. There are two independent questions involved in Langmuir's hypothesis, namely whether the maximum adsorption film is only one or two molecules thick and whether this layer forms a compound with the adsorbing agent. The first may be true and the second be wrong. Langmuir claims that the adsorbed film never exceeds two molecules in thickness for plane surfaces, though he is apparently willing to concede a greater thickness for porous substances. There is no question but that the adsorp- tion film is very thin. Mitscherlich 1 calculates a thickness of about 50fjLfj, for carbon dioxide adsorbed by charcoal, and the film of water round each particle of clay 2 is usually assumed to be of the same thickness, while L. J. Briggs 3 estimates about 27/^c for the film of water on quartz. It is difficult to be certain on these points because one does not know the surface with any degree of accuracy; but the evidence is in favor of Langmuir's contention. When it comes to the question of whether compounds are formed or not, it must be remembered that Langmuir uses the word molecule in a new sense. He considers that the oceans and all that in them is constitute one vast molecule and that catch- ing a fish is a chemical reaction because it breaks up the molecule. This may be perfectly true; but the usefulness of it has not been made clear. For the present, it seems much wiser to say that one substance adsorbs another selectively, meaning that the concentration at the surface of the phase has increased. When- ever there is no proof that a compound is formed of the type at 1 SITZUNGSBER-. Akad. Wiss. Berlin, 1841, 379. 2 Cf. DAVIS: Trans. Am. Ceramic Soc., 16, 69 (1913), 3 Jour. Phys. Chem., 9, 617 (1905), ADSORPTION OF GAS OR VAPOR BY SOLID 39 present recognized by chemists, it is safer not to postulate the existence of compounds. When the time comes that it is a dis- tinct advantage to distinguish an entirely new type of compound, an indefinite chemical compound, that can be done; but hopeless confusion will arise whenever a man uses a word in one sense and his listeners in another. It is quite probable that a more careful study of contact catalysis may force us to adopt Langmuir's view or some modification of it; but it will then be necessary to gather in a lot of the loose ends. In some cases compounds are undoubtedly formed. Thus when platinum black takes up oxygen, a portion of the oxygen is certainly present as oxide. On the other hand, it seems im- probable that it would be so difficult to make carbon tetrachloride direct if the chlorine, which is adsorbed by charcoal and there- by made active, 1 consisted of a normal compound of carbon and chlorine. While the oxidizing action of platinum may be due to a platinum oxide, this cannot be true in the case of charcoal and oxygen. Calvert 2 showed that oxygen adsorbed by charcoal will oxidize ethyl alcohol to acetic acid and ethylene to carbon dioxide and water, reactions which certainly cannot be due to any known oxide of carbon. It also seems unsatisfactory to postulate that the noble gases form definite chemical compounds with charcoal. Years ago Angus Smith 3 ignored the variation of adsorption with the pressure and concluded that at atmospheric pressure charcoal adsorbed gases in volumes which were multiples of the volumes ot hydrogen adsorbed. His figures were: hydrogen, 1; oxygen, 7.99; carbon monoxide, 6.03; carbon dioxide, 22.05; methane, 10.01; nitrous oxide, 12.90; sulphur dioxide, 36.95; nitrogen, 4.66. Smith lays stress on the fact that the 22 for carbon dioxide and the 10 for methane can be considered as 6 + 16 and 6 + 4 respectively, though it is difficult to see what this proves. In order to make the value for nitrogen seem plausible, Smith points out that 14 X 4.66 = (14) 1 ; but these calcula- tions are only of historical interest now. 1 DAMOISEAU: Comptes rendus, 73, 60 (1876). 2 Jour. Chem. Soc., 20, 293 (1867). 3 Chem. News, 18, 21 (1868); Proc. Roy. Soc., 28, 322 (1879). 40 APPLIED COLLOID CHEMISTRY CHEMICAL REACTIONS Since reaction velocity is probably directly proportional to the difference of chemical potential and inversely proportional to the chemical resistance, we can increase the reaction velocity either by increasing the difference of chemical potential or by decreasing the chemical resistance. The increased concentra- tion due to adsorption at the surface of a solid will necessarily cause an increase in reaction velocity quite apart from any catalytic action which the solid may exert. 1 In the form of wood ashes, charcoal is often used in outhouses and elsewhere to eliminate unpleasant odors. The charcoal adsorbs the odori- ferous substances and then accelerates their oxidation by oxygen which is adsorbed from the air. Shenstone 2 invented a respira- tor filled with powdered wood charcoal to be worn in hospitals and which should adsorb and then destroy germs of cholera and other diseases. This was the forerunner of the modern gas mask. It is not known whether the oxidizing action of the char- coal is due to anything more than the increase in concentration of the oxygen. As a matter of fact, the effect of concentration seems to be relatively unimportant in most of the cases studied hitherto. Patrick's silica gel has a surface of 2.5 X 10 8 cm 2 per gram and is an excellent adsorbent, but, so far, it does not have any marked catalytic effect except in the oxidation of nitric oxide. It would be interesting to try this substance at 100 with the mixed vapors of alcohol and acetic acid, because this reaction takes place at a measurable rate in the vapor phase in the absence of any catalytic agent. It is stated that oxyhydrogen gas has been found to be quite stable under pressures of two thousand atmospheres, which shows that the catalytic action of platinum on mixtures of hydro- gen and oxygen must be due to something more than a change in concentration. The marked difference in the behavior of platinum and charcoal toward oxyhydrogen gas is another argu- ment in favor of the action of a catalytic agent being specific. On the other hand Hempel 3 with S02 and 02 obtained a yield 1 See, however, LANGMUIR: Jour. Am. Chem. Soc., 38, 2286 (1916). 2 Pharm. Jour., 13, 454 (1854). 3 Ber. deutsch. chem. Ges., 23, 445 (1890). ADSORPTION OF GAS OR VAPOR BY SOLID 41 of 65 percent S0 3 without any catalytic agent by working at a pressure of thirty to forty atmospheres. If we have selective adsorption of the reaction products, this will cause a lowering of the chemical potential of the resultants 1 and will therefore increase the reaction velocity. It may even determine the direction of the reaction. Alcohol breaks down into acetaldehyde and hydrogen in presence of pulverulent nickel and into ethylene and water in presence of pulverulent alumina or silica. 2 Since nickel adsorbs hydrogen strongly and alumina adsorbs water, 3 it seems natural to assume that the selective adsorption is the determining factor. 4 While this seems very satisfactory, there are certain points which must not be overlooked. When making ethylene at Edge- wood Arsenal during the war, it was found advisable to have a large amount of steam present with the alcohol vapor in order to make temperature regulation easier. This undoubtedly decreased the rate of decomposition of the alcohol; but that difficulty was overcome by working at a higher temperature. It is very diffi- cult to see how alumina can dehydrate alcohol in presence of a large amount of water vapor if the reason the alumina acts is be- cause of its strong adsorption of water vapor. Since there seems very little probability of accounting for the various decomposi- tions of the esters on the basis of selective adsorption of the re- action products, it is probable that a complete explanation of the phenomena of contact catalysis is not to be found along these lines. If changes in the chemical potential are not the important factors in contact catalysis, the catalytic agent must change the chemical resistance in some way, and there are two possible ways in which this might happen. We may have more effective collisions between existing molecules or we may have a conversion of one or more of the reacting substances into an active form. According to the kinetic theory the reaction velocity is pro- 1 W. C. McC. LEWIS suggests the terms reactants and resultants for react- ing substances and reaction products. 2 IPATIEFF: Ber. deutsch. chem. Ges., 34,3579 (1901); 35, 1047, 1057, (1902) 36, 1990, 2003, 2014, 2016 (1903); SABATIER and SENDERENS: Comptes rendus, 136, 738, 921, 936, 983 (1903). 3 JOHNSON: Jour. Am. Chem. Soc., 34, 911 (1912). 4 ENGELDER: Jour. Phys. Chem., 21, 676 (1917). 42 APPLIED COLLOID CHEMISTRY portional to the number of collisions between possibly reacting molecules; but it does not follow at all that two molecules react every. time they collide. If a large number of collisions is neces- sary on an average before a pair of molecules react, anything which would make these collisions more helpful might increase the reaction velocity enormously. The first question is then whether there is any evidence of ineffective collisions. This matter has been studied by Strutt 1 who comes to the conclusion that a molecule of ozone reacts every time it strikes a molecule of silver oxide; but that a molecule of active nitrogen collides with a molecule of copper oxide five hundred times on an average before they react, while two molecules of ozone at 100 collide on an average 6 X 10 u times before they react. Without insist- ing on the absolute accuracy of these figures there is evidently plenty of margin for an increase in reaction velocity with ozone at 100 if one could produce more effective collisions. Langmuir 2 finds that 15 percent of all oxygen molecules at a pressure of not over 5 bars striking a tungsten filament at 1770K react with it to form tungstic oxide, WO 3 . This coefficient increases at higher temperatures and at 3300K about 50 percent of all the oxygen molecules which strike the filament react with it to form WO 3 . Since there are three atoms of oxygen in WO 3 and only two in the oxygen molecule, Langmuir considers that at least one-half of the tungsten surface must be covered at 3300K with oxygen in some form. Tt is possible that a catalytic agent may cause one molecule to strike another amidships instead of head-on and may thereby increase the effectiveness of the collisions. It is not impossible that part at least of the effect of solvents on reaction velocity may be due to some such thing as this. If we adopt Langmuir's views 3 of oriented adsorption, all sorts of things become possible. If ethyl acetate, for instance, attaches itself to one adsorbent by the methyl group, to another by the ethyl group, and to a third by the carboxyl group, it might very well be that bombard- ment of the captive molecule by free ones might lead to very different reaction products in the three cases. Such a suggestion 1 Proc. Roy. Soc., 87A, 302 (1912). 2 Jour. Am. Chem. Soc., 35, 105 (1913); 38, 2270 (1916). 3 Cf. DEBUS: Jour. Chem. Soc., 53, 327 (1888). ADSORPTION OF GAS OR VAPOR BY SOLID 43 is of very little value, however, unless it can be made definite. We do not know as yet whether ethyl acetate is actually adsorbed in one way by nickel, in another way by thoria, and in a third way by titania, nor do we know whether the difference in the manner of adsorption, assuming it to occur, is of such a nature as to account for the differences in the reaction products. It may well be that some entirely different hypothesis will prove neces- sary. The general problem of increasing the effectiveness of molecular collisions is distinctly an important one which has not been studied at all. There is one special case in which we can apparently identify one of the factors of chemical resistance. Buchbock 1 studied the decomposition of carbonyl sulphide and found that the reaction velocity decreased with increasing viscosity of the solution. Since the reaction products are gases, this suggests the possibility that the retardation is due to the greater difficulty of forming gas bubbles in a more viscous medium. This receives confirma- tion from Raschig's work on the preparation of hydrazine, 2 by the action of hypochlorite on ammonia, the evolution of nitrogen being kept down by something like starch to make the solution more viscous. The preliminary reaction is NH 3 + NaCIO = NH 2 C1 + NaOH and high viscosity promotes the reaction NH 2 C1 + NH 3 = N 2 H4HC1 as against the one 3NH 2 C1 + 2NH 3 = 3NH 4 C1 + N 2 . Instead of more effective collisions between unchanged por- tions of the reacting molecules we may have a partial conversion of one or more of the reacting substances into active forms either through the rupture of normal valences or of contra-valences, opening up fields of force as Baly 3 puts it. The conception of Baly that a catalytic agent activates the reacting substances by opening up fields of force covers the ground admirably as a purely formal statement. The trouble with it is that it is as yet too vague to be of any value as a working hypothesis, though it is undoubtedly the best starting-point. It is necessary to show in each particular case just what happens and why it happens. While the conversion of a reacting sub- 1 Zeit. phys. Chem., 23, 123 (1897); 34, 229 (1900). 2 HENDERSON: Catalysis in Industrial Chemistry, 60 (1919). 3 Jour. Chem. Soc., 101, 1469, 1475 (1912). 44 APPLIED COLLOID CHEMISTRY stance into an active form may very well*account for the effect of the solvent on reaction velocity in homogeneous systems, it must be shown what active form is produced and why one sol- vent produces more of it than another. Activation may lead to the formation of intermediate compounds or it may not; but we have got to get down to more definiteness in all the cases of contact catalysis. It is often assumed that definite intermediate compounds are formed and this is certainly true in some cases. In the catalysis of hydrogen peroxide by mercury, the interme- diate formation of mercuric peroxide can be detected by the eye because there is an intermittent formation of a film, which then breaks down, only to form again. 1 The formation of graphite is usually preceded by the formation of a carbide. The conversion of acetic acid into acetone 2 by passing the vapor over heated barium carbonate presumably involves the intermediate formation of barium acetate, though one cannot ear-mark the carbon dioxide and so cannot be absolutely certain that the barium carbonate is continually decomposed and regenerated. In the Deacon chlorine process, where a mixture of hydrochloric acid and oxygen is passed over heated cupric chloride, Hurter 3 assumes the existence of an intermediate compound. People are not agreed, however, whether the intermediate compound is cuprous chloride, cupric oxychloride, or both. The extreme sensitiveness of this reaction to poisons raises a question whether an intermediate compound really is formed. In the catalytic oxidation of carbon monoxide, it is usually believed that there is alternate reduction and oxidation of the oxides which act as catalytic agents. Sabatier 4 believes that^ the catalytic action of alumina on alcohol is due to the intermediate formation of an instable com- pound of alumina and water. On the other hand, Henderson 5 assumes the intermediate formation of aluminum ethylate. Neither one offers any experimental evidence in support of his 1 BREDIG and VON ANTROPOFF: Zeit. Elektrochemie, 12, 581 (1906); VON ANTROPOFF: Jour, prakt. Chem. (2) 77, 273 (1908). 2 SQUIBB: Jour. Am. Chem. Soc., 17, 187 (1895). 3 Jour. Soc. Chem. Ind., 2, 106 (1883). 4 Die Katalyse in der organischen Chemie, 240 (1914). 6 Catalysis in Industrial Chemistry, 4 (1919). ADSORPTION OF GAS OR VAPOR BY SOLID 45 belief, and neither one discusses why kaolin should act practi- cally as well as alumina. It seems probable that the best way to solve the problem of contact catalysis is to start with the side which has been neglected a good deal and to study the selective action of catalytic agents. Quite apart from any theory we may adopt, there are certain questions which ought to be answered experimentally and yet these questions' have not even been raised in a serious way. We know that ethyl alcohol can be decomposed into acetaldehyde and hydrogen or into ethylene and water but that is practically all that we know as yet about these reactions. With pulverulent nickel the reaction is almost completely CH 3 CH 2 OH = CH 3 CH 2 O + H 2 We do not know whether the hydrogen comes off in one stage or in two. If it comes off in two stages, is the first stage CH 3 CH 2 O or CH 3 CHOH? While it may not be easy to devise a method which will answer this question, the problem is not more difficult than many which the organic chemists have solved successfully. With pulverulent alumina the reaction is almost completely CH 3 CH 2 OH = CH 2 CH 2 + H 2 O. Does water split off as such or do we get the splitting off of hydroxyl or hydrogen? It seems improbable that water can split off as such because then it would be difficult to account for the intermediate formation of ether. If the first stage is a splitting off of hydroxyl does the other hydrogen come from the adjacent carbon atom giving ethylene direct or does it come from the same carbon atom, forming a substituted methylene, CH 3 CH, which then rearranges to ethylene? One would also like to know what nickel does with diethyl ether. With nickel we get the following decomposition of the esters: CH 3 C0 2 CH 2 CH 3 = CH 3 CH 2 CH 3 + C0 2 CH 3 CO 2 CH 3 = CH 3 CH 3 + CO 2 HC0 2 CH 3 (?) = GH 4 + C0 2 With thoria the decomposition is quite different: 2CH 3 C0 2 C 2 H 2 = CH 3 COCH 3 + C0 2 + (C 2 H 5 ) 2 O = CH 3 COCH 3 + CO 2 + C 2 H 6 OH + C 2 H 4 2CH 3 CO 2 CH 3 = CH 3 COCH 3 + C0 2 + (CH 3 ) 2 O 2H CO 2 CH 3 = HCHO + C0 2 + (CH 3 ) 2 O 46 APPLIED COLLOID CHEMISTRY With titania we have a third set of products: CH 3 C0 2 C 2 H 5 = CH 3 C0 2 H + C 2 H 4 2CH 3 CO 2 CH 3 = 2CH 3 C0 2 H + C 2 H 4 HC0 2 CH 3 = HCO 2 H + CH 2 = CO + CH 3 OH The decompositions are regular and characteristic with each catalytic agent and the molecules must break or slip at different points in the different cases; but we certainly ought to know what the hypothetical intermediate stages are. 1 Take the case of methyl acetate with nickel, for instance. One assumption might be that we have a methylene dissociation CH 3 C0 2 CH 3 = CH 3 C0 2 H + CH 2 = CH 3 CH 3 + CO 2 A second assumption is that methyl splits off: CH 3 C0 2 CH 3 = CH 3 C0 2 + CH 3 = C 2 H 6 + CO 2 A third assumption might be a splitting off of both methyl groups : CH 3 C0 2 CH 3 = 2CH 3 + C0 2 = C 2 H 6 + 2 . Is any of these what happens and how can we prove it? Also how does the final assumption tie in with the facts that in aqueous solution acetate ion goes to ethane and CO 2 2CH 3 C0 2 = C 2 H 6 + 2C0 2 and that dry lead tetra-acetate decomposes at about 140 into lead acetate, methyl acetate and carbon dioxide Pb(CH 3 C0 2 ) 4 = Pb(CH 3 C0 2 ) 2 + CH 3 CO 2 CH 3 + C0 2 The question also comes up as to the changes in the intermediate stages in case the esters are chlorinated more or less completely. The effect of the catalytic agent is very marked when it comes to the preferential oxidation of mixed gases. 2 At low temperatures, the nature of the catalytic agent may determine which of two combustible gases will burn the more rapidly. In presence of platinum, copper oxide, or chamotte, hydrogen burns more readily than methane, whereas the reverse is true in borosilicate glass bulbs at 300-400. Depending on conditions, either hydrogen or carbon monoxide will burn the more readily. 3 1 This is now being studied experimentally by PROFESSOR HOMER ADKINS. 2 HENRY: Phil. Mag., 65, 269 (1825); BONE: Jour. Chem. Soc., 81, 535 (1902); 85, 694 (1894); Proc. Roy. Inst., 19, 82 (1908); Phil. Trans. 215, 298 (1915). 3 RIDEAL: Jour. Chem. Soc., 115, 993 (1919). ADSORPTION OF GAS OR VAPOR BY SOLID 47 At very high temperatures the specific effect of the catalytic agent becomes negligible because the reaction will take place so rapidly in the gas phase. One very striking characteristic of solid catalytic agents is the ease With which they become inactive or are poisoned. Since the reaction takes place in or at the surf ace, it follows that any sub- stance which cuts down the rate at which the reacting substances reach the catalytic surface 1 or which prevents them from reach- ing it, will decrease the reaction velocity and may destroy the catalytic action entirely. Berliner 2 has shown that traces of fatty vapors from the air or from the grease on the stop-cocks will decrease the adsorption of hydrogen by palladium from nearly nine hundred volumes practically to nothing. Faraday 3 has shown that traces of grease destroy the catalytic action of platinum on oxyhydrogen gas. De Hemptinne 4 has apparently shown that carbon monoxide cuts down the adsorption of hydro- gen by palladium, though his method of presenting his results is very obscure. Lunge and Harbeck 5 found that carbon mon- oxide inhibits practically completely the catalytic action of platinum on a mixture of ethylene and hydrogen. Schonbein 6 pointed out that the hydrides of sulphur, tellurium, selenium, phosphorus, arsenic, and antimony act very energetically in cutting down the catalytic action of platinum on mixtures of air with hydrogen or ether. He considered that the hydride must decompose, giving rise to a solid film. This is not necessary in order to account for the phenomenon; but he seems to have been right in at least one case, for Maxted 7 has found that hydro- gen sulphide is decomposed by platinum black with evolution of hydrogen and that the platinum then does not adsorb hydro- gen. Langmuir 8 believes that oxygen prevents the dissociation of hydrogen by a heated tungsten filament because it cuts down the adsorption of the hydrogen. 1 TAYLOR: Trans. Am. Electrochem. Soc., 36 (1919). 2 Wied. Ann., 35, 903 (1888). 3 Experimental Researches on Electricity, 1, 185 (1839). 4 Zeit. phys. Chem., 27, 249 (1898). 5 Zeit. anorg. Chem., 16, 50 (1898). 6 Jour, prakt. Chem., 29, 238 (1843). 7 Jour. Chem. Soc., 115, 1050 U919). 8 Jour. Am. Chem. Soc., 38, 2272 (1916). 48 APPLIED COLLOID CHEMISTRY Harned 1 has shown that the rate of adsorption 2 of chloropicrin by a charcoal which has been cleaned by washing with chloropicrin is much greater at first than by a charcoal which has not been so cleaned, although the final equilibrium is apparently about the same in the two cases. This is analogous to the evaporation of water when covered by an oil film. The oil cuts down the rate of evaporation very much but has practically no effect on the partial pressure of water at equilibrium. Taylor points out that normally the time of contact between a gas and the solid catalytic agent is extremely small and consequently anything which decreases the rate of adsorption will cut down the reaction velocity very much. It is easy to see that the piling up of reaction products 3 will cut down the reaction velocity, if they prevent the reacting sub- stances from coming in contact with the catalytic agent. This has been observed in the contact sulphuric acid process. 4 The explanation that the decrease in the reaction velocity is due to a decreased adsorption of the reacting substances was first given by Fink 5 who is the real pioneer in this line. Although the reaction between carbon monoxide and oxygen is practically irreversible at ordinary temperature, Henry 6 recognized that the presence of the reaction product might slow up the rate of reaction and he proved his point by increasing the reaction velocity when he removed the carbon dioxide with caustic potash. Water vapor checks the catalytic dehydration of ether 7 and of alcohol 8 and hydrogen cuts down the catalytic dehydrogenation of alcohol. When catalytic poisons are present or are formed during the reaction, the apparent equilibrium may vary with the amount of the catalytic agent. With only a small amount present, the catalytic agent will be poisoned before the reaction has run very far. In the hydrolysis of ethyl butyrate by enzymes, the reaction 1 Jour. Am. Chem. Soc., 42, 372 (1020). 2 TAYLOR: Trans. Am. Electrochem. Soc., 36 (1919). 3 BUNSEN (1857). Quoted by DEACON: Jour. Chem. Soc., 23, 736 (1872). 4 BODLANDER and KOPPEN: Zeit. Elektrochemie, 9, 566 (1903); BERL: Zeit. anorg. Chem., 44, 267 (1905). 5 BODENSTEIN and FINK: Zeit. phys. Chem., 60, 61 (1907). 6 Phil. Mag. (3) 9, 324 (1836). 7 IPATIEFF: Ber. deutsch. chem. Ges., 37, 2996 (1904). S ENGELDER: Jour. Phys. Chem., 21, 676 (1917). ADSORPTION OF GAS OR VAPOR BY SOLID 49 apparently runs to different end-points, depending on the relative amounts of enzyme. 1 One may also get abnormal end-points in the hydrolysis of a solid, if the product of hydrolysis is also a solid and forms a film over the reacting substance. With one of the war gases, the amount of hydrochloric acid split off in presence of water was practically the same when the amount of water added varied in the ratio of one to four. In other words, an apparent equilibrium was reached at a given concentration of hydrochloric acid in one case and at one-fourth that concen- tration when four times as much water was added. The development of some of the technical contact processes has been so interesting that it seems worth while to discuss a few of them briefly, beginning with the contact sulphuric acid process. 2 CONTACT SULPHURIC ACID PROCESS In 1831 Phillips discovered that spongy platinum increases the rate at which sulphur dioxide combines with oxygen according to the equation 2SO 2 + 02 = 2SO 3 . He obtained a patent for this but could not develop a commercial process. In 1846 Jul- lion patented the use of platinized asbestos. In 1842 Wohler and Mahla discovered that the oxides of copper, iron, and chrom- ium act like platinum. In the Schroeder-Grillo contact material hydrated magnesium sulphate is soaked in a solution of a plati- num salt and then heated in presence of sulphur dioxide. A thin layer of platinum is thus precipitated on the surface of the porous, dehydrated salt, 0.2-0.3 percent of platinum giving an excellent catalytic agent. 3 Knietsch found that a practically complete conversion of sulphur dioxide into sulphur trioxide could be obtained when sulphur dioxide was diluted with air, the conversion being more nearly complete the greater the ratio of oxygen to sulphur dioxide. When the experiments were repeated with purified burner gas on a commercial scale, the contact material soon became inert. It was found that this disastrous result was due to impurities in the burner gases. Even when present in apparently negligible 1 KASTLE and LOEVENHART: Am. Chem. Jour., 24, 491 (1900). 2 KNIETSCH: Ber. deutsch. chem. Ges., 34, 4069 (1901). 3 RIDEAL and TAYLOR: Catalysis in Theory and Practice, 86 (1919). 50 APPLIED COLLOID CHEMISTRY amounts, these impurities play havoc with the catalytic action of the platinum. Arsenic is the most dangerous and next mer- cury, while antimony, bismuth, lead, iron, zinc, etc., act in so far as they settle on and actually blanket the contact material. An amount of arsenic equal to 1-2 percent of the platinum in the contact material will destroy the catalytic action completely. In view of the enormous volumes of gas which are handled, it is easy to see to what an infinitesimal value the arsenic content must be reduced. By cooling the burner gases slowly and bringing them into intimate contact with sulphuric acid, it was found possible to obtain a gas mixture which passed all ordinary optical and chemi- cal tests and which appeared to be pure. When tested on the large scale, it was still found to contain arsenic. The gases had been cooled in long iron tubes and had reacted with these suffi- ciently to give traces of arsine. This difficulty was eliminated by not letting the gases come in contact with the iron. When the roaster furnaces were run to maximum capacity, however, a slight cloud was formed which could not be got rid of and which proved to consist of unburned sulphur. This in itself would have done no harm had not the particles of sulphur con- tained some arsenic. It therefore proved necessary to admit enough air so as to be sure that the sulphur was burned com- pletely. While no one has shown that the presence of arsenic cuts down the adsorption by platinum, it is quite certain from Berliner's experiments on palladium and hydrogen 1 that this must be the case. In American practice the burner gases carry arsenic, lead sulphate, silicon tetrafluoride, hydrochloric acid, sulphur, and selenium. Reese 2 calls the first two permanent poisons and the others temporary poisons. When working with platinum on a magnesium sulphate base, it is possible to regenerate the contact material after it has been poisoned with arsenic by heating with hydrochloric acid. The assumption was that the arsenic was volatilized as the trichloride. Analysis showed, however, that there might be 2-3 percent of arsenic in the regenerated catalyst without its having any apparent effect. It must be that the 1 BERLINER: Wied. Ann., 35, 804 (1888). 2 Jour. Soc. Chem. Ind., 22, 351 (1903). ADSORPTION OF GAS OR VAPOR BY SOLID 51 arsenic had agglomerated in some way so that it did not coat the platinum. In spite of Knietsch's statement to the contrary, it is possible to filter out all the solid poisons. Under ordinary conditions sulphur dioxide and oxygen com- bine very slowly and sulphur trioxide dissociates very slowly in the absence of catalytic agents. In presence of catalytic agents sulphur trioxide begins to dissociate perceptibly at about 400 and is practically completely dissociated at 1000. At first sight it would therefore seem that the process should be run below 400. The time factor is a complicating one, however. With platinum as catalytic agent the reaction begins at about 200, but is relatively slow at that temperature, and under commercial con- ditions there is practically no reaction below 400. It has proved best technically to work at about 425-450, the increased reaction velocity compensating for the decreased chemical efficiency. At 425 the conversion is 98 percent, dropping to 91 percent at 500. Ferric oxide can be used as a catalytic agent; but the reaction velocity is less and consequently it is necessary to work at a higher temperature with a correspondingly lower chemical efficiency, perhaps 60 percent. Ferric oxide is much less susceptible to arsenic poisoning than is platinum, presumably because of the higher temperature at which it acts. Since arsenic oxide is itself a catalytic agent when free from iron, 1 there must be temperatures at which it will have little or no deleterious action. If one could make a ferric oxide catalyst which would work at as low a temperature as platinum, it is probable that it would be about as sensitive to arsenic poisoning as platinum. It is entirely a question of the way in which the arsenic precipitates on the surface of the catalytic agent. The heat of reaction for the conversion of sulphur dioxide into sulphur trioxide is given as 22,600 cal per mol of gaseous sulphur trioxide. Since this heat is evolved at the surface of the contact material, it is quite a serious matter to keep the temperature within the proper range. The technical manufacture of sulphur trioxide is therefore very largely an engineering problem and the recent improvements have been chiefly changes in design 2 so as to give more efficient temperature control. 1 Cf. BERL: Zeit. anorg. Chem., 44, 267 (1905). 2 ZEISBERG: Trans. Am. Electrochem. Soc., 36 (1919). 52 APPLIED COLLOID CHEMISTRY HARGREAVES SALT-CAKE PROCESS The Hargreaves and Robinson process 1 for making salt-cake was established on a practical basis in 1870. The upper tem- perature limit is about 600 owing to the fusion of the salt. Origi- nally the process involved no catalytic agent but consisted simply in the passage of pyrites gases over suitably prepared salt at 500-550, the reaction being 4NaCl + 2S0 2 + 2 + 2H 2 = 2Na 2 S0 4 + 4HC1 In 1886 small amounts of iron oxide were added as a catalytic agent. Krutwig 2 showed that without ferric oxide about 28 percent of the sulphur in pyrites was converted into sodium sulphate after four hours at a red heat. With ferric oxide a 70 percent yield was obtained at high temperatures and 90 percent at lower ones. The action is also accelerated by cupric oxide and, since 1907, 0.1-1.0 percent of copper or iron salts or both have been added to the salt before the latter is moulded, the amount added depending on the required purity of the product. When the salt-cake is to be used in glass works, only a small amount of the catalytic agent can be added; but the maximum amount is permissible when the salt-cake is intended for the LeBlanc soda process. Increasing the amount of catalytic agent increases the reaction velocity and lowers the temperature at which the action starts. In this particular process there is little -or no difficulty with the poisoning of the catalytic agent because the oxides are not used over and over again. DEACON CHLORINE PROCESS In 1845 Oxland 3 discovered that chlorine is formed to some extent when a mixture of hydrochloric acid and oxygen is passed over a heated porous substance. The reaction is 4 HC1 + 2 = 2H 2 O + 2C1 2 ; but the reaction does not proceed fast enough to be 1 THORPE: Dictionary of Applied Chemistry, 6, 25 (1913); JOBLING: Catalysis and its Industrial Applications, 32 (1916;; RIDEAL and TAYLOR: Catalysis in Theory and Practice, 89 U919J. 2 Recueil Trav. chim. Pays-Bas, 16, 173 (1897). 3 THORPE: Dictionary of Applied Chemistry, 2, 18 (1918); JOBLING: Catalysis and its Industrial Applications, 26 (1916); RIDEAL and TAYLOR: Catalysis in Theory and Practice, 99 (1919). ADSORPTION OF GAS OR VAPOR BY SOLID 53 practicable commercially. In 1855 Vogel treated cupric chloride alternately with hydrochloric acid and oxygen. In 1888 Deacon and Hurter put the process on a successful technical basis by passing a mixture of hydrochloric acid and oxygen over pumice impregnated with cupric chloride which acts as a catalytic agent. It is usually assumed that this is not a case of adsorption catalysis but that the cupric chloride breaks up into chlorine and cuprous chloride which latter combines with oxygen to form cupric oxy chloride, CuO.CuCl2. This reacts with hydrochloric acid to form cupric chloride, after which the whole series of reactions take place again and again. There is no real proof of this 1 and it is really quite as probable that the oxygen does react direct with the hydrochloric acid,which is what it must do when there is no cupric chloride present. The catalytic agent is very sensitive to the effect of impurities, such as sulphur dioxide, sulphur trio- oxide, and arsenic. It is because of this sensitiveness to sulphur dioxide that the hydrochloric acid from the Hargreaves salt-cake process cannot be used in the Deacon process. Water is also bad for the reaction. The gases are therefore dried at 40 before entering the decomposer, as it is called. The temperature at which the best results are obtained is about 450-460. Above this temperature the cupric chloride volatilizes too rapidly and at lower temperatures the reaction velocity is too low. About two-thirds of the hydrochloric acid is converted into chlorine. CHANCE-GLAUS PROCESS In the LeBlanc process for making sodium carbonate, calcium sulphide is a by-product. As far back as 1837 Gossage 2 patented a process for using this tank waste. His plan was to treat the calcium sulphide with carbon dioxide and then to burn the resulting hydrogen sulphide to sulphur according to the equations, CaS + H 2 + C0 2 = CaC0 3 + H 2 S 2H 2 S + 2 = 2S + 2H 2 Gossage failed because he could not get a sufficiently concentrated CO 2 to start with or a sufficiently concentrated hydrogen sulphide for the second stage of the process. This problem was solved by Chance in 1887. 1 HURTER: Jour. Soc. Chem. Ind. 2, 106 (1883). 2 JOBLING: Catalysis and its Industrial Applications, 33 (1916). 54 APPLIED COLLOID CHEMISTRY The burning of the hydrogen sulphide was carried out accord- ing to a process patented by Glaus but developed further by Chance. 1 The Glaus kiln consists of a cylinder provided inter- nally with a grating upon which rests a layer of broken fire-brick, with a further layer of bog-iron ore superimposed. The combus- tible mixture enters at the top, passes through the porous layers, where it is burned to sulphur vapor and steam with but little sul- phur dioxide and sulphuretted hydrogen as impurity, and is drawn from the bottom into a series of condensing chambers, where part of the sulphur vapor condenses to liquid sulphur and the remainder as flowers of sulphur along with the condensed steam. The reaction is exothermic, so that no external heating is necessary. To start the reaction, a few shovelfuls of red-hot coal are thrown on to the oxide; whilst the temperature is regulated by controlling the speed of the gas. During the process the ferric oxide becomes transformed into pyrites, but the pyrites so formed has some peculiar property attached to it which ordi- nary pyrites do not possess, for the latter is quite useless as a catalyst. When once the necessary temperature has been reached, the reaction is found to proceed fairly satisfactorily with other contact material such as broken brick, though the working temperature then becomes somewhat higher than when ferric oxide only is employed. For starting the reaction, however, ferric oxide or similar contact material must be employed; hence the two layers in the Glaus kiln. Salamon 2 says that the ferric oxide undergoes no change. If that is so, it is not necessary to account for the difference in behavior between natural pyrites and that formed in the Glaus kiln. If iron sulphide is actually formed, the alleged difference in behavior is undoubtedly due to a difference in surface and con- sequently of adsorption. The natural pyrites is of course only very slightly porous. Rideal and Taylor 3 consider that the reaction takes place in two stages, Fe 2 3 + 3H 2 S = 2Fe S + 3H 2 O + S 4FeS + 30 2 = 2FeO + 48. They do not discuss the behavior of natural pyrites. 1 JOBLING: Catalysis and its Industrial Applications, 33, (1916). 2 THORPE: Dictionary of Applied Chemistry, 6, 294 (1917). 3 Catalysis in Theory and Practice, 112 (1919). ADSORPTION OF GAS OR VAPOR BY SOLID 55 HABER AMMONIA PROCESS In the contact sulphuric acid process it was found that sul- phur trioxide dissociates practically completely into sulphur dioxide and oxygen, at 1000, while recombination is practically complete at 400 if equilibrium is reached. Matters are not so favorable in the case of the synthesis of ammonia from nitrogen and hydrogen. At 1000 ammonia is practically completely dissociated just as is sulphur trioxide; but at 550 the theoretical amount of recombination at atmospheric pressure is about 0.08 percent instead of 100 percent. Increase of pressure displaces the equilibrium so that more is formed of the system occupying the lesser volume, namely ammonia. At 550 and 200 atm. the amount of ammonia formed is almost 12 percent. The data 1 are given in Table XVIII. TABLE XV11I. EQUILIBRIUM DATA FOB AMMONIA SYNTHESIS Reaction is N 2 + 3H 2 = 2NH 3 Pressure in atm. 550 650 750 850 1000 Percentage ammonia 1 0.077 0.032 0.916 0.008 0.006 100 0.7 3.02 1.54 0.87 0.54 200 11.9 5.71 2.99 1.68 1.07 Though the yield of 12 percent is low, the ammonia can be removed as ammonia of crystallization by ammonium nitrate at low temperatures, 2 a fresh supply of the reacting gases can be added to the unchanged nitrogen and hydrogen, and the whole can be passed repeatedly through the reacting chamber until the concentration of methane and argon becomes too high. 3 What catalytic agent to use is a question of first importance. The first catalytic agents tried were nickel, manganese, iron and chromium; afterwards experiments were made with osmium and 1 HABER: Jour. Soc. Chem. Ind., 33, 53 (1914). 2 THORPE: Outlines of Industrial Chemistry, 151 (1916). 3 RiDEAL and TAYLOR: Catalysis in Theory and Practice, 241 (1919). 56 APPLIED COLLOID CHEMISTRY uranium. 1 While metallic osmium and uranium carbide are apparently the best for laboratory work, they are not well adapted for technical use. 2 Judging by the most recent patents, pure iron 3 is the catalyst generally employed in this process. Using this material and with the aid of various minor improve- ments, it is said to be found possible to lower the working pres- sure to the neighborhood of 50 atmospheres. In all cases, it is essential in the interests of efficiency, that the catalyst be prepared at a temperature not greatly exceeding 600. Of the remaining substances which catalyze the reaction, the most interesting are molybdenum and molybdates, tungsten and its alloys, and cerium with its congeners. Platinum, on the other hand, though related to osmium, possesses but a slight catalytic activity. It was found that the activity of the catalysts can be increased by the addition of certain foreign bodies. These " promoters, " as they have been called, include various metals, the oxides, hydroxides, and salts of the alkali and alkaline earth metals, as well as many other substances of the most varied nature. In many cases just a trace of the contaminating body enhances considerably the activity of the catalyst. Moreover, all the catalysts appear to possess the capacity for invigoration. 4 There are no data to show whether these so-called promoters change the degree of adsorption of any of the substances concerned in the reaction or whether they prevent a possible sintering of the catalytic agent. Mannich and Thiel 5 state that charcoal increases the adsorbing power of palladium for hydrogen. Rideal and Taylor 6 consider that the promoters may cause the relative adsorption of the reacting substances to approach the theoretical ratios more closely. The contact ammonia process is very susceptible to poisons. Among the harmful substances are sulphur, selenium, tellurium, phosphorus, arsenic, boron, and their compounds; many carbon compounds; lead, zinc, bismuth, tin, etc. 7 Such minute quanti- 1 HABER: Jour. Soc. Chem. Ind., 33, 54 (1914). 2 JOBLING: Catalysis and its Industrial Applications, 42 (1916). 3 Cf. JELLINEK: Zeit. anorg. Chem., 71, 121 (1911). 4 PEASE and TAYLOR: Jour. Phys. Chem. 24, 241 (1920). 5 Jour. Soc. Chem. Ind., 35, 548 (1916), 6 Catalysis in Theory and Practice, 29 (1919). 7 JOBLING: Catalysis and its Industrial Applications, 44 (1916). ADSORPTION OF GAS OR VAPOR BY SOLID 57 ties of any of the above-mentioned substances as are to be found almost always in the purest commercial products and so-called pure gases are sufficient to diminish the catalytic efficiency very seriously. For instance, an impurity of 0.01 percent sulphur in iron renders it nearly useless. The Haber process was of enormous value to the Germans during the war. OSTWALD NITRIC ACID PROCESS In 1830 Kugelmann 1 observed that ammonia can be oxidized to nitric acid in presence of platinum. About eighty years later the process was put on a semi-commercial basis by Ostwald. The nominal reaction may be written NH 3 + 2O 2 = HNO 3 + H 2 O; but the real reaction is the formation of nitric oxide, which is then oxidized to nitrogen peroxide, and afterwards converted into nitric acid, 4NH 3 + 70 2 = 4NO + 6H 2 O + 2O 2 = 4NO 2 + 6H 2 O = 2HNO 3 + 2HNO 2 + 4H 2 O The problem is complicated by the apparently simultaneous oxidation to nitrogen, a reaction which is usually written 4NH 3 + 3O 2 = 2N 2 + 6H 2 O It is more likely that the disturbing reaction is an oxidation of ammonia by nitric oxide, 2 perhaps 4NH 3 + 6NO = 5N 2 + 6H 2 0. Both the reactions are accelerated by platinum; but, fortunately, the reaction yielding nitric oxide takes place much more rapidly than the reaction giving rise to nitrogen. The best results are obtained by keeping the gases in contact with the catalytic agent for as short a time as possible, which is an argument for the reaction producing nitrogen being a secondary one. If the two reactions took place simultaneously, the ratio of the two sets of products should be independent of the rate of passage of the gaseous mixture. A mixture of ammonia with ten or more volumes of air is passed at a velocity of 1-5 meters per second 1 JOBLING: Catalysis and its Industrial Applications, 45 (1916). 2 SABATiER and SENDERENS: Comptes rendus, 135, 278 (1902). 58 APPLIED COLLOID CHEMISTRY through a plug of platinum gauze, 1-2 cm in length and kept at 300. This gives a period of contact not exceeding 0.01 sec. Under favorable conditions an 85 percent conversion of ammonia is claimed. Better results have been obtained in this country, 1 with several layers of platinum gauze, 2 the yield running 92 percent, with a flow of 250 cubic feet per minute. The average concentration of ammonia is 10.7 percent. A smooth platinum gauze is not effective. After some time it becomes roughened and activated. Contrary to what Ostwald had believed, a high temperature, say above 825, is better than a lower one. Phosphine is a bad poison, having a distinct effect even when present three parts per hundred million, but cyanogen compounds are not bad because they oxidize readily. Iron oxide, grease, oil or tar must be kept away from the platinum gauze and a new gauze should be washed with pure gasolene or ether before being activated. Ammonia from any source can be used though Landis 3 claims that the platinum must be activated anew when changing to a different ammonia. SAJBATIER'S HYDROGENATION PROCESS Sabatier and Senderens 4 discovered, a number of years ago, that reduced nickel will cause hydrogen to combine with ethyl- ene or acetylene forming ethane. This has since developed into a general hydrogenation process. 5 The principle of the method is a simple one. The vapor of the substance to be reduced is mixed with hydrogen and passed directly over specific metallic catalysts maintained at temperatures usually held between 150 and 300. The process is extremely rapid, since complete hydro- genation occurs during the short time the mixture is passing over the catalyst. Rapidity, however, is not the only advantage, for Sabatier and his collaborators, Senderens, Mailhe, and others, who have worked on this method for some fifteen years, find it to be one which, whilst requiring the minimum of attention, furnishes very high yields. LARSONS: Jour. Ind. Eng. Chem. 11, 541 (1919). 2 PERLEY: Jour. Ind. Eng. Chem. 12, 5, 119 (1920). 3 Trans. Am. Electrochem. Soc., 35, 300 (1919). 4 Comptes rendus, 124, 1358 (1897); 128, 1173 (1899). 5 JOBLING: Catalysis and its Industrial Applications, 68 (1916); SABATIER: Die Katalyse in der organischen Chemie, 44 (1914). ADSORPTION OF GAS OR VAPOR BY SOLID 59 The metals which are suitable as catalysts are nickel, cobalt, platinum, iron and copper, of which nickel is far and away the most active, while the others are arranged in diminishing activity. As would be anticipated, the practicability of the method depends largely upon the condition in which the metal is used. It must be prepared by reduction of the oxide, the activity depending upon the nature of the oxide employed and the temperature at which it is reduced. In general the greater the surface exposed the greater the activity, so that the conditions must be carefully arranged to satisfy this requirement. Reduction at a low tem- perature is found to yield a metal which is too active and too sensitive to external influences, while reduction at a high tempera- ture reduces the catalytic action almost to zero. Consequently, a suitable intermediate temperature must be chosen. Sabatier finds that the best nickel for the purpose is produced by dissol- ving the metal in nitric acid, calcining the nitrate at a dull red heat, and then reducing the oxide slowly at a temperature of about 300-325, until water is no longer evolved. After the nickel has been prepared as described, it is necessary to keep it out of contact with the air, as it is extremely pyrophoric and quickly loses its activity on exposure. Weight for weight, the efficiency of pure nickel is not so great as that which has had its active surface increased by the aid of some suitable carrier. Hence many proposals have been made for combining the catalyst with a great variety of supporting bodies, ranging from pumice and kieselguhr to charcoal and sawdust. The materials used, whether hydrogen, metal, or organic com- pound, must be as pure as can be obtained, for any impurity is readily adsorbed by the catalyst to the detriment of its ac- tivity. Sulphur and its volatile compounds, as well as the halo- gens, 1 are particularly to be avoided in this respect, and, to a lesser degree, arsenic and phosphorus with their volatile com- pounds. Electrolytic hydrogen, or hydrogen of equivalent purity, appears to be essential. Even then the catalyst must, sooner or later, succumb to the toxic effects of minute impurities impossible to remove; in consequence, the desirability of fre- quent renewals needs emphasis. 1 The nickel oxide must not be prepared from the chloride because of the great difficulty of washing out the last traces of adsorbed chloride. 60 APPLIED COLLOID CHEMISTRY Of practical importance is the preparation of numerous arti- ficial perfumes by the aid of this reaction. Moreover, it permits of the production of cyclohexanol and p-methyl cyclohexanol, easily obtained from phenol and p-cresol respectively, which are used in the manufacture of isoprene and butadiene, and seem, therefore, destined to play an important part in the synthesis of rubber. Probably the most important technical application of Saba- tier's hydrogenation process is to the conversion of liquid and unsaturated fatty acids such as oleic acid 1 into solid and saturated acids such as stearic acid. As the acid is not vaporized, this process does not belong strictly speaking under the general head of gas reactions accelerated by solid catalytic agents, so I merely refer to it because it is a further development of the same principle. SABATIER'S DEHYDEOGENATION PROCESS Attention has already been drawn to the fact that alcohols may split off hydrogen or water at a given temperature, depend- ing on the nature of the catalytic agent. Finely divided metals, 2 such as copper, cobalt, nickel, iron, platinum, and palladium catalyze almost exclusively the dehydrogenation process. So also do some anhydrous oxides, chief among which are the lower oxides of manganese, tin, uranium, molybdenum, vanadium, and cadmium, though the activity of these is less than that of the metals. The dehydrbgenation of alcohols, both primary and secondary, is most readily effected by reduced copper. If a fatty alcohol be passed over this catalyst at 200-300, a yield of at least 50 percent aldehyde or ketone can usually be condensed. Cyclic alcohols may be dehydrogenated in a similar way. Borneol, for instance, can be transformed into camphor by pas- sage over reduced copper at 150, whilst geraniol under similar conditions gives citral. The former fact is utilized industrially in the treatment of certain camphors which are sold very cheaply *Cf. ELLIS: Jour. Soc. Chem. Ind., 31, 1155 (1912); SHAW: Ibid., 33, 771 (1914); ELLIS: Jour. Ind. Eng. Chem., 8, 886 (1916); ELLIS: The Hydro- genation of Oils, 2nd Ed., 108 (1918). 2 JOBLING: Catalysis and its Industrial Applications, 82 (1916). ADSORPTION OF GAS OR VAPOR BY SOLID 61 on account of their large content of borneol. As the usual proc- ess for removing borneol by means of nitric acid is very slow and unsatisfactory, the above dehydrogenation method, which yields nearly pure camphor, has found ready application in this field. Of the other catalysts, cobalt, iron, platinum, as well as the above-mentioned oxides, are less advantageous. Zinc is stated to possess high activity. Nickel is ruled out on account of the violence of its reaction, for it pushes decomposition too far. The most important of the industrial applications of this reaction is the preparation of formaldehyde from methyl alco- hol. 1 The earliest recorded production of formaldehyde is to be found in the classic experiment of Hofmann's (1867), in which a red-hot coil of platinum wire suspended over methyl alcohol in a beaker continued to glow there so long as it is in contact with a mixture of the vapor and air. This method forms the basis of most of the subsequent attempts and of the present commercial process. Air is driven through wood spirit kept at a temperature suitable for saturation and the mixture then passed into the catalyzer, usually copper gauze. In the so-called oxidation of methyl alcohol, the primary reaction is the splitting-off of hydrogen 2 and the practically si- multaneous depolarization by oxygen, 2CH 2 OH + O 2 = 2HCHO + 4H + 2 = 2HCHO + 2H 2 0. Hofmann 3 found that osmium tetroxide oxidizes alcohol to aldehyde but not to acetic acid, acting apparently as a carrier for oxygen. It is probable that we have here a catalytic splitting of the alcohol, the nascent hydrogen reacting with the osmium tetroxide. Wieland 4 considers that the contact sulphuric acid process really involves a dehydrogenation, the reactions being: S0 2 + H 2 = H 2 S0 3 H 2 S0 3 = S0 3 + Hi 2H 2 + O 2 = 2H 2 O 1 THORPE: Dictionary of Applied Chemistry, 2, 592 (1912). 2 L,EBLANC and PLASCHKE: Zeit. Elektrochemie, 17, 47 (1911). 3 Ber. deutsch. chem. Ges. 46, 1657, 2854 (1913). * Ibid. 45, 685 (1912). 62 APPLIED COLLOID CHEMISTRY SURFACE COMBUSTION The subject of surface combustion has been developed by Bone in England 1 and by Lucke in America. Bone began by a study of the combination of hydrogen and oxygen in contact with hot surfaces of porcelain, magnesia, silver, gold, platinum, nickel, calcined spathic iron ore (containing 79.0 percent ferric oxide and 14.5 percent manganous oxide), nickelous oxide, and copper oxide. With each of these surfaces the rate is proportional to the pressure of the dry gas when electrolytic gas is used, is ap- proximately proportional to the partial pressure of hydrogen when either gas is present in excess, and is increased by a previous treatment with hydrogen. In the case of porcelain, Bone decides that porous porcelain occludes or condenses both hydrogen and oxygen at rates which depend to some extent upon the physical condition and past history of the surface. While the process is extremely rapid in the case of oxygen and the surface layer is soon saturated, the occlusion of hydrogen is slower and the limit of saturation much higher. Combination between the occluded gases occurs at a rate either comparable with, or somewhat faster than, the rate at which the film of occluded oxygen is renewed, but considerably faster than the rate of occlusion of hydrogen. In later papers Bone 2 showed that all incandescent substances are capable of accelerating gaseous combustion; and to an approx- imately equal degree, for the wide differences between the catalytic powers of various surfaces at low temperatures gradu- ally diminish as the temperature rises, until at incandescence they practically disappear. If an explosive mixture of a combus- tible gas and air is pressed through the pores of a granular, high- melting solid, and if the issuing gas be lighted, it will speedily raise the granular solid to incandescence if the conditions are right, after which the heated surface will accelerate the rate of combustion so much that the whole of the gas will be burned at the surface of the solid and we shall have what has been called flameless combustion. As a matter of fact there is flame, but the flame is so short and the light from the incandescent solid so intense that there seems to be no flame. The combustion 1 Phil. Trans., 206A, 1 (1906); Jour. Franklin Inst., 173, 101 (1912). 2 Summarized in Ber. deutsch. chem. Ges., 46,5 (1913); JOBLING: Catalysis and its Industrial Applications, 51 (1916). ADSORPTION OF GAS OR VAPOR BY SOLID 63 takes place only at the surface layer of the granular solid but all over the surface and not as flames between each solid particle. The advantages of this surface combustion are an intense and a localized heat, combustion with minimum excess of air, and the development of a large amount of the heat as radiant energy. By passing the explosive mixture through a porous diaphragm we get an incandescent surface which can, if need be, be held glowing side down over a solution which is to be evaporated. In this way, it is possible, for instance, to evaporate a water- glass solution without danger of bumping. The sodium silicate forms as a solid film on the surface of the solution and can be removed from time to time. This form of heating has proved very successful in making candy. 1 The granular mass may be piled in round a crucible or around boiler tubes giving an extremely high efficiency. The maximum temperature depends on the working conditions. The choice of refractory material is determined by the temperatures involved. Under ordinary conditions calcined fire-clay, ganister, etc., are used; but when high temperatures are in question, practically all solids are eliminated except calcined magnesia and carborun- dum. Attempts are now being made to substitute an atomized liquid fuel for a gaseous fuel. It is stated that the most success- ful applications of surface combustion to large scale work are where the heat is radiated very quickly so that the temperature of the refractory does not rise unduly with steam boiler tubes packed in refractory material for instance. 2 Surface combustion furnaces have proved very useful in annealing shell. Since an explosive gas mixture is used, it is necessary to keep the flame from working back and burning in the mass of the refractory instead of on the surface. Bone does this by regulat- ing the rate of flow of the gaseous mixture. Lucke 3 has described several ways of doing this by special construction of the heater. WELSBACH GAS MANTLE The invention of the gas mantle made it possible for gas to continue to compete with electricity as an illuminant. The 1 SEAQER: Jour. Franklin Inst., 177, 367 (1914). 2 Reports of Progress of Applied Chemistry, 1917, 26. 3 Jour. Ind. Eng. Chem., 5, 801 (1913). 64 APPLIED COLLOID CHEMISTRY best results are obtained when the Welsbach gas mantle consists of 99 percent thoria and 1 percent ceria. 1 Pure thoria gives a relatively poor light, but the continuous addition of ceria gradually increases the luminosity to a maximum of ten-fold when 1 percent is present. Any further addition then dimin- ishes the luminosity until, with 10 percent the difference in effect is inappreciable. In present-day manufacture, the cotton, silk or ramie web is impregnated with a 25-33 percent aqueous solution of thorium and cerium nitrates in the proportion of 99:1, together with a little hardening medium, e.g., beryllium nitrate, if necessary. We are dealing with surface combustion both in the case of the lime light and of the Welsbach mantle, as is shown by the rapid decay of the light of the lime light owing to sintering 2 and the similar though much slower decay of the light of the Welsbach mantle. Mantles of magnesia and lanthana give a good light when new; but the light decreases as the surface becomes glassy. While a thoria mantle reaches a high temperature, 3 its emissive power is low. Ceria has a high emissive power in the visible spectrum; but the energy of radiation is so great that the temper- ature of a ceria mantle does not rise high enough. Starting with thoria and increasing the amount of ceria increases the emissive power but lowers the temperature of the mantle. The maximum luminous efficiency is obtained when the rise of visible emission due to the ceria is balanced by the drop in the temperature of the mantle due to the increased radiation. The ceria may form a solid solution with the thoria but there seem to be no data to show whether equilibrium is actually reached in the mantle. 1 JOBUNG: Catalysis and its Industrial Applications, 62 (1916). 2 STEINMETZ: Radiation, Light and Illumination, 92 (1909). 3 IvES, KINGSBURY and KARRER: Jour. Franklin Last., 186, 401, 585 (1918). CHAPTER II ADSORPTION OF VAPOR BY LIQUID AND OF LIQUID AND SOLID BY SOLID AND LIQUID ADSORPTION OF VAPOR BY LIQUID Campbell 1 believes that one source of error in the method of determining molecular weights by the so-called air-bubbling method is due to the formation of a film of adsorbed air on the surface of the solution which interferes with the evaporation. He deduced the existence of an adsorbed gas film from his quanti- tative measurements on evaporation. The existence of such a film is shown directly by Rayleigh's experiments on liquid jets. 2 A vertical jet of water or any liquid slows up and breaks into drops when the head is sufficiently great relatively to the dia- meter of the jet. These drops scatter because they rebound when they strike one another. If a feebly electrified body is brought close to the jet, the jet becomes coherent. Under more powerful electrical action the scattering of the drops becomes even greater than when there is no electrification. What happens is that there is formed round each drop a film of adsorbed air which prevents two drops from coalescing when they collide. When the drops are electrified, the air film is removed to such an extent that the drops come more nearly in contact and coalesce. With higher electrification, a marked charging of the drops takes place and the repulsion of the like charges causes the drops to scatter. Nipher 3 states that drops of 1 mm radius do not attract each other when charged to 0.0031 volt. That there is an air film between the drops can also be shown by letting two jets im- pinge under such conditions that they bound apart. It was found that whenever the jets rebounded, the electrical insulation was practically perfect. Rayleigh found that presence of dust 1 Trans. Faraday Soc., 10, 197 (1915). 2 RAYLEIGH: Proc. Roy. Soc., 28, 40G; 29, 71 (1879); 34, 130 (1882). 3 Science, 34, 442 (1911). 5 65 66 APPLIED COLLOID CHEMISTRY or addition of a drop of milk had a marked effect on the ease with which the drops coalesce. The reason for this has not been worked out. In Worthington's experiments on splashes 1 the adsorbed air film plays an important part. A drop of water may fall on a sheet of water in such a way that the drop passes below the sur- face, becoming completely submerged, and then emerges at the head of a column of adherent liquid with the upper portion apparently un wetted by the liquid with which it has been covered. When the water drops from an oar on the recover, we often get what are known as rolling drops 2 which depend on the presence of adsorbed air films and which are also sensitive to the presence of electrified substances. If the surface of a sheet of water is covered with oil of with a scum of oily soot or of plankton organisms, a mist or fog may deposit drops on the surface which will be extremely persistent. This phenomenon is not at all uncommon at Lake Mendota at Madison and has frequently been observed elsewhere. 3 Since soap-bubbles are hollow drops, they should behave in some respects like liquid drops, though not necessarily like drops of a pure liquid. It is possible to press two soap bubbles together with considerable force without causing them to coalesce; 4 but if they are electrified slightly, they coalesce readily without bursting. It is interesting to note that water will apparently remove a film of adsorbed air much more readily from a polished surface of ivory or marble than it will from a water surface; 5 but we do not know whether this is true for all solids. Sakulka 6 obtained somewhat different results with lead shot, but this may have been because they were not polished. If a vapor is adsorbed by a liquid, it will condense to a liquid film at a temperature above that of the normal dew-point. Can- 1 WORTHINGTON: Proc. Roy. Soc., 34, 217 (1882). 2 OSBORNE REYNOLDS: Chem. News, 44, 211 (1881); KAISER: Wied. Ann., 53, 681 (1894). 3 JuDAY: U. S. Monthly Weather Review, 44, 65 (1916); NAKAMURA: Jour. Meteorological Soc., Japan, 36 (1917). * BOYS: Phil. Mag. 25, 409 (1888). 6 WORTHINGTON: Proc. Roy. Soc., 34, 217 (1882). 6 Drude's Ann., 25, 892 (1908), ADSORPTION OF LIQUID BY SOLID 67 tor 1 realized this with water and mercury and Clark 2 with olive oil which is not a well-defined and reproducible liquid. ADSORPTION OF LIQUID BY SOLID If a liquid is adsorbed at a solid surface, it forms a liquid film and we say that it wets the solid. If a liquid is not adsorbed by a solid, it does not wet the latter. There has been a good deal of question whether the liquid meets the wall at a definite angle, the contact angle, or whether the contact angle is zero. Lord Rayleigh 3 states definitely that in his opinion the contact angle is zero, 4 and this must be true if we consider that the liquid is adsorbed by the solid. The formation of a liquid film over the surface of a wetted solid accounts for the generally accepted fact that the rise of a liquid in a capillary tube is inde- pendent of the nature of the walls of the tube. This has always seemed a very improbable state of things and one that could only be justified by the fact that it was so. The whole thing becomes quite simple when we consider that the rising liquid does not come in contact with the walls of the capillary tube at all. 5 We are really dealing with the rise of a liquid in a liquid tube and it makes no difference what material is used to support the walls of the liquid tube. That this is the real explanation may be seen from the fact that concordant results are not obtained when a liquid is allowed to rise in a dry tube. To get good results it is important to immerse the tube in the liquid and then to raise it. Bigelow and Hunter 6 claim that the nature of the walls does have an effect on- the capillary rise; but their experiments were not done in a capillary tube. On the assumption that the contact angle is zero, a thin film of liquid must rise to the top of the tube, irrespective of the height and diameter of the tube. This has been recognized explicitly by Lord Rayleigh. 7 When the liquid does not wet the solid, 1 Wied. Ann., 66, 492 (1895). 2 Proc. Am. Acad., 41, 361 (1906). 3 Scientific Papers, 3, 393 (1902). * Cf. HARKINS and BROWN: Jour. Am. Chem. Soc., 41, 522 (1919). 8 RAYLEIGH: Scientific Papers, 3, 421 (1902). 6 Jour. Phys. Chem., 15, 367 (1911). 7 Scientific Papers, 3, 523 (1902). 68 APPLIED COLLOID CHEMISTRY mercury and glass for instance, the assumption of a zero contact angle leads to at least one awkward conclusion. The upper part of the meniscus is certainly not in contact with the glass and therefore none of the mercury is in contact with the glass. If we have a glass tube, infinitely long, closed at the bottom, and filled with mercury, we must conclude, if we are going to be strictly logical, that the column of mercury does not touch the glass at any point and that it stands alone, surrounded on all sides by a film of air, mercury vapor, or something which is not glass. Personally, I feel no desire to be so logical as that. It seems to me probable that at some point the force of gravity squeezes out all the vapor layer and that the mercury does come in contact with the glass. This is the more certain because we know that the mercury sometimes tends to hang in a freshly-made barometer and this could not happen if the mercury were not in contact with the glass. For a liquid to wet a solid in the presence of air, the liquid must be adsorbed more strongly than the air and must displace it. Hofmann 1 gives a number of details as to the slowness with which some liquids displace adsorbed air from certain solids. In some cases heating is necessary to get quick results. Lycopodium powder is distinctly difficult to wet. Gum arabic adsorbs air strongly and is not wetted readily by water. If ground with alcohol, the alcohol displaces the air and then the gum is wetted more easily by water. The floating of metallic powders or of pieces of metal on water is due to the slowness with which water wets them. The Cape Cod lighter consists of a por-ous stone set in a brass handle. When this is dipped into a can of kerosene, the kerosene is sucked rapidly into the pores displacing the air. The light is taken out, lighted, and placed under the logs of a wood fire. The adsorbed kerosene burns for about ten minutes, enabling one to do away with paper and even kindling wood. Another interesting application of the adsorption of a liquid by a solid is the case of solidified bromine. 2 It consists of siliceous earth saturated with bromine, which can be driven out by heat. Of more technical importance is the adsorption of fused magne- 1 Zeit. phys. Chem., 83, 385 (1913). 2 BRAND: Zeit. anal. Chem., 26, 222; Jour. Chem. Soc., 52, 688 (1887). Cf. THORP: Outlines of Industrial Chemistry, 251 (1916). ADSORPTION OF LIQUID BY SOLID 69 tite by chromite. 1 The simple form of magnetite arc consists of a lower, negative, magnetite terminal and an upper non-consuming terminal of copper, of such size that it does not get so hot as to oxidize or evaporate, but sufficiently hot to avoid condensation of magnetite vapor. When the arc is struck, a pool of melted magnetite forms on the surface of the magnetite terminal and the arc-conductor issues from a depression in this melted pool. The straight magnetite arc flickers badly; but this can be overcome by mixing the magnetite with some more refractory material, such as chromite. The chromite remains solid and holds the melted magnetite as a sponge does water. In this way, these arcs have been made steady and thereby suitable for industrial use. We are quite familiar with the fact that glass must be cleaned before it can be wetted satisfactorily by water; but it is not so well known that platinum is only wetted by mercury under special conditions. 2 "Mercury, which holds even a very little sodium in solution, has the power of wetting platinum in a very re- markable manner. The appearance of the platinum is quite similar to that presented by amalgamable metals in contact with mercury. But the platinum is in no wise attacked. Further, the amalgam may be washed off by clean mercury, and the latter will also continue to adhere closely to the platinum. All the phenomena of capillarity are presented between the two. The surface of the mercury in a platinum cup so prepared is quite concave; and a basin of mercury may be emptied if a few strips of similarly prepared platinum foil be laid over its edge just as a basin of water may be emptied by strips of paper or cloth, and under the same condition, namely that the external limb of such capillary siphon be longer than the internal one. The sodium removes the oxide film and permits the mercury to come in contact with the platinum." It is found that iron, copper, bismuth, and antimony are also wetted by mercury if their surfaces are first touched with sodium amalgam. A few years later a technical application of this observation was made. Kirchman 3 reports that sodium amalgam . can be 1 STEINMETZ: Radiation, Light, and Illumination, 110 (1909). 2 GUTHRIE: Proc. Roy. Soc., 13, 458 (1864). 3 Jour. Chem. Soc., 26, 418 (1873). 70 APPLIED COLLOID CHEMISTRY used advantageously in simplifying the method of dry-gilding iron. Surfaces of iron and similar metals, even though oxidized, are immediately amalgamated by simply rubbing with amalgam. Gold chloride in concentrated solution is then applied quickly on the amalgamated surface, and the mercury volatilized by the heat of lamp or fire. A very uniform gilding is thus obtained, permitting a high polish. Corresponding results are obtained with silver and platinum salts. Wald 1 has pointed out that there must be a lowering of the vapor pressure of a liquid, when a liquid wets a solid. Conse- quently, if water really wets ice, or if water rises in a capillary tube made of ice, the vapor pressure of dry ice will differ from that of wetted ice at the same temperature. This point is not taken into account in books on the phase rule because effects due to capillarity, gravity, etc., are usually ruled out explicitly. The adsorption of a liquid film may introduce quite an error in the determination of the specific gravity of bodies when they are weighed in a highly divided state. As long ago as 1848 Rose 2 pointed out that platinum in the state of foil has a specific gravity between 21 and 22 while a value above 26 was found for platinum precipitated from the chloride by sodium carbonate and sugar. Similar though less extreme differences were found with gold, silver, and barium sulphate. This can be accounted for if we assume that the powder is not weighed alone in water but in conjunction with a film of condensed water. 3 It was Edward Lear who wrote about the Jumblies who went to sea in a sieve. He considered that a nonsense rhyme; but modern science is almost equal to the problem, for we can carry water in a sieve. If we take a fine-meshed metal sieve and oil the meshes, air can still pass readily through the holes; but water does not wet the wires and consequently surface tension pre- vents it from passing through the holes. 4 Gasolene may be freed from water by filtering through a 200 mesh metal sieve. We can carry water in such a sieve 5 if we do not fill it too full; 1 Zeit. phys. Chem., 7, 514 (1891). 2 Pogg. Ann., 73, 1 (1848); Jour. Chem. Soc., 1, 182 (1849). 3 See however JOHNSTON and ADAMS: Jour. Am. Chem. Soc., 34, 563 (1912), 4 CHWOLSON: Lehrbuch der Physik, 1 III, 613 (1907). 6 CAMPBELL: Jour. Ind. Eng. Chem., 6, 762 (1919). ADSORPTION OF LIQUID BY SOLID 71 but an oiled sieve is not a perfect substitute for a pail. This principle has received an important technical application in the shower- proofing of textile fabrics. 1 The fibers are coated with some water-repellent hydrocarbon, wax, or salt, in which case the fabric will be porous to air but will shed water up to a certain point. All campers are familiar with the unpleasant fact that a tent will leak in a rain if one touches the canvas with one's finger. The tent is shower-proof and not water-proof. When one pushes against the tent, it either contracts or expands the air spaces. If the air spaces contract, air is driven out and when the finger is taken away, water enters. If the air spaces expand, water runs in. Either way the air is displaced at that point and water enters. What happens can best be seen from an analogy. If one dips a fountain pen filler into ink and presses the bulb, air is driven out and ink enters when the pressure is relieved. If one pinches the bulb before dipping the glass end into the ink, no air bubbles out but ink goes in as soon as the pressure is taken off. It makes no difference therefore whether one touches a tent from the inside or the outside. In either case, water is drawn into the pores of the fabric and the tent leaks continuously once the air is displaced. Awnings on a boat usually leak where they rest on a wooden or iron support. This is because there is always enough wind or vibration to be equivalent to the touching of the tent with the finger. For the same reason, a tent is apt to leak around the pole. The reverse case is to be seen in the ordinary water wings. They are made of a fine-meshed fabric which is first thoroughly wetted so that a film of water is formed in the pores. The strength of this water film is so great relatively that one can blow up the water wings without the air escaping and the water wings will support a grown person without difficulty. Ehrenberg and Schultze 2 point out that many powders, when once thoroughly dried, are not easy to wet. After a period of drought, drops of rain will often roll along the dust without wet- ting it. Even in the case of a shower the dust may be only wetted to a depth of less than a quarter of an inch. This is 1 C. O. WEBER: Jour. Soc. Dyers and Colourists, 17, 146 (1901). 2 Zeit. Kolloidchemie, 16, 183 (1914). 72 APPLIED COLLOID CHEMISTRY shown by Ehrenberg and Schultze to be due to the adsorbed air on the surface of the solid. Any treatment which cuts down the amount of adsorbed air makes the dust, or soot, or other substance, more easily wetted. It is not easy to wet carbon black with water; but if the oily material is driven out by a preliminary treatment with super- heated steam, carbon black is dispersed by water very rapidly and the liquid bubbles like soda water owing to the escape of the adsorbed air. If a 50 cc vessel is filled with good cocoanut charcoal and if water is then added until the vessel is full up to the mark, 150-200 cc of air will be driven off. A case analogous to that of the oiled sieve holding water has been described by Lord Rayleigh 1 who found that a piece of iron gauze pressed down upon the flat bottom of a glass vessel, holding a shallow layer of mercury, remained on the bottom of the vessel and did not rise through the mercury. It is quite clear from this that one could carry mercury in an iron sieve. With sodium in the mercury the amalgam would run through the gauze. Ray- leigh 's experiment of course would not succeed if tried with sodium amalgam instead of with mercury. Since the adsorption of liquids by solids is selective, it follows that one liquid will be adsorbed more by a solid than will a second liquid and that consequently the first will displace the second from contact with the solid. No systematic study of this phenomenon seems to have been made; but we know that kero- sene will displace water in contact with copper and that water will displace kerosene in contact with quartz, 2 while alcohol will displace oil in contact with metal 3 and linseed oil 4 will displace water in contact with white lead. According to this, one should use a rag dipped in alcohol with which to wipe off a lamp and it is for this reason that we cover metal surfaces with vaseline or with a heavy oil to prevent rusting. Parkerized or phosphatized iron is said to owe its non-rusting properties to an oil film. In the kitchen, people wet the moulds before putting in corn starch or gelatine, and they butter the pans before making fudge, or grease 1 Scientific Papers, 4, 430 (1903). 2 HOFMANN: Zeifc. phys. Chem., 83, 385 (1913). 3 PocKELs: Wied. Ann., 67, 669 (1899). 4 CRUICKSHANK SMITH: The Manufacture of Paint, 92 (1915). ADSORPTION OF LIQUID BY SOLID 73 the griddles before making flapjacks. Jt is claimed that alumi- num griddles do not need to be greased because the air in the oxide film keeps the cakes from sticking; but the same reasoning ought to apply to frying-pans when used for the same purpose and apparently does not. In making pottery, it is the practice to oil the dies when making dust-pressed ware. 1 Infusorial earth is excellent 2 for cleaning glass plates for photographic pur- poses. Even very greasy plates become clean very rapidly when rubbed with infusorial earth moistened with water. Some of the methods of removing grease spots from clothes depend on selective adsorption. 3 There are almost no quantitative data on selective adsorption. Graham 4 made a few rough measurements on the selective ad- sorption of liquids by palladium. A piece of foil was immersed in a liquid for an hour and afterwards dried by pressure for a few seconds between folds of blotting paper. One thousand parts of palladium foil were found to retain 1.18 of water, 5.5 of alcohol, 1.7 of ether, 0.54 of acetone, 4.5 of gylcerine, 3.5 of benzene, 10.2 of castor oil, and 18.1 of oil of sweet almonds. Quite recently Mathers 5 has shown that lead adsorbs certain essential oils from aqueous solution. Since the adsorption is selective, a vapor or liquid will be ad- sorbed more readily by one surface than by another. A striking application of this is to be found in the old daguerreotype proc- ess. The latent image is formed on a plate of silver coated with silver iodide and it is developed by means of mercury vapor. The mercury condenses on the places which have been exposed to light. I have seen a patched cement cellar floor on which mois- ture condensed so markedly on the weathered portion that it seemed as though a pipe had broken under the floor. If the adsorption is very marked we may get intense strains. 6 Gelatine is adsorbed so strongly by glass that the gelatine film on drying will often tear off the surface of the glass. On the other 1 SIMCOE and SMITH: Trans. Am. Ceramic. Soc., 14, 558, 567 (1912). 2 ROLLER: The Utilization of Waste Products, 310 (1915). 3 LAKE: Jour. Phys. Chem., 20, 761 (1916). 4 Jour. Chem. Soc., 20, 275 (1867). 5 Trans. Am. Electrochem. Soc., 31, 271 (1917). 8 GRAHAM: Jour. Chem. Soc., 17, 320 (1864). 74 APPLIED COLLOID CHEMISTRY hand, it does not adhere to plates of mica or calcite. With agar- agar we get a state of things which appears to vary with the concentrations. Certain agar-agar jellies will adhere readily to glass while others apparently do not. People have claimed that the substitution of agar-agar for gelatine in photographic plates would be impossible on a technical scale, because the agar-agar film would slip off if the plates were set vertically to dry; but it is probable that this difficulty would be overcome if that were the only drawback. The whole theory of adhesives depends in part on the fact that the cementing material adheres strongly to the two surfaces and hardens there. It is therefore possible that one agglutinant may be useful for a number of different materials, such as wood, glass, metal, ivory, etc., while others give good results only with special materials. Since the books give different recipes for cements for glass, cements for metals, cements for metals and glass, etc., the differences in adsorption are real ones though no one has ever made a careful study of agglutinants from this point of view. A good glue joint in wood will easily stand a strain of 6000 Ibs. per square inch. We are quite familiar with this; but we do not always realize that the tensile strength of liquid water is pretty high if a decrease in the cross-section can be prevented. As far back as 1850 Berthelot 1 filled a glass tube nearly full of liquid, removed the air, sealed the tube, heated the tube until it was entirely full of liquid, and allowed the whole thing to cool. The liquid continued to fill the tube until the volume was larger than it should be for the temperature by 1/420 in the case of water, 1/93 in the case of alcohol, and 1/50 in the case of ether. This indicates a breaking strain of about 50 atm. for water and over 100 atm. for alcohol and ether. Worthington 2 obtained 17 atm. for alcohol. Stefan 3 and Tumlirz 4 consider that the calculated internal pressure of a liquid is a measure of the theoretical tensile strength and they obtain in this way a value for alcohol 5 of about 2000 atm. Some direct measurements of the tensile 1 Ann. Chim. Phys. (3) 30, 232 (1850). 2 Phil. Trans., 183A, 355 (1892). 3 Wied. Ann., 29, 655 (1896). Sitzungber. Akad. Wiss. Wien., 109 Ha, 837 (1900) ; 110 Ha, 437 (1901). 6 Cf. HULETT: Zeit. phys. Chem., 42, 353 (1903). ADSORPTION OF LIQUID BY SOLID 75 strength of liquid films have recently been made by Budgett, 1 who found a maximum breaking strength of about 87 pounds per square inch for steel end-gauges held together by a water film. On the assumption that the water occupied only one-tenth of the surface, this gives a tensile strength of nearly 900 pounds per square inch, or about sixty atmospheres. Rice 2 reports that certain plastic clays from Montana, when worked in the brick-making machines adhered to iron in a remarkable way whenever they came into contact with that metal. Rice does not explain why these clays should stick to iron. One of the clays was very high in silica, 76.8 percent. The formation of glazes on pottery and still more of enamels on metals are cases in which the fused or sintered mass must not only adhere to the base after cooling but must also have nearly enough the same coefficient of expansion to prevent the glass or enamel from going to pieces as the temperature changes. 3 The adsorption of water vapor by gelatine presents some pecu- liarities. Working in Ostwald's laboratory, von Schroeder 4 found that gelatine contained a thousand percent of water when immersed in liquid water and less than 400 percent when in saturated water vapor at the same temperature. From these and other experiments von Schroeder concluded that the vapor pressure of water in gelatine must be higher than that of pure water, because water distills from the gelatine to the vapor phase. The reason for this unexpected phenomenon is possibly that gelatine has a cellular structure. If a dry paper bag be placed in a space saturated with water vapor, the paper will adsorb a certain amount of water; but there will be little or no tendency for the water to condense inside the bag and fill it with liquid water. If a similar bag is immersed in liquid water, the paper will take up water as before; but water will also pass into the bag and fill it. If the bag is then lifted into the vapor phase, and if the bag is so constructed that the actual dripping is negligible, the water will distill from the curved surface in the bag to the Proc. Roy. Soc., 82A, 25 (1912). 2 Trans. Am. Ceramic Soc., 14, 610 (1912). 3 RANDAU: Enamels and Enamelling, 101 (1912). Zeit. phys. Chem., 46, 109 (1903). 76 APPLIED COLLOID CHEMISTRY plane surface in the containing vessel. The smaller the bag is, the greater will be the curvature of the water within and conse- quently the higher the vapor pressure. With the microscopic spaces in solidified gelatine, the vapor pressure of the drops of water will be enough higher to counterbalance the difference of levels in von Schroeder's experiments. While this explanation accounts for the phenomena provided gelatine really has a cellular structure, a point which is not yet generally accepted it is superfluous if the facts are not as stated. Wolff and Buchner 1 claim that the loss of water when gelatine is placed in the vapor phase is merely due to experimental error. 2 They have certainly shown the possibility of error in von Schroe- der's results; but they did no experiments involving as much error as one must assume if von Schroeder's data are worthless. The whole matter is therefore up in the air for the present. Wash- burn 3 found that moistened clay will dry pretty completely if suspended in a closed vessel above water. He attributes this to the action of gravity. This could only be true in case there were no soluble salts in the clay and in case the pores were so coarse that the capillary rise was not equal to the distance above the water level. There are other substances which seem to behave like gelatine, though no experiments like von Schroeder's have been performed. Schwalbe 4 states that cotton cellulose takes up 21 percent moisture in presence of saturated water vapor. Masson and Richards 5 find 17.7 percent adsorbed moisture when the vapor phase is 97 percent saturated. By extrapolation to 100 percent this might easily give about 21 percent moisture; but Masson and Richards incline to the view that the curve is asymptotic, in which case the value would be infinite, which seems absurd. Vignon 6 finds that one gram of cotton takes up about five grams of liquid water, while a fine-grained sponge will hold thirty grams of water per gram of sponge, and a coarse-grained sponge 1 Kon. Akad. Wet. Amsterdam, 17, May 30 (1914); Zeit. phys. Chem., 89, 271 (1915), 2 See also ARISZ: Kolloidchemische Beihefte, 7, 87 (1915). 3 Jour. Am. Ceramic Soc., 1, 25 (1918). 4 Die Chemie der Cellulose, 10 (1911). 5 Proc. Roy. Soc., 79A, 421 (1906). 6 Comptes rendus, 127, 73 (1898). ADSORPTION OF LIQUID BY SOLID 77 only about six grams. Leighton 1 states that absorbent cotton will hold four grams of water when centrifuged at 4000 R.P.M. Some of this water is held in capillary spaces; some is adsorbed on the surface of the cotton; a part is probably dissolved in the cotton; and some may be held in cells. We have no satis- factory way of differentiating between these possibilities. Dry caseine takes up about 1.25 g water per gram of caseine. 2 Infusorial 3 earth is able to take up more than double its weight of liquid. Wood flour will take up over double its weight of nitro- glycerine without running. 4 Fibrous peat, fresh from the bog, may contain 90 percent of water and Sphagnum moss may im- bibe as much as two hundred times its own weight of water, 5 which is the reason that this moss is used largely in the trenches for dressing wounds. If a freshly-broken piece of porous earthenware be pressed against the tongue, water will be sucked up into the pores of the earthenware and the piece will tend to stick to the tongue. When water is drawn up into a capillary space, air is driven out and one at once wonders how much pressure the water will exert under these circumstances. Jamin 6 has made some pre- liminary experiments along this line. A hole was bored in a piece of dried chalk. Into this hole was dipped one end of a manometer and the hole was then closed. When the chalk was placed in water, the air was displaced from the pores and a pres- sure of 3-4 atmospheres was obtained. This is not the limiting pressure because the amount of dead space in the manometer was large. A better method would be to determine the pressure necessary for the air to force the water out of the pores of the chalk. Spring 7 obtained pressures of one-fifth of an atmosphere when water displaced air in contact with sand. Jamin, Askenasy, and Wiedeburg 8 have described different 1 Jour. Phys. Chem., 20, 32 (1916). 2 VAN SLYKE and PUBLOW: Science and Practice of Cheese- Making, 179 (1913). 3 ROLLER: Utilization of Waste Products, 310 (1915). 4 KRESSMAN: Met. Chem. Eng., 14, 372 (1916). 6 Percy's Metallurgy: Fuel, 208 (1875). 'CHWOLSON: Traite" de Physique, 1 III, 622 (1907). 7 Mem. Soc. beige Geologie, 97, 13 (1903). 8 Wied. Ann., 69, 764 (1896). 78 APPLIED COLLOID CHEMISTRY forms of apparatus in which water evaporates through a porous tube or a porous cup of plaster into dry air. As the water evapor- ates off, mercury is drawn up into the tube reaching a height of 90 cm without difficulty. If bubbles of air did not form, water or mercury could be drawn up to incredible heights, limited by the tensile strength of the liquids. When considering the adsorption of water vapor by cotton, reference was made to the heat developed. Attention might perhaps be called here to Meissner's experiments on the heat developed when powders are wetted by liquids. 1 He found that there was no appreciable difference in the heat evolved when water was taken above and below 4. His quantitative results were rather discouraging, however, because he found no apparent relation between the amounts of heat evolved and the amounts of powder taken, which must be the result of experi- mental error. The more careful experiments of Parks 2 showed that when silica, sand or glass is brought in contact with water at approximately constant temperature, the development of heat at 7 is approximately 0.00105 cal. per square centimeter of surface wetted. Foote and Saxton 3 find that water adsorbed by lamp-black supercools readily to 6 and is not all frozen at 35. When once frozen, the water does not begin to melt until nearly in the case of lamp-black. Water adsorbed in gelatinous pre- cipitates may start melting at 20. Mclntosh and Edson 4 find that when a salt solution is plunged into liquid air or solid carbon dioxide, the resulting solid forms a constant-melting mixture, melting completely at the temperature at which the solution should have started freezing. It is not clear why this should be so. ADSORPTION OF SOLID BY SOLID Dried glue may be considered either as a solid or as a super- cooled liquid. If we consider it as a solid, then the adsorption of gelatine by glass is a case of adsorption of a solid by a solid. It is more satisfactory to have cases where there can be no doubt I MEISSNER: Wied. Ann., 29, 114 (1886). 2 Phil. Mag. (6) 4, 247 (1902). 3 Jour. Am. Chem. Soc., 38, 588 (1916); 39, 627 (1917). 4 Ibid., 38, 613 (1916). ADSORPTION OF LIQUID BY SOLID 79 but that both substances are solids. We get marked adsorption of rouge by metals. 1 If a metal is polished with rouge which is not kept sufficiently moist, the rouge adheres so firmly to the metal surfaces that it cannot be removed without spoiling the polish. Blue powder is another case of adsorption. Here we have grains of zinc coated with zinc oxide which adheres even when the zinc is melted. Aluminum and nickel are always coated with solid surface films which have marked effects on the properties of the metals. Passive iron owes its peculiar proper- ties to the adsorption and stabilization of a higher oxide, 2 either FeC>2 or FeOa. It seems probable that vermilion is adsorbed by copper because De la Rue 3 says that electroplated copper blocks cause vermilion to blacken while cast copper does not. We get what seems to be a case of adsorption of a solid by a solid if we shake a moderately coarse powder with a much finer powder. Instead of the finer powder filling the voids in the space occupied by the coarse powder, the former tends to coat the lat- ter. 4 If the two powders are of different colors, the color of the coarser powder may be masked completely by a relatively small amount of the fine powder. In one experiment 0.032 g of Prus- sian blue was mixed with 10 g dolomite which passed a 40 mesh sieve and did not pass a 100 mesh sieve. In the second experi- ment the same amount of Prussian blue was mixed with 10 g dolomite, all of which would pass a 200 mesh sieve. In the the first case, the powder was a deep blue and in the second it was practically white, though the percentage compositions were the same in the two cases. In another experiment 1 percent Fe-jOa colored the dolomite a full red when the rouge passed a 200 mesh sieve and the dolomite did not pass a 100 mesh sieve. When the dolomite passed a 200 mesh sieve and the iron oxide did not pass a 100 mesh sieve, the powder was practically white. In a third set of experiments 8 milligrams of lamp-black black- ened 10 g dolomite which was coarser than 40 mesh. When the dolomite was ground so that it all passed a 200 mesh sieve, 654 1 DESCH: Metallographie, 210 (1914). 2 BENNETT and BURNHAM: Jour. Phys. Chem., 21, 107 (1917). 3 Mem. Chem. Soc., 2, 305 (1845). FINK: Jour. Phys. Chem., 21, 32 (1917); BRIGGS: Ibid., 22, 216 (1918). 80 APPLIED COLLOID CHEMISTRY milligrams of lamp-black, or over eighty times as much, were needed to produce approximately the same tint. This behavior of powders is of great importance when using inert fillers in mixed paints. 1 The amount of barytes that can be mixed with colored pigments without injuring them is re- markably large. There are hundreds of brands of para-red paints made and consumed every year by the agricultural imple- ment trade which contain as high as 90 percent of natural barytes. The opacity of paper may be increased by grinding the filler more finely. Fink has shown that a given mixture of a white non-conducting powder like thoria with a black conducting powder like tungsten may either be black and conduct electricity or white and a non- conductor, depending on the relative coarseness of the two pow- ders. If one strews lycopodium powder over the surface of water and then puts one's finger into the water, the lycopodium powder sticks to the finger and keeps it from being wetted. 2 A coarse powder kept in a bottle will fall to the bottom of the bottle through the force of gravity while a very fine powder will stick to the walls of the bottle discoloring them. Writing on paper with a pencil or on a blackboard with chalk is a phenomenon of about the same type, though the matter is simplified by the roughened surface on which one writes. Solid particles are rub- bed off and stick. It seems probable that the possibility of using face powder is merely a special case of a very fine powder adhering to a curved surface. In phosphate baking powders, the monocalcium phosphate is protected from moisture by an admixture of starch and it is well-known in the trade that the calcium phosphate must not be too fine nor the starch too coarse; potato starch cannot be sub- stituted for corn starch. The retarding effect on the rate of setting of Portland cement, which is produced by the addition of gypsum is probably due to the fact that the finer gypsum powder coats the coarser cement powder and acts to some extent as a protecting film. In all these cases the powders do not come in actual contact, but are separated by a film of air which, however, holds them 1 TOCH: The Chemistry and Technology of Paints, 113, 138 (1916). 2 EHRENBERG and SCHULTZE: Zeit. Kolloidchemie, 16, 183 (1914). ADSORPTION OF LIQUID BY SOLID 81 together, thus behaving like a liquid film. Traces of moisture increase the adhesion, as in the case of sugar and blueberries. The fine particles will be held more firmly in hollows in the coarser powders than on projections and will therefore accumulate first in the concave portions, next on the plane surfaces, and last of all on the convex surfaces. Aitken 1 found that hot powders adhered to a cold rod of glass or metal; but fell off when powder and rod were of the same temperature and were repelled when the rod was hotter than the powder. This is due to a lowering of the surface tension of the condensed air film on the hotter side of the powder. It is necessary to work with such coarse particles that they fall off under the influence of gravity except when special conditions prevail. It has been shown that the time factor is an important one in the wetting of some solids by some liquids, the air film being displaced slowly. It is possible that in some cases adhering powders displace the air film slowly and become really adsorbed by the supporting solid. ADSORPTION OF LIQUID BY LIQUID The spreading of one liquid over another is usually considered to depend on the relative surface tensions. If oil is dropped on water, it is assumed that we have a water-air surface, a water- oil surface, and an oil-air surface as shown in Fig. 4. If the surface ten- /4//T* sion water-air is greater than the sum of the surface tensions oil-water and oil-air, the drop of oil spreads out over the surface of the water. If the water-air surface tension is less than the sum of the other two the oil FIG. 4. will draw up into a globule. While this may be a perfectly correct way of looking at things, it is not a useful one because Rayleigh 2 claims that three fluids cannot be in contact simultaneously. The simplest way to look at the 1 Trans. Roy. Soc. Edin., 32, 239 (1884); TAMMANN: Drude's Ann., 18, 856 (1905). 2 Scientific Papers, 3, 354, 413, 416 (1902). 82 APPLIED COLLOID CHEMISTRY matter is to consider the adsorption. If oil is adsorbed by water, it wets the water and spreads over the surface. Now the ad- sorbed layer may be very thin, not over 2/x/x in some cases. At greater thicknesses there is nothing to prevent the oil tearing loose from the oiled surface of the water and drawing up into drops under the influence of surface tension. This has been commented on by Rayleigh 1 and by Budgett. 2 The first portion of oil spreads over the water because it is ad- sorbed and held by the water. The second portion of oil flows out over the oiled surface, but is instable because the same mass of oil would have a less surface if present in drops and still more so if present in a single drop. The excess oil therefore draws up into one or more drops depending on conditions, thereby leaving bare oiled surfaces. The several drops will coalesce into one large one if brought in contact. If more oil is added the drops become larger and finally form so large a drop that under the influence of gravity, it covers the whole surface of the water and becomes a liquid layer of free oil. From this way of looking at it, we see that the phenomenon must be general and so it is. Budgett 3 found that it was quite impossible to get a thin continu- ous film of kerosene or water on the surface of steel blocks. Microscopic examination showed that the applied films were not continuous membranes but that the liquid collected into small drops.- This is undoubtedly exactly like Rayleigh's ex- periment. On the surface of the wetted steel the kerosene or water gathered in small drops. With different liquids, we should expect to find the drops first forming when the films reached different thicknesses. This has been noted by Hardy 4 who found that the thickness of the oil film varied from 2-200^^ with dif- ferent oils. With one sample of oil there was apparently no spreading at all. If an oil is not adsorbed by water, there is theoretically a film of air between the water and the drop of oil. It seems improbable that this air film can continue to exist when a thick layer of oil is present and yet, if not, there must be some point under some 1 Scientific Papers, 3, 424 (1902). 2 Proc. Roy. Soc., 85A, 30 (1911). 3 Ibid., 86A, 30 (1911). 4 Ibid., 86A, 612 (1912). ADSORPTION OF LIQUID BY SOLID 83 conditions when surfaces of oil, water, and air coexist. It seems probable that Rayleigh's conclusion, that such a state of things cannot exist, holds only when the effect of gravity is ignored. With increasing amount of oil there probably comes a time when the air film is squeezed out entirely. The breaking up of a liquid film into drops has been paralleled by the behavior of thin solid 1 films when heated. Beilby 1 showed that a very thin film of gold leaf became granular if heated moderately. ADSORPTION OF SOLID BY LIQUID A solid which is wetted by a liquid, adheres to that liquid. With two liquids and one solid we get selective adsorption of the solid and this leads to some interesting results, which have been studied carefully by Hofmann. 2 If finely divided red lead is shaken with water and benzene or chloroform, the red lead ad- heres to the surface of the benzene or chloroform, making a practically continuous coating around the organic liquid pro- vided proper relative amounts of red lead and organic liquid are taken. The theory of the phenomenon has been given byDes Coudres. 3 The solid particles tend to go into the water phase if they adsorb water to the practical exclusion of the other liquid; they tend to go into the other liquid phase if they tend to adsorb the other liquid to the practical exclusion of the water; while the particles tend to go into the dineric interface in case the adsorption of the two liquids is sufficiently intense to increase the miscibihty of the two liquids very considerably at the sur- face between solid and liquid. Of course the powder will only stay in the upper phase or the dineric interface in case it is so fine that the adsorption over- balances gravity. These conclusions as to the way a given pair of liquids will behave with a given solid refer to the relations pre- vailing under the conditions of the experiment. Hofmann's experiments with two liquid layers in glass test-tubes (soda-lime glass) showed that the time factor might be an important one. If a few cubic centimeters of an aqueous potassium bichromate 1 Proc. Roy. Soc., 72, 226 (1904). 2 Zeit. phys. Chem., 83, 385 (1913). 3 Arch. Entwicklungsmechanik, 7, 325 (1898). 84 APPLIED COLLOID CHEMISTRY solution are shaken up in a carefully cleaned glass test-tube, there remains on the walls of the test-tube a continuous uniform coat- ing of solution recognizable by its yellow color. This film gradu- ally becomes thinner and paler but does not break at any point until the water disappears completely through evaporation. If one adds a layer of xylene or kerosene to the aqueous bichro- mate solution in the test-tube, it is easy to see that there is a coating of aqueous solution between the organic liquid and the glass. If water is allowed to flow in at the bottom of the test-tube, the or- ganic liquid can be forced out at the top and there will be no sign of its having adhered to the glass wall at any point. If, however, the test-tube containing the aqueous bichromate solu- tion and the xylene or kerosene be left standing quietly for a long time, 6-10 hours, it will be found that the organic liquid has diffused through the water film to some extent and now ad- heres to the glass wall nearly everywhere, having displaced the aqueous film almost completely. If the contents of the test-tube are shaken violently, the water displaces the organic liquid from the glass surface practically completely. The glass is therefore wetted more readily and more rapidly by water than by xylene or petroleum; but a large mass of xylene or petroleum will gradu- ally displace a water film from the glass surface. One would therefore expect to get a slightly different behavior with glass powder according as one treated it first wi th water 01 with organic liquid. This was exactly what Hofmann found. If the glass powder were treated first with water and then shaken with kero- sene, practically all the glass went into the water phase. If the glass powder were allowed to stand a long time with kerosene, or if it were heated with it to hasten the removal of the air film, some of the glass powder would remain in the dineric interface. Vig- orous shaking would, of course, cause the water to displace the kerosene. Chloroform behaves towards glass and water some- what like xylene or kerosene though the tendency to displace the water layer from the glass is less than with kerosene. Small drops of chloroform do not break through the water film at all. With increasing amounts of chloroform the force of gravity causes it to break through at the bottom of the, test-tube, the surface wetted by the chloroform increasing as the column of chloroform increases in height. ADSORPTION OF LIQUID BY SOLID 85 In Table XIX are given some of Hofmann's results. The letter w means that the solid goes nearly completely into the water phase; while o denotes that it goes almost completely into the organic liquid; s means that the powder collects nearly quanti- tatively in the interface. Two letters, such as sw, mean that the powder goes partly into the interface and somewhat less into the water phase; the use of a parenthesis, s(w) or s(o\ signifies that there is a good deal more powder in the interface than in the bracketed phase. Benzene and xylene behave alike so only one is given. In the experiments tabulated, the powder was placed in water and then shaken with the organic liquid. TABLE XIX. PREFERENTIAL WETTING OF SOLIDS Water and Ether Chloro- form Butyl alco- hol Benzene Kero- sene Amyl alco- hol Paraffin oil CaSO 4 W w W W W W W SnO 2 Al(OH), SnS .() w(s) ws ws ws ws WS ws ws s(w) s s(w) s(w) 8(wf) s(w) sw- sw g SW s(w) WS BaSO 4 ws ws ws s g s(w^) s(w?) ZnS w(s) ws ws s s(w tf ) g 0(711?') ZnO ws ws s(w) s g g CaCO 3 WS i ws s s g g Mg(OH) 2 s(wl} ws s s g s(w) sw Al sw g(w) s(w) s g s(w) g BaCO 3 ws ws s s g sw g CuS ws s(w) s s g g g PbCrO 4 .. . . ws s(wf) s s g g g Cu 2 O(?) ws s s s g g g MoS 2 PbS 8(W?) ws s s s s s s s s s g Fe 3 O 4 ws s s g g BaCrO 4 ws s s s g g g Pb 3 O 4 sw s s s g g c sw s s g g g PbI 2 s s OS s g OS g HgS . s s s s g g g HeO s s s g g g g HeI 2 s g s g Agl. . s s Q g Q OS 86 APPLIED COLLOID CHEMISTRY Much the same conclusions have been worked out by Reinders 1 independently of Hofmann. Some of his data are given in Table XX. The powder was placed in the water phase and then shaken with the organic liquid. TABLE XX. PREFERENTIAL WETTING OF SOLIDS Water and Paraffin oil Amyl alcohol ecu Benzene Ether Kaolin w w(s) w(s) W w(s) CaF 2 ws WS w(s) w(s) w(s) Gypsum w ws w SW WS BaSO 4 w(s) ws ws sw- WS Magnesium ws ws ws ws ws PbO Malachite. s so s s sw s s sw sw ZnS . . s s s s sw PbS HffI 2 . so so so s s s s s s s Carbon so s s s Selenium Sulphur so so so so so o(s) s so s Similar results were obtained with colloidal solutions, Isobutyl alcohol was added to a colloidal gold solution obtained by reduc- ing gold chloride with carbon monoxide. When the two liquids are shaken, the gold forms a thin film at the interface. This film is violet blue to blue green by transmitted light and golden by reflected light. A thin water film forms between the isobutyl alcohol and the glass, and the gold concentrates in the dineric interface thus formed, making the alcohol appear uniformly gold-plated. With ether the gold film rises high above the level of the two liquids. With carbon bisulphide the adherent film of gold appears blue. When the carbon bisulphide is broken into drops by shaking, each drop appears blue. When a blue gold was obtained by reducing gold chloride with phosphorus dissolved in ether, the gold went into the dineric interface. When a brown- ish red gold was obtained in this way, it remained in the water phase and showed no tendency to pass into the interface. This difference is undoubtedly due to an adsorption of something at the 1 Zeit. Kolloidchemie, 13, 235 (1913). ADSORPTION OF LIQUID BY SOLID 87 surface of the gold because Reinders found that 0.005 percent gum arable prevents colloidal gold from passing into the ether- water interface. With carbon tetrachloride, carbon bisulphide, and benzene, the gold goes to the interface as before; but the gum arabic prevents its changing from red to blue. With colloidal ferric oxide practically no iron went into the interface with any of the organic liquids. It seems probable that this is due to Reinders using a ferric oxide solution which was a year and a half old, because hydrous ferric oxide is ordinarily a pretty good emulsifying agent. 1 Colloidal arsenic sulphide goes into the dineric interface with amyl alcohol or butyl alcohol ; but stays in the water phase when carbon tetrachloride, benzene, or ether is the second liquid. India ink goes completely into the interface with isobutyl alcohol and stays entirely in the water phase when ether is the second liquid. When crystallized copper eosinate is placed in water and shaken with ether, it goes entirely into the surface between the two layers, 2 giving a film which resembles the purple skin of a grape, by reason of the steel-blue color of the crystals. An interesting experiment is to shake copper powder or alumi- num powder with kerosene and water. The metallic powder goes into the kerosene and into the interface, producing an effect of molten copper or molten aluminum as the case may be. When the bottle is allowed to stand after having been shaken, the metallic powder in the interface creeps up the side of the bottle above the surface of the liquid, rising higher if a little alcohol has been added. I have seen an apparently coherent metallic film rise two or three inches above the surface of the upper liquid phase. If too much copper or aluminum be added the kerosene cannot hold it all up and a portion falls to the bottom of the flask carrying drops of kerosene with it. If the mixture be poured out on a piece of wood, the copper spreads over the surface of the wood just as it did over the surface of the glass. This experi- ment illustrates the principle involved in all bronzing liquids. A bronzing liquid consists of a volatile liquid which will hold up the metal, and some substance which will keep the metallic powder from rubbing off too readily after it has been applied. 1 BRIGGS: Jour. Phys. Chem., 19, 296 (1915). 2 GILBERT: Jour. Phys. Chem., 18, 602 (1914). 88 APPLIED COLLOID CHEMISTRY The second substance also serves the purpose of keeping the metallic powder in a better state of suspension in the volatile liquid. According to Worden 1 the best of these bronzing liquids consist essentially of pyroxylin dissolved in amyl acetate, to which the metallic powder is added. For bronzing radiators and steam pipes, a pure pyroxylin would not do since it would decompose, disintegrate, and allow the metallic coating to peel off. To prevent this, boiled linseed oil and rosin may be added. Such preparations are of course much slower in drying than pyroxy- lin bronzing liquids; but that is not a serious objection for these special purposes. A much cheaper form of bronzing liquid consists of rosin dissolved in benzene, while the cheapest of all is apparently a solution of sodium silicate in water. The aluminum and copper powders on the market are coated with stearin; but special experiments in my laboratory have shown that the behavior of copper and aluminum with kerosene is the same qualitatively whether the stearin coating is removed with ether or not. Barus 2 and Rayleigh 3 have each noticed that when fine powders (bole) or dust are shaken with ether and water the solid particles go into the water layer, leaving the ether layer astonishingly clear. In fact Rayleigh said that the ether layer was more free of motes than he had ever been able to obtain by repeated distillation of liquid in vacuum. White lead passes from water into linseed oil. 4 " During recent years the practice has been adopted largely among white lead corroders who grind their own white lead in oil of doing away with the final drying of the white lead pulp as it comes from the washing process, and grind- ing or beating up the pulp (exhausted of water until the propor- tion of the latter does not exceed about 20 percent) with a suitable quantity of refined linseed oil. This process depends on the greater surface attraction which white lead particles offer to linseed oil than to water. It enables considerable economies to be effected in the manufacture of 'ground white lead' and it eliminates risk of lead poisoning during one of the most dangerous 1 Nitrocellulose Industry, 1, 310-320 (1911). 2 Am. Jour. Sci. (3) 37, 124 (1889). 3 Scientific Papers, 3, 569 (1902). 4 J. CRUICKSHANK SMITH: The Manufacture of Paint, 92 (1915). ADSORPTION OF LIQUID BY SOLID 89 parts of the white lead manufacturing process." Zinc oxide stays in the water phase. The Elmore bulk-oil process of ore flotation depended on the fact that some sulphide ores can be shaken out of the water phase by means of suitable oils while the gangue stays in the water phase. When ground bituminous coal is stirred up with fuel oil and water, there is a good separation of coal from ash, the former going into the oil and .the latter into the water. The readiness with which solid films form over the surface of mercury is well known to everybody. Salts of mercury or of the baser metals coat mercury with great readiness, especially in the air. CHAPTER III ADSORPTION FROM SOLUTION ADSORPTION FROM SOLUTION BY SOLID Selective adsorption by a solid from a solution may show itself by relatively more of the dissolved substance being adsorbed than of the solvent, or by relatively less being adsorbed. In the first case, commonly called positive adsorption, the solution becomes more dilute and in the second case, commonly called negative adsorption, the solution becomes more concentrated. Both cases are known; but the first one is the more important and will be considered now. The decolorizing action of char- coal 1 has been known since 1791. In 1874 Bottger 2 pointed out that when an alcoholic solution of any of the aniline dyes is shaken up with a sufficient quantity of infusorial earth, some water added, and the whole thrown on a paper filter, the liquid which runs through is perfectly colorless, the pigment being retained by the siliceous earth. The adsorption of dyes by hydrous alumina, stannic oxide, and other mordants as they are called, is of great importance in dyeing. Here as in all other cases the adsorption is selective. Da vison 3 showed thatwhen alumina is precipitated from an aluminum acetate solution, it will decolorize dilute solutions of alizarine, alizarine green, benzo orange, and acid violet: it adsorbs fast green, acid green, and croceine orange considerably, and has practically no adsorbing action on emerald green or chrysoidine. Red phosphorus will decolorize a dilute solution of iodine in carbon bisulphide. 4 Hatschek 5 points out that iodides cannot be shown to be present in sea- water by any of the usual tests, and that we owe our supply of iodine to the adsorption of iodides by certain sea-weeds. 1 Cf. OSTWALD: Lehrbuch allgem. Chemie, 2nd. Ed., 1, 1093 (1891). 2 Jour. Chem. Soc., 28 170 (1875). 3 Jour. Phys. Chem., 17, 737 (1913). 4 SESTINI: Gazz. chim. ital., 1, 323; Jour. Chem. Soc., 24, 1006 (1871). 6 An Introduction to the Physics and Chemistry of Colloids, 6 (1913). 90 ADSORPTION FROM SOLUTION 91 When making density determinations 1 by weighing a solid in a solution, an error may be introduced because of the solid adsorb- ing some of the salt. This error is likely to be larger the finer the particles of the solid, because the ratio of surface to mass increases with increasing subdivision. With grains of quartz or glass varying in diameter from 0.015 mm to 0.9 mm, placed in solutions of iodides, the specific gravity of the latter can be adjusted so that the smaller particles will sink while the larger will float. Marble is so porous that its surface is almost propor- tional to its volume, and consequently grains of marble do not show this phenomenon. According to Skey 2 antimony and arsenic can be removed so completely from solutions of their oxides or chlorides in moder- ately strong hydrochloric acid (with a little tartaric acid in the case of antimony) by charcoal that neither of them can be de- tected therein by Reinsch's test, although before the treatment they could be detected without difficulty. Commercial sul- phuric and hydrochloric acids diluted with a little water can be purified from either of these substances by agitating them inter- mittently for a short time with fresh charcoal, and then filtering off; application of heat to the mixture expedites this result. Quite recently Lockemann and Paucke 3 have studied quanti- tatively the adsorption of arsenic acid by charcoal, hydrous aluminum oxide, hydrous ferric oxide, and albumin. They find that egg albumin and blood serum, when coagulating, carry down certain amounts of arsenic from aqueous solutions. Animal char- coal also adsorbs some amounts of arsenic from aqueous solutions. When hydrous aluminum oxide is precipitated by ammonia, it carries down very appreciable quantities of arsenic. Curiously enough more arsenic is carried down when the hydrous aluminum oxide is precipitated from hot solution than when precipitated from a cold one. This unexpected fact is probably due to the form in which the alumina is precipitated. At the temperature of the water-bath, arsenic in amounts up to 20 mg As per 100 cc can be precipitated completely but this requires the use of very 1 THOULET: Comptes rendus, 99, 1072 (1884); Jour. Chem. Soc., 48, 476 (1885). 2 Chem. News, 17, 157 (1868); 36, 6 (1877). 8 Zeit. Kolloidchemie, 8, 273 (1911). 92 APPLIED COLLOID CHEMISTRY Jarge amounts of alumina. When precipitated by ammonia, hy- drous iron oxide carries down large amounts of arsenic, more being carried down the lower the temperature and the smaller the excess of ammonia. When the concentration of arsenic does not exceed 10 mg As per 100 cc it can easily be precipitated quantitatively by means of iron. In 1845 Yorke 1 discussed the adsorption of lead oxide by filter paper, while Mil Her 2 and Weiske 3 have shown that filter paper takes up such appreciable quantities of barium, strontium and calcium hydroxides that dilute solutions of these salts should not be filtered if they are to be used for quantitative work. A special phenomenon in connection with adsorption by filter- paper has been studied by Bay ley, 4 Lloyd, 5 Gordon, 6 and others. 7 If a drop of a salt solution be allowed to fall on filter-paper the solution will tend to spread. If the salt is adsorbed very strongly it will not spread and we get a relatively wide water-ring. If the salt is not much adsorbed, it will spread as far as the water. If a strip of filter-paper be dipped into a solution we get a similar phenomenon; the water will rise much farther than the salt if the latter is adsorbed strongly. Gordon found that with a M/40 CuS(>4 solution the water rose about 9.2 cm in half an hour and the copper salt only 5.5 cm. With a M/2 solution the copper salt rose 9.2 cm and so did the water. Ordinary ink shows the phenomenon well with blotting-paper. Leighton 8 has determined the adsorption of caustic soda, hy- drochloric acid, sulphuric acid, and phosphoric acid by purified absorbent cotton. These experiments were interesting in two ways. In the first place, they showed that no compounds were formed. In the second place, data were obtained showing the possible error when determining adsorption. The usual method is to shake a known solution with a weighed amount of the adsorbed solid, pipette off some of the supernatant liquid, and 1 Mem. Chem. Soc., 2, 399 (1845). 2 Jour, prakt. Chem., 83, 384 (1861). 3 Jour. Chem. Soc., 30, 662 (1876). * Ibid., 33, 304 (1878). 6 Chem. News, 61, 51 (1885). 6 Jour. Phys. Chem., 18, 337 (1914). 7 See SAHLBLOM: Kolloidchemische Beihefte, 2, 79 (1910). 8 Jour. Phys. Chem., 20, 188 (1916). ADSORPTION FROM SOLUTION 93 analyze. The amount of adsorption is calculated from the de- crease in concentration. This calculation is accurate only in case no liquid is taken up by the solid adsorbing agent. If the cotton adsorbed the alkali and the water in the same ratio in which they occurred in the solution there would be no change in the concentration of the solution and the apparent adsorption would be zero. To avoid this error, Leighton centrifuged the cot- ton for an hour in a centrifugal rotating about 4000 revolutions per minute, and then analyzed the cotton. This direct or "gravi- 0.20 o.40 o.eo ago GRAMS No OH ADSORBED /=>? G&AM COTTON FIG. 5. metric" method will give too high adsorption unless the centri- fuging removes all the unadsorbed solution. On the other hand, the indirect or "titration" method gives too low results unless no water is adsorbed by the cotton at all. Since this last is not true, the results by the titration method are necessarily wrong with cotton, while the results by the gravimetric method may be right. The differences between the two methods may be quite considerable. With a solution of 200 grams NaOH per liter Leighton found an adsorption of 0.27 grams NaOH per gram cotton by the titration method, and of 0.75 gram NaOH by 94 APPLIED COLLOID CHEMISTRY the gravimetric method. The true adsorption lies between these two values. 1 In Tables XXI-XXIV are given Leighton's data which are also reproduced graphically in Figs. 5-7. 700 i 1 1 ff 1 1 700 600 5 10 15 20 25 AWS HZ SO* ADSORBED RER JQQ GRAMS COTTON FIG. 6. $00 10 15 20 GRAMS HyF*O+ ADSORBED JOO GRAMS COTTOW FIG. 7. Kellner 2 has studied the adsorption of acids and alkalies by platinum black. One gram of platinum black adsorbed about 20 percent of the hydrochloric acid in 50 cc of a thousandth- iSee WEISER and SHERRICK: Jour. Phys. Chem., 23, 205 (1919). 2 Wied. Ann., 67, 79 (1895). ADSORPTION FROM SOLUTION 95 TABLE XXI. ADSORPTION OF SODIUM HYDROXIDE BY COTTON Volume of solution = 100 cc. Weight of cotton = about 1.0 g. Time of run = 3 hours. Solution grams NaOH per liter Grams NaOH adsorbed per gram cotton Solution grams NaOH adsorbed Grams NaOH adsorbed per gram cotton Gravimetric Titration Gravimetric Titration 474 0.916 230 0.664 450 0.28 212 0.662 427 0.887 210 0.19 415 0.875 190 0.594 400 0.27 189 0.579 379 0.846 184 0.572 350 0.27 175 0.12 323 0.782 140 0.08 300 0.27 138 0.461 276 0.733 125 0.433 275 0.25 90 0.01 231 0.665 40 0.00 TABLE XXII. ADSORPTION OF SULPHURIC ACID BY COTTON Volume of solution = 100 cc. Weight of cotton about 1.0 g. Time of run = 3 hours. Adsorption Solution g H2SO4 per liter Gravimetric: g HzSO4 per 100 g cotton Titration: g HzSO4 per 100 g cotton 747 7 2.04 521.7 459 2 23.70 1.53 12 413.9 20.99 0.0 352.3 15.36 0.0 302.3 12.55 0.0 252.3 10.00 0.0 202.3 7.35 0.0 151.7 5.77 0.0 101.2 50 4.65 0.0 0.0 96 APPLIED COLLOID CHEMISTRY TABLE XXIII. ADSORPTION OP PHOSPHORIC ACID BY COTTON Volume of solution = 100 cc. Weight of cotton about 1.0 g. Time of run = 3 hours. Adsorption Solution g HsPCh per liter Gravimetric: g HsPO4 per 100 g cotton Titration: g HsPOi per 100 g cotton 715.2 18.68 514.4 18.43 382.9 11.53 318.8 8.57 158.0 4.53 TABLE XXIV. ADSORPTION OF HYDROCHLORIC ACID BY COTTON Volume of solution = 100 cc. Weight of cotton about 1.0 g. Time of run = 3 hours. Adsorption Solution g HC1 per liter Gravimetric: g HC1 per 100 g cotton Titration: g HC1 per 100 g cotton 414.0 13. 97 1 373.6 9.1 11. 63 1 331.3 7.3 9.28 1 288.9 6.9 7.00 1 209.2 6.2 2.34 114.5 3.1 45.8 2.0 normal solution (about 0.36 mg HC1). The adsorption of caus- tic potash is greater. The bivalent bases and acids are adsorbed in general more than the univalent ones. Osaka 2 has made some experiments on the adsorption of potassium and sodium salts by blood charcoal. The charcoal takes up about 85 percent of its weight of water when suspended in saturated water vapor at 25. If no account is taken of this fact, a very large error 1 Probably wrong owing to volatilization of HC1. 2 Mem. Coll. Sci. Kyoto Univ., 1, 257 (1915). ADSORPTION FROM SOLUTION 97 may be introduced. When allowance was made for the amount of water adsorbed, it was found that the adsorption with salts of the same base decreased in the order I > NO 3 > Br > Cl > SO 4 . The results are shown graphically in Fig. 8, the abscissas being milligram equivalents of salt adsorbed per gram charcoal and the ordinates the concentration of the solution in gram equivalents per liter. It will be noticed that potassium salts are adsorbed more strongly than the corresponding sodium salts. The numerical values are given in Table XXV. The percentage adsorptions for normal solutions of KI, KNO 3 , KBr, KC1, and K 2 S0 4 are approximately 28, 16, 14, 7, and 5 respectively. Rona and Michaelis 1 find that 1.1 1.5 1.5 Of SALT AOSOf?BD 1.7 Fio. 8. the order of adsorption of the anions is OH>CNS>1>NO 3 > Br > Cl > HPO 4 > SO 4 and of the cations is H > Al > Cu > Zn > Mg, Ca>NH 4 , K, Na. With powdered quartz having an estimated surface of 1000cm 2 per gram, L. J. Briggs 2 found that 100 grams quartz adsorbed approximately the following amounts for M/10 solutions: Na- OH, KOH and NH 4 OH, 0.8 millimols; Na 2 CO 3 , K 2 CO 3 , (NH 4 )j- C0 3 , 0.2 millimols; NH 4 C1, MgCl 2 , NaCl, NaNO 3 , Na 2 S0 4 , nothing. An interesting and important case of selective adsorption has been worked out by Lloyd. From fuller's earth he has prepared a hydrous aluminum silicate 3 which adsorbs alkaloids. It has been suggested that the material be called Lloyd's reagent. 1 Biochem. Zeit., 94, 240 (1919). 2 Jour. Phys. Chem., 9, 623 (1905). 3 WALDBOTT: Jour. Am. Chem. Soc., 35, 837 (1913). 7 98 APPLIED COLLOID CHEMISTRY TABLE XXV. ADSORPTION OF SALTS BY BLOOD CHARCOAL Milligrams and milligram equivalents per gram charcoal Cone, of solution in g equiv./ liter Salt adsorbed by one gram charcoal Cone, of solution in g equiv./ liter Salt adsorbed by one gram charcoal mg mg equiv. mg mg equiv. Potassium sulphate Sodium sulphate 1.010 51.4 0.59 1.068 40.5 0.57 0.830 45.1 0.52 0.933 37.4 0.53 0.673 37.0 0.42 0.753 31.9 0.45 0.509 29.3 0.34 0.583 25.5 0.36 Potassium chloride Sodium chloride 0.966 68.6 0.92 0.998 47.4 0.81 0.765 50.8 0.68 0.896 44.9 0.77 562 41.8 0.56 0.697 35.3 0.60 0.470 33.2 0.45 0.597 29.0 0.49 Potassium nitrate Sodium nitrate 0.977 151.2 1.50 0.978 116.3 1.37 0.817 129.9 1.28 0.816 107.6 1.27 0.654 112.7 1.11 0.647 99.4 1.03 0.491 87.7 0.87 0.484 69.6 0.82 Potassium iodide Potassium bromide 0.967 277.9 1.67 0.970 139.2 1.17 0.833 250.8 1.51 0.717 98.1 0.82 0.622 207.4 1.25 0.526 75.7 0.64 0.464 167.8 1.01 Carey Lea 1 has shown that silver iodide adsorbs iodine strongly This property of silver iodide was verified by shaking portions of freshly precipitated and still moist Agl with iodine solutions. An alcoholic solution of iodine, diluted until it has a pale sherry wine color is quickly decolorized by Agl, and the same thing happens with a very dilute solution of iodine in KE, which in a few minutes becomes as colorless as water. It is probably this adsorption which makes silver iodide photographic emulsions apparently less sensitive than the silver bromide emulsions. If so, it should not be difficult to overcome this trouble. Reinders 2 has found that when silver chloride crystallizes 1 Am. Jour. Sci. (3), 33, 492 (1887). 2 Zeit. phys. Chem., 77, 696 (1911). ADSORPTION FROM SOLUTION 99 from an ammoniacal solution to which gelatine has been added, the crystals contain adsorbed gelatine. This fact is of great importance for the theory of the photographic emulsion. Much other work on adsorption by crystals is to be found in the papers of Marc. 1 The selective nature of the adsorption is shown by the fact that barium sulphate carries down much larger amounts of barium nitrate than of barium chloride. 2 In his presidential address before the mathematical and phys- ical section of the British Association, Trouton 3 says: "It is a very well known experiment in adsorption to run a solution such as that of permanganate of potash through a filter of sand, or, better, one of precipitated silica, so as to provide a very large surface. The first of the solution to come through the filter has practically lost all its salt owing to have been adsorbed by the surface of the sand. "I was interested in finding a few months ago that Defoe, the author of Robinson Crusoe, in one of his other books, depicts a party of African travellers as being saved from thirst in a place where the water was charged with alkali by filtering the water through bags of sand. Whether this is a practical thing or not is doubtful, or even if it has ever been tried; for it is only the first part of the liquid to come through the filter which is purified, and very soon the surface has taken up all the salt it can adsorb and after that, of course, the solution comes through intact. It is interesting, however, to know that so long ago as Defoe's time the phenomenon of adsorption from salt solutions had been observed. It is not so well known, that in the case of some salts under the circumstances mentioned above, the first of the solution to come through the sand filter is stronger instead of weaker. This, as already mentioned, is because water, or at least a weaker solution forms the adsorption layer." On or just back of the beaches at Galveston, Nantucket, and elsewhere, there are wells of fresh water, the levels of which correspond approximately to the ocean level at high tide. It is often believed that the water in these wells comes from the sea and that the salts have been removed in some mysterious 1 Zeit. phys. Chem., 61, 385 (1908); 67, 470; 68, 104 (1909); 73, 685 (1910)- 2 WEISER and SHERRICK: Jour. Phys. Chem., 23, 205 (1919). 3 Brit. Ass. Reports, 84, 287 (1914). 100 APPLIED COLLOID CHEMISTRY way. 1 Unfortunately this is not true. The water comes from inland but is backed up by the ocean and does not flow off so rapidly as it otherwise would. If for any reason the outward flow is checked for a sufficient time, the salt from the ocean does work back into the wells. This has been noticed at Galveston when excessive amounts of water are drawn from the wells. THE ADSORPTION ISOTHERM In all cases which have been studied quantitatively the form of the adsorption isotherm is similar to that obtained for the adsorption of gases by solids. Consequently an equation of the same type will represent both sets of phenomena. For solutions we write the equation (x/m) n = kc where x is the amount adsorbed by m units of the solid adsorbing agent, c is the concentration of the solution, and n is not neces- sarily an integer though experimentally never less than unity. The approximate accuracy of the formula is shown in Freund- lich's data, 2 Tables XXVI XXVIII. Other adsorption formu- las have been proposed or discussed by Freundlich, 3 McBain, 4 Schmidt, 5 von Georgievics, 6 and others; 7 but no formula is strictly accurate and the simplest one seems the best for the time being. If the curve really runs parallel eventually to the axis of ordinates, the simple formula is necessarily wrong. That this really does happen seems to be shown very conclu- sively in some experiments by Schmidt 8 on the adsorption of acetic acid by charcoal, Table XXIX. The data are plotted in 1 HILGARD: Soils, 267 (1906). 2 Kapillarchemie, 147 (1909); See also VON GEORGIEVICS: Monatshef t f ur Chemie, 34, 733 (1913). 3 Zeit. phys. Chem., 67, 385; 69, 284 (1907). < Jour. Chem. Soc., 91, 1683 (1910). 5 Zeit. phys. Chem., 74, 689 (1910); 77, 641 (1911); 78, 667 (1912); 91, 103 (1916). 6 Ibid., 83, 269 (1914). 7 MARC: Zeit. phys. Chem., 81, 641 (1913); DIETL: Monatsheft fur Chemie, 36, 784 (1914); TRUMPLER: Zeit. Kolloidchemie, 15, 10 (1914). 8 Zeit. phys. Chem., 77, 650 (1911). ADSORPTION FROM 101 TABLE XXVI. DISTRIBUTION OF ACETIC BLOOD CHARCOAL (x/m) n = kc. n = 2.35; log k = 0.98. c in mols per liter; x/m in millimols per gram charcoal. P IjijTVEftN, J Vf r 'ATER''*AND '' * - J> " c x/m found x/m calc. logk 0.0181 0.467 0.474 0.966 0.0309 0.624 0.623 1.029 0.0616 0.801 0.798 0.984 0.1259 1.11 1.08 1.006 0.2677 1.55 1.49 1.030 0.4711 2.04 1.90 1.055 0.8817 2.48 2.48 0.981 2.785 3.76 4.04 0.906 TABLE XXVII. DISTRIBUTION OF BROMINE BETWEEN WATER AND BLOOD CHARCOAL (x/m) n = kc. n = 2.4; logk = 0.974. c in millimols per liter, x/m in millimols per gram charcoal. c x/m found x/m calc. logk 0.92 2.07 2.09 0.965 2.59 3.10 2.96 1.032 6.69 4.27 4.10 1.028 17.08 5.44 5.64 0.930 29.75 6.80 6.80 0.974 TABLE XXVIII. DISTRIBUTION OF BENZOIC ACID BETWEEN WATER AND BLOOD CHARCOAL (x/m) n = kc. n = 2.53; logk = 2.32. c in millimols per liter, x/m in millimols per gram blood charcoal. c x/m found x/m calc. logk 6.18 0.437 0.445 2.299 25.00 0.780 0.776 2.329 53.13 1.04 1.04 2.318 117.73 1.44 1.43 2.330 102 APPI}I&V COLLOID CHEMISTRY ADSORPTION OP ACETIC ACID BY CHARCOAL 100 cc acetic acid solution and 10 grams charcoal. Ci = grams acetic acid per 100 cc. c 2 = grams acetic acid adsorbed. Sugar charcoal Animal charcoal Cl C2 Cl C2 0.009 0.052 0.04 0.20 0.032 0.101 0.19 0.49 0.037 0.126 1.20 1.19 0.212 0.322 1.88 1.42 1.16 0.588 3.18 1.68 3.76 0.795 5.09 1.96 3.75 0.811 13.63 2.37 5.60 0.828 14.39 2.47 9.18 0.901 21.52 2.56 12.65 0.953 32.06 2.44 16.60 0.904 38.89 2.50 25.73 0.882 29.38 0.902 30.60 0.903 Fig. 9. In both these cases people who only looked at the last five or six figures might easily decide that a definite, compound was formed though it would have to be a different one in each case. If one assumes that the animal charcoal is only 20 per- cent carbon, the hypothetical compound would be C 2 o(CH 3 CO2H). By assigning a higher molecular weight to carbon, the formula can be made anything one pleases. Schmidt 1 does not always give the units which he uses and his manipulation of decimal points is erratic; but he finds that the equation (x/m) 4 = kc represents pretty well the distribution of iodine between charcoal and alcohol or benzene over the range covered by the experiments. For -charcoal, water, and acetic acid or succinic acid, the equation takes the form (x/m) 2 = kc 1 Zeit. phys. Chem., 15, 56 (1894). ADSORPTION FROM SOLUTION 103 while the exponent becomes 10 in the case of oxalic acid. For the distribution of potassium chloride between silicic acid and water the exponent is unity, the equation becoming (x/m) = kc The same equation describes the distribution of potassium chlo- ride between stannic oxide and water 1 and between charcoal 40 16 k < SI SfSGArt, > AM/&7AL CHARCOAL CHAffCOAL y P KNO 3 >KBr>KCl>K 2 SO 4 , from which it follows, that, for equivalent concentrations, the sulphate ion is adsorbed the least of all. If we were to compare molecular concentrations the sulphate ion would be adsorbed as strongly as the bromide ion, but less than the nitrate or the iodide ion. Davis 4 found that the order of adsorption of iodine from different liquids was not the same with different kinds of charcoal. With animal char- coal there was decreasing adsorption in the order: chloroform, alcohol, ethyl acetate, benzene, and toluene; with sugar charcoal the adsorption decreased in the order: chloroform, toluene, ethyl acetate, benzene, and alcohol; whereas for cocoanut char- coal the order was toluene, chloroform, benzene, alcohbl, and ethyl acetate. There are at least two factors governing the effect of 1 SCHULZE: Jour, prakt. Chem. (2) 25, 43 (1882); 27, 320 (1884). 2 FREUNDLICH : Kapillarchemie, 354 (1909). . 3 OSAKA: Mem. Coll. Sci., Kyoto, 1, 267 (1915). 4 Jour. Chem. Soc., 91, 1682 (1907). ADSORPTION FROM SOLUTION 113 the solvent. The more soluble the dissolved substance is in a given solvent, the less readily will it be adsorbed, provided we can neglect the adsorption of the solvent itself by the solid. There are many illustrations of this, but one will suffice. Char- coal will decolorize aqueous solutions of iodine or of methyl violet, but alcohol will extract the color from the charcoal. The solu- bility cannot be the sole factor, however, because then the sol- vents could always be arranged in the same order for the same solute, regardless of the nature of the adsorbing agent. This is disproved absolutely by the experiments of Davis. One other factor is the adsorption of the solvent by the adsorbing agent. This factor was not taken into account at all by Davis, whose data are therefore not sufficient to enable us to tell whether there are other factors to be considered. In so far as adsorption is accompanied by an evolution of heat, the amount of adsorption must decrease with rising temperature. This is found to be the case experimentally, but the change is often a very small one. In some cases there is an apparently large increase in adsorption with rising temperature. Thus wool takes up very little acid violet at 20 and a great deal at about 95. Experiments showed 1 that this adsorption is not re- versible, for the acid violet adsorbed at the higher temperature cannot be washed out to any extent at 20. After adsorption the dye agglomerates or changes so that it becomes practically insoluble and consequently the wool takes up more dye. ADSORPTION OF SEVERAL SOLUTES Some years ago there was a widespread rumor that strychnine had been added to a certain breakfast cereal to make it more appetizing. The report was false and not even new, for in 1852 it was alleged that strychnine was being added to certain English pale ales. In order to show that strychnine could be detected in beer if present, Graham and Hofmann 2 shook two ounces of animal charcoal with half a gallon of beer to which 1/2 grain of strychnine had been added. The strychnine was removed practically completely by the charcoal, and was extracted from that with alcohol and identified. Jour. Phys. Chem., 20, 761 (1916). 2 Jour. Chem. Soc., 6, 713 (1853). 8 114 APPLIED COLLOID CHEMISTRY Skey 1 reports that dilute sulphuric acid can be freed from traces of nitric acid by shaking with charcoal, while concentrated sul- phuric acid cannot be. Dudley 2 showed that the rank and dis- agreeable odor of raw whiskey can be removed by leaching through charcoal. Schmidt 3 has studied the simultaneous adsorption of iodine and acetic acid by charcoal from solution in water and in ethyl acetate. With both solvents less of each substance was adsorbed than if the other had not been present. Schmidt believes that this is general, but this statement is un- doubtedly too broad, though it holds in many cases. Preund- lich and Masius 4 studied the adsorption of pairs of organic acids and obtained results similar to those of Schmidt. They also found that the acid, which is adsorbed more, is displaced less when the two acids are present in the solution. These experi- ments throw light on Skey's experiments. With increasing relative concentration of sulphuric acid, we should expect an in- creasing displacement of nitric acid from the charcoal, which is what actually occurred. Lachs and Michaelis 5 found that caus- tic potash cut down very much the adsorption of potassium chlo- ride by charcoal from water. In other words, the presence of the strongly adsorbed hydroxyl ion decreased the adsorption of the chloride very much. They also found that if sulphuric acid is added in small amounts to the chloride solution the adsorption of chloride ion is increased markedly. This is in accord with the results of Osaka, 6 that potassium chloride is adsorbed more than potassium sulphate, or sodium sulphate, and, presumably, hydrochloric acid more than sul- phuric acid. Thorium salts cut down the adsorption of uranium X by char- coal 7 and acetone and acetic acid decrease the adsorption of grape-sugar by charcoal 8 but albumin and acetone have no effect 1 Chem. News, 17, 217 (1886). 2 Jour. Am. Chem. Soc., 30, 1784 (1908). 3 Zeit. phys. Chem., 74, 730 (1910). 4 Van Bemmelen Gedenkboek, 88 (1910). 6 Zeit. Elektrochemie, 17, 1 (1911). 6 Mem. Coll. Sci. Kyoto Univ., 1, 267 (1915). 7 FREUNDLICH and KAEMPFER: Zeit. phys. Chem., 90, 681 (1915). 8 RON A and MICHAELIS: Biochem. Zeit., 16, 499 (1909). ADSORPTION FROM SOLUTION 115 on each other, nor does one acid dye apparently displace another on wool or silk. 1 If we have two salts with no common ion, it is easy to see that the most strongly adsorbed cation and anion will be taken up the most, while the other pair will be adsorbed the least readily. In the case of dyeing we have an admirable illustration of the fact, first recognized by Lachs and Michaelis, 2 and by Estrup, 3 that an anion is adsorbed more readily in presence of a readily adsorbed cation and a cation in presence of a readily adsorbed anion. In an acid dye the color is in the acid radical and in a basic dye the color is in the basic radical. An acid dye will therefore be taken up more strongly in an acid solution than in a neutral solution and will be taken up least in an alkaline solu- tion. A readily adsorbed anion will decrease the amount of dye taken up and a readily adsorbed cation will increase it. With a basic dye the reverse will be true. The dye will be taken up most readily in an alkaline solution but may be taken up in a natural or acid solution. A readily adsorbed cation will cut down the adsorption of the dye and a readily adsorbed anion will increase it. In Tables XXXI-XXXII are data by Pelet-Jolivet 4 for crystal ponceau, an acid dye. TABLE XXXI. ADSORPTION OF CRYSTAL PONCEAU BY WOOL Volume of solution = 200 cc. Weight of wool = 5 grams. x = percent crystal ponceau. Milligrams crystal ponceau adsorbed 3 cc N/10 H 2 S0 4 Water 0.12 g NazSOi 0.1 65 41.7 36.2 0.2 113 62.8 59.8 0.3 150 76.5 64.8 0.4 178 86.5 70.8 *LAKE: Jour. Phys. Chem., 20, 751 (1916). 2 Zeit. Elektrochemie, 17, 1 (1911). 3 Zeit. Kolloidchemie, 11, 8 (1912). 4 Die Theorie des Farbeprozesses, 94, 149 (1910). 116 APPLIED COLLOID CHEMISTRY TABLE XXXII. ADSORPTION OF CRYSTAL PONCEAU BY WOOL Volume of solution = 200 cc. Weight of wool = 5 grams. Crystal ponceau = 0.372 mg. Milligrams crystal ponceau adsorbed cc N/10 acid added HC1 H 2 SO4 H 3 P0 4 44 42 42 8 109 93 48 18 158 140 80 28 171 153 102 In acid solution crystal ponceau is adsorbed more strongly than in a water solution, while the presence of the strongly ad- sorbed anion of sodium sulphate cuts down the adsorption over that in water. In the acid solution the strongly adsorbed phos- phate anion works against the strongly adsorbed hydrogen cation and consequently the dye is least adsorbed in a phosphoric acid solution and most adsorbed in a hydrochloric acid solution. In Tables XXXIII and XXXIV are data by Pelet-Jolivet 1 TABLE XXXIII. ADSORPTION OF METHYLENE BLUE BY SILK AND COTTON A = 0.3 gram silk in 50 cc 0.2 percent methylene blue. B = 3 grams cotton in 100 cc 0.1 percent methylene blue. Fiber Methylene blue adsorbed, mg 2 cc N/10 HC1 Neutral solution 2 cc N/10 KOH Silk (A) 1.3 1.2 10.8 11.2 40.6 30.0 Cotton (B) for methylene blue, a basic dye. The most dye is taken up in the alkaline solution and the least in the acid solution which is just as it should be. For the same concentration of acid, more dye is taken up in a phosphate solution than in a sulphate solution while still less is adsorbed from a hydrochloric acid solution; and we should expect the phosphate radical to be adsorbed the iDieTheorie des Farbeprozesses, 95 (1910). ADSORPTION FROM SOLUTION 117 most and the chlorine radical the least. This comparison is not strictly accurate because equivalent concentrations of two acids do not necessarily mean equal degrees of acidity. On the other hand the amount of hydrogen as ion in 20 cc N/10 H 2 SO 4 TABLE XXXIV. ADSORPTION OF METHYLENE BLUE BY WOOL Volume of solution = 200 cc. Weight of wool = 3 grams. Concentration of methylene blue =0.8 percent. Methylene blue adsorbed, mg N/10 HC1 N/10 H 2 S0 4 N/10 HjP0 4 156 156 152 10 22 37 79 20 20 27 32 30 17 is unquestionably greater than in 10 cc N/10HC1 after both have been diluted to 200 cc, and, yet, more methylene blue is taken up in this sulphuric acid solution than in the hydrochloric acid solution. The difficulty in regard to the acidity can be got round by adding salts. Some data 1 are given in Table XXXV. At TABLE XXXV. ADSORPTION OF METHYLENE BLUE BY WOOL Volume of solution = 100 cc. Weight of wool = 1 gram. Methylene blue adsorbed, mg Methylene blue in solution, percent No salt 0.06 g Na"SO4 0.06 g Na z HPO 0.1 35 45.6 81.1 0.2 40 52.2 91.3 0.3 43 56.4 95.6 0.4 44 59.4 98.6 every concentration of dye more dye is fixed in the bath contain- ing sulphate than in the bath to which no salt is added, while still more dye is adsorbed in the phosphate solution. \Ve see 1 Pelet-Jolivet: Die Theorie des Farbeprozesses, 118 (.1910). 118 APPLIED COLLOID CHEMISTRY thus the amount of adsorption varying in the order Na 2 HPO 4 > Na 2 SO 4 > NaCl > H 2 O. Just as should be the case, sodium sul- phate increases the amount of a basic dye taken up, whereas it decreases the amount of an acid dye adsorbed. With 0.5 gram silk in 50 cc 0.2 percent methylene blue solution, the amounts of methylene blue taken up were 28.6, 18.4, 13.3, and 10.8 mg when sodium phosphate, sodium sulphate, sodium chlo- ride, and nothing were added respectively to the solution. 1 With cotton there was no difference under the same circum- stances; but there was a marked difference when the cotton had been treated with tannin. The effect of different cations was also studied by Pelet-Jolivet who found that silk took up 13.3 mg methylene blue from a solution containing sodium chloride and only 6.3 mg from a solution containing barium chloride, the strongly adsorbed barium cation thus cutting down the adsorp- tion of the basic dye. With wool the depth of color decreases in the following order: sodium phosphate, sodium sulphate, no addition, barium chloride, and platinum tetrachloride. Cotton takes up less methylene blue from a solution containing barium chloride than from one containing sodium chloride. With substantive dyes the effects of electrolytes are different because these dyes are in colloidal solution and the question of stability enters in. Abnormal results are also to be expected with acid and basic dyes which form colloidal solutions. When an acid dye is adsorbed nearly completely by the fiber it may easily happen that the differences between acids becomes so small as to be practically negligible. 2 The concentration of the anion can be increased very much relatively to the concentration of hydrogen as ion by adding salt, and thus the effect of the anion can be brought out. This is shown in Table XXXVI. In the third column are given the amounts of dye adsorbed when acids alone are added to the bath. In the fifth column are the amounts of dye adsorbed when the solution contains 10 cc N/10 acid + 10 cc N/l salt. With the acids alone the amount of adsorption of acid violet is practically independent of the nature of the acid. With croceine orange, sulphuric acid and phosphoric acid behave practically alike and cut down the adsorption as compared with 1 PELET-JOLIVET: Die Theorie des Farbeprozesses, 118 (1010). 2 LAKE: Jour. Phys. Chem., 20, 785 (1916). ADSORPTION FROM SOLUTION 119 hydrochloric acid. With acid violet the addition of sodium chloride has no appreciable effect. The slight apparent increase in adsorption is probably due to experimental error. With croceine orange the sodium chloride cuts down the adsorption slightly. With sodium phosphate the effect is quite extraordinary, TABLE XXXVI. ADSORPTION OF ACID VIOLET AND CROCEINE ORANGE BY WOOL Volum^ of solution = 250 cc. Weight of wool = 1 gm. Amount of dye = 40 mg. Temperature = 100. Time of run = 1 hr. Acid dye 10 cc N/10 acid Adsorbed dye, mg Acid + 10 cc N/l salt Adsorbed dye, mg Acid violet HC1 39.2 NaCl 39.5 Croceine orange. H 2 S0 4 H 3 P0 4 HC1 39.2 39.1 38.0 Na 2 S0 4 Na 2 HPO 4 NaCl 36.8 l.O 1 33.4 H 2 SO 4 H 3 P0 4 35.5 35.5 Na 2 S0 4 Na 2 HPO 4 20.0 O.I 1 the adsorption of acid violet and of croceine orange being cut almost to zero. The change in the hydrogen ion concentration cannot account for this because 38 mg acid violet and 32 mg croceine orange are adsorbed from a water solution under similar circumstances. We are therefore dealing with a specific effect of the phosphate radical. Qualitative experiments along the same lines have been ob- tained by Davison. 2 With acid green and emerald green on wool and with fast green and safranine on cotton, good illustra- tions are obtained of the generalization that acid dyes are taken up more completely in an acid solution than in a neutral one and least of all in an alkaline solution, while the reverse is true with basic dyes. Marked differences between the dyeing in acid and alkaline solutions can be obtained on wool with fast green, fast blue, acid violet, croceine orange, crystal ponceau, chrysoidine, safranine, and methylene blue; on cotton with 1 These experiments are probably not strictly comparable with the others owing to the change in hydrogen ion concentration. 2 Jour. Phys. Chem., 17, 737 (1913). 120 APPLIED COLLOID CHEMISTRY acid violet, croceine orange, and emerald green. The generaliza- tion that addition of sodium sulphate cuts down the adsorption of acid dyes and increases the adsorption of basic dyes is shown excellently on wool with acid green, fast blue, croceine orange, Victoria blue, chrysoidine, and safranine; on cotton with benzo- purpurine. The effect of continuously varying concentrations of sodium sulphate is shown well on wool with brilliant blue and safranine; on cotton with safranine. When alumina is precipi- tated from an aluminum acetate solution, it adsorbs fast green, acid green, and croceine orange considerably; but when precipi- tated from an aluminum sulphate solution it carries down practi- cally no fast green, acid green, or croceine orange. The presence of sulphate decreases the adsorption of acid dyes by alumina as well as by wool. Another illustration of this is that alumina precipitated from an aluminum sulphate solution can only decolorize dilute solutions of alizarine, alizarine green, benzo orange, coeruleine, and acid violet, while alumina precipitated from an acetate solution will decolorize more concentrated solu- tions. In these experiments with alumina the dye was added to the solution before the alumina was precipitated. It seems probable that if alumina were precipitated from a sulphate solution with barium hydroxide or calcium hydroxide, the effect due to the sulphate would disappear owing to the insolubility of the barium or calcium sulphate. This experiment was not tried. CHEMICAL ACTION Selective adsorption connotes the possibility of chemical decomposition as a result of adsorption. If a given substance adsorbs a base more strongly than an acid for instance, there will be a tendency for the salt of that base and that acid to hydro- lyze, the base being then adsorbed to a greater extent than the acid. Theoretically there is always some hydrolysis even with sodium chloride and water according to the equation Na' + Cl' + H 2 = Na- + OIF + IT + Cl' but this reaction does not run far because caustic soda and hydrochloric acid are strong electrolytes and we cannot have a high simultaneous concentration of hydrogen and hydroxyl as ions. If the caustic soda is removed by adsorption, the hydrol- ysis can go farther. How complete the hydrolysis will be ADSORPTION FROM SOLUTION 121 depends on the degree to which the base is adsorbed and on the strength of the acid. Any acid will tend to react with adsorbed caustic soda; but that tendency will be greater the stronger the acid. We thus see that a neutral solution will become acid if shaken with a substance which adsorbs the base much more strongly than the acid, that it will become alkaline if the substance adsorbs the acid much more strongly than the base, and that it will remain neutral in case the solid does not adsorb either base or acid at all, or in case it adsorbs the two in practically equivalent amounts either separately or as undissoci- ated salt. The amount of base or of acid taken up from a salt solution will be greater the weaker the unadsorbed acid or base. Many cases are known where the solution does become acid or alkaline when shaken with an adsorbing agent and this has nothing to do with any hypothetical acidity or alkalinity of the adsorbing substance. If fuller's earth be shaken with water and then filtered, the filtrate is neutral to litmus paper 1 or to phenolphthalein, showing that no soluble base or acid is present. If fuller's earth be shaken with a sodium chloride solution and filtered, the filtrate is acid to litmus or to phenolphthalein. This is because fuller's earth has adsorbed the base. If one presses litmus paper against moistened fuller's earth, the litmus paper turns red, and if one adds fuller's earth to a faintly alkaline solution of phenolphtha- lein, the red color disappears. This is not because the fuller's earth is acid, but because it takes the base from the sodium chlo- ride, the litmus, or the phenolphthalein. I have been told that the adsorbing power of fuller's earth is so great that an acre-foot as soil would adsorb 30,000 Ib. of lime and that this would make fuller's earth about equivalent in acidity to a two percent sulphuric acid solution. Silk has such a marked selective adsorption for rosaniline (magenta) that it will not only decompose the hydrochloride, leaving the acid behind, but will also convert the carbinol base back into the color base in presence of an excess of ammonia, 2 the silk being dyed red from a colorless ammoniacal solution. 1 Cf. CAMERON: Jour. Phys. Chem., 14, 400 (.1910). 2 jACQUEMiN: Comptes rendus, 82, 261 (1876); see also MILLS: Jour. Chem. Soc., 35, 27 (1879); FORTUYN: Zeit. phys. Chem., 90, 236 (1915.) 122 APPLIED COLLOID CHEMISTRY This is the more remarkable because the free color base is in- stable and cannot be isolated in a pure state. This makes it seem certain that silk stabilizes the free color base. Another similar instance is the mordanting of wool with copper salts, coloring it green. This green does not change to black when the wool is heated to boiling, though hydrous copper oxide by itself changes very readily. Blucher and Farnau 1 have extended Tommasi's 2 experiments on the stabilization of hydrous copper oxide by manganese salts and find that a number of other metallic hydroxides are also effective. Bayliss 3 has obtained results, which indicate that alumina stabilizes the free acid of Congo red, and Schaposchnikoff and Bogojawlenski 4 have isolated this metastable form by allowing the pyridine salt to effloresce. Pas- sive iron is an instable oxide, presumably Fe0 2 or Fe0 3 , stabilized when adsorbed by metallic iron. 5 When discussing the adsorption of gases by solids, numerous instances were given of so-called contact catalysis. With solu- tions the effect of contact materials is as yet relatively unim- portant. Lassar-Cohn 6 gives a few instances where yields have been increased by the use of porous masses. When two parts of acetanilide were heated with one part of zinc chloride, a yield of 3.5 percent flavaniline hydrochloride was obtained. In presente of two parts of infusorial earth the yield increased to about 7.5 percent. When benzene and chloroform react in presence of aluminum bromide, the yield increases from 33 percent *to 44 percent when infusorial earth is added. The sulphonation of benzene, xylene, naphthalene, anthraquinone, aniline, etc., is simplified very much by the addition of infusorial earth or of charcoal. When a-dinitronaphthalene is treated with sulphuric acid under suitable conditions in presence of infusorial earth or charcoal, oxidation takes place with elimination of the nitro groups and formation of naphthazarin. 1 Jour. Phys. Chem., 18, 629 (1914). 2 TOMMASI: Bull. Soc. chim. Paris (2) 37, 197 (1882); Comptes rendus 99, 37 (1884). 3 Proc. Roy. Soc., 84B, 881 (1911). 4 Jour. Russ. Phys. Chem. Soc. 44, 1813 (1913).' 6 BENNETT and BURNHAM: Jour. Phys. Chem., 21, 107 (1917). 6 Arbeitsmethoden fiir organisch-chemische Laboratorien, 584, 72, 1079, 923 (1903). ADSORPTION FROM SOLUTION 123 Gurwitsch 1 reports that amylene polymerizes readily on stand- ing in contact with " floridin," which is a hydrous silicate of some sort. Playfair 2 found that a solution of nitric acid could be pre- pared which would only bleach a piece of cloth dyed with indigo when the cloth was in contact with alumina, chromic oxide, or stannic oxide. Richardson 3 believes that petroleum and bitumens are formed by the polymerizing or condensing action of clays or sands. A case which has been studied a great deal quantitatively is the decomposition of hydrogen peroxide solutions by platinum. One of the most interesting things about this reaction is its ex- treme sensitiveness to so-called poisons. 4 The rate of decomposi- tion of hydrogen peroxide by a given solution of colloidal plati- num was reduced approximately to one-half by M/20,000,000 HCN, M/2,000,000 HgCl 2 , and M/300,000 H 2 S. Curiously enough the catalytic action of platinum and of ferments on hydro- gen peroxide is decreased in both cases by very slight traces of poisons, such as prussic acid, hydrogen sulphide, etc. It has also been shown by Schonbein 5 that these same substances cut down the catalytic action of red blood corpuscles on hydrogen peroxide. In Table XXXVII are given data 6 for the concentrations neces- sary to cut down the rate of decomposition of hydrogen peroxide to one-half in the case of colloidal platinum and of haemase, the active enzyme in the red blood corpuscles. While many of the substances that cut down the platinum catalysis also decrease the catalytic action of haemase, there are many cases where no parallelism occurs. Thus carbon monoxide is very toxic to plati- num and has no effect on haemase. On the other hand, nitric acid, sulphuric acid, potassium nitrate, and potassium chlorate have practically no effect on platinum and are quite toxic to haemase. Even where there is a general parallelism, one must not follow it too closely. , With platinum the prussic acid solution 1 Zeit. Kolloidchemie, 11, 18 (1912). 2 Mem. Chem., Soc. 3, 354 (1847). 3 Jour. Ind. Eng. Chem., 8, 4 (1916); Met. Chem. Eng., 16, 25 (1917). 4 BKEDIQ and VON BERNECK: Zeit. phys. Chem., 31, 258 (1899); BREDIQ and IKEDA: Ibid., 37, 1 (1901). 5 Jour, prakt. Chem., 105, 202 (1868). CENTER: Zeit. phys. Chem., 61, 701 (1905). 124 APPLIED COLLOID CHEMISTRY has one-tenth the concentration of the mercuric chloride solution, while with haemase it has double the concentration. The tabu- lated concentration for mercuric chloride is the same for platinum as for haemase, while that of iodine dissolved in potassium iodide is one hundred times as great for haemase. Kastle and Loeven- hart 1 point out that prussic acid is a strong poison for colloidal TABLE XXXVII. THE POISONING OF COLLOIDAL PLATINUM AND HAEMASE Concentrations at which poisons reduce the rate of the catalytic decom- position of hydrogen peroxide approximately to one-half. Poison Colloidal platinum Haemase H 2 S M /300,000 M / 1,000, 000 HCN M /20,000, 000 M/1,000,000 HgCl 2 M /2, 000, 000 M /2, 000, 000 He-Bra M/300 000 Hg(CN) 2 M /200, 000 M /300, 000 I 2 in KI M/5,000,000 M /50,t)00 NH 2 OH.HC1 M /25, 000 M /80, 000 Phenyl hydrazine . M /20, 000 Aniline M /5, 000 M/400 Arsenious acid CO M/50 Very poisonous No poisoning at M/2,000 No poisoning HC1 1 M /3, 000 M/ 100,000 NH 4 C1 HNO 3 M/200 No poisoning M/1,000 M /250,000 H 2 SO 4 KNO 3 KC1O 3 No poisoning No poisoning Slight poisoning? M/50,000 M/40,000 at M /40,000 at platinum and silver but accelerates the catalysis of hydrogen peroxide by iron and copper. It seems probable that the poisons are adsorbed strongly by the catalytic agent and therefore prevent the adsorption of hydrogen peroxide. 2 This adsorption will be specific and will therefore in general not be the same quantitatively for different catalytic agents. On the other hand, there may well be a qualitative agreement in some or even in many cases. Owing to the extreme 1 Am. Chem. Jour., .29, 397 (1903). 2 Cf. SENTEB: Zeit. phys. Chem., 61, 702 (1905); DENHAM: Ibid., 72, 689 (1910). ADSORPTION FROM SOLUTION 125 flexibility of this hypothesis there is nothing as yet to prevent its accounting for all the phenomena observed. 1 On the other hand it must be remembered that proof is still lacking that platinum adsorbs prussic acid for instance, very much more strongly than hydrogen sulphide, and that there is no difference with hae- mase. There may be other factors which must be taken into account. Thus Bredig 2 points out that when colloidal platinum is allowed to stand in contact with hydrogen peroxide and con- centrated potassium cyanide, the platinum flocculates and pre- cipitates. The agglomerated platinum causes the hydrogen peroxide to decompose, thus showing that the cyanide does not poison precipitated platinum black. There seem to be only two possible explanations. One is that the adsorption of potassium cyanide by platinum falls off very much more rapidly with in- creasing size of the platinum particles than the adsorption of hydrogen peroxide by platinum. The other explanation is that, through oxidation or otherwise, there is formed what might be called an anti-body, which cuts down the adsorption of the cya- nide. Neither hypothesis is very satisfactory and there is no experimental evidence for either, except that Jablczynski 3 found that platinized platinum decomposes chromous chloride three times as fast and hydrogen peroxide one hundred times as fast as does smooth platinum foil. The matter should be taken up because Rosnyak 4 claims that the catalytic action of platinum on hydrogen peroxide decreases with increasing dispersity of the platinum and Felgate 5 states that pulverulent nickel reduces nitric oxide while colloidal nickel does not. In both these cases it is probable that some factor has been overlooked or not stated. The more finely divided the platinum the more likely the surface is to be coated with oxide and it does not follow at all that the catalytic action of metallic platinum is the same as that of an oxide of platinum. If platinum is kept in suspension by means of gelatine or some similar substances, this will have an effect on the adsorption and the catalytic action. 1 Cf . FREUNDLICH and KAEMPFER: Zeit. phys. Chem., 90, 681 (1915). 2 Zeit. phys. Chem., 31, 332 (1899). 8 Ibid. 64, 751 (1908). Ibid. 85,68 (1913). 6 Chem. News, 108, 178 (1913). 126 APPLIED COLLOID CHEMISTRY A comparatively recent technical development in contact catalysis is the hardening of oils, by the conversion of oleic acid or olein, for instance, into stearic acid or stearin by the action of hydrogen on the liquid unsaturated oil in presence of a suitable catalytic agent. 1 In 1908 Paal 2 proposed using a colloidal metal as the catalytic agent at ordinary temperatures. Colloidal solutions of platinum, palladium, osmium, and iridium have been made, of which the palladium hydrosol appears to possess the greatest value. When hydrogen at ordinary temperatures is passed into glycerides of the unsaturated fatty acid to which some palladium hydrosol has been added, reduction takes place. It is more satisfactory, however, to precipitate the metal on an inert base such as magnesia or kieselguhr. A cheaper and apparently better method is to use nickel as a catalytic agent for the hydrogenation of liquid, unsaturated acids, the reduction taking place at about 250-300. There has been some controversy whether the catalytic agent was really metallic nickel 3 or an unspecified oxide or oxyhydride ; 4 but there seems to be little doubt but that the metal is the catalytic agent. Some details in regard to the preparation of the catalytic agent are given by Ellis. 5 Paal and Karl 6 report that the oxides of lead, cadmium, zinc, aluminum and iron cut down the catalytic action of palladium, while the oxides of magnesium, nickel, and co- balt do not. ADSORPTION FROM SOLUTION BY LIQUID There is no reason why there should not be adsorption by a liquid just as well as by a solid, except that the matter is com- plicated by the possibility of the solute dissolving in the second 1 JOBLING: Catalysis and its Industrial Applications, 65, 75 (1916). 2 Ber. deutsch. chem. Ges., 41, 2282 (1908). 3 MEIGEN: Jour, prakt. Chem. (2) 92, 390 (1915). 4 BEDFORD and ERDMANN: Jour, prakt. Chem. (2) 87, 425 (1913); SIEG- MUND and SAIDA: Ibid. (2) 91, 442 (1915); ERDMANN: Ibid. (2) 91, 469 (1915); GLASER: Zeit. anorg. Chem., 36, 1(1903); BELLACCI and CORELLI: Ibid., 86, 88 (1914); SABATIER and ESPIL: Comptes rendus, 158, 568; 159, 137 (1914). 5 Jour. Soc. Chem. Ind., 31, 1163 (1912); The Hydrogenation of Oils, 2d. Ed., 108 (1918). . 6 Ber. deutsch. chem. Ges., 46, 3069 (1913). ADSORPTION FROM SOLUTION 127 liquid. An interesting case of such adsorption was studied by Wilson, 1 Swan, 2 and Twomey. 3 When a drop of chloro- form or carbon tetrachloride is placed on the bottom of a glass dish and water or dilute acid poured over it, the drop is well rounded. If the solution is made alkaline, the drop flattens out. The flattening is clearly not due to a change in surface tension of the water phase, because Wilson found that the effect of alkali was the same qualitatively whether sodium chloride was present or not, whereas the addition of salt increases the surface tension of the water phase. What happens is that hydroxyl is adsorbed at the dineric interface, lowering the surface tension and causing the organic liquid to flatten. This is confirmed by the experi- ments of von Lerch 4 who determined the surface tension between benzene and water by means of the rise in capillary tubes. 5 The surface tension between pure benzene and pure water was found to be 32.6 dynes/cm at 15-16. With M/4 NaOH the value dropped to 20.7 and with M/2 NH 4 OH to 27.3. The lowering of the surface tension is greater with sodium hydroxide than with ammonia owing to the greater electrolytic dissociation of the former. The surface tension was also lowered by hydrochloric acid but only slightly, M/2 HC1 bringing it down to 31.5. This is in accord with Wilson's qualitative conclusions that acids do not increase the surface tension quite as much as pure water. Von Lerch also found a marked effect due to acetic acid; but this acid is miscible in all proportions with benzene and causes benzene and water to become more miscible, so there is nothing surprising about its behavior. The surface tension measurements are con- firmed by the electrical ones, for von Lerch found that drops of benzene in caustic soda moved fairly readily under electrical stress, whereas there was almost no movement in presence of sodium chloride or hydrochloric acid. It is possible that other ions might be adsorbed markedly by benzene, but I know of no experiments to show this. Patrick 6 found that new fuchsine, 1 Jour. Chem. Soc., 1, 174 (1849). 2 Phil. Mag. (3) 33, 36 (1848). 3 Jour. Phys. Chem. 19, 360 (1915). * Drude's Ann., 9, 434 (1902). 6 HARKINS states that VON LERCH'S experiments with dilute caustic soda solution are inaccurate. 6 Zeit. phys. Chem., 86, 545 (1914). 128 APPLIED COLLOID CHEMISTRY picric acid, salicylic acid, and mercurous sulphate are adsorbed from aqueous solution by mercury, the surface tension of the mercury being lowered as a result of the adsorption. In all four cases the amounts adsorbed varied with the concentration approximately according to a formula of the type found to hold for adsorption from solution by a solid. The adsorption of hydroxyl means practically a concentrating of alkali at the surface of the benzene and consequently a closely adhering film of water. This accounts for Wilson's observation that the drops of chloroform were readily wetted by an alkaline solution. The presence of the surface film of caustic soda solu- tion also accounts for the decreased mobility observed by Wilson. CHAPTER IV SURFACE TENSION BROWNIAN MOVEMENTS ADSORPTION AND SURFACE TENSION When a substance dissolves in a liquid, the surface tension between the liquid and the vapor changes, increasing in some cases and decreasing in others. Two solutions of the same den- sity will not necessarily have the same surface tension. For this reason a hydrometer graduated for sulphuric acid solutions will not be accurate for hydrochloric acid solutions, because the amounts of solution which will rise up the stem of the hydro- meter will depend on the surface tensions and will therefore not be the same in the two cases. In most cases the surface tensions of solutions will lie between the surface tensions of the constitu- ents; but this is not necessarily true. 1 Rontgen and Schneider 2 found a maximum surface tension for sulphuric acid and water at about 48 percent H 2 SO4. Whatmough 3 obtained minima for mixtures of acetic acid with benzene, methyl iodide, ethyl iodide, chloroform, or carbon tetrachloride ; for carbon bisulphide and ethylene chloride; for benzene and ethyl iodide; for carbon tetrachloride and chloroform, for ethyl acetate and amyl alcohol. Most salts increase the surface tension of the solution, the in- crease being very nearly proportional to the concentration. The strong bases increase the surface tension but ammonia lowers it and so also do nitric, hydrochloric, and hydrobromic acid. Worley 4 has discovered an interesting fact. Although sodium chloride increases the surface tension of water, the addition of salt to a solution of aniline in water decreases the surface tension. The sodium chloride decreases the solubility of aniline in water and therefore increases the partial pressure of the aniline vapor. 1 See FREUNDLICH : Kapillarchemie, 58 (1909). 2 Wied. Ann., 29, 165 (1886). 3 Zeit. phys. Chem., 39, 129 (1902). 4 Jour. Chem. Soc., 105, 260, 273 (1914). 9 129 APPLIED COLLOID CHEMISTRY The vapor pressures therefore correspond to a solution richer in aniline than that actually present and the surface tension relations run approximately parallel with the vapor pressures. Worley noticed another curious thing in regard to change of surface tensions of solutions with rising temperature. Since liquid and vapor are identical at the critical temperature, the surface ten- sion becomes zero at the critical point and the surface tension of pure liquids therefore decreases with rising temperature. One might reasonably have expected that the surface tension tem- perature curves for solutions would run somewhere near parallel to the corresponding curves for the pure constituents; but this is not the case for mixtures of water with aniline or phenol. The surface tension of the phenol-rich solutions vary relatively little with rising temperature, while the corresponding aniline solu- tions actually show an increase of surface tension with rising temperature, The explanation seems to be that the ratio of water to aniline or phenol in the vapor increases with rising temperature and that the solutions therefore show surface tensions corresponding to what one might expect of solutions richer in water than they actually are. It is a pity that Worley did not also study some such case as ether and water, where the partially miscible liquid with the lower surface tension has the higher vapor pressure. The change in surface tension when a solution is formed is also accompanied by a change in concentration, the surface film of the liquid having a different concentration from the mass of the liquid solution. The simple rule in regard to this is that the concentra- tion in the film tends to change so as to decrease the surface tension. 1 Consequently the surface film will be more dilute than the mass of the solution if the dissolved substance increases the surface tension, and will be more concentrated than it, if the solute lowers the surface tension of the solution. An equilibrium will be reached when the change in the surface tension is balanced by the difference of osmotic pressure between the surface film and the mass of the solution. This can be expressed mathe- matically by the equation 2 da ^~o> 1 J. WILLARD GIBBS: Scientific Papers, 1, 219. 2 Cf. DONNAN and BARKER: Proc. Roy. Soc., 85A, 557 (1911). SURFACE TENSION BROWNIAN MOVEMENTS 131 where a is the interfacial tension, /i is the chemical potential of the dissolved substance in the aquoeus phase, y is the mass of solute per unit area of interface in excess of that corre- sponding to the uniform bulk-concentration of the solute in the solution. It will be seen that y is, in fact, the amount of solute per unit area of interface adsorbed or concentrated in the interfacial transition layer. If we are justified in applying the laws of dilute solutions, the above equation can be written c da ~RTdc where R is the constant of the simple gas equation, T is the abso- lute thermodynamic temperature and c is the concentration. It is to be noted that these two equations apply only to the case of a substance which is really in solution, a point which is apt to be overlooked in papers on colloids. On the other hand, there are many substances such as soap, saponin, gelatine, which are probably not soluble in water to any appreciable extent; but which form apparent or colloidal solutions with surface tensions lower than those of pure water. In these cases the concentration of the added substance will be higher in the surface film than in the mass of the liquid, but the Gibbs formula will not apply. As a matter of fact with sodium glycocollate, Congo red, methyl orange, and sodium oleate, the adsorptions in the surface are 20-100 times those calculated from the formula. 1 If the difference in concentration between the mass of the solution and the surface layer is set up slowly, one ought to get one value for the surface tension when equilibrium is reached and another if rapid measurements are made on a fresh surface. This has been done by Lord Rayleigh, 2 though not for a case of true solution. 3 The equilibrium or static method consisted in measuring the rise in capillary tubes. The rapid or dynamic method depends on the principle that a jet of liquid issuing under moderate pressure from an elongated, e.g., elliptical, aperture 1 WILLOWS and HATSCHEK: Surface Tension and Surface Energy, 46 (1915). 2 Proc. Roy. Soc., 47, 281 (1890). 3 See however HABKINS and BROWN: Jour. Am. Chem. Soc., 41, 522 (1919). 132 APPLIED COLLOID CHEMISTRY perforated in a thin plate, assumes a chain-like appearance, the complete period, X, corresponding to two links of the chain, being the distance traveled over by a given part of the liquid in the time occupied by a complete transverse vibration of the column about its cylindrical configuration of equilibrium. Since the phase of vibration depends upon the time elapsed, it is always the same at the same point in space, and thus the motion is steady in the hydrodynamical sense, and the boundary of the jet is a fixed surface. Measurements of X under a given head, or velocity, determine the time of vibration, and from this, when the density of the liquid and the diameter of the column are known, follows in its turn the value of the capillary tension to which the vibrations are due. By this method the liquid surface may be tested before it is Mooo second old. Rayleigh's data have been recalculated by Freundlich 1 who added some of his own. These are given in Table XXXVIII. TABLE XXXVIII. STATIC AND DYNAMIC SURFACE TENSIONS OF SOLUTIONS Solution Concentration Surface tensions in dynes/cm Static Dynamic Water 75 55 26 26 26 52 54 75 79 79 62 58 73 68 Sodium oleate 0.025 percent . 25 percent 1 . 25 percent 2.5 . percent Sodium oleate Sodium oleate. Sodium oleate Saponin Heptylic acid O.OOoN Except in the case of pure water the surface tensions by the static method are lower than those by the dynamic method. One curious point is that the surface tensions by the dynamic method are higher than that of pure water in the case of the two very dilute sodium oleate solutions. Lord Rayleigh did not discuss this point and Freundlich leaves it untouched. In the actual measure- ments an error of about two percent would account for the dif- ference, which may be due to experimental error. There is another possibility, however, that the sodium oleate is completely 1 Kapillarchemie, 56 (1909). SURFACE TENSION BROWN1 AN MOVEMENTS 133 hydrolyzed in the dilute solutions and that what is measured by the dynamic method is really the surface tension of a caustic soda solution. While this set of experiments gives more or less the results that one would expect, it seems impossible to account for the difference between the static and dynamic methods in the case of the ad- sorption of gases by mercury. Stockle 1 measured the surface tension of mercury in a vacuum of 0.004-0.001 mm and ob- tained a value of 436 dynes/ cm at 15 regardless whether the measurement was made at once or not. When hydrogen was present, the time factor was important, a value of 470 dynes/ cm being obtained when the measurement was made quickly, but dropping to 434 dynes/cm if the mercury were allowed to stand in contact with hydrogen. Similar results were obtained with oxygen, nitrogen, carbon dioxide, and air, Table XXXIX. The general results were confirmed by G. Meyer, 2 Table XL, who made TABLE XXXIX. SURFACE TENSION OF MERCURY Data in dynes /cm with fresh and old surfaces Medium Rapid After 60' Vacuum (15) 436 434 Hydrogen (21) 470 436 Oxygen (25) 478 432 Nitrogen (16) Carbon dioxide (19) 489 438 429 Dry air (17) 476 429 Moist air (17) 481 TABLE XL. SURFACE TENSION OF MERCURY Data in dynes /cm with fresh and old surfaces Medium Vibrating jet Curvature of surface Hydrogen 554 470 Oxygen 504 478 Carbon dioxide 487 480 Nitrogen 496 489 Wied. Ann., 66, 49 (1898). Ibid., 66,523 (1898). 134 APPLIED COLLOID CHEMISTRY use of Rayleigh's jet method for the rapid measurements and consequently obtained higher values than Stockle. It may be easy to see that the adsorption of a gas by mercury should lower the surface tension, but it is not clear why different gases should bring the surface tension down practically to the same value as that found when there is supposed to be no gas present. The dynamic methods should start at the value for mercury in a vacuum if no factor has been overlooked and there is no apparent reason why an unadsorbed gas should raise the sur- face tension of mercury so much. Lenard 1 has brought up a point which may have a bearing on this. If we have a partially polymerized liquid such as water, the modification having the lower surface tension will concentrate in the surface. If we form a new surface suddenly, we shall get a higher concentration, temporarily, of the particular modification which has the higher surface tension. If the equilibrium is reached relatively slowly, the dynamic method will give a higher value for the surface tension than the static methods. If equilibrium is reached instantaneously, there will be no dif- ference. We can account for the facts observed with mercury if we make the assumptions that mercury is a partially polymer- ized liquid, that equilibrium between the modifications is reached practically instantaneously in a vacuum, and that equilibrium is reached relatively slowly in presence of gases. The difficulty with this is that we have no independent proof of these assumptions. Another hypothetical explanation becomes possible, if we assume that there is not a sharp discontinuity at the surface between liquid and vapor. The Laplace theory of surface tension assumes that there is a perfectly sharp line of demarcation be- tween the two media bounding the surface, for instance, 2 between liquid and air, while van der Waals postulates a continuous transition. According to the latter way of looking at things, mercury in a vacuum adsorbs its own vapor forming a thin transition layer varying from the density of liquid mercury at one side to the density of mercury vapor at the other side. If this transition 1 Cf. AGANIN: Drude's Ann., 45, 1020 (1914). 2 WILLOWS and HATSCHBK: Surface Tension and Surface Energy, 33 (1915). SURFACE TENSION BROWNI AN MOVEMENTS 135 film forms instantaneously in a vacuum and relatively slowly in presence of a gas, the surface tension of a fresh surface of mercury in presence of a gas will be higher than the equilibrium surface tension of mercury in a vacuum, and this higher surface tension thus measured will decrease, if the surface is not renewed, down to the equilibrium surface tension of mercury in a vacuum, or below it if the gas is adsorbed markedly at the mercury surface. This explanation seems somewhat more plausible than that of Lenard; but it is open to the same objection that there is as yet no independent proof of the assumptions involved. Ferguson 1 has shown that the surface tension of benzene is different in air from what it is in carbon dioxide. It is a question of course to what extent this difference is due to the solubilities of the gases in the benzene. Ferguson considers that the nature of the gas is more important than the solubility effect; but he gives no evidence for this. He does, however, mention one interesting and surprising fact that the absolute value of the surface tensions increased slightly as the liquid became more and more saturated, finally reaching a steady value. From the fact that the lowering of the surface tension is accompanied by an increase in the concentration of the surface film of a true solution, Freundlich 2 draws the conclusion that this is true in all cases, and that if a dissolved substance lowers the surface tension between the solid and the solution, the dis- solved substance is adsorbed; conversely that the dissolved substance is adsorbed only in case it lowers the surface tension at the interface. The conclusion may or may not be right; but the logic is faulty because the two cases are not parallel. The Gibbs relation holds for a true solution, where the difference of concentration is entirely inside the solution phase. Freundlich is extending the relation to a heterogeneous system in which the adsorbed substance is presumably on the outer surface of the solid and by definition cannot diffuse into it because we should then have a solid solution. It is very difficult to prove or dis- prove Freundlich's conclusion experimentally; but it is a great pity to have people believe, as many now do, that the generaliza- tion is based on sound thermodynamics. 1 Phil. Mag. (6) 28, 403 (1914). 2 Zeit. phys. Chem., 67, 424 (1907). 136 APPLIED COLLOID CHEMISTRY The surface tension of solids cannot be measured directly; but Tangl 1 has made some experiments on the surface tension of such pseudo-solids as rubber and paraffin against water. Solids undoubtedly do have definite surface tensions. In fact, one might consider the melting point as the temperature at which the surface tension overcomes the crystalline forces. In the liquid crystals and the flowing crystals, we have interme- diate stages. Reference has already been made to Beilby's experiments on thin gold leaf which became granular when heated. 2 We shall see later that the phenomenon of sintering depends on surface tension. BROWNIAN MOVEMENTS If a stone be dropped into water, it sinks rapidly; but if it be ground into fine particles, the surface is much greater and consequently the particles sink slowly. If the stone were ground into very fine particles, we should expect these to sink very slowly, the rate being a function of the diameter and density of the particles. A formula, ' 2r*(8-S')g &t for this case has been deduced by Stokes, 3 where V is the constant rate of fall, r the radius of the particles, S their specific gravity, S' the specific gravity of the liquid, rj its viscosity coefficient and g the gravitational constant. Hatschek 4 has applied this formula to several special cases. For particles 5 of gold of I/* radius the rate of fall in water is calculated to be 2.4 mm per minute. With gold particles of 10/zju radius the rate of settling in water is about 10 mm per month. For a particle of lOju/* radius and a specific gravity of only 3 the calculated rate would be about 1 mm per month. Very small rates of fall have actually been observed. Crude's Ann., 34, 311 (1911); cf. JONES: Zeit. phys. Chem., 82, 448 (1913); RONTGEN: Wied. Ann., 3, 321 (1878). 2 Proc. Roy. Soc., 72, 226 (1904). 3 Mathematical and Physical Papers, 3, 1 (1901). For modifications and limitations see CUNNINGHAM: Proc. Roy. Soc., 83A, 357 (1910); Jour. Am. Chem. Soc. 41, 319 (1919); MILLIKAN: Phys. Zeit., 11, 1097 (1910). * An Introduction to the Physics and Chemistry of Colloids, 24 (1913). 6 Elutriation, or the difference in the rate of settling, is much used in grading powders such as emery, carborundum, ultramarine, kaolin, etc. SURFACE TENSION BROWNI AN MOVEMENTS 137 Even in air fine particles settle very slowly. A sandstorm in the Grand Canyon may make the air a brick-red color for two or three days. 1 Similar phenomena are to be observed all over the world. 2 Ebell 3 states that ultramarine will stay suspended in pure water for months when obtained in a very finely divided state by grinding and elutriation. Mlilhauser 4 found that when very finely ground carborundum powder is treated with water, a por- tion of it does not settle in months and behaves in that respect like colloidal metallic silver. Brewer 5 points out the slow rate at which clays settle, and raises the question whether extremely finely divided particles settle at all. If the very finely divided particles do not really settle at all, there must be some factor which neutralizes the action of gravity more or less completely. The experiments of an English botanist, Brown, 6 give us a clue to the difficulty, though the importance of these experiments was not appreciated until very muqh later. Brown was studying under the microscope the grains of pollen from Clarckia pulchella, and observed that some of them moved about in the liquid. Further experiments showed that prac- tically all finely divided solids kept in continuous motion. Brown considered that he was watching the movements of active molecules. The zig-zag movements of suspended particles are known as Brownian movements. They are apparently due to the bom- bardment of the suspended particles by the molecules of the liquid. 7 On this hypothesis the motion should be more rapid, the smaller the bombarded particles and this proves to be the case. The work of Gouy showed that the hypothesis of molecular agitation gave a satisfactory explanation of the Brownian move- 1 ELLSWORTH and EMERY KOLB : National Geographical Magazine, 26, 134 (1914). 2 See EHRENBERG: Die Bodenkolloide, 209 (1915). 3 Ber. deutsch. chem. Ges., 16, 2429 (1883). 4 Zeit. anorg. Chem., 6, 117 U894). 6 Mem. Nat. Acad. Sci., 2, 165 (1884); Am. Jour. Sci. (3) 29, 1 (1885). 6 Phil. Mag., 4, 161 (1828); 6, 161 (1829). 'WIENER: Pogg. Ann., 118, 79 (1863); GOUY: Jour, de Phys. (2) 7, 561 (1888); Comptes rendus, 109, 102 (1889); Revue g4n6rale des Sciences, 1895, 1. 138 APPLIED COLLOID CHEMISTRY ments and that all the other tentative explanations were faulty. He showed that the Brownian movements are not due to vibra- tions transmitted to the liquid under examination, since they persist equally as well, for example, at night in the country as during the day near a populous street where heavy vehicles pass. They are not due to convection currents existing in fluids where thermal equilibrium has not been attained, for they do not change ap- preciably when plenty of time is given for equilibrium to be reached. Any comparison between Brownian movements and the agitation of dust particles dancing in the sunlight must therefore be set aside. In addition, in the latter case, it is easy to see that the neighboring dust-particles move in general in the same sense, roughly tracing out the form of the common current which bears them along, whereas the most striking feature of the Brownian movements is the absolute independence of the displacement of neighboring particles, so near together that they pass by one another. The movements cannot be due to an effect of light because Gouy was able to vary the intensity of the illumination abruptly a thousand-fold, or to change the color considerably without modifying the phenomena. The nature of the particles does not appear to be of any importance, and it is therefore difficult not to believe that these particles reveal an internal agitation of the fluid, the better the smaller they are, much as a cork follows better than a large ship the movements of the waves of the sea. The distance between the two end-points as a particle zig- zags forward and back is usually defined as twice the amplitude. Svedberg 1 found that for different solutions the product of the amplitude, A, into the viscosity, 17, is a constant and that the amplitude divided by the time which the particles take in passing through that distance is also a constant. These results are in qualitative agreement with a theoretical formula deduced by Einstein. 2 If one knows the amplitude and the time, one can calculate the speed of the particles. For platinum particles with a diameter of 10-50/i/*, Svedberg found speeds of 200-400^ per second. Perrin 3 considers that errors of measurements vi- 1 Zeit. Elektrochemie, 12, 835, 909 (1906). 2 Drude's Ann., 17, 549 (1905); 19, 289, 371 (1906). 3 Brownian Movement and Molecular Reality, 22 (1910). SURFACE TENSION BROWNIAN MOVEMENTS 139 tiate these data. He says that the speed of a granule in suspen- sion cannot be measured accurately. Even though quantitative measurements may be inaccurate, it is desirable to have a general idea of the magnitude of the phe- nomenon. From Exner's 1 experiments it appears that particles with a diameter larger than 4/* show no perceptible Brownian movements. Particles smaller than O.lju begin to show lively movements, 2 while particles with diameters of about lO^Ai give apparent trajectories up to 20^. Perrin 3 came to the conclusion that the gas laws must apply to dilute emulsions. "Let us suppose that it is possible to obtain an emulsion, with the granules all identical, an emulsion which I shall call for shortness, uniform. It appeared to me, at first intuitively, that the granules of such an emulsion should dis- tribute themselves as a function of the height in the same manner as the molecules of a gas under the influence of gravity. Just as the air is more dense at sea-level than on a mountain-top, so the granules of an emulsion, whatever may be their initial dis- tribution, will attain a permanent state where the concentration will go on diminishing as a function of the height from the lower layers; and the law of rarefaction will be the same as for air." To test this hypothesis Perrin prepared a uniform suspension of purified gamboge in water by means of fractional centrifug- ing. The results confirmed the hypothesis. With gamboge particles 0.3ju in diameter a rise of 30/* was sufficient to lower the concentration to a half of its value, 30/x in the gamboge suspen- sion being equivalent to 6 kilometers in the air. 4 Satisfactory- results have also been obtained by Zangger 5 for drops of mercury, by Brillouin 6 for gamboge in glycerol solutions having a viscosity 160 times that of water, and by Iljin. 7 This cannot be true over any appreciable range because it would mean that the concentra- tion would drop to one-billionth in a rise of about a millimeter, which would mean that no colloidal solution would appear uni- 1 Sitzungsber. Akad. Wiss. Wien., 56 II, 116 (1867). 2 ZSIGMONDY: Kolloidchemie, 18 (1912). 3 Brownian Movement and Molecular Reality, 22 (1910). 4 PERRIN : Brownian Movement and Molecular Reality, 43 (1910). 6 Zeit. Kolloidchemie, 7, 216 (1911). 6 Ann. Chim. Phys. (8) 27, 412 (1912). 7 Zeit. phys. Chem., 83, 592 (1913). 140 APPLIED COLLOID CHEMISTRY form. It has been suggested by Burton that the concentration increases with increasing depth only up to about a certain value and then remains practically constant. Einstein 1 assumes that suspended particles will behave exactly like dissolved particles, exerting an osmotic pressure and diffusing through the solvent. 2 Instead of the van't Hoff expression that the osmotic pressure, P, of a dissolved substance is given by the equation PV = nRT Einstein deduces the equation PV ET N where n is now the number of suspended particles and N is Avogadro's constant or the actual number of molecules in a gram-molecule. Following this up he deduces for the rate of diffusion the equation r> /RT t_ : \ NT Sri, where D* is the displacement in a given direction, t is the time, R is the gas constant, T the absolute temperature, N is the Avo- gadro constant, r is the radius of the particles a*id 77 is the vis- cosity of the liquid. This has been confirmed quantitatively by Perrin who gets a mean value 3 of about 68 X 10 22 for N. In regard to this he says: 4 "Thus the molecular theory of the Brownian movement can be regarded as established experi- mentally, and it becomes very difficult to deny the objective reality of molecules. At the same time we see the law of gases, already applied by van't Hoff to dilute solutions, extended to uniform emulsions. The Brownian movement offers us, on a different scale, the faithful picture of the movements possessed, 1 Drude's Ann., 17, 549 (1905); 19, 280, 371 (1906). See also SMOLU- CHOWSKI: Ibid., 21, 756 (1906). 2 It must be remembered that Einstein has ignored all adsorption of the liquid by the suspended particles. He did the same thing when deducing a formula for the viscosity of colloidal solutions and came to the erroneous conclusion that the viscosity depends only on the volume and not on the size of the suspended particles. 3 PERRIN: Die Atome, 123 (1914). 4 PERRIN: Brownian Movement and Molecular Reality, 46 (1909). SURFACE TENSION BROWNTAN MOVEMENTS 141 for example, by the molecules of oxygen dissolved in the water of a lake, which, encountering one another only rarely, change their direction and speed by virtue of their impacts with the molecules of the solvent. "It may be interesting to observe that the largest of the granules, for which I have found the laws of perfect gases fol- lowed, are already visible in sunlight under a strong lens. They behave as the molecules of a perfect gas, of which the gram- molecule would weigh 200,000 tons. "I add lastly that all the measurements detailed in this para- graph have been made on dilute emulsions, which in the parts richest in granules only contain a thousandth part of resin, and where the osmotic pressure does not reach a thousand millionth of an atmosphere." If we substitute four times the amplitude for D x (D a) = 4A) in Einstein's formula, t will be the time required for the particle to make one complete oscillation. 1 Einstein's formula there- fore becomes ...^ TrRT t (4A) -N"3Si If we are considering particles of the same size in various liquids we may write the equation A 2 = k.t/rj if we are working at constant temperature. " Since, by the kinetic theory, the average kinetic energy of a particle is the same as the average kinetic energy of a molecule, and the molecules of all fluids at the same temperature have the same average kinetic energy, we should expect the average velocity of all the particles to be the same, provided their masses are the same, whatever the liquid in which they are moving." Einstein's formula therefore leads to the set of relations: Kr\ = const.; D x = \/T/5 an d D/\/t' = const. As has been pointed out, Svedberg had discovered independ- ently that the product of the amplitude into the viscosity was a constant at constant temperature. Exner 2 found that tl\e ratio of displacements at 20 and 71 is 1:1.6 while the ratio of the corresponding values of x/T/r? for water at these two tempera- 1 BURTON: The Physical Properties of Colloidal Solutions, 76 (1916). 2 BURTON: The Physical Properties of Colloidal Solutions, 77 (1916). 142 APPLIED COLLOID CHEMISTRY tures is 1:1.7. Seddig 1 has shown by means of a photographic method in conjunction with the microscope, that finely divided cinnabar in water obeys Einstein's formula, the greatest depar- ture from the theoretical value being 6 percent. Chaudesaigues, 2 working with an emulsion of gamboge, varied the viscosity of a solution by adding sugar. In this way he carried out observations on two solutions containing the same-sized particles at the same temperature, the viscosities of which were in the ratio 1 : 4 and found that the motion was twice as fast in the less viscous solution as it was in the solution of higher viscosity. Chaudesaigues observed the velocities of particles of gamboge with radii respectively 4.5 X 10~ 5 cm and 2.3 X 10~ 5 , practically a ratio of 2:1, and found that these velocities varied inversely as the square of the radii. He also observed the distances described by 50 grains, each of radius 2.13 X 10~ 5 cm during successive intervals of 30 seconds each and obtained the following results: on the average the particles moved 6.7, 9.3, 11.8 and 13.95 microns in 30, 60, 90 and 120 seconds. The square roots of the times are proportional to the numbers, 6.7, 9.46, 11.6 and 13.4, giving an ideal confirmation of this point. It was first pointed out 3 by Smoluchowski 4 that we should expect to have the Brownian movements in gases as well as in liquids and he quotes from Bodaszewski 5 and 0. Lehmann 6 as to the dancing movements executed by the particles of fumes of ammonium chloride, acids, phosphorus, etc. which they com- pared to the Brownian movements in liquids and interpreted as molecular movements. 7 The formulas of Einstein and Smolu- chowski are based on the Stokes law, which does not hold for gases under low pressure where the mean free path of the gas molecules 1 Phys. Zeit., 9, 465 (1908). 2 Comptes rendus, 147, 1044 (1908). 3 BURTON: The Physical Properties of Colloidal Solutions, 85 (1916). 4 Drude's Ann., 21, 756 (1906). 6 Kosmos, 7, 177 (1882); Beibl., 8, 488 (1883). 6 Molekularphysik, II, 5. 7 " The sun," says Daniel Culverwell, "discovers atoms, though they be invisible by candle light, and makes them dance naked in his beams." SURFACE TENSION BROWNIAN MOVEMENTS 143 is large relatively to the radius of the particles. Ehrenhaft 1 was the first to carry out direct measurements on the Brownian movement of particles in gases. He found that, as theory pre- dicts, there is a much livelier motion in gases than in liquids; at the same time the action of gravity in causing vertical descent of the particles is also much more apparent in gases than it is in liquids. For comparatively large particles in the smoke of cigars and cigarettes and in the fumes of ammonium chloride, he ob- served an undoubtedly zigzag Brownian movement, while with smaller particles obtained by a spark discharge between zinc, platinum, or silver electrodes, Ehrenhaft concluded that he has proved without a doubt the existence of a motion in gases completely analogous to the Brownian movement in liquids. It is more difficult to recognize this motion in gases on account of disturbing convection currents and the action of gravity. For the heavier particles, the velocity due to the force of gravity completely overshadows that due to the molecular shocks, while the opposite is the case with the very small particles. In Table XLI are given the velocities impressed by molecular shocks and by the force of gravitation, respectively, on silver particles of TABLE XLI. BROWNIAN MOVEMENTS IN GASES Radius of silver particles Velocity due to molecular shocks Velocity due to gravity IMM 63 . OM/SCC. 1 . 2ju/t/sec. SMM 28 . (V/sec. 30. (W /sec. 1(W 20.0 M /sec. 120 . 0/z/i/sec. 5(W 8 . 9M/sec. 3.0 M /sec. 10(W 6.3 M /sec. 12.0 M /sec. 50(W 2 . 8/i/sec. 300. M /sec. In 2.(V/sec. 1200.0 M /sec. various sizes (density 10.5) . These numbers show that when we reach particles having diameters of the order of the wave-length of light the two velocities do not differ materially. As we deal with smaller particles, the velocity induced by gravity soon becomes negligible, while with particles increasing above O.lju, the motion due to gravity soon dominates the situation. This is 1 Sitzungsber. Akad. Wiss. Wien., 116 Ho, 1139 (1907). 144 APPLIED COLLOID CHEMISTRY in keeping with what Ehrenhaft found. Particles, the linear dimensions of which were of the order of the size of the mean free path of gas molecules (0.1/x), and somewhat larger particles, fall in a zig-zag line, the velocity due to gravitation being greater than that due to molecular shocks. Particles, which are near the limit of visibility in the ultra-microscope (IMM), are in such lively molecular motion that the vertical gravitational velocity is completely masked. Ehrenhaft viewed silver particles that remained in lively motion in the air of his ultra-microscope cell for some thirty minutes. His measurements on the cigarette smoke particles gave a mean velocity of 25/*/sec. and on the smaller particles of silver, 46/-i/sec. Much interesting work has been done in this field by De Broglie. 1 Both Ehrenhaft and De Broglie have found that these silver particles suspended in air, are charged and, by measuring the velocity impressed on them by a known electric field, have come to the conclusion that the charge is that of one electron. The values that they find for this charge, 2 e, are respectively 4.6 X 10- 10 and 4.5 X lO" 10 electrostatic units. Wells and Gerke 3 have measured the velocity of smoke particles in an electrical field and have determined the size of the particles from the resulting data. It seems to be accepted pretty generally that the work of Perrin, Svedberg, and others has established the practical con- tinuity between suspended particles and dissolved substances; but this seems to be an overhasty conclusion. The molecular weight of a dispersed substance, as calculated from any of Ein- stein's formulas means something entirely different, if it means anything at all. We can determine the molecular weight of liquid benzene approximately from a measurement of the surface tension; but it is absurd to say that suspending fine drops of benzene in water causes the molecular weight of benzene to be equal to 200,000 tons. 4 We are talking about entirely different things in the two cases. What we mean is that dissolved ben- 1 Comptes rendus, 146, 624, 1010 (1908); 148, 1163, 1315 (1909). 2 See also MILLIKAN: Phil. Mag. (6) 19, 209 (1910). 3 Jour. Am. Chem. Soc., 41, 312 (1919). 4 Such a hypothetical gas would have a density roughly a million times that of metallic copper. SURFACE TENSION BROWNIAN MOVEMENTS 145 zene has a molecular weight of 78, and that liquid benzene sus- pended in water behaves or may behave as if it were a dissolved substance having a molecular weight of 200,000 tons, more or less. As a matter of fact the experiments bring out clearly the enor- mous difference between a solution and a suspension. The safe ground to take is : that the Brownian movements are due to the incessant movements of the molecules of the fluid; that the Brownian movements tend to make finely-divided, suspended particles distribute themselves uniformly throughout the liquid; that the uniform distribution is affected by the force of gravity as in the case of a gas ; and that the Brownian move- ments, though causing diffusion, give rise to no appreciable osmotic pressure. Very finely divided particles (less than 0.5/* for instance) will be kept in suspension indefinitely by the Brownian movements, so long as the particles remain finely divided. If, however, two or more particles agglomerate or coalesce, the force of gravity may cause the particles to settle to the bottom of the containing vessel. 10 CHAPTER V COALESCENCE COALESCENCE OF LIQUIDS If two small drops of a pure liquid are brought in contact, they run together and form a larger drop. The two small drops are therefore instable with respect to a larger drop. Conse- quently we should expect the small drop to distill over to, and condense on, a larger drop. This is equivalent to saying that the greater the curvature of the drop, the higher the vapor-pressure. This conclusion was drawn by Lord Kelvin, then William Thom- son, 1 many years ago. He reached his conclusion in a slightly different way. If we have a capillary tube dipping into a liquid, the whole being in a closed vessel containing only the liquid and its vapor, the raised liquid surface in the capillary must be in equilibrium with the main surface of the liquid if the system is all at the same temperature. The pressure of the vapor varies, however, at different levels in consequence of gravity. Con- sequently the concave surface of the liquid meniscus must have a lower vapor-pressure than the plane surface of the liquid. The converse must also be true that the convex surface of a drop must have a higher vapor-pressure than the plane surface of the liquid. This difference will be greater the greater the curvature of the surface or, in other words, the smaller the drop. The actual difference in vapor pressure between the highly curved surface in a capillary tube and the nearly plane surface in a wide vessel is very small. It amounts to one-thousandth of the vapor-pressure in a capillary tube in which water would rise thirteen meters. Lord Kelvin 2 considered that this theory, originally due to Laplace, is applicable to cases in which the radius of curvature is 1 Phil. Mag. (4) 42, 448 (1871). 2 Ibid. (4)42,451 (1871). 146 COALESCENCE 147 as small as 1.2ju; but he thinks that we are not entitled to push it much further. Consequently he believes that the formulas he has deduced are not applicable to the vapor pressures of moisture retained by such substances as cotton cloth and oatmeal at temperatures far above the dewpoint of the surrounding atmos- phere. He believes, however, that the difference is quantitative and not qualitative; that the adsorption of vapor by fibrous and cellular organic structures is a property of matter continuous with the adsorption of vapor into a capillary tube. The converse of the theory is also true that liquid will distill into a capillary tube closed at the lower end until the height of the meniscus above the level of the liquid is the same as the height to which the liquid would have risen in the capillary tube if the lower end had been open. Bacon 1 has done the experiment successfully with ether, though the rate of distillation is very low. The rate at which small drops coalesce depends on the mobility of the liquid. With viscous liquids such as cooling lavas or slags, we know that we get all sorts of phenomena which do not correspond to equilibrium relations, and it is probable that jellies are rarely in true equilibrium. COALESCENCE OF SOLIDS While it is perfectly familiar to all of us that two small drops of a pure liquid will coalesce readily to form a larger drop, all our ordinary experience is the other way in regard to solids and yet, theoretically, finely divided solids behave like finely divided liquids in many respects. A very finely divided powder must have a higher vapor-pressure, a lower melting-point, and a greater solubility 2 than the same substance in a more coarsely crystalline form. When yellow phosphorus is placed in a tube, the air removed, and the tube then sealed, it is said by Herman 2 and by J. Lawrence Smith 3 that crystals form in a short time on the walls of the tube and may grow to a diameter of 3-5 mm in the course of a month or six weeks. Crystals of iodine having a diameter of 2-3 mm were kept in the dark for eight years, 1 Phys. Rev., 20, 1 (1905). 2 OsTWALD: Zeit. phys. Chem., 36, 495 (1900). 3 Jour. Chem., Soc., 27, 869 (1874). 148 APPLIED COLLOID CHEMISTRY the temperature varying from to 24. At the end of that time von Jonstorff 1 found that the smaller crystals had mostly disap- peared and that the crystals were now about 4-5 mm in diameter. Hautef euille and Perrey 2 found that at a dull red heat hydrochloric acid gas at a pressure of three atmospheres causes amorphous alumina to change into corundum, titania into anatase and zir- conia into a crystalline form. The crystallization of the alumina is believed to be due to the successive formations and decomposi- tions of an aluminum chloride. Emden 3 has shown that ice grains as large as hazel nuts may be grown from snow slush with original small ice nuclei, kept in sealed vessels for several weeks at a constant temperature of 0. Pawlow 4 states that crystals of salol which were 2,u in diameter melted 1.1 lower than crystals 40/4 in diameter. Comparing a fine dust of less than 2/u diameter with crystals 0.5-2.0 mm diameter, differences of 4-7 were obtained with salol, antipyrin, and phenacetin; but it seems probable that part of these differ- ences is due to impurities and part to experimental error. Hulett 5 found practically no difference for the solubility of calcium sulphate in water until the radius of the crystals was less than 2/4. Changing from 2/4 to 0.3/4 caused an increase in solu- bility of nearly 20 percent. A coarsely crystalline barium sulphate (1.8/4) was soluble 2.29 mg per liter at 25. When ground in an agate mortar to O.l/i the solubility rose to 4.5 mg per liter. The corresponding figures with natural barytes were 2.38 mg and 6.18 mg. Red mercuric oxide has a solubility of 50 mg per liter at 25; when ground to a fine powder the color becomes yellow and the solubility increases to 150 mg per liter. Jones and Partington 6 made conductivity measurements on saturated gypsum solutions. They found no variation with size of crystals when the radius exceeded 50/i. This parallelism between small grains and small drops makes one wonder whether the apparent failure of solid particles to 1 Ber. deutsch. chem. Ges., 10, 866 (1877). 2 Comptes rendus, 110, 1028 (1890). 3 VON ENGELN: Am. Jour. Sci. (4) 40, 461 (1915). 4 Zeit. phys. Chem., 65, 1 (1908); 68, 366 (1909). 6 Ibid., 37,385 (1901). 6 Jour. Chem. Soc., 107 II, 1019 (1915). COALESCENCE 149 coalesce may not be due to experimental error. If we precipitate copper electrolytically on copper, we get a mass which does not differ materially in properties from cast copper having the same size of crystals. 1 The successive layers of copper coalesce to a uniform mass. On the other hand if we electroplate nickel on nickel without taking special precautions, we get a film of nickel which does not adhere. The reason for this is that the original nickel had an oxide film over the surface and the electro- lytically precipitated nickel never really came in contact with the nickel electrode. If the oxide film is removed, the plating of nickel on nickel 2 is as simple as the plating of copper on copper. A silver halide adheres well to a silver anode if the conditions are such that the halide is formed in actual contact with the metal. 3 If a plate were broken without chipping, the two pieces would fit together perfectly and should cohere, showing no signs of a break. This does not happen because of the film of condensed air which acts as an elastic cushion and keeps the two parts from actually being brought in contact. 4 A corollary from this is that if one were to break a plate in an absolute vacuum and were to bring the pieces together in the same vacuum, the plate should theoretically be as good as new. This is not likely to happen practically, because we cannot get an absolute vacuum. If we heat the fragments of any broken object, several things happen. At the higher temperature less gas is adsorbed and it is held less firmly. The solid is likely to become more malleable and it is therefore easier to make contact at a number of points. All chemists know that it is easy to make two glass rods unite at temperatures at which the glass is still very viscous. It may be urged that glass, at those temperatures, is unquestionably a liquid and not a solid. This criticism does not apply to plati- num. To weld platinum it is only necessary to put two clean pieces of platinum in contact, to heat them to a moderate tem- perature far below the melting-point of platinum, and to give them a light tap with a hammer. Since the object in these cases 1 BENNETT: Jour. Phys. Chem., 16, 294 (1912). 2 SNOWDON: Jour. Phys. Chem., 10, 134 (1906). 3 REEDY: Am. Jour. Sci. (4) 40, 407 (1915). See BREUER: Kitte and Klebstoffe, 23 (1907). 150 APPLIED COLLOID CHEMISTRY is to bring two like surfaces into intimate contact, it is easy to see why clean surfaces are essential when soldering or brazing metals, and why it is important to use a flux. When iron is treated with sodium amalgam as a preliminary to dry-gilding, 1 part of the effect of the sodium amalgam is to clean the surface of the iron. Cold sodium cuts readily while warm sodium sticks to the knife. 2 In some cases two surfaces will unite fairly readily. This is very noticeable with unvulcanized rubber from which it should follow that vulcanized rubber adsorbs air much more strongly. I do not know of any experiments along this line. Willows and Hatschek 3 point out that copper can be polished to such a degree that a cube of this metal will support eleven others merely placed in contact. This means that 1 cm 2 of surface carries a mass of copper 11 cm long and 1 cm square, a weight of about 98 grams. Polished plane surfaces of glass are apt to stick together. Usu- ally this is to be avoided; but Parker and Dalladay 4 have taken advantage of this property. Glass surfaces can be polished without difficulty so as to be so nearly plane or so nearly of the same curvature that they show less than one interference fringe per inch when placed together and examined by light reflected from the interfaces. These surfaces may be brought into optical contact so that the two pieces of glass act optically as if they were one block by applying a pressure of several pounds per square inch. The two pieces can, however, be separated easily by sudden heating or cooling, by a shearing force, or by the applica- tion of liquids which creep in between the surfaces. By selection of the proper temperatures and pressures it has been found pos- sible to weld two such pieces. It was found essential that the temperature at which union took place should not only be far below the melting-point of the glasses to be joined, but also below the annealing temperature, for even at that temperature a very slight strain would cause deformation of worked glass surfaces. At 60 below the annealing temperature it was found that two pieces of glass in optical contact would unite within an 1 KIRCHMAN: Jour. Chem. Soc., 26, 418 (1873). 2 MCDONALD: Chem. News, 97, 73 (1908). 3 Surface Tension and Surface Energy, 8 (1915). 4 Engineering, 103, 23; Jour. Franklin Inst., 183, 369 (1917). COALESCENCE 151 hour and that any slight deformation produced by the compres- sion could be remedied by a few strokes of the polishing tool. . Mr. Jerome Alexander has called my attention to the fact that freshly split pieces of mica can be made to coalesce by pressing them between the fingers. After they have been exposed to the air for five or ten minutes, it is impossible to make them unite in this way. This is undoubtedly due to the formation of a surface film, because Lord Rayleigh 1 found that freshly split surfaces of mica attract to themselves at once a coating of moisture which is replaced or supplemented in a few hours by a layer of grease. The fresh surface conducts electricity, while an old surface, in which the moisture has presumably been replaced by grease, insulates well. I have been told that some of the electrical companies make mica blocks by compressing scrap mica; but I do not know any of the details of the process. Horn waste and tortoise-shell waste may be worked up into solid masses by high pressure and moderate heat. 2 While it is not possible as a usual thing to mend a broken object by pressing the two pieces together at ordinary temperatures, Spring 3 has shown that powders may be welded into massive blocks by the use of sufficiently high pressures. This means that the pressure has been sufficient to squeeze out the air films coating the powders. Since solids will coalesce when heated and subjected to pres- sure, there is no reason why some solid powders should not coa- lesce to some extent under the influence of gravity when heated. This is known as sintering or sometimes as fritting. The two terms are not distinguished as sharply as they might be. 4 Sinter- ing should be used when a powder coalesces without melting or with but slight fusion, so that the product is still granular and more or less porous. Fritting should be used when the fusion is quite marked and the product is distinctly glassy. The forma- 1 Scientific Papers, 3, 523 (1902). 2 ROLLER: The Utilization of Waste Products, 62 (1915). 3 Bull. Acad. roy. belg. (2) 49, 323 (1880); Ann. Chim. Phys. (5) 22, 170 (1881); Ber. deutsch. chem. Ges. 16, 395 (1882); Bull. Soc. chim., Paris, (2) 40, 520 (1883); FRIEDEL: Ibid. (2) 39, 626; 40, 526 (1883); JANNETZ, Neeland Clermont: Ibid. (2) 40, 51 (1883); Tammann: Zeit. Elektrochemie, 15, 447 (1909). 4 Percy's Metallurgy: Fuel, 46, 280 (1875). 152 APPLIED COLLOID CHEMISTRY tion of bricks and pottery involves sintering; the materials for glazes are fritted. There has been some question whether sinter- ing could take place without at least incipient fusion. Lucas 1 maintained that it was a property of amorphous substances only, in which case it would be nothing more than the slow and partial coalescence of very viscous drops. Day and Allen 2 found that powdered glasses sinter slowly or rapidly several hundred degrees below the melting temperature of crystals of the same composition. The first traces of sintering appear earlier the finer the powder and the slower the heating. There is no doubt that the sintering of powdered glasses is due to flow in a super- cooled liquid and that it is a phenomenon of viscosity and surface tension. Powdered crystalline feldspars do not sinter readily below their melting temperature and, at first, Day and Allen were inclined to suspect that pure, dry, stable crystals would not sinter at all. Their final experiments, however, showed that the purest crystalline feldspar which they could prepare still sintered, though very slowly. Kohlschutter 3 found that porous silver obtained by reducing silver salts in different ways, sintered at temperatures from 200. Instead of concluding that a crystal- line substance can sinter, he drew the surprising conclusion that Lucas was right and that therefore the degree of sintering of the silver was a measure of the amount of amorphous silver present, even though he points out the relation between the porosity and the sintering. Other things being equal the contraction or sintering is greater the higher the temperature, and consequently the higher the temperature the denser the solid. Bricks are less porous the higher the temperature of burning. Of course if the temperature is reached at which vitrification takes places a new phenomenon occurs. Magnesia is more dense the higher it is heated 4 and so is lime. 5 While it is unquestionably true that sintering will take place less readily with a pure crystalline substance than with an amorphous one or with one containing impurities which 1 Zeit. phys. Chem., 62, 327 (1905). 2 Carnegie Inst. Pub., 31, 59 (1905); Zeit. phys. Chem., 64, 39 (1906). 3 Liebig's Ann., 398, 37 (1913); REEDY: Am. Jour. Sci. (4) 40, 407 (1915). 4 DITTE: Comptes rendus, 73, 270 (1861); MOISSAN: Ibid., 118, 506 (1894). 6 WEIGHT: Zeit. anorg. Chem., 68, 397 (1910). COALESCENCE 153 lower the melting-point 1 and cause a portion of the mass to soften long before the whole mass is fused, this does not prevent sintering occurring with perfectly pure crystalline substances. The technical advantage of impurities is shown in the case of magnesia. 2 The crystalline Styrian magnesite is less pure than the amorphous Grecian magnesite, containing more iron. It therefore sinters more readily and at a lower temperature. It is often preferred for this reason to the purer Grecian magnesite, which must either be burned at a higher temperature or must have some bonding material added which will cause sintering at lower temperatures. Of course, the material which sinters at a low temperature will not stand up at so high a temperature as a purer substance which has been shrunk in an electric furnace at a higher temperature. The sintering of tungsten 3 and of tantalum 4 powder is a tech- nical process carried on far below the melting-point of tungsten, and not depending, so far as we know, on the presence of impuri- ties, though these may be present. Another interesting case of sintering occurs in the gold assay. 5 After parting, the cornets are of a dull brown color and very friable. When annealed at a bright-red heat they sinter very much and become a pure gold- yellow color. This has been studied in detail by Hanriot 6 who took an alloy containing four parts of silver to one of gold and dissolved the silver with nitric acid. The gold was left in a spongy brown form which decreased rapidly in bulk on heating and became gold color. Wohler 7 has shown that the glowing of heated chromic oxide jellies is the result of rapid sintering and not of the formation of an allotropic modification. Soot becomes sandy if heated too long or too hot. The density of red phosphorus depends on the heat treatment. Many pre- cipitates sinter or agglomerate on standing, though the change is not usually called that. The agglomeration is often accompanied 1 ENDELL: Silikat-Zeitschrift, 2, 1, 25 (1914). *HAVARD: Refractories and Furnaces, 89 (1912). 8 RUFF: Zeit. angew. Chem., 25, 1889 (1912). 4 SIEMENS and HALSKE: Jour. Soc. Chem. Ind., 23, 206 (1906). 8 Percy's Metallurgy: Silver and Gold, 268 (1880). 6 Comptes rendus, 151, 1355 (1910); 152, 215 (1911); cf. PRAT: Ibid., 70, 841 (1870). 7 Zeit. Kolloidchemie, 11, 241 (1912). 154 APPLIED COLLOID CHEMISTRY by increased resistance to chemical action. Strongly heated lime reacts very slowly with carbon dioxide. 1 After short ignition, silica dissolves in a boiling solution of potassium or sodium carbonate, but does not if ignited for a long time. Freshly precipitated arsenic sulphide gives off much hydrogen sulphide when boiled with water; but does not after being heated- at 125. Pulverulent iron burns in the air while massive iron does not. The action of alkali on hydroxides, and of gelatine and ammonia or potassium bromide on silver bromide is de- creased enormously if the precipitates are allowed to stand over night. Holmes and Fall 3 have made a quantitative study of the change in peptizability of ferric arsenate, so-called, with time of standing. A plastic substance is one which can be moulded by pressure. 4 This connotes that any break due to the deformation is self- healing. A mobile liquid is not plastic, because it will not retain its shape. With increasing viscosity a liquid becomes plastic. Molasses candy is plastic until it crystallizes. Semifused glass is plastic. The time factor may be important. Sealing-wax is brittle if the pressure is applied rapidly; but is plastic if the rate of deformation is low. Since metals and other solids coalesce under high pressure, there is no reason why they should not be plastic under the same conditions, as indeed they are. We make use of this property when squirting metals into rods or pipes. Under high pressure ice is also plastic. It has been claimed by Spring 5 and by Kahlbaum 6 that solids become almost like mobile fluids under high pressure; but it seems to have been proved conclusively by Spezia 7 that this is not true when the pressure is uniform and that the apparent fluidity occurs only when there is a shearing force. The question of unequal pressures has been 1 RAOULT: Comptes rendus, 92, 189 (1881); SOSMAN, HOSTETTER and MERWIN: Jour. Washington Acad. Sci., 5, 563 (1915). S RAMMEL,SBERG: Jour. Chem. Soc., 26, 242 (1873); cf. LAUFER: Ber. deutsch. chem. Ges., 11, 60, 935 (1878). s Jour. Am. Chem. Soc., 41, 713 (1919). 4 DUFF: A Text-Book of Physics, 4th Ed., 119 (1916). 6 Bull. Acad. Soc. belg. (2) 49, 364 (1880). 6 Verh. Naturforscherges. Basel, 15, 14 (1903). 7 Atti. Accad. Sci. Torino, 46, 525 (1911); 46, 682 (1912). COALESCENCE 155 discussed in some detail by Johnston. 1 Iron filings in presence of a magnet might be considered a special case of plasticity, though not one of any great interest. Another special case of partial plasticity is the mineral itacolumite 2 which is said to have something akin to a series of ball-and-socket joints. Each ball and socket does not admit of a great play, and is not smooth and perfect like that of the joint of a limb. The stone is built up of grains and masses of grains, which are loosely coherent and form irregular cavities in which are engaged projecting parts of other grains or masses of grains of sand, susceptible of a slight motion in the cavity, in some cases in one direction, and in others in several or in all directions. This freedom of motion is of the true quality of a ball-and-socket joint. Perhaps it would be better to call this mineral flexible rather than plastic. Between the extreme cases of a solid such as a metal and a liquid such as glass or molasses candy, we have the intermediate case of putty which is whiting and oil, or a solid with a liquid film around it. If a liquid is adsorbed strongly by a solid, thin films of the liquid will hold the solid particles together, while still permitting them to move relatively to one another. The tendency of liquid surfaces to coalesce causes any break to heal. A liquid film may therefore act as a bond for solids and may make the mass plastic. Oil is used as a binder in roads and to make putty out of calcium carbonate. Everybody knows that dry sand cannot be moulded whereas wet sand can. Fondant consists of very small crystals of sugar made plastic by a syrupy film of invert sugar and water. Plastic dynamite is made by adding three parts of nitroglycerine 3 to one part of diatomaceous earth. Bingham 4 has made a preliminary study of the concentration at which plasticity begins or ends. If a finely powdered solid is added to a liquid, the viscosity of the liquid is increased or the fluidity, which is the reciprocal of the viscosity, is decreased. The fluidity was calculated from the rate of flow of the liquid through a viscosimeter. At temperatures between 25 and 60 1 Jour. Am. Chem. Soc., 34, 788 (1912); JOHNSTON and ADAMS: Am. Jour. Sci. (4) 40, 351 (1915). 2 WETHERILL: Chem. News, 22, 266 (1870). 3 THORP: Outlines of Industrial Chemistry, 3d Ed., 482 (1916). 4 Am. Chem. Jour., 46, 278 (1911); Jour. Franklin Inst., 181, 845 (1916). 156 APPLIED COLLOID CHEMISTRY the fluidity dropped to zero at the same concentration, so the concentration for zero fluidity is independent of the temperature over the range studied. With infusorial earth in water, zero fluidity was reached at a volume concentration of about 13 percent solids; with China clay at about 4 percent; with the graphite used in Acheson's aquadag zero fluidity was reached at a volume concentration of about 5.5 percent graphite; and with an unspecified clay at about 19.5 percent clay. With infusorial earth in alcohol, zero fluidity was reached at a volume concentra- tion of about 12 percent solids. The mixtures having zero fluidity are not stiff and will not maintain their shape. At higher concentrations there is a change from viscous flow to plastic flow. The distinction made by Bingham is that with viscous flow any shearing force no matter how small will produce permanent deformation, whereas, in the case of plastic flow, it is necessary to use a shearing force of finite magnitude in order to produce a permanent deformation. It seems reasonable to assume that we reach zero fluidity when liquid enough is added to the solid to begin to scatter the particles, in other words, when about enough liquid is added to fill the voids. This is true in one case studied by Bingham. The clay referred to contained 81.6 percent voids and required 80.5 volume percent of water to bring it to zero fluidity. The two values would only be identical in case the volume of the adsorbed water just equalled the volume of the adsorbed air. Of course this very important generalization of Bingham ; s must be tested in more cases before it can be considered as definitely established; but it is so obvious, after it has been pointed out, that it must be approximately true. This discovery of Bingham's may be of distinct importance in the paint industry. The oil requirement for a given pigment is a very arbitrary amount and experts often differ widely in their values. Reproducible figures could be obtained if in each case there was determined the amount of oil necessary to give zero fluidity. Since this is more oil than painters wish to use, it might be advisable to adopt as the standard, some definite frac- tion of the amount of oil necessary to produce zero fluidity. Emley 1 has made some determinations on the viscosity of lime pastes. Judging from the diagram given, the viscosity 1 Trans. Am. Ceramic Soc., 15, 401 (1913). COALESCENCE 157 would become infinite or the fluidity zero when there was about 50 percent of hydrated lime in the paste. Since Emley used the Clark viscosimeter, one does not know to what extent his results are comparable with those of Bingham. Since the concentration producing zero fluidity apparently depends on the volume of the voids, it is necessary to consider approximately what this may be. If equal spheres are taken and are piled so that each sphere is directly on top of another minimum or cubical piling the voids constitute theoretically about 48 percent of the whole, irrespective of the size of the spheres. If the spheres are piled so that each one rests equally on the four in the layer below maximum or hexagonal piling the theoretical percentage of voids is nearly 26, again irrespective of the size of the spheres. With solids of different sizes and different shapes, the voids may theoretically be made almost anything. If the solids have condensed air films round them so that they do not come actually in contact, the percentage of voids will go up with decreasing size of the solids. As a matter of fact, we have seen that the voids in carbon black may run up to 95 percent or thereabouts. If the air-adsorption were negli- gible and if the particles were cubes of the same size and arranged in an orderly fashion the voids would be practically zero. As a matter of fact round-grained water-worn sands have 2-5 percent less voids than corresponding sharp grains of sand 1 because the former pack better. The difference between the theoretical values and the actual values as found under certain experimental conditions is shown by the statement that large enough masses of equal-sized marbles cannot be poured or tamped into a large vessel so as to give less than 44 percent voids. Undoubtedly these would have jarred down in time so as to occupy a lesser volume; because this result comes very close to that required for cubical piling. The effect of moisture on sand is also unexpected at first. 2 Moist sand occupies more space and weighs less per cubic foot than dry sand. The explanation of this apparent anomaly is that a film of water coats each grain of sand and separates it from the adjacent ones. Since fine sand has a large number of grains and consequently more surface area, addition 1 TAYLOR and THOMPSON: Plain and Reinforced Concrete, 174 (1912). 2 TAYLOR and THOMPSON: Plain and Reinforced Concrete, 176 (1912). 158 APPLIED COLLOID CHEMISTRY of water to it causes a greater increase in bulk than in the case of coarse sand. If a small quantity of water is poured into a vessel containing dry sand, the friction of the particles keeps the volume from increasing; but the bulk increases if the moistened sand is dumped out and then turned back into the vessel with a shovel or trowel. The percentage of water by weight which will give the greatest volume corresponding to the largest percent- age of voids varies with different sands from 5 to 8 percent. Since the densest mixture occurs with particles of different sizes, it is important, in making concrete, to grade the materials so as to get the maximum density. 1 On the other hand for a sand filter the grains of sand should be of approximately the same size, 2 as otherwise the filter clogs too readily. The question has also been raised but not answered, as to what are the proper sizes for the balls in the ball mill. Since there should probably be more of the small balls than the large ones, the proper ratio cannot be obtained if only large balls are added as the others wear down. 3 A number of curious things in regard to wetted sand have been pointed out by Osborne Reynolds. 4 If we have sand in maxi- mum piling in a rubber bag and enough water to form a thin layer over the surface, squeezing the sand will necessarily cause it to expand or dilate, and the water will be sucked down into the dilated mass. We are familiar with this experiment in another form. The wet sand below high water mark is in the position of maximum piling. When one walks over this sand, the sand whitens momentarily and appears to dry around the foot. The sand is full of water, the surface of which is kept up to that of the sand by capillary attraction. The pressure of the foot dilates the sand and since the water does not suck up rapidly enough, the water level is lower in the dilated sand than before and conse- quently the sand appears dryer until more water has been brought up from below by capillary action. On raising the foot, the sand in and around the foot-print usually becomes wet momen- tarily, because the sand contracts when the shearing pressure is taken off and the excess of water is left momentarily on the surf ace. 1 UPTON: Materials of Construction, 304 (1916). 2 MENDELEEFF: Jour. Chem. Soc., 48, 114; LUEGER: Ibid. 48, 198 (1885). 3 Chem. Met. Eng., 14, 285, 292 (1916). * Phil. Mag. (5) 20, 469 (1885); Scientific Papers, 2, 217 (1901). COALESCENCE 159 In a filter press the solid to be filtered tends to go into cubical pil- ing because that lets the water through most readily. If the crys- tals are displaced so that they pass into hexagonal piling, water will be set free. Hatschek 1 says in regard to this that it is a perfectly familiar fact that apparently dry cakes of calcium sulphate, for instance, which break with a clean fracture can be reduced to an almost liquid mass by mere shaking It seems probable that in quicksands the sand is in a state of cubical piling, making it possible for a man to sink in it and raise the water without raising the sand. To show some of the phenomena due to dilatancy Osborne Reynolds took a thin, transparent, india-rubber balloon, filled it with about six pints of sand and water, and closed it so that it contained no air. There was more than enough water to fill the voids when the sand was at maximum density but not enough to fill them when the sand was at minimum density. When this rubber sack was laid on the table, the elasticity of the rubber gave it a rounded shape. The sand settled into the condition of maximum piling and there was consequently a layer of water above the surface of the sand. The bag can be squeezed and altered in shape as though it had no firmness, so long as the sand does not rise to the surface of the water. If the bag is placed between two vertical boards and squeezed, it changes shape at first apparently without resistance, but soon comes to a dead stop. Turning it on its side and laying a 56 pound weight on it produced no further change in shape; but on removing the weight the bag goes back to its almost rounded shape. If the bag is shaken slightly while it is being squeezed, the sand keeps at its densest and the bag can be pressed out until it is a broad flat plate. The bag is soft so long as it is squeezed; but if the pressure is removed, the elasticity of the rubber tends to draw the bag back to its rounded form. This dilates the sand and causes the excess of water to be sucked into the pores. So soon as there is no excess of water, no further change takes place and the bag remains a flat cake with peculiar properties. It yields at once to pressure on its sides, such pressure having nothing but the elasticity of the bag to overcome 1 because pressure in that direction causes the sand to contract. It is perfectly rigid, 1 Jour. Soc. Chem. Ind., 21, 540 (1908). 160 APPLIED COLLOID CHEMISTRY however, to radial pressures because such pressures tend to dilate the sand still farther. When placed on its edge, it will hold up a hundred pounds. If, however, it is pressed sufficiently on the sides while holding up such a weight, all the strength vanishes and it is again a rounded bag of loose sand and water. By shaking the bag into a mould, it can be made to take any shape; then by drawing off the excess of water and closing the bag, the sand becomes perfectly rigid and will not change its shape unless the bag is torn, Since the sand is at its maximum density, no amount of shaking will produce a change. In this way bricks can be made of sand or fine shot with water in the thinnest india- rubber envelope and they will withstand as much pressure as an ordinary brick without change of shape. The clay workers use the word plasticity in a special sense to denote that a clay is plastic and will burn to a coherent mass. From this point of view wet sand is not plastic because it falls to pieces when dried. To meet this definition we must have in the water some bonding material presumably in a gelatinous form. The film of water about the clay particles is usually assumed to be 50^ in thickness. 1 There is no difficulty in accounting for a gelatinous film. 2 Cushman 3 has shown that the cementing power of rock powders is due to the formation of gela- tinous silica, ferric, oxide, etc. If we have a gelatinous film which is adsorbed strongly by the solid particles and which can take up and lose water, we shall have plasticity. In the case of clay the gelatinous material may be hydrous alumina, hydrous silica, or some intermediate composition. The difficulty, however, is to account for its remaining plastic after moderate drying and for its readsorbing water to form a gelatinous mass. This is probably due to the presence of some salt; but we do not know what salt and consequently we cannot tell why one clay is plastic and another not, and we do not know what we should add to the non-plastic clay to make it plastic. People have never studied the problem from a rational point of view. From this study of coalescence we conclude that liquid drops 1 GROUT and POPPE: Trans. Am. Ceramic Soc., 14, 73 (1912); DAVIS: Ibid., 16, 19 (1914). 2 QuiNCKE: Drude's Ann., 7, 74 (1902). 3 Jour. Am. Chem. Soc., 25, 451 (1903). COALESCENCE ' 161 or solid grains will coalesce or agglomerate if brought into actual contact. To keep them from doing so we must have a surface film of some sort. Since this coalescence or agglomeration takes place spontaneously, work must be done on the liquid or solid to disintegrate it. It is easy to see that a film might keep drops separated and yet not break up a large drop into a lot of smaller ones. Dirty mercury will remain indefinitely as a single liquid mass so far as we know. If we disintegrate the mass by shaking, a film will form over each drop and will keep the drops from coa- lescing. This "flouring" of mercury, as it is called, is a distinct source of trouble when extracting gold by the amalgamation process. In most cases the film is oxide, chloride, or grease, though it may consist of oxides of other metals if one starts with an amalgam . Anything that will dissolve the surface film will cause the drops of mercury to run together. Treatment with potassium cyanide is therefore very effective. Emulsions of oil in water are obtained if some substance like sodium oleate is present to form a film over the drops of oil, thus keeping them from coales- cing. 1 It is easy to see that the work necessary to break up a liquid or solid mass into drops or grains will be less if we add some- thing to the solution which will lower the surface tension of the liquid or solid. In some cases water alone will cause the disinte- gration. With gelatine we get reversible disintegration at very moderate temperatures, say 30; but cellulose requires a very much higher temperature. We do not know at all to what extent this matter of disintegration depends upon the structure; but a single illustration from the field of true solutions will show the possibility of such a factor. If we place a crystal of rock candy in cold water the crystal dissolves so as to keep its shape approxi- mately. It grows smaller but does not fall to pieces. If we put a lump of sugar into cold water, the smaller crystals which act as binding material dissolve first and the lump disintegrates to a mass of grains. TYPES OF PRECIPITATES Sufficiently finely divided particles will be kept in suspension in a liquid indefinitely by the Brownian movements, provided coalescence and the resulting agglomeration are prevented. 1 Cf. DONNAN and POTTS: Zeit. Kolloidchemie, 7, 208 (1910). 11 162 APPLIED COLLOID CHEMISTRY Such a colloidal solution is called a sol. One may use the words hydrosol, alcosol, etc., to show that the dispersing liquid is water, alcohol, etc. When the colloidal particles agglomerate and precipitate, the precipitate is called a gel, or a hydrogel in case the dispersing liquid is water. The word gel is also used to include a jelly, which differs from a gelatinous precipitate in that there is no supernatant liquid when first formed. It is really the general term for a solid or semi-solid colloid. The change from a gel to a sol is usually called peptization, a term which we owe to Graham. The reverse change from sol to gel was called pectization by Graham; but this term has never become popular. A stable colloidal solution will consist of very fine particles kept from agglomerating by a protecting film 1 of some sort, electrifica- tion being equivalent to a film. 2 Small particles 3 may be ob- tained either by condensation from vapor or solution (usually called precipitation) or by disintegration of larger masses. The disintegration is often called peptization especially when done chemically, as opposed to mechanical or electrical disintegration. People have distinguished eight types of precipitates: crystalline; colloidal or non-settling; flocculent; curdy; sandy; gelatinous; jelly-like; fluid. A crystalline precipitate is one that is distinctly crystalline, and settles rapidly, as when KOH is neutralized by H 2 SO 4 or when moderately concentrated solutions of lead nitrate and potassium chromate are mixed. A non-settling precipitate may be anhydrous or hydrous; there is little or no agglomeration. A typical case is the precipitation of lead chromate from very dilute solutions of lead nitrate and potassium chromate. 4 A flocculent precipitate is obtained when a non-settling, non-gela- tinous precipitate agglomerates to a bulky mass. It is easy to obtain a flocculent precipitate of silver chloride from suitable solutions of silver nitrate and potassium chloride. Haber and Oordt 5 call the precipitate of beryllium hydroxide flocculent; but I am not certain whether they are limiting the term sufficiently sharply. A curdy precipitate is obtained when a flocculent 1 Cf. HOBER: Physikalische Chemie der Zelle und der Gewebe, 233 (1914). 2 Cf. HARDY: Proc. Roy. Soc., 86A, 601 (1912). 3 FREUNDLICH: Kapillarchemie, 527 (1909). 4 FREE: Jour. Phys. Chem., 13, 114 (1909). 5 Zeit. anorg. Chem., 38, 377 (1904). COALESCENCE 163 precipitate agglomerates to a fairly dense mass. The typical case is, of course, the precipitation of caseine by acids, but a similar precipitate can be obtained with AgNO 3 and HC1. There is no sharp dividing line between a flocculent and a curdy precipitate. A sandy precipitate is obtained when a flocculent or a curdy precipitate agglomerates to a dense mass. The alumina from hot sodium aluminate in the Bayer process is a sandy precipitate. 1 A granular or sandy alumina is said to be precipi- tated by ammonia 2 at 66. A gelatinous precipitate is always viscous; it involves agglomeration to a bulky form. Hydrous alumina and hydrous ferric oxide, when precipitated cold, are typical gelatinous precipitates. When a sparingly soluble salt is precipitated from very concentrated solutions we apparently always get a gelatinous precipitate. Thus gelatinous barium sul- phate is obtained by mixing concentrated solutions of barium sulphocyanate and manganese sulphate, 3 and gelatinous calcium carbonate by mixing concentrated solutions of calcium chloride and sodium carbonate. The difference between a gelatinous precipitate and a jelly is that the liquid phase is visible as such in the case of the gelatinous precipitate and is not in the case of the jelly. A jelly is always hydrous, viscous, and bulky; it may consist of a number of coalescing drops or rods, in which case both phases are continuous; or it may consist of coalescing films, in which case we have a honeycomb structure, with water as the internal phase. A typical jelly can be obtained with sodium silicate and hydrochloric acid. A fluid precipitate is one which is too mobile to be classed as a gelatinous precipitate. Sodium phenolate or sodium oleate with an excess of HC1 gives a fluid precipitate. The mean mass of the crystals in a crystalline precipitate is given by the total weight of the precipitate divided by the number of crystals. Experimentally we find that the number of nuclei increases less rapidly than the concentration at first and more rapidly later. When very dilute solutions are mixed, the number of nuclei is so large relatively to the weight of the precipitate that the particles themselves are very small. With increasing 1 Russ: Zeit. anorg. Chem., 41, 216 (1904). 2 VON WEIMAEN: Griindziige der Dispersoidchemie, 67 (1911). 3 TAYLOR: Chem. News, 103, 169 (1911). 164 APPLIED COLLOID CHEMISTRY concentration we get increasing size of crystals for a while and then decreasing size, because the number of nuclei then increases more rapidly than the concentration. At very high concentra- tions a gelatinous precipitate is always obtained, as has been pointed out. It is possible that we have here the appearance of a viscous and instable second liquid layer; but this has not yet been proved. According to Clowes a jelly is formed when a viscous phase separates slowly in presence of a substance which tends to keep it dispersed. 1 To obtain large crystals one must have a solution so slightly supersaturated that no nuclei form spontaneously and that no branched crystals form in presence of a single crystal. 2 Von Weimarn 3 cites the case of a spring in which crystals of barium sulphate had grown 5 cm long ! Geolo- gists have prepared certain substances in distinct crystals by arranging for the slow diffusion of dilute solutions of two salts which form the desired substance by metathetical reaction. Johnston 4 has improved on the technique by allowing the two solutions to diffuse into a large vessel containing water, causing a further dilution. Dreaper 5 obtained distinct crystals by letting solutions diffuse through sand, the rate of diffusion being rela- tively slow on account of the capillary spaces. Holmes 6 ob- tained crystals of silver bichromate in flat needles one cm long by this method, substituting aluminum powder for sand. He filled a small test-tube full of N/10 potassium iodide, covered the mouth of the tube with gold-beater's skin, and immersed the tube in a small beaker containing a saturated lead acetate solu- tion. At once an almost amorphous precipitate of lead iodide appeared on the under side of the membrane and in less than a minute crystals of lead iodide fell in a gleaming shower to the bottom of the test-tube. If the same solutions are mixed in a *Cf. HARDY: Jour. Physiology, 24, 167 (1899); Proc. Roy. Soc., 66, 95 (1900); 87 A, 29 (1912); HOLMES and RINDFUSZ: Jour. Am. Chem. Soc., 38, 1970 (1916); HOLMES and FALL: Ibid., 41, 763 (1919). 2 Cf. HOPKINS: Am. Chem. Jour., 25, 413 (1907); BOWMAN: Jour. Soc. Chem. Ind., 26, 143 (1906); HOSTETTER: Jour. Washington Acad. Sci., 9, 85 (1919); MOORE: Jour. Am. Chem. Soc., 41, 1060 (1919). 3 Griindziige der Dispersoidchemie, 55 (1911). 4 Jour. Am. Chem. Soc., 36, 16 (1914). 6 Jour. Soc. Chem. Ind., 32, 678 (1913) 6 Jour. Phys. Chem., 21, 709 (1917). COALESCENCE 165 test-tube without using a membrane, a yellow powder is obtained and no easily recognized crystals. This method as developed by Holmes seems the most promising of any because it appears to give results with more concentrated solutions and would pre- sumably work even better with more dilute solutions. Excessive supersaturation and consequently the formation of very fine precipitates can be obtained by using a solvent in which the resulting salt is practically insoluble. 1 It is easier to avoid supersaturation when the substance is more soluble. Thus one can get crystals of silver chloride without trouble from an ammoniacal solution. 2 The effect of othefr substances in the solu- tion depends on their effect on the supersaturation and whether they are adsorbed by the precipitate. If a precipitate forms in the presence of a substance which is adsorbed strongly by it, the adsorption will prevent the growth of the crystals and the precipitate will therefore come down in a finer state of subdivision. A typical case is the precipitation of silver bromide in presence of gelatine. The addition agents in electrolytic refining are substances which move to the cathode under the influence of the current and which are adsorbed strongly by the metal as it comes out. Barium sulphate precipi- tates in a finely divided form in presence of barium chloride 3 or potassium sulphate, both of which it adsorbs, and more coarsely crystalline in presence of sulphuric acid or hydrochloric acid. 4 Barium sulphate shows selective adsorption for its own ions, which accounts for the behavior of barium chloride 5 and potas- sium sulphate. The peptizing action of sulphate ions in sulphuric acid is cut down by the tendency of hydrogen ion to be adsorbed strongly. The precipitate is therefore more coarsely crystalline. The solvent action of hydrochloric acid is a factor in making the barium sulphate precipitate more coarsely crystalline. 6 The question has been raised whether the finest precipitates are crystalline or amorphous. Von Weimarn 7 calls them all crystal- 1 BYK: Zeit. Elektrochemie, 13, 38 (1907). 2 REINDERS: Zeit. phys. Chem., 77, 696 (1911). 3 FOULK: Jour. Am. Chem. Soc., 18, 793 (1896). * MAR: Jour. Chem. Soc., 60, 1137 (1891). 6 WEISER: Jour. Phys. Chem., 21, 314 (1917). 6 See also OSBORNE: Jour. Phys. Chem., 17, 629 (1913). 7 Zeit. Kolloidchemie, 3, 166 (1908), 166 APPLIED COLLOID CHEMISTRY line; but he considers that liquids and gases are crystalline, which throws him out of court, because he is not using the word in the same sense as other people. A supercooled liquid like glass is amorphous in the ordinary sense of the term. Whether a precipitate is amorphous or not depends upon its being a super- cooled liquid or not. Von Weimarn is undoubtedly right in saying that many precipitates are really crystalline, though they are classed as amorphous; but there is no question that there are amorphous precipitates. In many cases there is a gradual change from an amorphous precipitate to a crystalline one on standing 1 and still more on heating. 2 At higher temperatures precipitates are less hydrous and tend to be more coarsely crystalline. There are no data in regard to a substance like sodium sulphate which is less soluble at 80 than at 40. When barium sulphate is precipitated hot it does not run through the filter so readily as when precipitated cold. Other conditions being the same, a non-settling silver 3 is obtained by reduction with hydrazine at 16 and pulverulent silver by reduction at 80 to 90. THEORY OF PEPTIZATION We find experimentally that certain solutions will peptize or disintegrate certain precipitates so that we get colloidal solutions. This is always the result of adsorption. Theoretically there are three possibilities. If an adsorbed film has a low surface tension on the water side and a high one on the other side, it will tend to scrunch up and to peptize the solid as internal phase. If the reverse is the case, the solid will tend to form the external phase. If the two surface tensions are practically equal, neither will prevail. Some adsorbed substances seem to have practically no peptizing action; 4 but these cases should be studied again in the light of what we now know about colloid chemistry. Tenta- 1 WATSON: Jour. Chem. Soc., 60,875(1891); GILBERT: Jour. Phys. Chem., 18, 86 (1914). 2 MOURET: Comptes rendus, 123, 54 (1896); DOELTER: Zeit. Kolloid- chemie, 7, 29, 86 (1910); GALPIN: Trans. Am. Ceramic Soc., 14, 337 (1912). 3 GUTBIER: Zeit. Kolloidchemie, 4, 308 (1909). 4 DOYLE: Jour. Phys. Chem., 17, 390 (1913). COALESCENCE 167 tive theories of peptization have been discussed by Lottermoser, l Jordis, 2 Mecklenburg, 3 and von Weimarn. 4 If we adopt Freund- lich's view that adsorption always lowers the surface tension, a theory of peptization follows at once. We may have peptization by a liquid, by a non-electrolyte, by an adsorbed ion, by a salt, or by a peptized colloid. When a liquid is adsorbed by a solid, it will tend to peptize it and in some cases will do so. Water peptizes tannin, amyl acetate peptizes pyroxylin, and fused baths peptize metals. At higher temperatures the peptizing action increases, and we may get glass peptized by water 5 or vulcanized rubber by various organic liquids. 6 If the amount of substance peptized by a given mass of water under given con- ditions is fairly constant, it will look as though we were dealing with true solubility. In fact, data are to be found in the litera- ture for the solubility of tannin in water. There are a number of cases where mixed solvents will peptize a solid much better than either one alone cellulose nitrates in ether and alcohol, caseine in pyridine and water, 7 and probably cinchonine in chloroform and alcohol, 8 as well as phloretine in ether and water. 9 The theory of this has not been worked out. Cellulose nitrate swells in alcohol and not in ether; 10 but it is not known whether this is universal. We do not know whether alcohol peptizes cellulose nitrate at higher temperatures. Zein is also peptized in mixed solvents. 11 Larguier des Bancels 12 claims that gelatine is peptized more readily by aqueous alcohol or aqueous acetone than by water alone. 1 Jour, prakt. Chem. (2) 68, 341 (1903); 72, 39 (1905); 73, 374 (1906); Zeit. phys. Chem., 62, 371 (1908). 2 Van Bemmelen Gedenkboek, 215 (1910). 3 Zeit. anorg. Chem., 74, 260 (1912). 4 Lehre von den Zustanden der Materie, 1, 160 (1914). 5 BABUs: Am. Jour. Sci. (3) 38, 408 (1869); 41, 110 (1891); (4) 6, 270 (1898); 7, 1 (1899); Phil. Mag. (5) 47, 104, 461 (1899). 6 BARUS: Am. Jour. Sci. (3) 42, 359 (1891). 7 L,EViTEs: Zeit. Kolloidchemie, 8, 4 (1911). 8 OUDEMANNS: Jour. Chem. Soc., 26, 533 (1873). 9 SCHIFF: Zeit. phys. Chem., 23, 355 (1897). 10 Private communication from Professor Chamot. 11 GALEOTTI and GIAMPALMO: Zeit. Kolloidchemie, 3, 118 (1908). > 2 Comptes rendus, 146, 290 (1911). 168 APPLIED COLLOID CHEMISTRY Relatively little work has been done on direct peptization by means of a dissolved non-electrolyte, but a good deal of stress has been laid on the cases where a non-electrolyte prevents the formation of a visible precipitate. A concentrated solution of sugar in water will prevent the precipitation of calcium silicate, 1 silver chromate, and silver chloride; 2 also of lime and of the hydrous oxides of copper 3 uranium, and iron. 4 Invert sugar is about seven times as effective as cane sugar in holding up hydrous ferric oxide. Grimaux 5 showed that glycerine prevents the pre- cipitation of hydrous ferric oxide by caustic potash. If one ion of an electrolyte is adsorbed more than the other ion, it will tend to peptize the adsorbing material and to give rise to a colloidal solution containing positively or negatively charged particles according to the nature of the adsorbed ion. Freshly precipitated silver halides are peptized by dilute silver nitrate or the corresponding potassium halide, 6 the silver and the halide ions being adsorbed strongly. Many oxides are peptized by their chlorides or nitrates, forming so-called basic salts. 7 Sul- phides are peptized by hydrogen sulphide. 8 Gelatine is liquefied or peptized by a potassium iodide solution. The peptization of hydrous oxides by caustic alkali can be considered as a case of common ion or as the preferential adsorption of hydroxyl ion. Hydrous chromic oxide gives an apparently clear green solution when treated with an excess of caustic potash; but the green oxide can be filtered out completely by means of a collodion 1 WEISBERG: Bull. Soc. chim., Paris, (3) 16, 1097 (1896). 2 LOBBY DE BRUYN: Ber. deutsch. chem. Ges., 35, 3079 (1902). 3 GRAHAM: Jour. Chem. Soc., 15, 253 (1862). 4 RIFFARD: Comptes rendus, 77, 1103 (1873). 6 Comptes rendus, 98, 1485, 1540 (1884). 6 LoTTERMosER: Jour, prakt. Chem. (2) 60, 341 '(1903); 72, 39 (1905); 73, 374 (1906); Zeit. phys. Chem., 62, 371 (1908). 7 GRAHAM: Jour. Chem. Soc., 15, 254 (1862); MULLER: Ber. deutsch. chem. Ges., 39, 2856 (1906); Zeit. anorg. Chem., 52, 316 (1907); SZILARD: Jour. Chim. phys., 5, 488, 636 (1907). 8 SPRING: Ber. deutsch. chem. Ges., 16, 1142 (1883); PROST: Jour. Chem. Soc., 54, 653 (1888); WINSSINGER: Bull. Soc. chim., Paris (3) 49, 452 (1888); LINDER and PICTON: Jour. Chem. Soc., 61, 116 (1892); MEUNIER: Comptes rendus, 124, 1151 (1897); YOUNG: Jour. Phys. Chem., 21, 141 (1917). COALESCENCE 169 filter, a colorless solution passing through. 1 Hantzsch 2 considers that hydrous beryllium oxide is peptized by caustic alkali 3 and so is cobalt oxide. 4 In ammoniacal copper solutions part of the copper oxide it apparently colloidal and part dissolved. 5 Freshly precipitated zinc hydroxide is peptized by alkali; but the solution is very unstable, the zinc hydroxide often coagulating inside half an hour. The relatively small amount of zinc remaining in solution is present chiefly or entirely as sodium zincate. 6 The bulk of the evidence seems to be that alumina is not peptized appreciably by alkali and that it goes into solution as sodium aluminate, 7 though the other view has been supported. 8 Adsorp- tion of hydroxyl ion accounts for the peptization of silicic acid 9 and of caseine by alkalies. Caseine can also be peptized by acids. A. Miiller 10 has prepared colloidal solutions of aluminum, iron, cobalt, thorium, and yttrium oxides by peptization with dilute hydrochloric acid, and Bentley and Rose 11 have peptized freshly precipitated alumina with 8 percent acetic acid. It is possible but not probable, that the peptization is done by a trace of metallic salt formed by the acid and not by the hydrogen ion. There are no cases where it has been shown conclusively that 1 FISCHER and HERZ: Zeit. anorg. Chem., 31, 352 (1902); FISCHER: Ibid., 40, 39 (1904); NAGEL: Jour. Phys. Chem., 19, 331, 569 (1915). 2 Zeit. anorg. Chem., 30, 289 (1902). 3 LoEw: Zeit. anal. Chem., 8, 463 (1870); FISCHER: Zeit. anorg. Chem., 40, 39 (1904). *TTJBANDT: Zeit. anorg. Chem., 46, 368 (1905). 6 PELIGOT: Ann. Chim. Phys. (3) 63, 343 (1861); GUIGNET: Comptes rendus, 66, 741 (1862); GRIMAUX: Ibid., 89, 1434 (1884). 6 HANTZSCH: Zeit. anorg. Chem., 30, 289 (1902); 76, 371 (1912); FISCHER and HERZ: Zeit. anorg. Chem., 31, 352 (1902); KLEIN: Zeit. anorg. Chem. 74, 157 (1912). 7 HERZ: Zeit. anorg. Chem., 26, 155 (1900); HANTZSCH: Zeit. anorg. Chem., 30, 289 (1902); RUBENBAUER: Zeit. anorg. Chem., 30, 331 (1902); SLADE: Jour. Chem. Soc., 93, 421 (1908); Zeit. anorg. Chem., 77, 457 (1912); Trans. Faraday Soc., 10, 150 (1914); BLUM: Jour. Am. Chem. Soc., 35, 1499 (1913). 8 MAHIN, INGRAHAM AND STEWART: Jour. Am. Chem. Soc., 36, 30 (1913). 9 GRAHAM: Jour. Chem. Soc., 17, 324 (.1864). IO SVEDBERG: Die Methoden zur Herstellung kolloider Losungen anor- ganischer Stoffe, 400 (1909). 11 Jour. Am. Chem. Soc., 35, 1490 (1913). 170 APPLIED COLLOID CHEMISTRY peptization is due chiefly to adsorption of undissociated salt, 1 but undoubtedly such instances will be found. Water-peptizable colloids like gelatine, 2 gum arabic, 3 dextrine, 4 soap, 5 or saponin 6 will peptize many precipitates and they are often called protect- ing colloids because they prevent the agglomeration and conse- quent settling of finely divided precipitates. Caseine is not peptized by water; but acts as a protecting colloid when peptized by acids or alkalies. Hydrous chromic oxide when peptized by caustic potash, can then prevent the precipitation of hydrous ferric oxide, etc. If too much ferric oxide is present, all the chromic oxide is carried down by it. 7 Solutions of copper oxide in ammonia will peptize chromic oxide. 8 Molybdic acid is not precipitated from its salts by uranyl salts, but tungstic acid is. In presence of tungstic acid, practically all the molybdic acid is precipitated. This is obviously a case of adsorption and the converse is also true that no tungstic acid would be precipitated in presence of a sufficient excess of a molybdate. 9 Aniline dyes, which are insoluble in benzene can be peptized by a benzene- soluble colloid such as the so-called zinc or magnesium resinate. 10 Since a colloidal solution is one in which a finely divided phase is kept from coalescing in some way, it is clear that there may be any number of colloidal aluminas, for instance, varying from anhydrous alumina (A1 2 O 3 ) up to the most highly hydrous form that can be obtained. As a matter of fact, people have generally been satisfied with distinguishing only two sets of colloidal solu- tions, which they have called solutions of alumina and metal- 1 It seems probable that the peptization of caseine by acetic acid is due largely to the undissociated acid. 2 Eder's Handbuch der Photographic, 5th Ed., 3 I, 28 (1902); LUPPO- CRAMER: Phot. Correspondenz, 44, 578 (1907). 3 LEFORT and THIBAULT: Jour. Chem. Soc., 42, 1322 (1882). 4 LACHAUDI Bull. Soc. chim., Paris (3) 16, 1105 (1896). 6 SPRING: Zeit. Kolloidchemie, 4, 161 (1909); 6, 11, 109, 164 (1910). 6 ScHiAPARELLi: Jour. Chem. Soc., 46, 333 (1884). 7 NORTHCOTE and CHURCH: Jour. Chem. Soc., 6, 54 (1854); NAGEL: Jour. Phys. Chem., 19, 331 (1915). 8 PRUDHOMME: Jour. Chem. Soc., 25, 672 (1872). 9 Miss HITCHCOCK: Jour. Am. Chem. Soc., 17, 483, 520 (1895); WOHLER: Zeit. Elektrochemie, 16, 693 (1910); WOHLER and ENGELS: Kolloidchem- ische Beihefte, 1, 454 (1910). 10 SOXHLET: Art of Dyeing and Staining Marble, etc., 76 (1902). COALESCENCE 171 lumina, of stannic and metastannic acids, etc. 1 While adsorption will cause peptization under suitable conditions, the disintegrat- ing power of the adsorbed substance is relatively small and often will not be sufficient to break up solid masses. A protecting colloid, for instance, will prevent the formation of a precipitate when it may not be able to disintegrate a massive precipitate. 1 HANTZSCH: Zeit. anorg. Ghem., 30, 338 (1902). CHAPTER VI PREPARATION OF COLLOIDAL SOLUTIONS The methods of making colloidal solutions can be grouped under two heads, the condensation methods and the dispersion methods. The condensation methods will be considered first. CONDENSATION METHODS In condensation methods we must have strong adsorption (including protecting colloids), or we must keep down the con- centration of agglomerating agents either by choice of reagents or by dilution. A few instances will be given of the preparation of colloidal solutions in the presence of a strongly adsorbed sub- stance. Colloidal palladium may be obtained by reduction with hydrazine in presence of gum arabic. 1 Colloidal solutions of silver chloride, silver bromide, or Prussian blue may be ob- tained by precipitation in presence of gelatine. 2 Sulphide solu- tions may be stabilized by gum arabic 3 or by sodium sulphide, but not by ammonium sulphide. 4 Silver chromate, hydrous copper oxide, 5 and hydrous ferric oxide may be obtained colloidal in presence of sugar. 6 Invert sugar holds up about seven times as much iron as cane sugar. Glycerol also enables us to make a colloidal solution of ferric oxide. 7 Gengou 8 prepared colloidal alumina and barium sulphate in presence of sodium citrate, and Schiaparelli 9 held up lead sulphide with saponin. A solution 1 SVEDBERG : Die Methoden zur Herstellung kolloider Losungen anorgan- ischer Stoffe, 105 (1909). 2 SVEDBERG, 323. 3 LEFORT and THIBAULT: Jour. Chem. Soc., 42, 1322 (1882). 4 HAUSSMANN: Zeit. anorg. Chem., 40, 121 (1904). 5 SVEDBERG, 224, 304. 6 RiFFARo: Jour. Chem. Soc., 27, 292 (1874). 7 GRIMAUX: Jour. Chem. Soc., 46, 96 (1884). 8 Zeit. Kolloidchemie, 9, 88 (1911). 9 Jour. Chem. Soc., 46, 332 (1884). 172 PREPARATION OF COLLOIDAL SOLUTIONS 173 obtained by adding sulphuric acid to barium arabinate was milky after standing four years. l Cadmium sulphide can be obtained colloidal in presence of caseine, 2 and so can copper oxide, 3 while mercurous chloride can be held up by albumin. 4 All the metals and many other substances can be kept in colloidal solution by sodium protalbinate and lysalbinate, which are salts of somewhat hypothetical acids obtained by saponification of albumin. 5 It is an open question whether they are any more effective than gelatine. Colloidal solutions of silver in organic solvents can be obtained with wool-fat as protecting colloid. 6 Carey Lea prepared colloidal silver of various colors by reduction of a silver salt with ferrous citrate or tartrate. 7 Both red and blue copper solutions have been obtained in presence of lysalbinates. 8 Col- loidal gold and silver can be kept from coagulating by means of starch. 9 Mercury and bismuth may be kept up by stannic oxide 10 and gold by ceria, 11 while sodium sulphate keeps up sulphur in strongly acid solution and potassium sulphate precipitates it. 12 Acheson aquadag is kept in colloidal solution by tannin and iron oxide. 13 In a number of other cases the concentration of the agglomerat- ing agent is kept low either by selection of the reaction or by dilution. Since electrolytes are very effective in producing pre- cipitation in many cases, this usually means keeping the concen- tration of electrolytes very low. It may make matters clearer to classify the methods according to the type of reaction involved. 1. Reduction without Electrolysis. At 50 hydrogen gas reacts slowly with silver oxide in presence of water, the dissolved 1 Jour. Chem. Soc., 30, 538 (1876). 2 SVEDBERG, 325. 'RITTENHAUSEN: Jour, prakt. Chem. (2) 5, 215 (1873); 7, 361 (1874). 4 SVEDBERG, 325. 6 SVEDBERG, 125, 295, 327. 6 AMBERGER: Zeit. Kolloidchemie, 11, 99 (1912). 7 SVEDBERG, 211. 8 SVEDBERG, 166, 170. 9 GUTBIER and WEINGARTNER: Kolloidchemische Beihefte, 6, 211, 244 (1913). 10 SVEDBERG, 204. 11 SCHOTTLANDER: Jour. Chem. Soc., 68 II, 218 (1895). 12 SVEDBERG, 249; Cf. ODEN: Zeit. Kolloidchemie, 8, 186 (1911). " Cf. DOYLE: Jour. Phys. Chem., 17, 390 (1913). 174 APPLIED COLLOID CHEMISTRY silver oxide being the substance which reacts. The reaction is Ag 2 O + H 2 = 2Ag + H 2 O, so the silver is formed in the absence of electrolytes and tends to come down colloidal, though some does precipitate as a mirror. 1 From dilute solutions of gold salts, colloidal gold can be obtained by reduction with carbon monoxide, sulphur dioxide, formaldehyde, pinene, alcohol, hydrazine, and phenyl hydrazine. 2 In these cases the concentration of electro- lytes is kept low by dilution. On the other hand, colloidal cop- per can be obtained by the action of alcohol on anhydrous copper sulphate and concentrated sulphuric acid, 3 a reaction which cannot be classified without further study. Colloidal gold can also be obtained by reduction by means of Aspergillus oryzae.* The effect of concentration has been studied by Liversidge. 5 Colloidal metals may also be obtained by reduction in flames. 6 2. Reduction with Electrolysis. When very dilute solutions of gold, silver, and mercury salts are electrolyzed, some of the metal comes out in colloidal form. 7 Billitzer 8 obtained colloidal solu- tions in a similar way with salts of the less noble metals, but it is probable that the dispersed phase consisted of oxide and not of metal. 9 Lecoq 10 has prepared colloidal arsenic electrolytically, and Sansonoff 11 colloidal UC>2,the latter being stabilized byuranyl chloride. 3. Oxidation. Colloidal sulphur can be obtained from the reaction 12 between H 2 S and SO 2 , which involves no high concentra- tion of electrolyte. In presence of gelatine as stabilizer, col- loidal sulphur 13 can be obtained by oxidation of H 2 S. Wohler and Witzmann 14 have prepared iridium dioxide as a violet or 1 SVEDBERG, 38. 2 SVEDBERG, 60-71, 101, 112. 3 RASENFOSSE: Jour. Soc. Chem. Ind., 30, 133 (1911). 4 SVEDBERG, 199. 5 Jour. Chem. Soc., 60, 401 (1891). 6 DONAU: Zeit. Kolloidchemie, 16, 81 (1915). 7 SVEDBERG, 206. 8 Ber. deutsch. chem. Ges., 35, 1929 (1902). 9 HABER: Zeit. Elektrochemie, 8, 551 (1902). 10 Comptes rendus, 160, 700 (1910). 11 Zeit. Kolloidchemie, 8, 96 (1911). 12 SVEDBERG, 243. 13 HIMMELBAUR: Zeit. Kolloidchemie, 4, 307 (1909). 14 Zeit. anorg. Chem., 57, 328 (1908). PREPARATION OF COLLOIDAL SOLUTIONS 175 blue solution. Bredig and Marek 1 have obtained colloidal manganese dioxide by the action of hydrogen peroxide on per- manganate. When aurous chloride breaks down into auric chloride and gold, the latter can be obtained colloidal under suit- able conditions. 2 4. Hydrolysis. Water will hydrolyze any salt until the pro- duct of concentrations of the hydroxyl and hydrogen ions reaches a value of about 10~ 14 . If either base or acid is very sparingly soluble, the hydrolysis will run farther than if both are strong electrolytes. Equilibrium will be reached more rapidly if the solution is heated. If a dilute solution of ferric chloride is boiled for a few minutes and then cooled, it will not be blackened by hydrogen sulphide 3 and it will not form Prussian blue with potassium ferrocyanide. 4 On the other hand, addition of sodium sulphate or of sulphuric acid will cause a precipitation of hydrous ferric oxide. Colloidal solutions of the hydrous oxide can be obtained by hydrolysis of the acetates, the nitrates, or the chlo- rides, but not in general by the hydrolysis of sulphates, because sulphuric acid precipitates the colloidal solutions more readily than hydrochloric, nitric, or acetic acid. The so-called chrom- sulphuric acids are apparently special cases of colloidal solutions due to hydrolysis in sulphate solutions. 5 Colloidal solutions can also be obtained by hydrolysis of a few compounds which are not electrolytes, 6 such as silicon sulphide, ferric ethylate, and copper succinimide. 5. Metathetical Reactions. When hydrogen sulphide acts on arsenious oxide, arsenious sulphide is formed practically in the absence of electrolytes, As 2 3 + 3H 2 S = As 2 S 8 + 3H 2 O. This colloidal solution can be made of the very high concen- tration of 3 grams As 2 S 3 per 5 grams water. 7 The reaction be- tween mercuric cyanide and hydrogen sulphide gives prussic acid in solution; but this is a very weak acid and has but slight 1 Van Bemmelen Gedenkboek, 342 (1910). 2 VANINO and ROESSLER: Zeit. Kolloidchemie, 6, 289 (1910). 3 WRIGHT: Jour. Chem. Soc., 43, 156 (1883). 4 SVEDBERG, 260. 6 DENHAM: Zeit. anorg. Chem., "67, 378 (1908); STRONG: Comptes rendus, 160, 1172 (1910). c SVEDBERG, 276. 7 SVEDBERG, 297. 176 APPLIED COLLOID CHEMISTRY precipitating power. 1 Colloidal sulphides can also be obtained with salts of strong acids by working in sufficiently dilute solu- tions. 2 In some cases sodium chloride has relatively little pre- cipitating power, and it is therefore possible to prepare colloidal solutions of silicic acid from, sodium silicate, of the so-called stannic and metastannic acids from stannic chloride and sodium stannate respectively; 3 and of tungstic acid from sodium tung- state. The tungstic acid solution can be concentrated until it contains 80 percent tungstic acid and has a density of over 3.2, so that glass will float in it. 4 If M/20 AgN0 3 is mixed with M/20 KC1, KBr, or KI, so that there is a slight excess of one or the other solution, 5 more or less of the silver halide stays in suspension, being charged positively in the one case owing to Adsorption of silver ion, and charged negatively in the other case owing to adsorption of halide ion. A salt which is readily soluble in water can be obtained in a colloidal state by working in a solvent in which the salt is practi- cally insoluble and therefore precipitates in an extremely finely divided state. 6 Thus colloidal sodium chloride can be obtained by the interaction of sodium malonic ester and chloracetic ester in ligroin. 7 6. Decomposition by Light. Colloidal solutions can be ob- tained when substances are precipitated by the action of light. Svedberg 8 has studied the action of light on gold solutions, and Wigand 9 the behavior of sulphur solutions. There are also several other interesting papers along similar lines. 10 1 SVEDBERG, 302. 2 WINSSINGER: Jour. Chem. Soc., 54, 911 (1888). 3 SVEDBERG, 303. MULLFR: Van Bemmelen Gedenkboek, 416 (1910). 6 LoTTERMOSER: Jour, prakt. Chem. (2) 68, 341 (1903); 72, 39 (1905); 73, 374 (1906); cf. DRUCKER: Zeit. Kolloidchemie, 4, 216 (1910). 8 BOTTGER: Jour. Chem. Soc., 36, 107 (1878); VON WEIMARN: Zeit. Kolloidchemie, 7, 92 (1910); 8, 216 (1911); 11, 651 (1912); cf. PIERONI: Gazz. chim. ital., 43 I, 197; PATERN& and MEGIGRECIANU, Ibid., 43 I, 439 (1913). 7 SVEDBERG, 349. 8 Zeit. Kolloidchemie, 6, 238 (1910). 9 Zeit. phys. Chem., 77, 423 (1911). IO AMANN: Zeit. Kolloidchemie, 8, 197 (1911); GALECKI: Ibid., 10, 149 (1912); HARTWAGNER: Ibid., 16, 70 (1915). PREPARATION OF COLLOIDAL SOLUTIONS 177 DISPERSION METHODS Five different types of preparation of colloidal solutions by peptization may be distinguished: 1. Removal of Agglomerating Agent. If a precipitate has settled from a colloidal solution owing to the addition of too much of an agglomerating agent, the precipitate may go back into apparent solution if the excess of agglomerating agent is washed out. No colloidal solution will be obtained if it is impossible to wash out the coagulating agent or if the agglomeration has gone too far. 1 If a silver halide precipitate is washed on a filter at once, the silver salt is apt to run through the filter when the excess of potassium salt has been removed and there is present only the amount which would have kept the silver halide in sus- pension originally. When the rare earth nitrates are precipitated by ammonia and then washed, they are liable to stay suspended in the liquid when the ammonium nitrate is nearly all removed. In alloy work stannic oxide should be washed with dilute nitric acid and not with water. Zinc sulphide is apt to form a colloidal solution when the ammonium salt is washed out, 2 and copper ferrocyanide does the same thing if all the copper sulphate is removed by washing. 3 In fact, Chatard 4 claims that the quickest way to wash a gelatinous precipitate is to evaporate the solution to dryness and heat before trying to wash. Merely evaporating on a water-bath is not always sufficient. 5 2. Addition of Peptizing Agent. Instead of washing out a pre- cipitating agent, a peptizing agent may be added. Ammonia is very effective in suspending clay, 6 silicic acid is peptized readily by caustic soda, 7 and Prussian blue is peptized by oxalic acid or by potassium oxalate. In some cases the concentration of the peptizing agent has to be high, as when oxides are to be peptized by alkali, and people usually assume the formation of compounds. 1 Cf. ABEGG and VON SCHROEDER: Zeit. Kolloidchemie, 2, 85 (1907). 2 DONNINI: Jour. Chem. Soc., 66 II, 318 (1894). 3 BERKELEY and HARTLEY: Phil. Trans., 206A, 486 (1906). * Jour. Chem. Soc., 26, 527 (1873). 6 WRIGHT: Jour. Chem., 43, 156 (1883); KRATZ: Jour. Phys. Chem., 16, 121 (1912). 6 SKEY: Chem. News, 17, 164 (1868); 22, 236 (1870); 34, 142 (1876). See also DOELTER: Handbuch der Mineralchemie, 2, 122 (1912). 7 GRAHAM: Jour. Chem. Soc., 17, 324 (1864). 12 178 APPLIED COLLOID CHEMISTRY Jowitschitsch 1 describes conditions under which chromic oxide is apparently completely soluble in ammonia. This is probably a case of peptization. The action of soap on rouge or carbon black 2 looks like a dis- integration; but it is not. If a suspension of carbon black in water be filtered several times through filter-paper, the water will finally run through clear, and the carbon black will be held back by the filter-paper. If a soap solution be poured on the filter, a black filtrate is obtained and the filter-paper is no longer black. All the carbon black has passed through the filter-paper. The same thing can be done with rouge, except that a red filtrate is obtained instead of a black one. At first sight it seems as though the soap must have broken up the carbon or the rouge into finer particles, which then passed through the filter, but this is probably not so. The filter-paper is porous enough at first to let through the particles of carbon or rouge, as is shown by the fact that some of the suspended matter does pass through the filter at first. The cellulose adsorbs the carbon black or the rouge, and and this clogs the filter to such an extent that the pores are not large enough to let the remaining particles through. The soap removes the rouge or the carbon black from the paper because it adsorbs these substances more strongly, and everything, there- fore, goes through the paper. That this is the true explanation can be shown in two ways. In the first place, the experiment does not succeed if the rouge or the carbon is too coarse. In the second place, Spring showed that we are dealing with an adsorption of carbon black by filter-paper. If the black filter- paper be reversed and washed with water, the water removes only the black which is not in immediate contact with the paper. 3. Mechanical Disintegration. If a solid be ground up suffi- ciently fine, it will necessarily form a colloidal solution for a time. This has been done experimentally by Wegelin 3 in the case of a number of metals. The addition of gelatine makes it easier to disintegrate ductile metals. 4 It must be remembered that a 1 Monatsheft fur Chemie, 34, 225 (1913). 2 SPRING: Zeit. Kolloidchemie, 4, 161 (1909); 6, 11, 109, 164 (1909). 3 Zeit. Kolloidchemie, 14, 65 (1914). 4 Metallic inks are said to be made by grinding with honey. THORPE: Dictionary of Applied Chemistry, 3, 138 (1918). PREPARATION OF COLLOIDAL SOLUTIONS 179 powder made with gelatine will not necessarily have exactly the same properties as the same powder when pure. Tyndall 1 has called attention to one classical instance of an error due to over- looking the effect which a little gum might have on the adsorption of dark heat rays. Melloni found the same radiant and absorb- ent power for chalk and lamp-black. Masson and Courte'pe'e studied a number of chemical precipitates and found that all manifested the same power of radiation. The conclusion was drawn that the influence of the fine state of division is so power- ful as entirely to mask and over-ride whatever influence might be due to chemical constitution. As a matter of fact, the precipi- tates were mixed with a solution of gum in water and were then painted on the side of a cube with a brush. The gum coating is very transparent to light but as opaque as pitch or lamp-black to the longer non-luminous rays. The measurements all re- ferred to a surface of gum and consequently all came out alike. Since a finely divided solid is more soluble, it is possible that it may go into solution and then precipitate in another form. This seems to happen with quartz. When reduced to an impalpable powder by long grinding, quartz can be converted into colloidal hydrous silicic acid merely by boiling with water. 2 4. Electrical Disintegration. When a direct current arc is formed under water between two wires, the metal is disintegrated and colloidal solutions of platinum, iridium, palladium, gold, silver, and cadmium may be obtained in this way. 3 Satisfactory condi- tions are obtained with 30 to 40 volts and 5 to 10 amperes. A trace of alkali in the water causes formation of finer particles, presumably owing to the stabilizing effect of the hydroxyl ion. The disintegration is chiefly at the cathode. The method is not satisfactory with organic liquids, because too much decomposition of the liquids takes place. Svedberg found empirically that this decomposition could be decreased very much if the current density were made as small as possible. 4 1 Fragments of Science: Radiant Heat and its Relations. 2 DESCH: The Chemistry and Testing of Cement, 58 (1911). 3 BREDIG: Zeit. Elektrochemie, 4, 514 (1898); Zeit. phys. Chem., 31, 258 (1899). 4 SVEDBERG, 424. 180 APPLIED COLLOID CHEMISTRY He therefore used an oscillatory discharge from an induction coil with a condenser in parallel or in series. The best results are obtained with large capacity, small self-induction, low resist- ance, and short arc. 1 By this improved method Svedberg suc- ceeded in preparing colloidal solutions of all the metals, including the alkali metals. Liquid methane, ether, and isobutyl alcohol at low temperature were especially satisfactory with the metals of the alkalies and the alkaline earths. The order of disintegra- tion of some of the metals under similar conditions was found to be Fe, Cu, Ag, Al, Ca, Pt, Au, Zn, Sn, Cd, Sb, Tl, Bi, Pb, the iron being the least rapidly disintegrated and the lead the most rapidly. There is no apparent relation either with the order of the boiling- points or with the order of disintegration by cathode rays or canal rays. 5. Electrochemical Disintegration. With a lead cathode in caustic soda solution, the lead disintegrates when the current density exceeds a critical value, and the solution is colored black like ink with fine particles of metallic lead. 2 This is due to the temporary formation of a sodium-lead alloy, which then disin- tegrates in contact with water. Similar results can be obtained with cathodes of tin, bismuth, thallium, arsenic, antimony, and mercury. E. Mtiller 3 obtained colloidal solutions of tellur- ium with a tellurium cathode. This seems to be due to the formation of polytellurides, which break down and set free tellurium. In the presence of oxygen there may be also an oxida- tion of a telluride. Fischer 4 has obtained metallic copper in the solution by using a high current density with a copper anode in sulphuric acid. Cuprous sulphate is formed, which breaks down to metallic copper and cupric sulphate. This experiment has not yet been made to give colloidal copper, but this could prob- ably be done if one were to add a suitable protecting colloid. 1 See also MORRI&-AIREY and LONG: Jour. Chem. Soc., 104 II, 1033 (1913); LONG: Zeit. Kolloidchemie, 14, 136 (1914); KTTTSCHEROFF : Ibid., 11, 165 (1912); ZAWRIEFF: Zeit. phys. Chem., 87, 506 (1914). 2 REED: Jour. Franklin Inst., 139, 283 (1895); BREDIG and HABER: Ber. deutsch. chem. Ges., 31, 2741 (1898); HABER and SACK: Zeit. Elektrochemie, 8, 243 (1902); Zeit anorg. Chem., 34, 286 (1903). 3 Zeit. Elektrochemie, 11, 521, 701 (1905); HABER: Ibid., 11, 660, 827 (1905). 4 Ibid., 9, 507 (1903). PREPARATION OF COLLOIDAL SOLUTIONS 181 The disintegration of all electrodes by an alternating current when the current density is high is undoubtedly due to the tem- porary formation and subsequent breaking down of a hydrogen or metallic alloy. 1 1 HABER: Trans. Am. Electrochem. Soc., 2, 192 (1902); VAN NAME: Zeit. Elektrochemie, 10, 303 (1904). CHAPTER VII PROPERTIES OF COLLOIDAL SOLUTIONS GENERAL PROPERTIES The whole theory of colloidal solutions is based on the assump- tion that we are dealing with two-phase systems. Since this is not obvious and is still disputed by some people, it is important to see what independent proof we have of the existence of a second phase. If a beam of light strikes small particles in sus- pension in a transparent medium, the light is scattered and polar- ized, provided the refractive index of the particles differs from that of the transparent medium. The intensity of the scattered light is inversely proportional to the fourth power of the wave- length and the scattering by each particle is proportional to the square of the volume of the particle. 1 The scattered light is therefore chiefly blue. 2 This is known as the Tyndall phenomenon and it was used by him to determine whether air was free from dust and was optically empty. The same principle has been perfected and applied in the ultra-microscope, 3 so-called. In this instrument a powerful beam of light is focussed on the liquid to be examined and is looked at from the side through a powerful microscope. 4 If there are suspended particles in the liquid, they appear as bright points. One does not see the particles themselves but merely the light diffracted by them. With colloidal gold in glass it is possible to see that we have a two-phase system. By means of some form of this instrument there is no difficulty in demonstrating the existence of a second phase in many colloidal solutions. In fact the instrument is so delicate that one must guard against the danger of casual im- i Phil. Mag. (4), 41, 107, 447 (1871). 2 Instead of calling this by the clumsy term of the blue of turbid media, it would be very appropriate to speak of Tyndall blue. 3 SIEDENTOPF and ZSIGMONDY: Drude's Ann., 10, 1 (1903). 4 For a general discussion of the ultra-microscope, see BURTON: The Physical Properties of Colloidal Solutions, 28 (1916). 182 PROPERTIES OF COLLOIDAL SOLUTIONS 183 purities making one see a second phase when it is not there. This difficulty becomes very serious with liquids other than water. The recent work of Martin 1 seems to prove that some organic liquids scatter light even when absolutely pure. In so far as this may be true for a pure liquid or a true solution, the evidence of the ultra-microscope is not conclusive as to heterogeneity. Although the particles themselves are not seen, it is possible to estimate their size if one knows the number and mass, and as- sumes that the particles are practically spherical. It is custom- ary to use the word " microns" for particles with diameters exceeding 250^, while those particles ranging from 6-250/xju are called submicrons, and the term amicrons is used for particles with a diameter less than 6/^. The existence of these smallest particles can be shown in certain cases because the particles may serve as nuclei for crystallization. Kogonossoff 2 claims to have followed the electrolysis of silver nitrate solutions with the ultra-microscope. Since the visibility of particles depends on the difference in index of refraction of the particles and the me- dium, it is evident that the more hydrous the second phase, the less easily can it be seen. With many organic colloids the particles can be seen only when they exceed 40^ in diameter. The Brownian movements are consequently much less marked. It is possible to prepare salt solutions which are optically empty. On the other hand Lobry de Bruyn 3 claims to have found that cane sugar solutions were optically inhomogeneous. This has been cited as a case of a large molecule being visible; but it is much more likely that the scattering was due to an impurity and not to the sugar itself. It apparently seems obvi- ous to most people that if one continues to subdivide a precipi- tate sufficiently, one will pass from a suspension to a true solution without a break. 4 The fallacy lies in the assumption that one can have all sizes of particles in equilibrium with the liquid. Since the solid becomes more soluble the more finely subdivided it is, there will be particles of a certain size which will dissolve 1 Jour. Phys. Chem., 24, 478 (1920). 2 Zeit. Kolloidchemie, 7, 129 (1910). 3 Rev. Trav. chim. Pays-Bas, 23, 218 (1904). 4 See, however, VON WEIMARN: Zur Lehre von den Zustanden der Materie, 161 (1914). 184 APPLIED COLLOID CHEMISTRY and reprecipitate in a coarser form and which therefore cannot coexist for any length of time with the liquid. This comes out clearly in the case of smokes, it being very difficult to obtain a stable smoke of satisfactory concentration with particles less than lOju//. If one takes definite amounts of benzene and water, it is possible theoretically to subdivide the benzene so that it will dissolve completely in the water; but the true solution thus formed would have different properties from the corresponding suspension. It would show characteristic osmotic pressure phenomena and it would be instable in quite a different way from the suspension. While it is perfectly possible to treat a solution formally as the end term in a series of suspensions, there is nothing as yet to show that this is a desirable hypothesis. 1 The presence of a second phase in a colloidal solution can also be shown by what is called ultra-filtration. A colloidal solution goes right through an ordinary filter; but Bechhold 2 showed that it was possible to form, on filter paper or wire gauze, collodion films with varying amounts of acetic acid or gelatine films hard- ened with formaldehyde, such that pores of almost any desired size can be obtained. By taking a filter with sufficiently fine pores, almost any colloidal solution can be filtered so as to leave the suspended matter behind. With a more porous filter the finer particles go through and the coarser ones are held back. In this way it is possible to get an approximate idea of the rela- tive sizes of the particles in two different colloidal solutions. Solu- tions made up by Bechhold are arranged in Table XLII in the order of increasing fineness of particles. Blue colloidal solu- tions of litmus have finer particles in them than the corresponding red solutions. Solutions made up in other ways would give different results, because colloidal gold solutions, for instance, can be made with particles of very different sizes. The deep green solution formed when chromic oxide is peptized by alkali comes through color- less when a suitable ultra-filter is used. On standing, the chro- mic oxide eventually becomes so coarse that it can be separated by means of an ordinary paper filter. In all cases of ultra-filtra- *Cf. LEBLANC: Zeit. Elektrochemie, 19, 794 (1913). 2 Zeit. phys. Chem., 60, 257 (1907); 64, 328 (1908); SCHOEP: Zeit. Kolloid- chemie, 8, 80 (1911); MALFITAMO: Comptes rendus, 131, 1229 (1904). PROPERTIES OF COLLOIDAL SOLUTIONS 185 tion, care must be taken to ensure that a real nitration is taking place and that the results are not vitiated by adsorption. It is to be noted that the behavior of ultra-filters can be accounted for on the basis of size of pores and size of particles, which is not true for semi-permeable membranes. 1 On the other hand there seems to be some difficulty in determining the absolute size of the pores. Experiments by the bubble method with the ultra-filter indicated that the size of the colloidal silver (or collar- gol) particles was 200^490juju whereas experiments with a Cham- berland filter indicated that the coarsest collargol particles were less than 170/i/x in diameter. In Table XLII they are called 20/iM- TABLE XLII. ORDER OF ULTRAFILTRATION Coarse suspensions Prussian blue Platinum sols (Bredig) Iron oxide Caseine Arsenious sulphide Gold (about 4(W) Bismuth oxide Lysargin (silver) Collargol (silver, 20/i/i) Gold 1 percent gelatine 1 percent hemoglobin Serum albumin Diphtheria toxin Protalbumoaes Silicic acid Lysalbinic acid Deutero albumoses Litmus Dextrin Crystalloids With a suitable ultra-filter it is possible to filter ordinary water so that it is absolutely sterile. It is also possible to obtain in this way water which can be used for work with the ultra-micro- scope. The behavior of mixtures is distinctly interesting. If the two substances do not either one adsorb the other and if the particles are of different sizes, a separation can be effected. The green mixture of Prussian blue and hemoglobin filters red with a suitable filter, the fine hemoglobin going through and the coarse Prussian blue remaining behind. In a mixture of Prussian blue with arsenic sulphide or ferric oxide, there is adsorption and everything is stopped by a filter which would let through either the arsenic sulphide alone or the ferric oxide alone. By adding a protecting colloid, sodium lysalbinate, to Prussian blue further dispersion takes place and the Prussian blue will pass through a filter which just stopped it before. If serum albumin is added 1 WALDEN: Zeit. phys. Chem., 10, 699 (1892). 186 APPLIED COLLOID CHEMISTRY to Prussian blue and then oxalic acid, the Prussian blue particles are found to be smaller than before. If oxalic acid is added and then the serum albumin, no marked change in the size of particles takes place. Undoubtedly the oxalic acid is adsorbed by the Prussian blue and prevents the adsorption of the albumin to a considerable extent. Since the existence of a colloidal solution depends on the Brownian movements counteracting the force of gravity, it follows that if we increase the force of gravity, a colloidal solution will become less stable and may precipitate. This can be realized experimentally by centrif uging. l The best known case of this is the separator in which the cream, being lighter, goes to the center while the milk is thrown to the outside. A few precipita- tions of other solutions have been made; but the subject has not been studied systematically. At 4000 revolutions per minute gold submicrons are precipitated in a few minutes while the fine amicrons of Fe 2 O 3 are not precipitated. With a super-centrifuge running 40,000 revolutions per minute and more, it is possible to precipitate many more colloidal solutions. The converse is also true, that a solution becomes more stable against centrifug- ing if the particles are made smaller. A separator does not give cream with homogenized milk in which the fat particles are less than 0.8/-6 in diameter. 2 The fact that a true solution, if practi- cally saturated, can be made to crystallize by centrif uging, 3 does not indicate any real analogy between true and colloidal solutions. In the case of the true solution, only a small amount of the dissolved substance can be precipitated while with a colloidal solution practically all the suspended material comes down. 4 A coarse suspension, if entirely insoluble and anhydrous, should give rise to no osmotic pressure, no diffusion except as a result of electrical charges, and no change in surface tension or 1 FRANKLIN and FREUDENBERGER: Trans. Am. Electrochem. Soc., 8, 29 (1903); GIOLITTI: Gazz. chim. ital., 36 II, 159 (1906); DUMANSKI: Van Bemmelen Gedenkboek, 421; FRIEDENTHAL: Zeit. Kolloidchemie, 15, 75 (1914); AYRES: Met. Chem. Eng., 14, 500 (1916). 2 FLEISCHMANN: Lehrbuch der Milch wirtschaft, 393 (1906). 3 BREDIG: Zeit. phys. Chem., 17, 459 (1895); LOBRY DE BRUYN and VAN CALCAR: Rec. Trav. chim. Pays-Bas, 23, 218 (1904). 4 BECHHOLD: Zeit. Kolloidchemie, 2, 3 (1907). PROPERTIES OF COLLOIDAL SOLUTIONS 187 conductivity. If the suspension is hydrous it may have a marked effect on the surface tension. In fact Lottermoser 1 has suggested that the hydrous colloid may sometimes be the external phase. In general the apparent osmotic pressures of colloidal solutions are very small and this has led to absurd molecular weights. 2 Such values as 30,000 for albumin, 700,000 for glycogen, over 48,000 for silica and " enormous" for Fe 2 3 mean nothing whatso- ever. This can be seen readily if we consider the case of boiling water to which scrap platinum is added to prevent overheating. Nobody would deduce a molecular weight of platinum from such an experiment, not even if the platinum were sub-divided a good deal, and yet this is just what people do continually with colloids. If one adds water to benzene, a little will dissolve and will change the freezing point of the benzene. From this one can calculate the molecular weight of water dissolved in benzene; but the calcu- lation means nothing so soon as there is an excess of water. It would still mean nothing if one were to break the water up into drops which stayed suspended for a while in the benzene. It is not denied that colloidal solutions may show osmotic phenomena. This may happen in case the suspended material is somewhat soluble or contains adsorbed soluble material. 3 It is wrong, however, to refer the observed osmotic pressure to the total amount of added material. It is also wrong to study the swelling of jellies and to speak of the results as measurements of osmotic pressure. The low rate of diffusion of colloids was the phenomenon which first started Graham on his studies in regard to colloids. There is no question but that colloids do diffuse. This may be due in part to adsorbed material, but we also get a diffusion due to the Brownian movements, since the bombardment by the molecules of the medium apparently tends to produce a uniform distribu- tion of the suspended particles, except in so far as this is inter- fered with by gravity, surface tension, electrical stresses, etc. 1 Van Bemmelen Gedenkboek, 152 (.1910). 2 GLADSTONE: Phil. Mag. (5) 2, 838 (1889); SABANEEFP: Jour. Chem. Soc., 58, 1215 (1890); REID: Jour. Physiology, 31, 438 (1904); GATIN- GRUZEWSKA: Jour. Phys. Chem. 8, 515 (1904); EULER: Pflanzenchemie, 2, 33 (1908): CHWOLSON: Jour. Phys. Chem., 16, 88 (1912). 3 BAYLISS: Proc. Roy. Soc., 81, 269 (1909); BILTZ and VON VEGESACK: Zeit. phys. Chem., 68, 357 (1909). 188 APPLIED COLLOID CHEMISTRY The dilfusion of a colloid through certain membranes is practically zero, while true solutions go through readily. This property has been made use of to purify colloidal material from dissolved substances. The process is called dialysis. In its simplest form a tube, closed at one end with a parchment paper or animal membrane, is filled with the solution to be dialyzed and is placed in a larger vessel filled with distilled water. The dissolved sub- stances diffuse out and the colloidal material is left behind. Wintgen 1 claims that the reciprocal of the density, the specific volume, is practically a linear function of the percentage concen- tration by weight. This was tested with arsenic sulphide, anti- mony sulphide, silicic acid, molybdic acid, ferric oxide, tannin, starch, gelatine, albumin, sodium caseinate so-called, and serum globulin. If extrapolation is made to 100 percent colloid the value for the specific volume comes out too low and conse- quently the value for the density too high. This is probably due to the existence of a film of condensed water around each particle. If the suspended matter be in practically stable equilibrium, the density as determined by the hydrometer will be the same as that determined by weighing a given volume. 2 If the suspended material is settling or rising with sufficient rapidity, the density shown by the hydrometer will be that of the liquid medium. It is all a question of the extent to which the immersion of the hydrometer displaces water or displaces water and suspended material. It is another form of the prob- lem whether a beaker weighs more in case a fly, which is heavier than air, hovers in the beaker without touching it. So long as the fly is there, the fly and the air are equivalent to a medium which is denser than air and consequently the downward thrust is greater when the fly is there than when it is not. If anybody doubts this, let him consider the case where a glass plate is laid over the mouth of the beaker imprisoning the hovering fly. The product of the specific volume into the index of refraction is said to be a linear function of the percentage composition by weight. 3 1 Kolloidchemische Beihefte, 7, 251 (1915). 2 BONDY: Pogg. Ann., 121, 314 (1865); OWENS: Geographical Journal, 37, 68 (1911); EHRENBERG: Die Bodenkolloide, 84 U915). 3 WINTGEN: Kolloidchemische Beihefte, 7, 268 (1915). PROPERTIES OF COLLOIDAL SOLUTIONS 189 An anhydrous substance in suspension, such as ferric arsenic sulphide, gold, etc., appears to have no effect on the sur- face tension of the medium; but some of the gelatinous colloids have a marked effect. Soap lowers the surface tension a great deal. Addition of 0.5-0.8 g gelatine to 100 cc water causes a marked decrease in the surface tension of water, while addition- of further amounts has practically no effect. 2 Addition of gum arabic increases the surface tension within the same limits. We do not know why this last should be so. It has already been shown that when oleic acid is dropped on a water surface a portion is adsorbed and spreads over the surface, while the excess, if slight in amount, draws up into drops. The thickness of the adsorbed oil film will vary from nothing up to that where the drop formation occurs. If water peptized oleic acid appreciably, we should find that, for a given small amount of oil, the surface film would be thinner the greater the peptiz- ing action of the liquid or the greater the surface relatively to the mass. Conversely, if we 'started with a peptized mass, some of the oil would go into the surface and form a film there. This has been studied in some detail by Metcalf, 3 who did not realize that he was dealing with colloidal solutions. He found that peptone solutions gradually formed an elastic surface film with properties depending on the temperature. As the rate of diffu- sion was very slow, this film could be prepared more readily by dropping a concentrated peptone solution on the surface of pure water. The difference in concentration between the sur- face and the mass of the liquid was too great to be accounted for by the Gibbs formula for the relation between change of surface tension and change of concentration in the surface film. Conse- quently Metcalf assumed the existence of an insoluble form of peptone. This cannot be the true explanation because the phenomenon is general. The difficulties disappear the moment one realizes that it is a question of colloidal solution and not of a balancing of osmotic pressures. Saponin 4 also forms a markedly I LINDER and PICTON: Jour. Chcm. Soc., 60, 72 (1895); ZLOBICKI: Ibid., 87, 1924 (1905). 2 ZLOBICKI: Bull. Acad. Sci. Cracovie, 1906, 497. 3 Zeit. phys. Chem., 62, 1 (1905). 4 SHORTER: Phil. Mag. (6) 11, 317 (1906); 17, 5GO (1909). 190 APPLIED COLLOID CHEMISTRY viscous surface film. It is familiar to everybody that a skin forms readily over boiled milk or over cocoa. The more rapid formation after heating is undoubtedly due to the colloidal solu- tion becoming less stable, though this has not been proved ex- perimentally, nor has it yet been shown that the film is thinner the greater the peptizing action of the liquid. Since the film formation will occur more readily and more rapidly if the colloidal substance adsorbs air to a marked extent and since air is always present in actual experiments, a further discussion of film formation from colloidal solution will be left till we take up emulsions and froths. VISCOSITY The viscosity of a liquid is the resistance offered to shearing, to stirring, or to the flow through a capillary tube. 1 The viscos- ity coefficient is the force required to move at unit velocity a plate of unit surface separated from another plate of the same size by a layer of liquid of unit thickness. In the case of water- soluble colloids the viscosity may increase enormously with the concentration, a 1 percent solution of agar forming a solid jelly. This is undoubtedly in part a question of structure just as we know it is when 99 percent of kerosene is emulsified in a 1 percent of a soap solution. Here we have minute glob- ules of oil coated by viscous soap films which coalesce to form a honeycomb structure. The metal and the sulphide sols have only a very slight effect on the viscosity of water. This varies with the nature and the sub-division of the solid particles. According to Hatschek 20-30 percent of precipitated calcium carbonate or barium sulphate does not increase the viscosity very much. On the other hand, Bingham 2 found zero fluidity or infinite viscosity with 4 percent of china clay or 5.5 percent of graphite. Einstein 3 has deduced 4 a formula for the viscosity (771) of a suspension of rigid spheres in any liquid, in terms of the viscosity 1 HATSCHEK: An Introduction to the Physics and Chemistry of Colloids, 21 (1916). 2 Am. Chem. Jour., 46, 278 (1911); Jour. Franklin Inst., 181, 845 (1916). 3 Drude's Ann., 19, 289 (,1906). 4 BURTON: The Physical Properties of Colloidal Solutions, 162 (1916). PROPERTIES OF COLLOIDAL SOLUTIONS 191 (r?) of the pure liquid and the ratio (/) of the volume of the sus- pended matter to the total volume, 171 = 17(1 + kj) where A; is a constant which Einstein at first considered equal to unity but which he afterwards made equal to 2.5. A similar formula has been deduced by Hatschek 1 for cases where the suspended material occupies less than 40 percent of the total volume. It differs from Einstein's formula by having k = 4.5. This formula was tested by Harrison 2 for starch and water and was found to hold fairly well up to 30 percent concentration. According to these formulas, the viscosity depends only on the volume of the dispersed phase and not at all upon the degree of dispersity. Oden 3 found, however, an approximately 50 per- cent greater viscosity with sulphur sols in which the particles had a diameter of about 10,uju than with sols in which the sulphur particles had a diameter of about lOOju/*. Hatschek 4 attributes this very properly to the existence of an adsorption film of liquid around the sulphur particles, a phenomenon which was not considered when the formula was deduced. On the assumption that the thickness of the adsorbed film is independent of the size of the particles, Hatschek deduces that the thickness is about 0.87ju/z which corresponds to an increase in volume of about 62 percent for particles having a diameter of 10^. Hatschek 5 has also deduced a formula for the viscosity of two liquid phases in which the volume of the dispersed phase is the larger, where A = 1/f and is the ratio of the total volume to the volume of the dispersed phase. From this formula he deduces that one volume of caseine binds more than eleven volumes of water and that in a 0.4 0.5 percent rubber sol, the rubber binds 75-100 volumes of liquid. 1 Zeit. Kolloidchemie, 7, 301 (1910). 2 Jour. Soc. Dyers and Colorists, 27, April (1911). 3 Zeit. phys. Chem., 80, 709 (1912). 4 Zeit. Kolloidchemie, 11, 280 (1912). o Ibid., 8, 34 (1911); 11, 284 (1912); 12, 238 (1913). 192 APPLIED COLLOID CHEMISTRY We get rather curious results if we apply Hatschek's views on viscosity to Bingham's experiments on zero fluidity. If we make the assumption that plastic flow is reached when the surfaces of adsorbed water are in contact and if we make the further as- sumption that we are dealing with spheres in open piling, the voids will then be 48 percent of the whole and, in the case of graphite for instance, the amount of water adsorbed by the graph- ite must be 94.5 48 = 47.5 volume percent, or each volume of graphite must adsorb about nine volumes of water. If we assume close piling or different sizes of graphite powder, the voids will be less and the amount of water to be adsorbed will be greater. Since the volumes of two spheres are proportional to the cubes of the radii, one volume of graphite will hold seven volumes of water if the thickness of the water film is equal to the radius of the graphite particles. If the thickness of the water film is 1.2 times the radius, the graphite will hold eleven volumes of water. This is the same type of calculation that Hatschek made and it shows that it is theoretically possible on this assumption to account for zero fluidity in a graphite-water mixture containing 5.5 percent graphite. It has not been shown, however, that the adsorbed films of water on the graphite particles are of the desired thickness, nor has it been shown that 47 percent of the water in the mixture is in a different state from the rest of the water. It might be possible to do this last by measuring the vapor pressure curve for graphite- water mixtures. If the suspended particles aggregate into chains, the viscosity will be increased very much. If the particles form larger spheri- cal particles which are homogeneous, there will be a decrease in the viscosity, because of the decrease in the surface and conse- quently in the amount of bound water. If, however, the particles simply agglomerate loosely into spherical masses, the viscosity will increase because the water in the voids inside the spherical agglomerates no longer counts as free water. We shall therefore expect to get an increase of viscosity as a result of agglomeration when the effect of agglomeration is not to increase the size of homogeneous drops. One very interesting case of the effect of viscosity and pre- sumably of the adsorbed water film occurs in the settling of slimes or pulps. Here the particles vary in size from those PROPERTIES OF COLLOIDAL SOLUTIONS 193 coarse enough to settle moderately rapidly down to those so fine that the Brownian movements tend to keep them up com- pletely. If there is a large amount of water present and the pulps are consequently very dilute, each particle separates practically unaffected by the others. The upper part of the liquid remains cloudy owing to the presence of the very fine parti- cles, there is no sharp line of demarcation in the liquid, and there is, or may be, a sharp line at the bottom separating the coarser particles which have settled and the supernatant liquid. 1 With increasing concentration the coarser particles tend to interfere with the finer ones and to carry them down. The suspended pulp seems to settle as a whole, the upper portion of the liquid is practically clear with a sharp line separating it from the settling mud, and there is no visible dividing line at the bottom of the vessel between the slimes which have settled and those which are settling. According to Free the change from one type to the other came at about 8-9 percent with the kaolin suspension that he took, the higher concentrations giving what he calls consolida- tion settling and the lower concentrations subsidence settling. When a given volume of pulp settles, an equal volume of water must rise and when the pulp is fairly viscous, it acts practically as a filter, removing all the finest particles. 2 Deane says that the effect of reducing the resistance to the upward flow of the water by creating a channel for the passage of the liquid can be shown very well by placing a sample of pulp in two test-tubes, one of which is held vertically, while the other is inclined slightly. In-addition to the increased clarification in the inclined tube due to the increased settling area, it will be noticed that the pulp in settling away from the upper surface of the tube leaves an open channel which facilitates the free passage of the liquid. The rising current of the liquid on the upward surface and the settling of the solid on the lower surface is quite marked. If wires are placed vertically in a tank and are given rapid vibrations of low amplitude, this will cause channelling and will hasten thickening by making the upward flow of the liquid easier. The dairymen have noticed that, contrary to what one would 1 OWENS: Geographical Journal, 37, 68 U911); FREE: Eng. Min. Jour., 109, 601 (1916). 2 DEANE: Trans. Am. Electrochem. Soc., 37 (1920). 13 194 APPLIED COLLOID CHEMISTRY expect, cream rises more rapidly and more completely in a deep vessel than in a shallow pan. 1 The finer particles of butter fat are caught by the coarser ones and swept upwards because the concentration of the coarser particles per unit of cross-section soon becomes high enough to have a filtering effect. The differ- ence between this and the ore pulp is that the cream rises and filters upwards, while the pulp settles and filters downwards. In neither case has anybody made any experiments as to any possible relation between the concentration at which the filtra- tion becomes effective and the concentration for zero fluidity. Arisz 2 has made a number of valuable experiments on the vis- cosity of gelatine peptized at 50-60 by glycerine of 1.176 specific gravity, corresponding to about 32 percent of water in the glycerine. He finds that above 65 there is a continuous decrease in the viscosity of the gelatine sols. At 95 the viscosity of a 10 percent sol dropped in thirty minutes to 92 percent of the viscosity which it had three minutes after reaching 95. At 85 the drop was only to 97 percent in thirty minutes and at 75 to 99 percent. At 65 there was no change in viscosity in twenty-four hours. Arisz believes, apparently properly, that the change above 65 is due to a progressive decomposition of the gelatine which becomes negligible at and below this temperature. At 65 equilibrium is reached in less than three minutes and at lower temperatures with ever-increasing slowness. For 10 percent gelatine in the standard glycerine, Arisz considers that the equilibrium values for the viscosity as determined in the Ostwald viscosimeter are: 222 at 65; 415 at 55; about 950 at 50; less than 4200 at 47; about 5000 at 46; and more than 30,000 at 44, which is about the temperature at which the sol sets to a jelly. Since equilibrium is reached slowly at the lower temperatures, the viscosity of the sol will be less than the equilib- rium value if the sol is cooled say from 65 to 55 and the viscos- ity will increase with time. If a sol in equilibrium at say 46 is heated to 55, the viscosity will be greater than the equilibrium value and will decrease with increasing time. Until equilibrium is reached, the viscosity of a given gelatine-glycerine sol will thus 1 McKAY and LARSEN: Principles and Practice of Butter-making, 125 (1906). 2 Kolloidchemische Beihefte, 7, 1 (1915). PROPERTIES OF COLLOIDAL SOLUTIONS 195 depend at temperatures below 65 on the previous history of the sample and on the time that it has been held at the temperature in question. These gelatine sols show the Tyndall beam and Arisz finds that the intensity of the scattered light increases with falling temperature and with increasing time, provided the viscosity is below the equilibrium value. Cooling a 1 percent gelatine- glycerine sol from 70 to 20 caused a four-fold increase in the intensity of the scattered light. When a similar solution was cooled to 18 and held there, the intensity of the scattered light at the end of nine days was nearly six times the initial value at 18. There is therefore a parallelism between the changes in the Tyndall effect and the changes in the viscosity of these sols, and consequently part at least of the viscosity changes must be due to an agglomeration of the particles. Arisz considers that the small drops or particles have run together to form larger drops; but this cannot be true because that would mean a decrease in the viscosity and not an increase. The increase in the apparent size of the particles which one must assume in order to account for the Tyndall effect can be due equally well to an agglomeration of particles or drops without having them coalesce to a homo- geneous particle or drop. That this is what actually happens is made probable by the experiments of Meyer 1 on starch. This of course does not exclude the possibility of viscosity changes due to displacement of equilibrium between two forms of gelatine 2 in case these exist. The important point is that the increase in viscosity does go hand in hand with an increase in agglomeration which'must mean that increasing agglomeration involves decrease in the amount of available free water. The phenomenon is apparently general because .Freundlich 3 has made use of the increase in viscosity as a means of studying the rate of agglomera- tion of alumina sols. With colloidal solutions of gelatine and other substances of the same type, the viscosity changes with the time, showing that there is a gradual change in structure. This is further confirmed by the fact that the viscosity of such solutions changes when they 1 Kolloidchemische Beihefte, 5, 1 (1913). 2 SMITH: Jour. Am. Chem. Soc., 41, 135 (1919). 3 Trans. Faraday Soc., 9, 66 (1913). 196 APPLIED COLLOID CHEMISTRY are shaken violently. Until we know more than we do now about the physical state of these colloidal solutions, it is distinctly unwise to predict anything very definite about their viscosities. Where a film formation is possible, the surface viscosity will be quite different from that in the mass of the liquid. This is very striking in the case of colloidal solutions of saponin, peptone, etc. With saponin the surface viscosity is 60 times that of the mass of the liquid. 1 With peptone the film is so solid that the surface viscosity cannot be measured. COLOR It will simplify matters not to limit the discussion to the colors of colloidal solutions, but to consider something about the colors of colloids in general. The colors which depend on phys- ical structure and not on pigments are known as structural colors and are often connected with the presence of colloidal material. An absolutely smooth reflecting surface is invisible. It is only when the specular reflection is changed to some extent to diffuse reflection that we see the surface/It was suggested that the Zeppelins should be coated with a polished surface of metallic chromium so as to make them invisible. While it would be impossible to see such a surface, almost anybody would realize that something queer was going on if he should happen to see a reflected portion of the earth's surface apparently moving along in the sky./ While this last might never happen, the reflection of light from the polished surface would more than counter- balance the advantages of the theoretical invisibility. A substance is also invisible if there is no reflection or refraction at its surfaces. If we suspend a transparent powder in a trans- parent liquid having the same index of refraction, the powder will be invisible. Some glass powders in carbon bisulphide are practi- cally invisible. If a glass rod is dipped into a mixture of olive oil and cassia oil, the rod appears only to reach to the surface of the liquid for it cannot be seen in the liquid. 2 It may happen, however, that the dispersion of the liquid mixture differs from that of the solid, in which case there will be an agreement of 1 MALCOLM: Phil. Mag. (6) 12, 508 (1906). 2 VON BEZOLD: Die Farbenlehre, 62 (1874). PROPERTIES OF COLLOIDAL SOLUTIONS 197 refractive index only for certain wave-lengths, which will be transmitted. We may therefore get the transmission of colored light, a phenomenon first studied by Christiansen. 1 When clean powdered glass is immersed in carbon bisulphide and ben- zene is added, the transmitted light is first red, then yellow, green, and blue in succession. For the same mixture, the trans- mitted light changes towards the blue with rising temperature. 2 In some cases the light is extremely pure, the range of ref rangibility being only two and a half times that included by the two D lines. 3 If we have a thin film with light reflected from the front and the back surfaces, it may happen that the crest of certain waves reflected from one surface will coincide with the hollow of waves reflected from the other surface, in which case this particular color will disappear and the place where this disappearance oc- curs will appear to have the complementary color. The colors of thin films 4 are often known as Newton's Rings. We are quite familiar with them in the case of soap bubbles and with oil films on the ground. In the streets of Washington the colors of the oil films are exceptionally brilliant; but this is not due to any particular brand of oil which is spilled there. It is because the streets are asphalted and consequently there is practically no reflection of light from the asphalt surface to interfere with the colors of thin films. This is a very good illustration of the im- portance of a dark background in the case of structural colors. The colors of tempered steel, of lead skimmings, of Nobili's metallochromes, and the flashing colors of certain insects' wings are further illustrations of the colors of thin plates. If black ice is struck with a heavy stick, the colors of thin plates appear in great beauty at the fractured surfaces. We may get colors by reflection and refraction. The rainbow is an illustration of this. The old-fashioned cut-glass chandelier gives a good example of the colors that may be produced by light passing through a prism. If we have a number of lines ruled parallel and very close together on a smooth surface, this gives us what is known as a diffraction grating, which may spread a ray 1 Wied. Ann., 23, 289 (1884). 2 WOOD: Physical Optics, 110 (1911). S RAYLEIGH: Proc. Roy. Inst., 16, 116 (1889). * TTNDALL: Electricity and Light, 97 (1895). 198 APPLIED COLLOID CHEMISTRY of light out in a way very similar to a prism. Agates are com- posed of thin films about 0.001 of a millimeter in thickness, and when the agate is properly cut it may behave like a diffraction grating. 1 In the case of mother-of-pearl, we have alternate layers of calcium carbonate and an organic material. These layers overlap very slightly in a regular fashion and consequently act to some extent like a diffraction grating, giving rise to the peculiar effect known as pearly lustre. 2 The reflection of light by powdered material may give rise to a number of interesting color phenomena. When white light is scattered from a surface instead of being reflected as in a mirror, it gives rise to the sensation of white. A block of ice is not white, because it does not scatter the light. If the ice is powdered, or still better if we have snow, the light is scattered and we call the snow white. Instead of having solid particles of ice in air we may have air bubbles embedded in liquid, and then we get a white froth or foam. If the blue crystals of copper sulphate are ground to a fine powder, the light passes through such fine layers of the material that it does not become colored blue to any appreciable extent. On the other hand, the light is scattered from the pow- dered surfaces, and the powdered copper sulphate looks white instead of blue. When silver is precipitated by an electric current, it does not come down with a mirror-like surface, but rather in a mass of tiny crystals which scatter the light in every direction. Consequently electrolytic silver is white and only has the characteristic appearance of silver after it has been bur- nished. There is no white pigment in any white flowers, or in white hair, and probably not in white bark. The white color of the lily is due to the presence of innumerable air bubbles and the same is true of white hair. In order that hair may turn white in a single night, it is not necessary for the original pigment to disappear and for white pigment to develop. It is only necessary for a mass of minute air bubbles to be formed in the hair as a result of worry. While this accounts for the physical change, we are as much at a loss as ever to know why intense grief should develop air bubbles in the hair. The scattering of light by a powder or by the fibers of a sheet 1 GOODCHILD: Precious Stones, 165 (1908). 2 Pfund: Jour. Franklin Inst. 183, 453 (1917). PROPERTIES OF COLLOIDAL SOLUTIONS 199 of paper is due to the difference in the index of refraction between the solid particles and the air. If we replace the air by some liquid having very nearly the same index of refraction as the solid particles, these latter will cease to be white and will become more transparent. It is well known that oiled paper is distinctly translucent. If powdered copper sulphate were immersed in a liquid having about the same index of refraction as the crystals, these would probably become blue. This experiment has not been tried; but there is a more interesting one which has puzzled people for a good while. In the zoological gardens some flamingoes are much redder than others. At one time it was thought that the redness was connected with a fish diet; but flamingoes have been raised which had bright red feathers even though the birds had never eaten fish, so this explanation had to be given up. It is now believed that the brilliancy of the color is due to the presence of oil in the feathers, this making them more translucent and bringing out the red color. 1 Water colors always become paler as the water dries out, because more light is then scattered. Some people have gone so far as to say that white pigments do not occur in nature in animals or plants; 2 but this is an over-state- ment. Some butterflies have a whitish pigment in their wings 3 and the silvery whiteness 4 of many fishes is due to the presence of a substance called guanin, and riot to air bubbles. Complete absorption of light gives the sensation known as black. Pulverulent silver is black because the crystals are so arranged that all the light is absorbed by reflection forward and back. This can be shown in another way. If we take a bunch of steel needles and place them with their points side by side facing the observer, the mass will appear black, 5 although we know that no single needle is black in itself. The rich color of velvet is due to its trapping the light, and the peculiar effect known as damask is due to the fibers being arranged in two different ways, so that one set absorbs more light than the other. Trees and grass may act as light traps when seen from above and they are said to look 1 KRUKENBERG: Vergleichend-physiologische Studien, 1 V, 90 (1887). 2 GADOW: Proc. Zool. Soc., London, 1882, 410. 3 HOPKINS: Phil. Trans., 186B, 661 (1895). 4 NEWBIGIN: Color in Nature, 37, 219 (1898). 5 WOOD: Physical Optics, 449 (1911). 200 APPLIED COLLOID CHEMISTRY black to aviators. 1 If we have a material which absorbs light so completely that the powder is black, we shall get all gradations between the original color and black if we grind up such a sub- stance, just as we got all variations between blue and white by grinding crystals of copper sulphate. If we start with yellow gold and make it more porous, it will become brown before becoming black. This happens unintentionally in the assaying of gold when a brown powder is obtained which people have thought was an allotropic form of the metal. It is merely a porous gold and when it is heated until the crystals sinter together and become less porous, the yellow color of gold reappears. If we have very fine particles suspended in a transparent solid, liquid, or gas, these particles scatter blue light much more than they do red light, and consequently such a mass appears red by transmitted light and blue by reflected light. Skimmed milk is an instance of this sort, being distinctly bluish by reflected light and reddish by transmitted light. Tobacco smoke is blue by reflected light and red by transmitted light. The blue color of the sky is due to light which is scattered by drops of liquid or by particles of dust in the air. The intensity of the color is un- doubtedly increased by the fact that we see this against the black background of infinite space. The sunset colors are due to the light which is transmitted and not scattered. There is no blue pigment in blue eyes and it was pointed out by Tyndall 2 years ago that the blue of the eye is really the blue of turbid media, and is thus analogous to the blue of the sky or the blue of skimmed milk. At the back of the iris there is a dark pig- ment known as the uvea, which prevents the reflection of light and prevents the color of the blood from being seen. When this dark pigment is absent, we have an albino with a pink eye. The various stages between the blue and the gray eye are due to differences in the coarseness of the particles giving rise to the blue color, the blue color being the most intense the finer the particles. This is probably the reason why babies' eyes are so very blue, because the suspended particles tend to grow coarser with increasing age. All other people have a yellowish-brown pigment in the front of the iris and the combination of the struc- : Jour. Franklin Inst., 187, 303 (1919). 2 Fragments of Science : The Sky. PROPERTIES OF COLLOIDAL SOLUTIONS 201 tural blue with the yellowish-brown pigment gives rise to the green, hazel, brown and black eyes. Except with people who have very black eyes, the pigment in the front of the iris does not develop at birth, just as the teeth do not come till later. Consequently many infants have blue eyes, which change to hazel, brown or black as they grow older, while the reverse change never occurs. f Once in a while we see a man whose eyes are colored differently^ This means that pigment has developed in one eye and not in the other, or that the pigment has developed unequally in the two eyes. The green eye is due to the combination of structural blue with the pigment yellow. While this is not a common type in human beings, we get it very markedly in the case of the green tree-frog, who has no green pigment and whose green color is due to struc- tural blue with a yellow pigment overlaying it. If we scrape the pigment layer off the back of an unfortunate frog, he turns blue. In fact, the grass-green Australian tree-frog is called Hyla coerulea, because he is blue when he comes to us preserved in alcohol, the alcohol having dissolved the yellow pigment. There is a wide-spread belief that people with blue or gray eyes are better rifle shots than people with dark eyes. Since we do not see through the iris, this could not be cause and effect, though it is theoretically possible that the two characteristics might be correlated. Colonel Chamberlain has made a study of tjiis matter and finds that of three thousand soldiers who had qualified for the grade of marksman or better, approximately two thousand had blue or gray eyes, a result which seems at first to bear out the popular belief. He finds however that of one hundred thousand white soldiers approximately two thirds have blue or gray eyes. People have made the mistake of considering abso- lute figures and not percentage. If we leave out of account feathers with a blue metallic lustre, which may constitute a special case, we may say that there is no blue pigment in the feathers of any bird. This means that the blue of the kingfisher, the indigo bunting, the blue-jay, and the blue-bird, is not due to blue pigment; but is a structural color. The only pigment in the blue feathers of these birds is a dark brown one which apparently serves merely as a background, just as in the case of the blue eyes. The best explanation of the blue 202 APPLIED COLLOID CHEMISTRY feather 1 is that the horny matter is filled with an enormous num- ber of minute air bubbles, which scatter blue light and transmit red. The blue feather is therefore the same in principle as, or the opposite in detail from, the blue sky. The color of the sky is due to the scattering of light by particles of liquid or solid suspended in a gas, the air. The blue of the feathers is due to the scattering of light by bubbles of gas, air, suspended in a solid medium. While this explanation is undoubtedly the right one, nothing analogous to a blue feather has yet been made in the laboratory. This is one of the things that people must do in the future. Apparently nobody has made any experiments to de- termine whether the blue light from the blue eyes and the blue feathers is polarized. This should be the case. In this country the tufted titmouse is a gray bird with no brilliant colors; but the German titmice show great differences in color. One variety is green on the back, due to structural blue overlaid by a pigment yellow, and yellow on the belly. In another variety the brown pigment which is essential to the structural blue has not developed and the bird is consequently yellow all over. There is a third form in which the yellow pig- ment has not developed and the bird is therefore blue on the back and white underneath. One would like to think that the differences in color between the male and the female redstart were due to the presence in the male of a single color which was lacking in the female; but this cannot be the case, unless it can be shown that the black on the male redstart is due to a red pigment just as hematite is black in mass. Tyndall, 2 Aitken 3 and Spring 4 consider that the color of water is blue, while Rayleigh 5 believes that the blue of the sea is merely the light reflected from the blue sky. Assuming that water is in itself blue, it would look black if no light were re- flected back from the interior. A difference in the index of refraction due to difference of temperature might be sufficient ; Zool. Jahrbiicher, Abt. syst. Geog. Biol. Thiere, 16, 267 (1902). 2 Fragments of Science : Voyage to Algeria. 3 Proc. Roy. Soc. Edin., 11, 472 (1882). 4 Bull. Acad. roy. belg. (3) 5, 55 (1883); 12, 814 (1886). 6 Nature, 83, 48 (1910). PROPERTIES OF COLLOIDAL SOLUTIONS 203 but a more usual cause is the presence in suspension of reflecting particles. If these are few in number the water will appear indigo to blue. If they are many in number we shall begin to get the yellowish to red color of light transmitted through a turbid medium and the water will appear green. Muddy waters may appear yellow or brown. As a matter of fact, the deep sea which contains little suspended matter, is blue, while near shore where there is more suspended matter, the color is often a brilliant green, especially in the tropics. The presence of air bubbles will act in the same way and in an indigo sea the crest of a wave may be green. The Lake of Constance is quite green because of the suspended calcium carbonate, while the Lake of Geneva is more blue because there is less suspended matter. The Rhine at Strasbourg is green and contains 70 percent more calcium carbonate than the blue Rhone at Geneva. The Gulf Stream is blue because there is less carbon dioxide in the warm water and consequently less calcium carbonate. The clear brown color of some New England brooks is probably due to the presence of tannin-like materials. The colloidal metals are a special and very interesting case of the colors of colloidal solutions. Colloidal gold solutions 1 can be prepared which are red, violet or blue by transmitted light; and colloidal silver solutions 2 which are yellow, red or blue. In general the blue gold solutions contain coarser particles than the red ones, but this is not always true. The peculiarities in thp colors of colloidal metals seem to be connected intimately with the fact that the metals show selective reflection. In the case of selective reflection, some or all of the light which is adsorbed very strongly by a given substance is reflected strongly from a polished surface of that substance. The surface color of magenta is green, while the body color or the color due to the transmitted light is red. With indigo the surface color is red and the body color blue. The surface color is believed to be due to resonance, the substance emitting the rays which it adsorbs very strongly. It seems probable that a very small granular particle might emit surface color by resonance on the opposite side from the source of light, in which case we should have the surprising phenomenon 1 FARADAY: Phil. Trans., 147, 145 (1857). 2 CAREY LEA: Am. Jour. Sci. (3) 37, 475 (1889), 204 APPLIED COLLOID CHEMISTRY of a substance apparently transmitting the light which it adsorbs most strongly. On the other hand, the amount of reflection from the back surface of a particle of the light ordinarily trans- mitted will be greater, the less the thickness of the particle. Consequently, with fine particles there will be a more or less complete reversal of the usual colors. The light which is ordinar- ily reflected selectively will be transmitted by resonance, while the light which is ordinarily transmitted is scattered. This is known to be true for indigo and for gold. Colloidal indigo solutions transmit red light and not blue. Massive gold reflects yellow when compact and brown to black when porous. Particles which do not resonate, are yellow or brown by reflected light and transmit blue light. Massive gold is red by multiple reflection and thin films are green by transmitted light. 1 Col- loidal solutions when containing very fine particles reflect green and transmit red. In the red gold solutions the particles are green, while they are yellow to brown in the blue solutions. The violet solutions are a mixture of red and blue solutions. A thin film of gold does not resonate readily and consequently transmits green light. If heated, it becomes granular and then transmits purple light. Gelatine films containing gold may be red by transmitted light when moist and blue when dry. The contrac- tion due to drying brings the particles of gold sufficiently near together to be equivalent to reversible agglomeration, thereby destroying the resonance. In a colloidal gold solution a single particle of a given size might resonate and transmit red light, while a cluster of finer particles might not resonate and would therefore transmit blue light. This is probably the explanation for some of the abnormal results observed. Silver reflects yellow after multiple reflection, while with fewer reflections the light is said to be red. 2 A very fine colloidal solution transmits yellow and scatters blue, while somewhat coarser particles transmit red and are blue by reflected light. The coarsest colloidal silver does not resonate and consequently is yellow by reflected light, transmitting blue light. This is exactly what should happen on the facts as stated. The pink color in a given sample of white lead was shown by 1 BEILBY: Proc. Roy. Soc., 72, 226 (1903). 2 REGNAULT-BETTON: Elements of Chemistry, 366 (1874). PROPERTIES OF COLLOIDAL SOLUTIONS 205 Baker 1 to be due to silver, which was probably present as col- loidal silver. The colors of Carey Lea's photohalides are known to be due to colloidal silver 2 and the difference in the color of photographic plates 3 with different developers is due to the degree of fineness of the precipitated silver. Colloidal platinum 4 has been obtained blue, green, yellow, red, brown and black; but there are no data which will enable us to discuss these colors satisfactorily. Owing to differences in the variation of the absorption with varying thickness of the absorbing agent, a substance may be one color in thin films and another in thicker masses. Thin layers of a cyanine solution appear blue and thick layers red, 5 and the same is true of cobalt glass. Potassium ferricyanide is a brown-red in coarse crystals and yellow when powdered. Andersen 6 has shown that thin plates of hematite (0.1/x) are yellow by transmitted light, the color varying with increasing thickness through reddish brown to deep brown red or blood red. An anhydrous, yellow ferric oxide is said to be obtained by the oxidation of ferric sulphide 7 and we get yellow bricks on burning a highly calcareous clay even though the iron content is higher than that which would give a red clay if the lime were not pres- ent. 8 It seems probable that the ferric oxide is present in such a fine state of subdivision that it is yellow and not red. Mott 9 has prepared both red and yellow ferric oxide by volatilization of iron in the electric arc, the yellow being the finer powder. Addition of a liquid having approximately the same index of refraction as yellow ferrocyanide or yellow rouge should make 1 Phil. Mag. (4) 37, 344 U869). 2 CAREY LEA: Am. Jour. Sci. (3) 33, 349, 489 (1887). SLUPPO-CRAMER: Zeit. Kolloidchemie, 7, 99 (1910); 8, 240 (1911); CHAP- MAN JONES: Phot. Jour., 51, 159 (1911). 4 GARBOWSKI: Ber. deutsch. chem. Ges., 36, 216 (1903); GUTBIER: Zeit. Kolloidchemie, 5, 50 ^1909); WOHLER and SPENGEL: Ibid., 7, 243 (1910); BREDIG: Zeit. phys. Chem., 31, 258 (1899) 6 WOOD: Physical Optics, 438 (1911). 6 Am. Jour. Sci. (4) 40, 370 (1915). 7 DIAMOND: Jour. Soc. Chem. Ind., 37, 451 R (1918). "KEANE: Jour. Phys. Chem., 20, 734 (1916); SCHEETZ: Ibid., 21, 570 (1917). 9 Trans. Am. Electrochem. Soc., 34, 292 (1918). 206 APPLIED COLLOID CHEMISTRY the powder more transparent and redder. Experiments by Schenck 1 on the precipitation of hydrous copper and aluminum oxides make it seem probable that cupric oxide is blue in thin films and not black. Mott obtained a brilliant blue by volatiliz- ing a mixture of equal parts of manganese and copper. ELECTRICAL PROPERTIES If a particle suspended in a liquid adsorbs a free ion, the charged particle will then behave to some extent like an ion and will tend to move under electrical stress toward the cathode if the adsorbed ion is positive and toward the anode if the adsorbed ion is negative. 2 Reference has been made to the fact that silver bromide particles move to the cathode when stabilized by silver ions and to the anode when stabilized by bromine ions. By following the electrolysis under the microscope it is possible to determine the rate of movement of a suspended particle for a given difference of electrical potential. The rate of migration of colloidal particles is of the same order as that of ions. 3 Thus colloidal silver moves with a velocity of 2.0-3.8/i/sec. under a potential drop of one volt per centimeter, and colloidal gold at about 4.0^/sec., while the chlorine ion gives a value of G.S/x/sec. Billitzer 4 has attempted to determine absolute potential differ- ences by finding solutions in which a given colloidal metal did not move under electrical stress. While the method is good in principle, one cannot reason from the adsorption of a finely divided metal to the adsorption of massive metal. It seems pos- sible that definite results might be obtained if the problem were attacked with a better knowledge of colloid chemistry. Since the colloidal particle moves in one direction with reference to the water, the water moves in the opposite direction relatively to the particle. This can be shown if we keep the colloidal particles stationary in the form of a diaphragm. The electrical transfer of particles through the liquid is called cataphoresis and the transfer of liquid through a stationary dia- phragm is called electrical endosmose. It was first observed by Mour. Phys. Chem., 23, 284 (1919). 2 See LINDER and PICTON: Jour. Chem. Soc., 71, 568 (1897). 3 ZsiGMONDY: The Chemistry of Colloids, 46 (1917). 4 Drude's Ann., 11, 902, 927 (1903). PROPERTIES OF COLLOIDAL SOLUTIONS 207 Reuss at Moscow in 1808. Working with so-called pure water it is found that water passes to the cathode when the porous diaphragm is made of glass, shellac, earthenware, carborundum, asbestos, wool, cotton, platinum, silver, arsenic sulphide, or sulphur; and to the anode when the diaphragm is chromic chloride, cobalt oxide, zinc oxide, or barium carbonate. 1 When turpentine was substituted for water, it moved to the anode with all diaphragms except sulphur. Coehn 2 has attempted to account for this by postulating empirically that when two non- miscible substances are in contact, one being a pure liquid, the substance with the higher dielectric constant is charged positively with reference to the one with the lower dielectric constant. This rule applies in the case of water and turpentine where the dielec- tric constants are 81 and 5 respectively, but it is by no means absolute even for pure liquids, 3 and breaks down completely when applied to aqueous solutions. The phenomenon depends on the relative adsorption of ions and Coehn' s rule is a useful first approximation in certain cases. The first quantitative study of electrical endosmose with solu- tions was made by Wiedemann 4 who deduced two empirical generalizations : 1. The mass of liquid transported through a porous diaphragm in a given time is directly proportional to the current. For a given diaphragm material and given current, the amount is independent of the length and cross-section of the diaphragm. 2. For a given diaphragm material the difference in hydrostatic pressure maintained between the two sides of a porous diaphragm is proportional to the applied potential and is independent of the dimensions of the diaphragm. Helmholtz 5 considers that under most circumstances liquids and solids become charged electrically when brought into con- tact, the surface of the solid having the opposite charge from that of the liquid in contact with it, thus giving rise to a so-called 1 QUINCKE: Pogg. Ann., 107, 1 (1859); 110, 38 (1860). Actually Quincke forced water through the diaphragms and noticed the charge. 2 Wied. Ann., 64, 227 (1898); Zeit. Elektrochemie, 16, 586 (1910). 3 BRiGGs: Jour. Phys. Chem., 21, 204 (1917). Pogg. Ann., 87, 321 (1852); 99, 177 (1856). * Wied. Ann., 7, 337 (1879). 208 APPLIED COLLOID CHEMISTRY double layer. If the solid cannot move because it is a capillary tube or a diaphragm, the introduction from outside of a potential gradient will tend to cause a displacement of the electrically charged layer of liquid. If the liquid is not a perfect insulator, the displacement results in a continuous flow of liquid along the surface of the solid. 1 Since the movement of a suspended and electrically charged particle is in accord with Faraday's law and Kohlrausch's law, it follows that the movement of water relatively to the diaphragm will not be. Freundlich 2 has deduced an equation for the amount of liquid V transported in unit time through a porous diaphragm: E is the total potential drop through the diaphragm, D is the dielectric constant and 77 the viscosity coefficient of the liquid, e is the potential of the Helmholtz double layer at the solid- liquid interface, q is the cross-section and I the thickness of the diaphragm. If 7 is the specific resistance of the liquid, then the total resistance, R, of the diaphragm will be l/yq. Substituting for E by means of Ohm's law we may write the equation For a given liquid and diaphragm at constant temperature, e, rj, 7 and D are constant, so that V is proportional to the current and the term for the diaphragm disappears. This is Wiede- mann's first empirical generalization. The other can be deduced by introducing an expression for the hydrostatic pressure. Since the charge on the diaphragm depends on the nature of the ion adsorbed, it follows that an aqueous solution may flow one- way or the other through a given diaphragm, depending on the nature of the dissolved substance. This was brought out clearly by Parker 3 and was emphasized by Perrin, 4 who found a reversal in many cases when the solutions were alkaline instead of acid. J LAMB: Phil. Mag. (5) 25, 52 (1888). 2 Kapillarchemie, 226 (1909). 3 Johns Hopkins Dissertations, 31, 23 (1901); BRIGGS: Jour. Phys. Chem., 21, 235 (1917). 4 Jour. Chim. phys., 2, G01 (1904). PROPERTIES OF COLLOIDAL SOLUTIONS 209 Every diaphragm tends to become charged positively in an acid solution and negatively in an alkaline solution. Every ion of unlike sign tends to neutralize the charge on the diaphragm. Perrin believed that the effect of other ions was more marked the greater the valence; but this is true only in so far as the ion with the higher valence is adsorbed more strongly than the ion with a lower valence, 1 in other words, in so far as Schulze's so-called law holds. In some cases the agreement is very good. Thus in acid solutions, the flow through the diaphragm to the anode increases as the anion changes from quadrivalent ferrocyanide to bivalent sulphate to univalent chloride, because the adsorption of the anion decreases in the same order and the diaphragm be- comes charged more positively. In alkaline solution the flow to the cathode increases as the cation changes from trivalent lanthanum to bivalent barium to univalent sodium. On the other hand, with a porous cup in a slightly acidified copper sulphate solution the flow is to the cathode, apparently because copper ions are not adsorbed to any great extent, while the ad- sorption of sulphate ions exceeds the adsorption of hydrogen ions because the concentration of the first is so much greater than that of the second. With a slightly acidified copper nitrate solution the flow is to the anode, because the nitrate ions are not adsorbed very much and the hydrogen ions determine the direction of the flow. Much to Perrin's surprise a barium carbonate solution acted in one case as though it were an acid. This is undoubtedly because of the preferential adsorption of barium ions which charge the diaphragm positively. In electrolytic reductions of organic substances in aqueous alcoholic solutions using a porous cup, the flow of liquid to the cathode may be quite annoying. Owing to the lower dielectric constant of alcohol, the normal tendency of the liquid to go to the cathode becomes less, but the increased resistance would increase the flow and nothing is known in regard to adsorption from these solutions. Billitzer 2 found that colloidal platinum is charged positively in aqueous alcohol though negatively in pure water. 1 ELISSAFOFF: Zeit. phys. Chem., 79, 285 (1912). 2 Zeit. phys. Chem., 43, 312 (1903). 14 210 APPLIED COLLOID CHEMISTRY The amount of liquid transferred by electrical endosmose increases with rising temperature, the increase being nearly pro- portional to the change in the fluidity 1 (the reciprocal of the viscosity). This would follow from Freundlich's equation if one assumes that D and c change but slightly over the tem- perature range. The commercial applications of electrical endosmose have not been very successful. Count Botho Schwerin 2 took out a number of patents for removing water from peat. Starting with a peat mud containing about 90 percent moisture, he claimed that the moisture content could be reduced to 65 per- cent at an expenditure of about 15 kw.-hr. per cubic meter of water removed, the best potential gradient being 4-5 volts per centimeter. The cost of drying from 65 percent moisture to 20 percent moisture was so great that the process was not a success. It is reported that electrical endosmose has been used successfully in the dye industry in the drying of pastes. Elec- trical tanning has also been tried, the tanning liquor being carried into the hides by electrical endosmose. This seems to have failed in part on account of the oxidation of the tannin at the anode. Cataphoresis is made use of in electrolytic refining and plating, the use of addition agents being essential in some cases. 3 There are no very definite results as to the effect of colloids on the conductance of solutions, 4 but it is evident that an ad- sorbed ion cannot have as high a migration velocity as an un- adsorbed ion, and that consequently the presence of a colloid must decrease the conductance to some extent, depending on the amount and nature of the ions adsorbed. If a clay suspension is stirred, 5 the conductance may increase because the adsorbed ions which are carried down when the suspension settles are brought up between the electrodes. McBain 6 found that certain solutions of sodium palmitate conducted well, although the 1 BRIGGS, BENNETT and PIERSON: Jour. Phys. Chem., 22, 256 (1918). 2 Zeit. Elektrochemie, 9, 739 (1903). 3 MULLER and BAHNTJE: Zeit. Elektrochemie, 12, 317 (1906). 4 Wo. OSTWALD: Van Bemmelen Gedenkboek, 267 (1910). e BLEININGER: Trans. Am. Ceramic Soc., 15, 343 (1913). 6 Jour. Chem. Soc., 106, 965 (1914); Jour. Am. Chem. Soc., 42, 426 (1920). PROPERTIES OF COLLOIDAL SOLUTIONS 211 freezing points of the solutions were only enough to account for the sodium ions. He assumes that the palmitate ions polymerize to large hydrated units which he calls micelles and which he considers colloidal and to which he applies Stokes' law although he does not state specifically why or how the polymerized ions separate as a second phase. It is probable that we have hydro- lysis to caustic soda and colloidal palmitic acid, the latter 1 then adsorbing hydroxyl ions, just as we know that silicic acid adsorbs hydroxyl ions and is peptized by them. This suggests an inter- esting theoretical possibility. Suppose we have two sets of finely divided particles neither of which adsorbs the other appreciably and let us also suppose that one set of particles adsorbs a given cation very strongly while the other set of particles adsorbs a given anion very strongly. If we take a mixture of these two sets of particles and add a small amount of the salt of the given base and the given anion, we shall have a colloidal solution which will conduct electricity very well but which will contain no free ions to speak of because, by definition, the cations have been practically completely adsorbed by one set of particles and the anions by the other set of particles. It is not known whether these limiting conditions can all be fulfilled simultaneously. A possible case would be the mixing of dilute solutions of silver sulphate and lead chloride. The lead sulphate would adsorb sulphate or lead ions strongly and the silver chloride would adsorb chloride or silver ions strongly. Nobody knows what each would do to the other. While an adsorbed ion may increase the conductance of a liquid, it will behave quite abnormally as regards electrometric measurements. An adsorbed chlorine ion will not give a test with silver nitrate and consequently will have only a negligible effect on a chlorine electrode. It is not probable that an ad- sorbed hydrogen or hydroxyl ion behaves electrometrically like a free hydrogen or hydroxyl ion. This possible error has been overlooked in many cases where the hydrogen electrode has been used. Another source of error in electrometric measurements with colloidal solutions is to neglect the fact that we are not dealing with true solutions. If we measure a copper electrode in a solution of alkaline copper tartrate we find that the concen- 1 There may also be adsorption by colloidal sodium palmitate. 212 APPLIED COLLOID CHEMISTRY tration of copper ions is extremely low. That is all that the measurement tells us. If the copper is in true solution, it must be present as a complex salt, as is unquestionably the case with potassium silver cyanide. If we have peptized copper oxide or hydroxide present, there is no need to assume the existence of a complex copper salt at all. The true interpretation of the phe- nomena cannot be made until we know whether a so-called alka- line copper tartrate solution is a true solution or not. We cannot answer this question at present. On the other hand, it seems practically certain that the effect of sugars in preventing the precipitation of the heavy metal hydroxides by alkalies is due to the formation of colloidal solutions. Consequently the conclusions drawn from electrometric measurements of such solutions are necessarily false. STABILITY OF COLLOIDAL SOLUTIONS 1. Sols stabilized by Ions. Three classes of colloidal solutions have been distinguished, in which the stabilization is due: to adsorbed liquid; to an adsorbed undissociated substance which may be in true solution or may be itself in colloidal solution; and to an adsorbed ion. The last case is the simplest theoreti- cally and the most varied experimentally. So long as the parti- cles are all charged positively or all charged negatively, they will tend to repel each other and will not coalesce unless the concen- tration is too high. If the charge is neutralized or counter- balanced in any way, the particles will agglomerate 1 unless some other factor comes into play. Such colloidal solutions will therefore be very sensitive to electrolytes having a readily ad- sorbed ion of the opposite charge from that which stabilizes the colloidal solution. Colloidal solutions which are peptized by the liquid will be relatively insensitive to electrolytes because these will act only in so far as they decrease the peptizing action of the liquid. The second case is intermediate with reference to the other two and its properties will vary between those of the other two types. We will consider first the case in which the colloid is stabilized 1 HARDY: Zeit. phys. Chem., 33, 285 (1900); BURTON: Phil. Mag. (6) 12, 472 (1906); 17, 583 (1909). PROPERTIES OF COLLOIDAL SOLUTIONS 213 chiefly or entirely by the presence of an adsorbed ion. Such solutions precipitate completely at the iso-electric point, as it is called, where the electrical charge is neutralized or counter- balanced. If a suspension is stabilized by the preferential adsorption of hydrogen ion from hydrochloric acid solution, the solution contains free hydrogen ions, free chlorine ions, and the adsorbed hydrogen ions which make the suspension behave like a cation though with a different migration velocity from that of hydrogen. If the suspension is made to adsorb an anion in an amount equivalent to the hydrogen ion adsorbed, the sus- pended particles will be electrically neutral. This can be done by adding an electrolyte with a readily adsorbed anion. Since this is a matter of selective adsorption, the concentration of the added anion necessary to cause an adsorption equivalent to the hydrogen adsorption will vary with each anion. To put the matter more generally, the amount of an electrolyte necessary to precipitate a colloidal solution will vary with the nature of the cation, the anion, and the dispersed phase. While it is generally true that an ion of higher valence will be adsorbed more strongly than one of lower valence, this so-called law of Schulze 1 is only a first approximation and should be considered only as a guide. That some univalent ions are adsorbed more strongly by some substances than some bivalent or trivalent ions is shown clearly in data by Ode*n 2 on colloidal sulphur, given in Table XLIII. In the second column are the liminal concentrations necessary to coagulate the sulphur, given in gram ions per liter of the cations; in the third column are given the reciprocals of these values, the so-called ionic precipitating power. Under the conditions of OdeVs experiments, sulphur is a negative colloid and the precipitation is therefore due to an adsorption of cations. The first thing to be noticed is that hydrogen ion is not adsorbed strongly by sulphur, the precipitat- ing power of hydrochloric acid being much less than that of lithium, ammonium, sodium, potassium, rubidium, or caesium chloride. Instead of these univalent cations precipitating at the same concentration, the required concentration of lithium chlo- ride is in round numbers one hundred times that of caesium 1 Jour, prakt. Chem. (2) 25, 431 (1881); 27, 320 (1884). 2 Der kolloide Schwefel, 156 (1912). 214 APPLIED COLLOID CHEMISTRY TABLE XLIII. COAGULATION OF SULPHUR AT 18-20 Salt Liminal value gram- ions cations per liter Ionic precipitating power of cation HC1 6.0 0.16 LiCl 913 1 1 NH 4 C1 435 2 3 (NH 4 ) 2 SO 4 0.600 1.7 NH 4 NO 3 0.506 2.0 NaCl 153 6.1 Na 2 SO 4 0.176 5.7 NaNO 3 0.163 6.1 KC1 . . . 021 47.5 K 2 SO 4 025 39 7 KNO 3 RbCl 0.022 0.016 45.5 63.0 CsCl 009 108 MgSO 4 0.0093 107.5 Mg(NOs)a 0.0080 125.0 CaCl 2 0041 245 Ca(NO 3 ) 2 0.0040 247.0 Sr(N0 3 ) 2 BaCl 2 0.0025 0021 385.0 475.0 Ba(NO 3 ) 2 0.0022 461.0 ZnSO 4 Cd(NO 3 ) 2 0.0756 0493 13.2 20.3 A1C1 3 0044 227 CuS0 4 Mn(NO 3 ) 2 Ni(NO 3 ) 2 0.0098 0.0096 0446 102.0 105.0 22.4 UO 2 (NO 3 ) 2 0.0137 73.0 chloride. The liminal values for barium and strontium are nearly equal, but calcium chloride requires a distinctly higher concentration. If we take the different bivalent ions the values range from 0.0756 for zinc to 0.0022 for barium, a ratio of over thirty to one. The univalent caesium ion has a greater precipi- tating power than the bivalent zinc, cadmium, nickel, and uranyl ions; and about the same precipitating power as the bivalent copper, manganese, and magnesium ions. The trivalent alumi- num ion has about the same precipitating power as the bivalent calcium ion and distinctly less precipitating power than bivalent strontium and barium ions. The specific nature of the adsorp- PROPERTIES OF COLLOIDAL SOLUTIONS 215 tion comes out extraordinarily clearly with sulphur, about the only orthodox thing being that nitrate, chloride, and sulphate behave practically alike, though' even here Ode*n considers that sulphate has a slight protecting action. This specific nature appears more clearly perhaps if we arrange the cations in order, the one with the greatest precipitating power coming first: Ba, Sr>Ca, Al>Mg, Cs, Mn, Cu>U0 2 >Rb>K>Ni, Cd, Zn>Na>NH 4 >Li>H. Sulphur is admittedly an extreme case, but Freundlich 1 gives data for colloidal platinum from which I deduce the order: Al, Pb>Ba, U0 2 >Ag>K, Na. Bivalent lead has practically the same precipitating power as trivalent aluminum. Univalent silver is nearer to bivalent uranyl and barium than to univalent potassium and sodium. If more cations had been studied we should very likely have got more distinct evidence of specific action. As it is, it takes 130 millimols NaOH per liter to coagu- late the platinum and only 2.5 millimols NaCl. The change from chloride to hydroxide has a more marked effect than the change from sodium to barium. It seems very probable that barium hydroxide would have no greater precipitating power than sodium chloride. From Pappada's experiments with colloidal silver 2 I deduce the following order of adsorption: Al>Ba, Sr, Ca>H>Cs>Rb>K>Na>Li. From these data Pappada con- cludes that the migration velocity is the determining factor with the univalent cations; but this cannot be true. The differ- ence between aluminum and hydrogen is not very great, one drop of M/10 HC1 producing a coagulation and one drop M/20 A1C1 3 . In tenth-normal solutions potassium iodide, nitrate, and sulphate produce no coagulation. The reason given by the author is that these anions react with the colloidal silver. In normal solutions the iodides, nitrates, and sulphates are said to precipitate at the same concentrations as the corresponding chlorides and bromides. The effect of concentration is a little obscure in other respects, since 5 or 6 drops of normal KC1 pre- cipitate 2 cc 0.06 percent Ag, whereas it takes only 30 drops N/10 KC1 to produce precipitation. The essential thing is that the different univalent cations have different liminal values; the 1 Kapillarchemie, 352 (1909). 2 PAPPADA: Gazz. chim. ital., 42 I, 263 (1912). 216 APPLIED COLLOID CHEMISTRY difference between hydrogen and lithium is greater than that between hydrogen and aluminum. Experiments might well have been made with barium hydroxide and with the sodium salts of organic acids. Freundlich and Leonhardt 1 find that with the negative colloid, MoO2.H 2 O, caesium is two hundred times as effective as lithium, and barium four times as effective as calcium. TABLE XLIV. COAGULATION OF MASTIC Salt Liminal value gram atoms Ionic precipitating power of cation NaCl 1.0 1 AgNO 3 125 8 HgNO 3 0.00125 800 HC1 CaCl 2 0.010 025 100 40 BaCl 2 025 40 ZnSO 4 A1 2 (SO 4 ) 3 . . 0.050 0004 20 2500 A1 2 (NO 3 ) 3 0004 2500 FeCl 3 0.0003 3300 From experiments on mastic 2 we get the data given in Table XLIV. If we consider the mercury in mercurous nitrate as a univalent ion, it- is very much out of place, precipitating at much lower concentrations than the barium, calcium and zinc salts. Of course the formula should be written Hg 2 (NO 3 )2 with Hg2 as a bivalent ion. 3 In this case the precipitating power becomes 1600 instead of 800, which puts it up much nearer the trivalent cations than the bivalent ones. The order of cations is: Fe, Al>Hg 2 >H>Ba, Ca>Zn>Ag>Na. Only three anions are given in the table, so it is impossible to tell what effect the anions have. A good many experiments have been made on mastic with different acids but the degree of electrolytic dissociation varies so as to make these results inconclusive. With Prussian blue Pappada 4 found the order of the cations to be : Fe, Al, Cr > 1 Kolloidchemische Beihefte, 7, 172 (1915). 2 FREUNDLICH: Kapillarchemie, 367 U909). 3 OGG: Zeit. phys. Chem., 27, 285 (1898). 4 Zeit. Kolloidchemie, 6, 83 (1911). PROPERTIES OF COLLOIDAL SOLUTIONS 217 Ba, Cd > Sr, Ca > H > Cs > Rb > K > Na >Li. Sulphates, nitrates, chlorides, bromides, and iodides all behaved alike. Practically the same order of the cations was obtained for copper ferrocyan- ide. 1 In the cases studied by Pappada the specific adsorption appears to play a very small part. The data for arsenic sulphide, 2 however, give variety enough. The order of cations is: Ce, In, benzidine, Al>new fuchsine, crystal violet > quinine > morphine, UO2, Sr, Ca>Be, Zn, Ba > Hg > p-chloraniline, toluidine > anil- ine > strychnine > guanidine > H > K > Na >Li. The organic cations come in where they please and play havoc with any rule as to valency. The chlorides and nitrates give practically the same values and the sulphates are not far out of line, though it seems probable that the restraining power of sulphate is rather greater than that of chloride or nitrate. The liminal values in gram molecules of the cation per liter are 0.0056, 0.0066, 0.0086, 0.110 and 0.240 for potassium nitrate, sulphate, formate, acetate, and citrate, from which one can deduce that the order of adsorp- tion of anion is : citrate, acetate, formate, sulphate, nitrate, chloride. It is a great pity that Freundlich did not try other combinations, such as barium acetate for instance. Addition of amyl alcohol 'or of phenol makes the arsenic sol more susceptible to univalent and trivalent cations 3 and less sensitive to bivalent and quadri- valent cations, but we do not know why. From the experiments on hydrous ferric oxide, 4 the order of adsorption of the precipitating anions appears to be Cr2C>7>SO4 > OH > salicylate > benzoate > formate > Cl > No 3 > Br > I, while the order for the cations is: H>Ba>Mg>Tl, Na>K. The univalent ions do not all behave alike and neither do the bivalent ones; but the upholders of Schulze's law can comfort themselves with the fact tfrat the two sets do not overlap except in the case of hydrogen. There is no such comfort in the case of albumin. The probable order of adsorption of anions, so far as known, is: sulphocyanate, iodide > chlorate > nitrate > chloride 1 PAPPADA: Zeit. Kolloidchemie, 9, 136 (1911). 2 FREUNDLICH: Kapillarchemie, 351 (1909); FREDNDLICH and SCHUCHT: Zeit. phys. Chem., 80, 564 (1912). 3 KRUYT and DUIN: Kolloidchemische Beihefte, 5, 269 (1913). 4 FREUNDLICH: Kapillarchemie, 352, 358 (1909); ZSIGMONDY: .Kolloid- chemie, 181 (1912); PAPPADA: Zeit. Kolloidchemie, 9, 233 (1911). 218 APPLIED COLLOID CHEMISTRY > acetate > phosphate > sulphate > tartrate, the sulphocyanate ion being adsorbed the most and the tartrate ion the least. Here there is nothing even to suggest Schulze's law and the firm belief which most people have in Schulze's law is probably one reason for the marked failure to account satisfactorily for the phenom- ena with albumin. With the cations albumin appears to be fairly orthodox for the order of adsorption appears to be: Th, UO 2 >Cu, Zn>Ca>Mg>Li>K, Na>NH 4 , though even here the lithium stands higher in the series than it has been found with other substances. Weiser and Sherrick 1 have measured the adsorption of ions by barium sulphate and find that the order is : f errocyanide > nitrate > nitrite > chlorate > permanganate > f erricyanide > chloride > bromide > cyanide > sulphocyanate > iodide, the f errocyanide ion being adsorbed the most and the iodide ion the least. The above order is obtained when the concentrations are given in gram equivalents. If they are taken as gram molecules, f erro- cyanide moves down between the chlorate and permanganate, and the ferricyanide between the chloride and bromide. On either method of tabulation the trivalent ferricyanide ion is adsorbed very much less than some of the univalent ions. The difference for different univalent ions is very great, 100 mols of barium sulphate taking up about 8.5 mols of nitrate as against about 0.06 of iodide. Weiser and Middleton 2 have determined the order of precipi- tation of colloidal ferric oxide by different anions and obtain results which agree with the previous data in so far as they deal with the same electrolytes, the order being: dichromate> tartrate > sulphate > citrate >oxalate>iodate> phosphate. They have also determined the amount of adsorption when ferric oxide is precipitated in presence of a definite excess of salt. This order is quite different, being : phosphate > citrate > tartrate >oxalate > sulphate >iodate>dichromate. The reason for the discrepancy is apparently that there is an adsorption of the electrically neutral salt in addition to the ion adsorption. If this is selective, it may change the order to any extent and apparently does change it very much in this case. The difficulty with this is that it 1 Jour. Phys. Chem., 23, 205 (1919). 2 Ibid., 24, 30 (1920). PROPERTIES OF COLLOIDAL SOLUTIONS 219 leaves us with no independent proof that the order of precipita- tion is really the same as the order of adsorption of the ions. This confirmation will undoubtedly be furnished some day; but, until it is, the connection between adsorption and precipitation can only be considered as extremely probable but as not proved. What we actually measure is the adsorption necessary to cause precipitation plus the adsorption by the precipitated substance. This latter item can theoretically be made as small as one pleases by adding only enough electrolyte to precipitate an infinitely small amount of the colloid; but the errors of analysis then become very large. While there is unquestionably a tendency for ions of a higher valence to be adsorbed more strongly than ions of a lower va- lence, the experiments which have been cited show that there are many exceptions 1 and that the fundamental rule is that the adsorption is specific both as regards the adsorbing substance and the ion adsorbed. Albumin is a specially interesting case because it is peptized readily by cations or anions. When it is negatively charged, as in a slightly alkaline solution, a strongly adsorbed anion will make it more negative and more stable. Consequently precipi- tation by a sodium salt will be more effective the less readily the anion is adsorbed. On the other hand, in acid solutions the sodium salt with the most strongly adsorbed anion will be the most effective in causing precipitation. Negatively charged albumin is precipitated readily by sodium chloride and not at all by sodium iodide, while positively charged albumin is precipi- tated by sodium iodide much more readily than by sodium chlo- ride. Potassium iodide 2 has more effect than potassium chloride with Rhodamine B while the reverse is true with wool violet S. Caseine has not been studied so intelligently as albumin and the results are consequently less clear. Caseine is insoluble in water but reddens litmus paper and sets free carbon dioxide from carbonates and bicarbonates. It is therefore usually considered an acid whereas the phenomenon is really due to a marked ad- sorption of the base. Owing to adsorption, indicators are usually unreliable in protein solutions as a means of determining hy- J See also Mines: Kolloidchemische Beihefte, 3, 212 (1912). 2 TRAUBE: Kolloidchemische Beihefte, 3, 241 (1912). 220 APPLIED COLLOID CHEMISTRY drogen ion concentration. Like albumin, caseine is carried into apparent solution by acids, alkalies, alkaline earths, and some salts. It is also peptized by caffeine, strychnine, etc. 1 The so-called solutions are opalescent and the caseine can be removed by ultra- filtration, showing that it is not in true solution. As one might expect, the peptizing action of the alkaline earths is less marked than that of the alkalies, because the adsorption of the bivalent cations cuts down the peptizing action of the hydroxyl. When peptized by hydroxyl, caseine is precipitated by solutions of calcium chloride, barium chloride, magnesium sulphate, zinc sulphate, copper sulphate, and alum. In these cases it seems probable that the negative charge is neutralized by the readily adsorbed cations. On the other hand, when peptized in acict solution, salts cause precipitation in the following order: ZnCl 2 >KBr>CuCl2>MnCl2. There is no apparent reason for this and there must be experimental error somewhere. In fact the whole behavior of caseine in presence of salts needs careful revision because the present data are quite unsatisfactory. Colloidal silicic acid behaves like caseine in one respect, that it adsorbs lime from suspended calcium carbonate. In accord- ance with this is the fact that silicic acid, though peptized by caustic soda or ammonia, is precipitated by barium hydroxide because of the marked adsorption of the alkaline earth cation. It is also precipitated by alum and by basic dyes, which is as it should be. With colloids stabilized by an electrical charge, we find that the amount of any salt necessary, to produce precipitation varies with the way in which it is added, more being necessary if the salt solution is added slowly. The explanation of this seems to be that the first addition of salt causes a slight agglom- eration which is not enough to cause precipitation, but which cuts down the adsorbing power of the colloid, so that more salt is necessary to cause the same adsorption. It is also true that a change in the adsorbing power of a colloidal substance will affect its stability. If we have anhydrous ferric oxide on the one hand and hydrous ferric oxide on the other, without specify- ing at present what we mean by hydrous, we shall expect to : Jour. Physiology, 20, 399 (1910); VAN SLTKE and PUBLOW: Science and Practice of Cheese-making, 142 (1913). PROPERTIES OF COLLOIDAL SOLUTIONS 221 have different degrees of stability in the two cases and so we do. Two colloidal ferric oxides have been distinguished and we speak of colloidal solutions of alumina and metalumina, tungstic acid and metatungstic acid, stannic and metastannic acid, etc., where our only real criterion is the difference in stability. Stannic acid is peptized by dilute HC1, HN0 3 , H 2 SO 4 and alkalies while metastannic acid is soluble in HC1, insoluble in HNO 3 , and swells in H 2 SO 4 . As a matter of fact one can have a whole series of stannic acids. 1 It is merely a question of how carefully one differentiates properties. By rapid dialysis of silicic acid solu- tions it is possible to get much higher concentrations than by dialysis because the silicic acid changes less. 2 With a colloidal solution stabilized by an adsorbed ion, the reversibility of the coagulation depends upon the possibility of washing out the precipitating agent and on the physical state of the precipitate. The more strongly the precipitate is adsorbed the more difficult it will be to wash it out. When albumin is precipitated by sodium chloride, the coagulation is ordinarily reversible. When it is precipitated by the salt of a heavy metal, the coagulation is irreversible. All precipitates tend to change on standing, becoming coarser or more firmly agglomerated. It may then happen that washing out the precipitating agent will not cause peptization. When albumin is precipitated by calcium chloride; the coagulation is reversible if the calcium chloride is washed out at once; but is not reversible if the precipitate is allowed to stand. If a precipitate of silver chloride be allowed to stand over night, it can then be filtered and washed much more satisfactorily than when first precipitated. The effect of temperature changes on the stability of a colloidal solution may be two-fold. Raising the temperature will in- crease the peptizing action of the solvent thereby increasing the stability, or it may decrease the adsorption of the stabilizing ion, thus making the solution less stable. The second factor is the more important with colloidal solutions stabilized by an adsorbed ion. 3 Thus albumin solutions coagulate on heating . * VIGNON: Jour. Chem. Soc., 66, 833 (1889): MECKLENBURG: Zeit. anorg. Chem., 74, 207 (1912). 2 ZSIGMONDY: Van Bemrnelen Gedenkboek, 7 (1910). and PICTON: Jour. Chem. Soc., 87, 1931 (1905). 222 APPLIED COLLOID CHEMISTRY and so do caseine solutions. When a solution is actually frozen solid, we have the possibility of the colloid agglomerating. If it does that, the colloidal solution will not be formed when the mass melts; but otherwise it may. The experiments along this .line are not very satisfactory because they were done without any knowledge of the general theory. l The main result is about what one would expect. In presence of a very small amount of peptiz- ing agent, the precipitation is likely to be irreversible because the precipitate agglomerates on standing. If there is a fair ex- cess of peptizing agent, the precipitation will be reversible. This is of course especially true if the peptizing agent is a water- soluble colloid. Ehrenberg 2 considers that the pressure due to the expansion when water freezes may be a factor in causing coagulation. Where the adsorption is very low, it may require a very high concentration of a given ion to cause peptization. This occurs in a number of cases of peptization by alkalies and is probably another reason why people have considered that alkali salts were formed and not colloidal solutions. If one over-runs the point of electrical neutrality, the suspen- sion may become stable again because it is stabilized by preferen- tial adsorption of an ion having the opposite sign. If one adds a little less than the equivalent amount of a dilute potassium bromide solution to a dilute silver nitrate solution, one gets a positively-charged colloidal solution of silver bromide because the silver bromide is stabilized by adsorbed silver ions. If one adds slightly more than the equivalent amount of the potassium bro- mide solution, there is obtained a negatively-charged colloidal solution of silver bromide, stabilized by adsorbed bromine ions. If more bromide solution is added, the silver bromide will coagu- late because there will come a point at which the slight adsorption of the potassium ion counter-balances the adsorption of the bromine ion. In other words, silver bromide precipitates when there is a distinct excess of silver nitrate; stays in suspension as a positively charged colloid when there is only a slight excess of I LJUBAVIN: Jour. Chem. Soc., 68, 685 (1890); LOTTERMOSER: Zeit. phys. Chem., 60, 462 (1907); Ber. deutsch. chem. Ges., 41, 3976 (1908); Zeit. Elektrochemie, 17, 317 (1911); FISCHER: Zeit. Kolloidchemie, 8, 292 (1911). 2 Die Bodenkolloide, 20 (1915). PROPERTIES OF COLLOIDAL SOLUTIONS 223 silver nitrate,- precipitates when the suspension becomes electri- cally neutral/ stays in suspension as a negatively charged colloid when there is a slight excess of potassium bromide/ and precipi- tates when there is a distinct excess of potassium bromide. At the two ends of this series there is also the possibility of true solution of silver bromide in silver nitrate or potassium bromide. An interesting case of the precipitation of a colloid by an electrolyte is to be found in the action of sea-water on muddy river water. Skey 1 pointed out that suspended mud is precipi- tated by electrolytes, and Waldie 2 has discussed the clearing of the water of the Hooghly. When a river flows into the ocean, the current becomes less and some of the suspended mud is dropped on this account; but this is relatively unimportant in many cases. Schloesing 3 called attention to the fact that the forma- tion of deltas is due primarily to the coagulation of the suspension by the electrolytes in the salt water. Another interesting case of the neutralization of an electrical charge is the precipitation of positively charged colloids by rays. 4 Frischauer 5 found that j8 rays increase the agglomeration of sulphur, while Doelter 6 reports they change selenium hydrosol to crystalline selenium. Farmer and Parker 7 report that ultra-violet light coagulates colloidal platinum. This seems to be in line with the action of ultra-violet light in causing metals to emit electrons. It would be interesting to know whether the action is confined to metals or whether ultra-violet light will coagulate any negatively charged sol. Svedberg 8 has prepared colloidal solutions ap- parently by the disintegrating effect of ultra-violet light; but later experiments 9 indicate that the silver is first oxidized and then 1 Chem. News, 17, 160 (1868). 2 Ibid., 30, 37 (1874). 3 Jour. Chem. Soc., 24, 750 (1871). 4 HARDY: Jour. Physiology, 29, 29 (1903); JORISSEN and WOUDSTRA: Zeit. Kolloidchemie, 8, 8 (1910); HOBER: Physikalische Chemie der Zelle und Gewebe, 332 (1911); BURTON: The Physical Properties of Colloidal Solu- tions, 172 (1916). 6 Comptes rendus, 148, 1251 (1909). 6 Das Radium und die Farben, 17, 95, 97 (1910). 7 Jour. Am. Chem. Soc. 35, 1524 (1913). 8 Zeit. Kolloidchemie, 6, 130 (1910); SCHULZE: Ibid., 16, 136 (1914). 9 NORDENSON: Kolloidchemische Beihefte, 7, 110 (1915). 224 APPLIED COLLOID CHEMISTRY reduced. Young and Pingree 1 have done some work on the action of light on colloidal solutions. There is a very important field here especially with reference to the colors of gems; but nobody has attacked the problem systematically. 2. Sols stabilized by Water. When we come to colloidal solutions peptized by liquids, the effect of salts is necessarily relatively small. Tannin is peptized by water and by glacial acetic acid. Apparent solubilities for tannin vary from 20-250 parts per hundred of water; but Navassart 2 has shown that tan- nin is miscible in all proportions with water, alcohol, acetone and glacial acetic acid. The apparent solubility has been due to lack of wetting of the tannin powder by the sol. It is probable also that the tendency to coagulate increases with increasing concentration and the presence of impurities might easily cause agglomeration and precipitation at apparently definite concen- trations in each case. Salts have relatively little effect on the stability of tannin sols, though the hydroxides of the heavy metals adsorb tannin strongly, the two precipitating together. Gela- tine swells in cold water but is not peptized by it. It is peptized by warm water, the apparent melting-point or maximum gelation temperature of a 10 grams per 100 cc gelatine 3 jelly being 30.0. Salts in relatively high concentration have an effect, and the anions may be arranged in the following order: sulphate > citrate > tartrate > acetate > chloride > nitrate > bromide > iodide >sul- phocyanate>benzoate>salicylate. The first three raise the apparent melting-point while the others lower it. From these data it seems probable that the liquefaction by sodium iodide, for instance, is due to a peptization by iodide ion superposed on the water peptization. Briggs 4 has shown that removing iodide by dialysis causes the gelatine to set again; but we have no trust- worthy data which show definite adsorption of iodide by gelatine. Glycerine and sugar behave like sulphates, raising the apparent melting-point, while such non-electrolytes as furfurol, urea, alco- hol, resorcinol, hydroquinone, and pyrogallol lower the apparent melting-point. Here again we have empirical observations 1 Jour. Phys. Chem., 17, 657 (1912). 2 Kolloidchemische Beihefte, 5, 299 (1913). 3 SMITH: Jour. Am. Chem. Soc., 41, 135 (1919). 4 Jour. Phys. Chem., 24, 74 (1920). PROPERTIES OF COLLOIDAL SOLUTIONS 22,5 and no real knowledge. Formaldehyde hardens gelatine so that it is not peptized by water. By soaps we usually mean sodium palmitate, stearate, or oleate, or some mixture of these salts. They are peptized by water, dilute solutions showing a rise of boiling-point and con- centrated solutions practically none by the usual method, which is in error owing to presence of air. By the dew-point method McBain 1 finds a rise of 0.2 for normal potassium stearate. There is, of course, a certain amount of hydrolysis and it seems probable that a sodium oleate solution contains sodium oleate, oleic acid, and caustic soda, the first two of which are not in solution. The hydroxyl ion is unquestionably adsorbed to some extent by the oleic acid or sodium oleate or by both and it is possible that the caustic soda is adsorbed in concentrated solutions It should be possible to verify this assumption by doing ultra-filtration of soap and analyzing the filtrate. A sufficient addition of sodium salts causes the precipitation of soap and this is made use of in soap manufacture. We do not know to what extent this effect of salt is due to a direct coagulation of the soap by adsorbed ions or to coagulation owing to a decrease in the peptizing action of water. Some unpublished experiments by Clowes seem to indicate that the concentration of sodium ions is an important factor, though Kurzmann 2 finds that equimolecular solutions of potassium carbonate and hydroxide behave alike and not equiva- lent solutions. 3. Sols stabilized by Non-electrolytes or by a Second Colloid. A colloid which is stabilized by a strongly-adsorbed non-elec- trolyte such as sugar will behave very much like a water-soluble colloid. There has been, relatively little work done on such sys- tems. There are many interesting possibilities, however, when we consider the mutual action of two colloidal solutions. If we mix colloidal solutions of two substances each of which adsorbs the other, the mutual adsorption will have some effect on the adsorption of the peptizing agent. If it causes a decrease in the adsorption of the peptizing agent, this means a decreasing stability of the colloidal solution and may lead to precipitation over some range of relative concentrations. In case this occurs, 1 Jour. Am. Chem. Soc., 42, 432 (1920). 2 Kolloidchemische Beihefte, 5, 427 (1914). 15 226 APPLIED COLLOID CHEMISTRY we shall have colloidal solutions of decreased stability when either colloid predominates sufficiently, and precipitation for inter- mediate relative concentrations. The range over which precipi- tation occurs may be great or small depending on the ease with which the peptizing agent is displaced or neutralized. The most familiar case of such precipitation is when we are dealing with two colloids having opposite electrical charges. This point was studied carefully by Biltz 1 and the general statement is to be found in all the text-books on colloid chemistry. The negatively charged colloids studied by Biltz were platinum, gold, selenium, cadmium sulphide, antimony sulphide, arsenic sulphide, stannic oxide, molybdenum blue (MO 3 O 8 ), tungsten blue ("W^Oa), and vanadium pentoxide. The positive colloids were the oxides of iron, aluminum, chromium, thorium, zirconium, and cerium. Precipitation occurs when each positive colloid is mixed with each negative colloid in proper proportions. Since complete neutralization takes place only when one sol has adsorbed the amount of the sol carrying an equivalent amount of the ion having the opposite charge, it follows that the amount of one sol neces- sary to precipitate a given amount of another sol will vary with the degree of adsorption. It will therefore be a specific property and not not an additive one. This can be tested experimentally on data by Biltz given in Table XLV. 2 Alumina is more effective than chromic oxide in precipitating antimony sulphide and much less effective in precipitating arsenic sulphide. The alumina must therefore be adsorbed more by antimony sulphide than chromic oxide, while the reverse must be true for arsenic sulphide. Cerium oxide is less effective than either of these in precipitating the sulphides of antimony and arsenic. The phenomenon is thus specific, varying with the nature of the two colloids. This seems not to have been realized before. In fact Freundlich says definitely 3 that one seems to find approximately the same order regardless of what sol is to be precipitated. This statement is true, but it missed the impor- tant thing in the experiments, which was that the order was not always the same. 1 Ber. deutsch. chem. Ges., 37, 1095 (1904). * 2 FREUNDLICH: Kapillarchemie, 445 (1909). 3 FREUNDLICH: Kapillarchemie, 445 (1909). PROPERTIES OF COLLOIDAL SOLUTIONS . 227 So-called irregular series 1 have been observed when a suspen- sion of a positively-charged colloid is added to a suspension of a negatively-charged colloid. When increasing concentrations of aluminum sulphate were added to a negatively-charged solution of mastic, there was at first no precipitation and the suspension was charged negatively because the mastic was present in excess. At higher concentrations of aluminum sulphate, the suspension became electrically neutral and complete precipitation took place. At still higher concentrations the mastic was held in suspension by the hydrolyzed alumina and the suspension was charged posi- TABLE XLV. PRECIPITATION OF NEGATIVE COLLOIDS 1.4 mg gold completely precipitated by CeOz Fe 2 O 3 ThO 2 ZrO 2 Cr 2 O 3 A1 2 O S 4 3 2.5 1.6 0.3 0.1-0.2mg 28 mg Sb 2 S 3 completely precipitated by Fe 2 Os ThO 2 CeO 2 ZrO 2 Cr 2 O 3 A1 2 O 3 32 20 11 6.5 3.0 2.0mg 24 mg As 2 s 3 completely precipitated by Fe 2 O 3 ThO 2 CeO 2 ZrO 2 A1 2 O 3 Cr 2 O 8 13 6 4 2 2 0.5 rag tively. At still higher concentrations of aluminum sulphate, complete precipitation occurred. This apparently abnormal result is due to the fact that the experiment is not exactly what it purports to be, the addition of one colloidal solution to another. Instead of adding a colloidal solution of alumina, there was added a so-called solution of aluminum sulphate, which 1 hydrolyzed to colloidal alumina and sulphuric acid, with possibly some aluminum sulphate left unchanged especially in the higher con- centrations. The amount of free sulphuric acid is always equiva- lent to the amount of alumina and the two concentrations increase proportionally. The so-called abnormal precipitation at the higher concentrations is merely a precipitation by sulphate ions and is absolutely normal. Since a colloid peptized by water may be charged positively or negatively, there is no reason why it should riot precipitate another colloid under suitable conditions. We usually consider the col- 1 NEissERand FRIEDEMANN: Zeit. angew. Chem., 1906, 1953; BECHHOLD: Zeit. phys. Chem., 48, 285 (1904); BUXTON and TEAGUE: Ibid., 67, 47, 64 (1907); FREUNDLICH: Kapillarchemie, 402 (1909). 228 . APPLIED COLLOID CHEMISTRY loids peptized by water solely as protecting colloids but this is clearly an inadequate view as is shown by the experimental data. Years ago Schulze 1 pointed out that small amounts of a gelatine solution were as effective as lime or alum in causing the rapid sedimentation of clay and that addition of minute quantities of gelatine to barium sulphate simplified the question of filtration and washing very much. He of course gave no adequate explana- tion of the phenomenon and the real explanation was given by Billitzer 2 who called attention to the fact that while gelatine, agar-agar, etc., ordinarily check the precipitation of colloidal solutions by electrolytes, small amounts of these same substances may have a precipitating action. This can be detected even when the gelatine produces no precipitation itself. Billitzer cites the experiments of Neisser and Friedemann in which it was found that a mastic emulsion containing a trace of gelatine was precipitated more readily by sodium chloride than when no gelatine is present. Billitzer finds that gelatine precipitates such negative colloids as antimony sulphide and arsenic sulphide in acid or neutral solution, but does not precipitate positively charged sols such as hydrous ferric oxide. Gelatine in ammoni- a v cal solution precipitates hydrous ferric oxide, though no precipi- tation occurs if ammonia is added to a mixture of gelatine and ferric oxide. Bismarck brown, which is a positive colloid, is precipitated by an alkaline gelatine solution, while eosine is precipitated by an acidified gelatine solution. Graham 3 points out that colloidal silica is precipitated by gelatine, the composition of the precipitate varying between 56 and 92 parts of gelatine per hundred of silicic acid. Lumiere and Seyewetz 4 have studied the action of salts of chromium upon gelatine because of its importance for the theory of chrome tanning. If a suitable amount of chrome alum is added to a gelatine sol or if sheets of gelatine are soaked long enough in a chrome alum solution, the gelatine becomes insoluble even in boiling water and does not putrefy on standing as the original gelatine would have done. They found first that gela- 1 Pogg. Ann., 129, 369 (1866). 2 Zeit. phys. Chem., 51, 145 (1905). 8 Jour. Chem. Soc., 16, 246 (1862). 4 Bull. Soc. chim. Paris, (3) 29, 1077 (1903). PROPERTIES OF COLLOIDAL SOLUTIONS 229 tine could be rendered insoluble by treatment with almost any salt of the formula Cr 2 X 6 : sulphate, sulphite, nitrate, chloride, fluoride, acetate, formate, citrate, or lactate. Since it had been shown by Namias that the power of chrome alum to make gela- tine insoluble was increased by addition of alkali up to the point where precipitation of hydrous chromic oxide occurred, experi- ments were made with the green basic chromic sulphate of Recoura. Gelatine treated with a solution of this salt resisted repeated washing with boiling water better than gelatine which had been treated with a solution which was not basic. On the other hand, gelatine is not made insoluble by treatment with a so-called solution of chromic oxide in potassium hydroxide; but this is probably due to the excess of alkali. If the proper amount of ammonia is added to gelatine and then a chrome alum solution, the gelatine becomes insoluble in boiling water; but this is not the case if ammonia is present in excess. The gelatine is insolu- ble when it contains 2.5-25 grams of chrome alum per hundred of gelatine. Experiments with chromic sulphate, chloride, nitrate, and acetate showed that approximately the same quantity of chromic oxide was necessary to make the gelatine insoluble irrespective of the nature of the acid. Since this implies that only the chromic oxide counts, experiments were made to see what became of the acid radical. Sheets of gelatine were immersed in a known chromic sulphate solution and the solution analyzed afterwards. The ratio of chromium to sulphate in the solution was practi- cally the same before and after treatment with gelatine, which meant apparently that the chromic sulphate was taken up as a whole. It could not be the chromic sulphate as such which made the gelatine insoluble because the same result could be obtained when chrome alum was added to an ammoniacal gela- tine. When the gelatine containing chromic sulphate is washed repeatedly with boiling water, acid is extracted very slowly; but after some washings the gelatine swells and a little is carried off in the wash water, leaving the residue relatively richer in chro- mium, the chromic oxide content running up to 5.4-4.6 grams per hundred grams of gelatine. If the gelatine is treated with dilute alkali, it is possible to remove the acid without causing any swell- ing or any solution of the gelatine. When the gelatine is treated 230 APPLIED COLLOID CHEMISTRY in this way the maximum amount of chromium found in the gelatine is 3.3-3.5 grams Cr 2 O 3 per hundred grams of gelatine. The explanation seems to be that the chromic oxide makes the gelatine insoluble and that the acid is adsorbed strongly by the chromic oxide, the gelatine, or both. When the acid is washed out at high temperatures, it causes a swelling of the gelatine. In fact, an insoluble gelatine, which has been freed from acid, swells rapidly and finally dissolves if boiled with water containing small amounts of acid or alkali. If the acid is adsorbed by the chromic oxide in the gelatine, it would be interesting to know whether there is a time factor and whether after standing for a month it would be easier to wash out the acid. Coming back to the general problem, if the ratio of gelatine to the other colloid is increased sufficiently, we shall pass through the precipitation range into the range where the colloidal solu- tion is stabilized by gelatine and then behaves more like a water- soluble colloid. Colloidal gold and colloidal silver solutions, when stabilized by gelatine can be evaporated and redissolved because the gelatine prevents the irreversible agglomeration. When less gelatine is used, it may retard, though not prevent, the change of red colloidal gold to blue. Zsigmondy^defines as the gold number the number of milligrams of a protecting^coiloid which just prevents the color change in a 10 cc red gold solution, containing 0.0053-0.0058 percent gold, when one cubic centi- meter of a ten percent sodium chloride solution is added. course a strongly adsorbed non-electrolyte, such as sugar, will act similarly to gelatine; but, in most cases, the adsorption is so much less that these substances are only interesting theoretically as stabilizers. Some of his results are given in Table XL VI. Most of the stable colloidal solutions depend for their stability on the presence of a so-called protecting colloid. There are several other ways in which this property can be made useful. If we have colloids like the so-called zinc and magnesium resinates which are peptized by benzene, we can use these as protecting colloids to stabilize pigments and thus form colloidal solutions which could not otherwise exist in benzene. There are other cases where one does not insist on a colloidal solution but wishes a fine-grained precipitate. The most important case of this sort 1 Zeit, anal Chem., 40, 697 (1901). PROPERTIES OF COLLOIDAL SOLUTIONS 231 technically is the photographic emulsion, which consists chiefly of silver bromide precipitated in presence of gelatine, so as to get a fine-grained precipitate, which is afterwards given the desired properties by agglomerating through suitable treatment. The gelatine plays the additional part of a depolarizer. TABLE XL VI. GOLD NUMBER FOR DIFFERENT COLLOIDS Colloid Gold number Gelatine 005-0 01 Caseiiie 01 Eg " albumin . . 0.06-0.3 Gum arabic 15-0.25 Gum tragacanth 2 Dextrin . . 10-20 Potato starch 25 Sodium oleate 0.4-1.0 Sodium stearate (added boiling hot) 001 Sodium stearate (added at 60) 10 Even though the gelatine or similar substance may not be present in sufficient amount to prevent precipitation, its pres- ence may make itself known by cutting down the agglomeration of the colloidal precipitate. When caseine is t precipitated from cow's milk by acid, it comes down in relatively coarse lumps which is one reason why cow's milk is bad for infants. By adding gelatine 1 to the milk, the growth of the curdy precipitate can be prevented. In human milk a protecting colloid is present and consequently coarse curds are not formed. This property of gelatine and other strongly adsorbed substances of keeping down the size of precipitates is made use of in the preparation of many colloidal solutions. By varying the amount of gelatme one can vary the fineness of the precipitate. The addition of gelatine to ice cream prevents the formation of coarse crystals. In electro- lytic refining of metals and in electroplating, addition agents are used to make the deposited metal less coarsely crystalline. An addition agent must not be carried away from the cathode by the current and must be adsorbed by the precipitating metal. 1 ALEXANDER: Jour. Soc. Chem. Ind., 28, 280 (1909). 232 APPLIED COLLOID CHEMISTRY Gelatine is used in lead refining and has a great effect in an acidi- fied copper sulphate solution. Many of the essential oils have proved satisfactory. Glucose can also be used in some cases, not because it is a colloid, but because it is adsorbed strongly by the precipitating metal. In the case of two colloids, each peptized by water, the mutual adsorption may cut down the adsorption of water sufficiently to cause precipitation. This seems to be the most plausible ex- planation for the behavior of gelatine and tannin. The only difficulty is that the adsorption is more marked when the gelatine is positive and the tannin negative, which is more what one would expect if the two were colloids peptized by ions. There seem to be at least two possible ways out of this difficulty. It may be that gelatine is not really peptized by water 1 and that we are dealing with a case of ion peptization where the range of in- stability is very small. The other possibility is that we get a different adsorption when the colloids are charged oppositely and that the resulting complex is less readily peptized by water under those conditions. This is not an impossibility, as is shown by the fact, previously mentioned, that ammoniacal gela- tine precipitates colloidal ferric oxide, whereas there is no pre- cipitation if the gelatine and ferric oxide sols are mixed before the ammonia is added. There seems to be no way at present to decide this question. Tannin and basic dyes give a precipitate but the range over which this occurs is very limited and in order to make use of tliis property in dyeing, it is customary to fix the tannin with a salt of antimony. Of course, in so far as the water peptization is cut down, the mixture will tend more and more to behave like a colloid peptized by ions. Tiebackx^ finds that when a half percent gelatine solution is mixed with a four percent gum arabic solution, the mixture is precipitated readily by acids. If the acid is washed out, the coagulum can readily be dispersed in water. Tiebackx points out that the properties of a gelatine and gum arabic mixture are almost those of a globulin or of caseine. "Caseine is peptized by dilute alkalies and somewhat stronger acids, the precipitation 1 Cf. LOEB: Jour. Gen. Physiology, 1, 365 (1919). 2 Zeit. Kolloidchemie, 8, 198, 238 (1911), PROPERTIES OF COLLOIDAL SOLUTIONS 233 range lying between these two limits. Sulphuric acid is also adsorbed by caseine somewhat more strongly than hydrochloric acid. With globulin and caseine the acids are also displaced by salts as can be shown by means of methyl orange. The dis- placement of acid also involves a peptization, for van Dam 1 has shown that presence of five percent sodium chloride increases the amount of caseine peptized by a given concentration of acid." Apparently gelatine and agar-agar are two practically non- miscible liquids, for Beijerinck 2 reports that it is impossible to mix a hot ten percent gelatine solution with a hot two percent agar solution, one of the liquids being emulsified in the other when the two are shaken together. The phenomenon is even more marked when ten percent soluble starch is shaken with ten percent gelatine. Biltz 3 obtained no precipitation upon mixing any two of his positively charged sols and no precipitation on mixing any two of the negatively-charged sols except when gold or selenium was mixed with the sulphides and then chemical action probably took place. Biltz concluded that sols of like charge had no effect upon each other; but this does not follow, because adsorp- tion of one colloid by the other would not neutralize the electrical charge and therefore would not cause precipitation. We know, however, that cases of adsorption are not limited to colloids or electrolytes having opposite signs. Charcoal adsorbs both bases and acids. Silver bromide adsorbs silver ions or bromine ions as the case may be. There is therefore no theoretical reason why precipitated hydrous ferric oxide might not adsorb chromic oxide and vice-versa. If the precipitated substance will do this there is no reason why the peptized substance should not. Nagel 4 has shown recently that this does occur and that it accounts for the behavior of mixtures of chromic and ferric salts with excess of alkali. Hydrous chromic oxide is peptized by caustic potash while hydrous ferric oxide is not. If the chromium salt is present in large amount relatively to the iron salt, the ferric oxide will adsorb the peptized chromic oxide and be pep- 1 Van Bemmelen Gedenkboek, 104 (1910). 2 Zeit. Kolloidchemie, 7, 16 (1910). 3 Ber. deutsch. chem. Ges., 37, 1095 (1904). * Jour. Phys. Chem., 19, 331 (1915). 234 APPLIED COLLOID CHEMISTRY tized by it, going apparently into solution. If the ferric salt is present in excess, it will adsorb the peptized chromic oxide, carrying it out of the liquid phase. It is to be noticed that the chromic oxide, when in excess, acts as a so-called protecting col- loid to the iron oxide. Other cases of the same type could be cited. The fact that gelatine protects ferric oxide when added before the ammonia shows that it must have been adsorbed by the ferric oxide. Bechhold 1 has shown that an ultra-filter which will stop Prussian blue and will let arsenic sulphide pass, holds back the arsenic sulphide when mixed with Prussian blue. Although both are negatively charged sols, it seems evident that there is a mutual adsorption. On the other hand, Bechhold found that hemoglobin can be separated from Prussian blue by filtration through a suitable ultra-filter, the hemoglobin going through the filter. In this case there is no appreciable adsorption. Since the precipitation of one colloid by another may take place within a relatively narrow range of concentrations, it is not surprising that people have postulated the existence of defi- nite chemical compounds in order to account for the precipitation. The literature on this subject is in a worse state even than that on the so-called basic salts. To cite a few cases out of many, ferric arsenate, aluminum silicate, aluminum stannate, stannic phos- phate, and cupric eosinate are not formed under ordinary condi- tions by precipitation from solution. The precipitation is due to the mutual coagulation of two colloids. If each adsorbs the other pretty strongly, it may easily happen that the precipitate can be peptized without change of composition under certain conditions, a behavior which one usually associates with com- pounds. The color lakes are an especially fruitful field for imagin- ary formulas. Unpublished work by Mr. Basseches of the Bureau of Standards shows that no ferric tannate is formed under ordinary conditions. The purple of Cassius is a typical case. It is usually prepared by the reaction of gold chloride with stannous chloride and con- sists of colloidal gold adsorbed by stannic oxide. Since it could apparently be carried into solution by ammonia and since mercury did not extract gold from it, Berzelius was positive that it must be a compound and he wrote for it the formula, Au 2 O2.Sn 2 O 3 . 1 Zeit. phys. Chem. 60, 299 (1907). PROPERTIES OF COLLOIDAL SOLUTIONS 235 xH 2 O. The final proof that it was a colloidal precipitate was given by Zsigmondy 1 who made it directly from colloidal gold and colloidal stannic oxide. Owing to the intensity of the color, the formation of the purple of Cassius may be used as a test for gold, 2 sensitive to one part in one hundred million. It has also been possible to obtain similar precipitates with alumina, barium sulphate, mercurous chloride, etc. 3 A corresponding brown 4 and red 5 precipitate has been obtained with platinum. When purple of Cassius is made with zirconia, the precipitate is not peptized by ammonia and aqua regia does not extract the gold. 6 * Liebig's Ann., 301, 375 (1898). 2 ROSE: Jour. Chem. Soc., 64 II, 245 (1893). 3 SCHWARZ: Jour. Chem., 26, 205 (1873); MULLER: Ibid., 48, 352 (1885); KRUSS: Ibid,. 62, 16 U887); ANTONY and LUCCHESI: Ibid., 72 II, 43 (1897); MARSHALL and WELKER: Jour. Am. Chem. Soc., 35, 822, 823 (1913). 4 DELACHANAL and MERMET: Comptes rendus, 81, 370 (1875). 5 WOHLER: Zeit. Elektrochemie, 14, 634 (1908). 6 BiLTz: Ber. deutsch. chem. Ges., 37, 1095 (1904). CHAPTER VIII GELATINOUS PRECIPITATES AND JELLIES i STRUCTURE When we add caustic soda to a ferric chloride or an aluminum chloride solution, we get a precipitate which varies from gelati- nous to sandy, depending on the temperature, being more sandy the higher the temperature of precipitation. In this case we have the gelatinous precipitate and a supernatant liquid. If we add caustic potash in excess to a chromic sulphate solution, we may get a gelatinous precipitate with a supernatant liquid or we may get the whole mass setting to a jelly, which may afterwards contract and squeeze out liquid. 1 To avoid confusion in the future it will be simpler to use the' word jelly only for those cases where there is no supernatant liquid at first. This is in accord with general usage for one speaks of a gelatine solution, for in- stance, setting to a jelly; but some writers have been so careless in their use of the word that it has not always been clear just what they did mean. Gel is the inclusive term, gelatinous precipitates and jellies being two forms of gels. A gelatinous precipitate is always viscous and contains liquid; it may be plastic. No one has any difficulty in recognizing a gelatinous precipitate; but we are not at all clear as to what gives it these physical properties. In the case of gelatinous ferric oxide there is every reason to suppose that none of the water is com- bined to form a definite chemical compound. Vapor pressure experiments on silicic acid jellies show no evidence of any definite chemical compound between silica and water. Le Chatelier 2 believes that a silicic acid jelly is merely grains of anhydrous silica and water. He claims to have proved this by polishing metals with colloidal silicic acid; but this is not conclusive be- cause anhydrous silica might have been formed as a result of the 1 Syneresis is the technical term. 2 La Silice et les Silicates, 76 (1914). 236 GELATINOUS PRECIPITATES AND JELLIES 237 pressure used in polishing. A better line of attack is to consider whether grains of sand mixed with water could give a precipitate having the properties of gelatinous silicic acid. Presence of water will make sand plastic, and addition of an excess of water will give a mass of wet sand with supernatant liquid, which is good so far as it goes. If the sand is shaken with an excess of water, the grains will be separated and will afterwards settle at a given rate depending on the conditions. That does not happen with the silicic acid jelly. There is a marked coalescence which we do not get with sand and water so far as we know. We must therefore assume one of two things. Either the sand grains are held together extraordinarily firmly by water when they are very fine or there is some other factor comes in. The first ex- planation cannot be the right one because, if it were, one ought then to be able to get a gelatinous precipitate of any colloid at ordinary temperatures without much difficulty, which is not the case. We never get gelatinous gold and while we can get gelat- inous calcium carbonate, we have to do it in a very special way. Consequently, Le Chatelier's hypothesis cannot be accepted without modification. We are not dealing with a solid solution because then the vapor pressure would be independent of the previous history of the sample. It seems to me that Oden's experiments on sulphur are very important in their bearing on this problem. Oden 1 found that the physical properties of sulphur precipitated from colloidal solution varied very markedly with the electrolyte used for precipitation. It came down as a hard precipitate with potas- sium salts, fine-grained with copper sulphate, plastic with barium salts, fluid with hydrochloric acid, and slimy with other salts. Similar but less marked phenomena have been observed with colloidal ferric oxide, 2 which comes down more sandy when precipitated by hydrochloric acid than when precipitated by sulphuric acid. 3 In the case of sulphur we are not very far below the melting-point and we know that it is a simple matter to supercool molten sulphur so as to get a plastic mass at ordinary 1 Der kolloide Schwefel, 134, 157 (1912). 2 GIOLITTI: Gazz. chim. ital, 35 II, 181 (1905); 36 II, 157, 433 (1906). 3 For alumina with chlorides and sulphates see ROSE : Zeit. Kolloidchemie, 16, 1 (1914). 238 APPLIED COLLOID CHEMISTRY temperatures. The simplest assumption is that the sulphur pre- cipitated from colloidal solution may contain enough water and electrolytes to keep it as a supercooled viscous liquid the drops of which have coalesced presumably to form filaments or perhaps nets, thus giving it a structure. The only apparent difference between the sulphur and the ferric oxide is that we never get as fluid a mass with the iron oxide as with the sulphur; but that may be due to the difference in the melting-points. It seems possible therefore that a gelatinous precipitate consists of partially coalescing drops or films of a viscous liquid containing more or less of the mobile liquid adsorbed or held in other ways. The super-cooled liquid is kept from solidifying by the water or the electrolyte or both. While this does not seem a very satisfactory hypothesis, it is not an impossible one. With benzoic acid and water we get two liquid phases which are stable. With salicylic acid and water the two liquid phases are instable but can be realized without any difficulty. On this basis rapid precipitation would be likely to get us down to the instable curve for the dineric system and that is exactly what happens. Von Weimarn 1 obtains any anhydrous inorganic salt as a gelatinous mass by precipitating under such conditions that the percentage super- saturation is enormous. It is also interesting to note that gelatinous precipitates are obtained only as a result of condensa- tion methods. Where the reverse seems to be the case, as in the peptization of silica, it is probable that true solution occurs, followed by precipitation. There is no doubt but that silicic acid can exist temporarily in true solution. Mylius and Groschuff 2 showed that when a sodium silicate solution was acidified with hydrochloric acid, the silicic acid set free diffused fairly readily and lowered the freezing-point. These two characteristics dis- appeared gradually as the free acid changed into colloidal silica. If one objects to the distinctly arbitrary assumption that we have viscous liquids and water in the gelatinous precipitates, one alternative is to assume that solid particles and water behave like a gelatinous precipitate when the solid particles are suffici- ently fine and provided they adsorb water sufficiently strongly. 1 Grundziige der Dispersoidchemie, 67 (1914). 2 Ber. deutsch. chem. Ges., 38, 116 (1905). GELATINOUS PRECIPITATES AND JELLIES 239 This is apparently what Zsigmondy 1 does; but he does not show why this should be so. Another alternative is to assume that the adsorbed ion makes the surface viscous. In the case of an emulsion this does happen. We may have drops of oil coated with a soap film and these may coalesce sufficiently to form a gelatinous mass. This is not helpful because the soap is gelat- inous in itself. It is possible, however, that there is an inter- mediate stage between that of peptization and that of irreversible coagulation, where there may be a surface which is gelatinous in its properties. While something of this sort may happen, it has not been shown to take place. The real test would be to make a gelatinous gold precipitate without any protecting colloid. Until something of this sort has been done, or until we know why it cannot be done, 2 we must admit that we know very little in regard to what constitutes a gelatinous precipitate. A possible explanation with ferric oxide is that we have grains of oxide with a gelatinous film of instable ferric hydroxide adsorbed on the surface and stabilized thereby. This would not be inconsistent with the vapor pressure data because those only show that ferric hydroxide does not exist in mass under the conditions of the experiment. We know that sand can be converted into gelatinous silicic acid if ground sufficiently fine and that clay particles have a gelatinous coating. The difficulty is that this explanation does not help us in the case of barium sulphate and we must either explain all gelatinous precipitates in the same way or we must divide them into groups and be able to distinguish between the groups. We do not know for certain what the structure of a gelatinous precipitate is. We may have viscous drops coalescing partly to form filaments or films, or we may have spherical drops of water coated with the gelatinous material. This second case is really water emulsified in water, which seems improbable; but a mass of foam is air emulsified in air, so that such a system is not impossible. This would constitute a honeycomb structure whereas the other type gives the interlacing or sponge structure, each phase being continuous. The general opinion at present 1 Kolloidchemie, 149 (1912). 2 It has been suggested that gold does not adsorb water sufficiently strongly to give a gelatinous precipitate. 240 APPLIED COLLOID CHEMISTRY is that gelatinous precipitates and jellies have a sponge structure. 1 On the other hand we know that certain emulsion jellies have the honeycomb structure. The fact that one can do ultra- filtration through a suitable gelatine or collodion membrane is a strong argument in favor of a porous structure in these particular cases. On the other hand, a collodion membrane can be made which is less permeable to air and more permeable to hemoglobin than the collodion ultra-filter, 2 thus indicating an entirely differ- ent structure. We can now consider the structure of jellies. It seems pro- able that in a jelly we have the concentration of the more vis- cous constituent equal to or greater than that necessary to give zero fluidity. It has already been shown that certain graphite powders give zero fluidity at a concentration of 5.5 .percent graphite and that this could be accounted for if the thickness of the adsorbed film were a little greater than the radius of the graphite particle. If there is partial agglomeration to larger masses, the viscosity will increase and zero fluidity can be reached at lower concentrations for the same relative thickness of ad- sorbed film or at the same concentration with a lesser relative thickness of adsorbed film. If the particles coalesce partially to form something analogous to a string of beads or if they form threads, the viscosity will be increased very much and jellies can be obtained at very low concentrations. All these things have been observed. Bachmann 3 has shown that when soap solutions set to a jelly, long threads are formed. In the case of sodium stearate a translucent jelly can be obtained containing less than one percent sodium stearate. These threads undoubt- edly go over finally to a net structure. With starch and water Meyer 4 has observed the formation of a net structure made up distinctly of drops or globules. In cases where there is a structure without the visible globules being in apparent contact, it is probable that the larger ones are ce- 1 QUINCKE: Drude'sAnn., 10, 482 (1903); 14, 489 (1904);ZsiGMONDY:Zeit. anorg. Chem., 71, 356 (1911); BACHMANN: Ibid., 73, 125 (1912); FLADE: Ibid., 82, 173 (1913). 2 BECHHOLD: Zeit. phys. Chem., 64, 337 (1903). 3 Zeit. Kolloidchemie, 11, 145 (1912). 4 Kolloidchemische Beihefte, 5, 1 (1913). GELATINOUS PRECIPITATES AND JELLIES 241 merited together by smaller ones which are invisible under the conditions of the experiment. It is interesting to note that the single drops are never homogeneous but are always turbid. With gelatine in aqueous alcohol it is possible to get a net structure made up of globules rather than filaments. Meyer considers that all the starch-water jellies are porous masses with both phases continuous. Hardy 1 believed that concentrated gelatine jellies consisted of drops of water suspended in a gelatine-rich phase; but this conclusion was based on the result of compressing jellies and consequently needs further confirmation. It seems probable that the concentrated gelatine jellies with drops of water suspended in a gelatine-rich phase should not be considered to have a honeycomb structure in the sense that an emulsion has a honeycomb structure. With the emulsion there is a film of a third substance around the dispersed drops and we should still have an emulsion if the outer phase could suddenly be made fluid. Under similar conditions the drops of water in a gelatine jelly would undoubtedly run together. A gelatine jelly of the type under consideration is merely a viscous medium in which water is dispersed. If we knead water into butter we do not get a true emulsion any more than we get a true froth when we knead air into dough or into soap. In the case of emulsions the type depends on the nature of the third substance and not on the relative masses of the two liquids. If a gelatine- water mixture were the water-in-oil type at any concentration, it would have to be a water-in-oil type at all concentrations. A good deal of stress has been laid at times on the alleged fact that ultramicroscopic examination has shown that jellies have a sponge structure and not a honeycomb structure; but it is at last admitted that we cannot distinguish between these two cases by means of the ultra-microscope. 2 It has always been assumed tacitly that each so-called phase in a gelatine jelly is truly homogeneous and attempts have been made to determine the composition of one of the phases by squeezing liquid out of a gelatine jelly and analyzing it. Quite apart from errors of manipulation, which are probably large, the underlying assumption is not necessarily sound. If we 1 Zeit. phys. Chem., 33, 326 (1900). 2 BACHMANN: Zeit. Kolloidchemie., 23, 89 (1918). 16 242 APPLIED COLLOID CHEMISTRY shake up fairly pure oil with fairly pure water, the mixture separates into two liquid layers, the oil layer containing some dispersed water and the water layer containing some dispersed oil. Since water peptizes gelatine under certain conditions, there is no reason liquid gelatine or a gelatine-rich phase should not peptize water. It is probable therefore that with gelatinous precipitates and with jellies we have a binary system in which each constituent peptizes the other. A necessary characteristic of such a system is that what we ordinarily consider the separate phases will never be homogeneous. The gelatine-rich phase will always contain peptized water and the water-rich phase will always contain peptized gelatine. Theoretically this can be carried on indefinitely because the peptized water may peptize gelatine and the peptized gelatine may peptize water; but practi- cally there will soon come a point beyond which no effect can be detected. The predicted lack of homogeneity in what would ordinarily be a separate phase has been commented on by Meyer 1 in his studies on starch. A dilute colloidal solution of gelatine in water will consist essentially of turbid drops of a gelatine-rich phase dispersed in water. With increasing concentration of gelatine, there will be a tendency for the separate drops to coalesce into larger drops. If they are too viscous to do this, they may coalesce partially forming threads or, to a still lesser extent, forming a chain of beads. In this latter case it is not necessary that the drops themselves should be in actual contact. In the preliminary stages it may only be the adsorbed films of water which are in contact, just as happens with a plastic mass of sand and water. If the dispersed drops are of different sizes, the smaller drops will probably group themselves around the larger ones, so that the latter will be held in place by a much less visible group of smaller drops. With increasing concentration of gelatine, loose chains may be formed which will then pass into a net or sponge structure of which a granular jelly is an extreme form. In jellies of this type, both phases are continuous and we are dealing with an interlacing system With still further increase in the gelatine we may get water dispersed as definite drops in the gelatine-rich phase. 1 Kolloidchemische Beihefte, 6, 1 (1913). GELATINOUS PRECIPITATES AND JELLIES 243 While this might be called a honeycomb structure, Meyer 1 is probably right in not so classifying it. It is probably better to reserve the term "honeycomb" for structures in which the cell walls are thin relatively to the diameter of the cells and it may be desirable to restrict it to cases like the emulsions where there is a distinct film round the drops. Owing to our comparative ignorance in regard to jellies, we do not know to what extent increasing the concentration of gelatine, for instance, causes the mass to fill up with chains or filaments of the same diameter or to what extent each filament increases in cross-section. It is the absence of a satisfactory working hypoth- esis that makes it so difficult to account for many of the properties of jellies. If a porous jelly of the interlacing type be allowed to dry at constant temperature, the walls may be rigid enough not to collapse completely, in which case we shall eventually get an interlacing system with air in the pores. If the walls are not rigid, they may collapse to such an extent that the pores will disappear practically completely. If the walls do not unite when thus brought in contact, or if there is formed something analogous to a scar, the behavior of the jelly, when placed in water, will vary with its previous history. If the walls coalesce absolutely and uniformly, the rate and amount of swelling will be independent of the previous history of the jelly. Since a gelatinous precipitate differs from a jelly jn^that the latter has not contracted and excreted liquid, it follows that we shall get jellies when the conditions for precipitation are such as to avoid marked contraction. With gelatine or agar the tend- ency to fonn jellies is so strong that no special precautions are necessary. In most cases, however, a slow precipitation is essential where there is a tendency to form a gelatinous precipi- tate. 2 A striking instance of this occurs in dialysis where the liquid in the dialyzer is very apt to set to a jelly just before the colloid is entirely pure, in other words just as the peptizing agent is brought slowly below a critical value. 3 Slightly sour milk coagulates on boiling and a very acid milk will set to a solid mass 1 Kolloidchemische Beihefte, 6, 30 (1913). 2 HOLMES and FALL: Jour. Am. Chem. Soc., 41, 763 (1919). 3 HOLMES and ARNOLD: Jour. Am. Chem. Soc., 40, 1014 (1918). 244 APPLIED COLLOID CHEMISTRY at room temperature. This is a case of a gradual production of precipitating agent and therefore fits in with our general point of view. This is not always understood and it is a popular belief in Sweden that theso-called"dicke Milch" cannot bemade in winter. All this means is that the kitchens are ordinarily so cold in winter that the lactic acid fermentation does not take place sufficiently rapidly. The same effect can be produced by a suitable addition of acetic acid. 1 A so-called calcium suc- rate solution sets to a jelly on addition of carbon dioxide. 2 When caseine is heated with water, the solution becomes more turbid and this turbidity is more marked in presence of calcium salts which tend to precipitate it. 3 With a suitable amount of calcium chloride the liquid clots. 4 Calcium chloride and milk require a higher temperature to cause clotting and calcium chlo- ride does not cause milk to set to a jelly at ordinary temperatures. It is not known what constituent in the milk checks the clotting. There are a number of things about the coagulation of milk by rennin which need further study. In the case of some inorganic jellies, the presence of certain ions seems to be necessary either to ensure sufficiently slow pre- cipitation or to prevent contraction. To get chromic oxide jellies 5 acetate or sulphate must be present. With cupric oxide jellies 6 a small amount of sulphate is necessary. It is surprising in what a number of cases it is possible to get inorganic jellies. Delafontaine 7 found that jellies are formed when ammonia is added to yttrium, didymium, lanthanum, and erbium acetates. Ferric oxide, glycerine and alkali form a jelly on dialysis which liquefies on adding alkali. 8 Zinc oxide 9 and mercuric oxide 10 : Zeit. phys. Chem., 41, 189 (1904). 2 HORSIN-DEON: Jour. Chem. Soc., 26, 612 (1872). 3 VAN SLYKE and HART: Am. Chem. Jour., 33, 467 (1905). * RINGER: Jour. Physiology, 11, 470 (1890). B BUNCE and FINCH: Jour. Phys. Chem., 17, 269 (1913); NAGEL: Ibid., 19, 331 (1915). 6 FINCH: Jour. Phys. Chem., 18, 26 (1914). 7 Chem. News, 73, 284 (1896). 8 GRiMAux: Comptes rendus, 98, 1485, 1540 (1884); FISCHER: Biochem. Zeit., 27, 223 (1910). 9 KuRiLOFr: Zeit. Elektrochemie, 12, 213 (1906). 10 BUNCE: Jour. Phys. Chem., 18, 269 (1914), GELATINOUS PRECIPITATES AND JELLIES 245 jellies have been made. Klemp and Gyrlai 1 discuss the forma- tion of ar senate jellies. Crum 2 made an alumina jelly con- taining 1/600 alumina and 1/7500 sulphuric acid. Bentley and Rose 3 dissolve hydrous alumina in 8 percent acetic acid and add acetic acid. Pickering 4 states that basic aluminum sul- phate can be obtained either flocculent or as a jelly. It is a little difficult to see how a jelly containing 1/600 alumina and 1/7500 sulphuric acid can possibly be solid if it has a sponge or net structure, whereas there would be no difficulty in case there was a honeycomb structure. It is also not clear how coagulation to a net structure can give a gel without any supernatant liquid. There is a good deal more careful work to be done on jellies before we can be sure of what happens in each particular case. With one percent sodium stearate jellies, it seems to be certain that the particles agglomerate to form filaments. 5 VAPOR PRESSURE RELATIONS If the inorganic jellies are not hydrated we should expect the vapor pressure concentration curve to be similar in type with that of charcoal and water, a smooth curve with no breaks. This is what happens with the exception that the pores in the jellies may change in size, in which case the curve will not be reversible. Over fifty years ago, Davies 6 showed that hydrous ferric oxide could be dehydrated completely by prolonged heating under water, which of course could not be the case if a definite hydrate were formed; but is just what might be expected if the ferric oxide changed in structure and became less adsorbent. Over forty years ago Ramsay 7 showed that the rate at which hydrous ferric oxide lost water was not constant and that consequently there could be no definite hydrate within the range studied. Twenty 1 Zeit. Kolloidchemie, 16, 202 (1914). 2 Jour. Chem. Soc., 6, 216 (1854). 3 Jour. Am. Chem. Soc., 35, 1490 (1913); ROSE: Zeit. Kolloidchemie, 16, 1 (1914). 4 Jour. Chem. Soc., 42, 700 (1882). 6 BACHMANN: Zeit. Kolloidchemie, 11, 145 (1912). 6 Jour. Chem. Soc., 19, 71 (1865). 7 Ibid., 32,395 (1877). 246 APPLIED COLLOID CHEMISTRY years ago van Bemmelen 1 determined the isotherm for ferric oxide and water, finding no indication of a definite compound. On the other hand, a definite compound, Fe 2 O 3 .H 2 O, is found in nature as the minerals, Gothite and Lepidocrite. 2 This can be made in the laboratory by decomposing sodium ferrite, Na 2 FeO 4 , but in no other way. This hydrate behaves on dehydration exactly as a definite compound should. The vapor pressure of the hydrous ferric oxide changes with the time and therefore depends on the previous history of the sample. There is no evidence of any hydrate of chromic oxide 3 between 15 and 28. The case of hydrous cupric oxide is a little more difficult because the vapor pressure becomes practically zero at the composition, CuO.H 2 O, and consequently one cannot tell from pressure measurements whether a definite compound occurs or not. If a definite compound existed with a practically zero pressure, it should form from cupric oxide in presence of water. Actually the reverse thing happens and the blue gelatinous cupric oxide goes over spontaneously into the black cupric oxide under water. This change can be prevented by the presence of hydrous oxides of nickel, cobalt, chromium, aluminum, mercury, magnesium, zinc, or manganese. 4 Consequently we conclude that no definite hydrate is formed under ordinary conditions. This does not exclude the possibility of a definite hydrate 5 being formed under special conditions. While these experiments show that some cherished hydroxides do not exist, there are others which stand the test. Van Bemme- len recognizes the existence of BeO.H^O, MgO.H 2 O, MnO.H 2 O, and A1 2 O 3 .3H 2 O. It is not clear why A1 2 O 3 .3H 2 O should be a definite compound and Fe 2 O 3 .3H2O and Cr 2 O 3 .3H 2 O not; but that seems to be the case. The case of silicic acid has been studied with great care by van Bemmelen. 6 A dilute jelly may contain about 300 mols of water to one of silica. Such a jelly flows together again when 1 Zeit. anorg. Chem., 20, 185 (1899). 2 POSNJAK and MERWIN: Am. Jour. Sci. (4), 47, 311 (1919). 3 VAN BEMMELEN: Die Absorption, 78 (1910). 4 BLUCHER and FARNATJ: Jour. Phys. Chem., 18, 629 (1914). 6 BONSDORFF: Zeit. anorg. Chem., 41, 134 (1904). 6 Die Absorption, 196, 214, 232 (1910). GELATINOUS PRECIPITATES AND JELLIES 247 broken into pieces. When the water content is only 30-40 mols the gel can be cut and will stand alone. With 20 mols the gel is stiff, and it is brittle when the content is 10 mols. At 6 mols the gel can be pulverized and the powder is apparently dry. The vapor pressure is practically that of pure water down to a water content of 6 mols per mol of silica. On further dehydra- tion the vapor pressure curve drops continuously with no sign of the existence of any definite compound. When the water con- tent drops to somewhere usually between 1.5 and 3.0 mols with different samples, the gel becomes opaque and chalky but clears up again when the water content reaches about 0.5-1.0 mols. The cloudiness begins when the first air bubbles appear in the pores and lasts until the pores are filled completely with air. Owing to the action of capillarity the water which evaporates from the outer surface of the capillaries is at once replaced from the inside of the gel, leaving a vapor space in the centre of the j elly . This produces an opacity which lasts until the pores are free from capillary water. The balance of the water 0.5 -1.0 mol is probably adsorbed so strongly on the surface of the capillaries that it is almost impossible to remove it, just as we have already seen that heating charcoal 100 above the boiling-point of ben- zene does not necessarily remove the last traces of benzene. As the water passes off, the volume of the gel shrinks until the clouding begins, after which there is practically no change in volume during further loss of water. This is shown by the data in Table XL VII. TABLE XLVII. VOLUMES OP SILICA JELLIES Mols H 2 per mol Si0 2 Volume Mols H 2 per mol Si0 2 Volume Mols HiO per mol SiO 2 Volume 122.0 75.7 29 18 11.3 2.8 3.0 1.0 clouding 1.0 0.73 45.2 11 2.2 0.86 0.39 0.73 23.2 4 1.7 0.75 0.3 t 0.73 In Fig. 10 is given a schematic representation of the pressure- concentratiori relations at 15 for a freshly prepared silica gel. 248 APPLIED COLLOID CHEMISTRY The A-curves represent the first dehydration over sulphuric acid. The Z-curves are for material which has been dehydrated once more or less completely. The direction of the arrows shows whether dehydration is taking place. Starting with a freshly prepared gel, the vapor pressure drops below that for pure water and the dehydration curve consists of the three portions Aft Aa/3 and Aa. There is probably an actual break in the curve at O, where the gel begins to cloud and it is doubtful whether there is a break at Oi where the clouding ends. Owing to the decrease in volume as far as O, the capillaries are entirely filled with water and the gel is transparent. Since the volume does not change much after reaching 0, the capillaries necessarily S/O2 FlG. 10. begin to fill with air as the water evaporates and the gel becomes turbid even up to the point of chalkiness. Owing to capillary action the liquid is always drawn to the outer surface of the capil- laries and the clouding starts at the center. By the time Oi is reached the capillaries are full of air except for a layer of strongly adsorbed water and the gel becomes clear again. Along the curve Aa the last traces of adsorbed water are driven off. If the dehydration is stopped at some point on A/8 and if the gel is exposed to higher partial pressure of water vapor, it is found that the hydration is not reversible and we get a curve Zy. While the gel shrinks along Aft it does not swell again to any appreciable extent and consequently it does not take up water under the same conditions. The curve Zy repre- GELATINOUS PRECIPITATES AND JELLIES 249 sents reversible phenomena at least until the dehydration has been carried beyond the point where Zy meets A/8. From O to Oi we have an irreversible pressure change. If the dehydration is stopped at any point we get curves for hydration like Z0 and Z5 which usually meet as drawn in the point O 2 . From O 2 to Oa the pressure-concentration curve is reversible. From O% we pass to on dehydration. It is possible to pass along the path OOiO 2 O as often as one likes but only in the one direction. The loop OOiO 2 O is therefore a hysteresis loop. Zsigmondy 1 considers that this is due to adsorbed air preventing the water from wetting the capillaries readily but this can hardly be so because the capillaries have not dried out. It seems more prob- able that the water condenses first at the outside of the capil- laries and that the hysteresis is due to the entrapped air or vapor. On this hypothesis the hysteresis should be less marked if the hydration and dehydration were carried on in the absence of air. This experiment has been tried 2 but without any striking results. The break at O was much less marked and the hysteresis loop was somewhat narrower; but it is not certain that this may not be due to differences in the preparation of the gel. Conse- quently there is no conclusive proof of the cause of the hysteresis. The matter is complicated further by the fact that O 2 does not always lie on the direct line from Os to O; but may sometimes be to the left of it. From the observed vapor pressures Anderson 3 calculates that the diameter of the pores is about 2.4-55/i/z. It has been suggested that OOi would be a flat if all the pores were of the same size. The water removed along OiOo is probably the last film of water adsorbed on or in the capillaries. Zsig- mondy considers that it may be water in solid solution in the silica; but this seems improbable because in that case this portion of the curve should be independent of the history of the sample which is not so, though there is nothing impossible about water forming a solid solution with quartz. Tammann* has shown that magnesium platinocyanide and zeolites do not lose their 1 Kolloidchemie, 161 (1912). Spear's translation of this is inaccurate. 2 ZSIGMONDY, BACHMANN and STEVENSON: Zeit. anorg. Chem., 76, 189 (1912). 3 Zeit. phys. Chem., 88, 191 (1914). 4 Wied. Ann., 63, 16 (1897); Zeit. phys. Chem., 20, 323 (1898). 250 APPLIED COLLOID CHEMISTRY water of crystallization at constant pressure and do not cloud. He assumes consequently that the water is present as a solid solution. This has been confirmed by Lowenstein 1 who found a similar behavior with strychnine sulphate and with the oxalates (but not the nitrates or sulphates) of cerium, lanthanum, erbium, yttrium, thorium, and zirconium. In these cases there is no evidence that the vapor pressures depend on the previous histoiy of the sample. If a silicic acid gel is heated, it sinters and contracts, thereby making the pores finer. They consequently do not hold so much water and owing to the change in structure they do not hold it so firmly. Consequently the whole diagram is moved to the left and the curve OOi becomes much shorter. If the silicic acid gel is allowed to stand for a long time at room temperature, there is a change in the walls of the capillaries such that the contraction of the gel on dehydration ceases at an earlier point. Under these circumstances the curve OOi becomes longer be- cause the pores are coarser when they begin to empty. Since there is less water adsorbed strongly by the capillaries more water has to be removed before the pores are entirely empty. That the curve OOi is not due to the presence of a definite hydrate is shown by the fact that the position and length of this curve are so variable and by the fact that similar optical phenom- ena can be observed when alcohol or benzene is substituted for water in the jelly. With alcohol two differences are noted in the pressure-concentration curve. 2 The break at O is not so marked and the point O 2 is far to the left of O, almost down to Oi. The significance of these two variations was not worked out by Anderson. Holmes, Kaufmann and Nicholas 3 have shown that when a suitable silicic acid jelly is allowed to form in a glass test-tube, it vibrates like a rigid solid when tapped. Some of these gels gave a tone two octaves above middle C. The vibration frequency varies directly as the concentration of silicic acid, increases with excess of mineral acids, and decreases with excess of organic acid. 1 Zeit. anorg. Chem., 68, 69 (1909). 2 ANDERSON: Zeit. phys. Chem., 88, 191 (1914). 3 Jour, Am. Chem., 41, 1329 (1919). GELATINOUS PRECIPITATES AND JELLIES 251 The pitch does not vary with the length of the column of gel alter a minimum length is exceeded. With gel columns 60, 90, and 120 mm in length and all 17 mm in diameter, the frequency was 1024 per second in each case, indicating that the vibrations are not longitudinal. With columns of the same length but of different diameters, the frequencies were 341, 640, 1024, and 1152 per second when the diameters were 34, 23, 17, and 15 mm respectively. The vibration frequency varies approximately inversely as the diameter and consequently the vibrations are transverse. When the gels were formed in vaselined tubes, they did not adhere to the walls and the vibration frequency was lower than for similar gels adhering to the glass walls. This shows that under ordinary circumstances the silica jellies are in a state of tension when formed in glass vessels and that this tension has an effect on the pitch of the tone which they emit when tapped. It has been shown that a silica gel containing a great deal of water shrinks very much when the water is removed down to a certain point; but the volume change is not reversible. With a gelatine jelly the change is more nearly reversible. If dried gelatine is placed in cold water it swells a good deal and may take up ten times its weight of water; but there are no experiments to show that it would ever go up, say to an eight percent jelly. On the other hand it is possible to start with an eight percent jelly and dry it to a ninety-six percent jelly, after which it will take up water rapidly to an eight percent jelly. This means that the structure of the gelatine plays an important part in the rate of swelling. This is confirmed by some unpublished pre- liminary results by Mr. Cartledge. Gelatine jellies were made up coritaining 8, 16, 24 and 32 percent of gelatine. These were all dried at room temperature to about 96 percent concen- tration. When water was added, each swelled rapidly to the original concentration and then took up water slowly. If these results are accurate, it means that the four 96 percent jellies were all different and that the 8 percent gelatine did not become like the 16 percent, 24 percent or 32 percent gelatines while being dried. This is very important if true, because this possi- bility is usually overlooked. Arisz 1 finds that at 20 a half 1 Kolloidchemische Beihefte, 7, 1 (1915). 252 APPLIED COLLOID CHEMISTRY percent gelatine jelly disintegrates completely in water, a ten percent jelly goes to a two percent jelly in four days, a twenty percent jelly to a six percent one, a fifty percent jelly to a sixteen percent one, and an eighty percent jelly to a twenty percent one. Although the twenty percent jelly will take up enough water so that its composition is equal to that of a ten percent jelly, the two do not then behave alike. The jelly which has swelled until its composition is ten percent will not then take up as much water in a reasonable time as a jelly which is made up at ten percent. This shows that there is probably a difference in structure and in the way in which the water is held, although Sheppard considers that these phenomena are due to changes in the shape of the mass and to a consequent unequal distri- bution of water. Until we have some satisfactory theory to account for this difference in behavior, all experiments on the swelling of gelatine jellies in solutions of salts, acids and bases are likely to be misleading because one does not know to what extent the electrolytes are causing a change in structure. That a change in structure takes place even with water is shown by the experiments of Arisz on intermittent soaking. If two identical gelatine jellies are placed in water and one is kept in the water six days, while the other is in the water only on the first, fifth, and sixth days, the amount of swelling will be the same in the two cases within the limits of experimental error. While the partially swollen jelly is out of the water, some change in struc- ture takes place such that it takes up water so much more rapidly than the other that the total swelling is the same in the two cases, although one jelly was soaked twice as long as the other. This seems to be a case without a moral because the hare caught up with the tortoise. Mr. C. R. Smith tells me that ash-free gelatine does not set to a jelly in cold water but comes out as a precipitate. This fact must be kept in mind in any discussion of the theory of the swelling of gelatine. In certain salt solutions gelatine swells more than in water. With sodium salts the swelling decreases in the order CNS > I > Br > NO 3 > C1O 3 > Cl > acetate > citrate >tartrate>SO 4 . The last four cause a lesser swelling than in pure water. At a con- centration of 13.8 percent NaCl the amount of water taken up passes through a maximum. Calcium chloride causes more swell- GELATINOUS PRECIPITATES AND JELLIES 253 ing than sodium chloride. 1 Gelatine swells more in dilute acid and dilute alkali solutions than in water or in salt solutions, the order for the acids being HC1>HNO 3 >CH3CO2H>H2SO4>H 3 BO3. With the strong acids the swelling passes through a maximum for a given concentration and then decreases. Salts cut down the swelling both in hydrochloric acid and in caustic soda solutions, the order being the same in both cases. The chlorides, bromides, and nitrates have less effect in cutting down the swelling than do the sulphates and acetates. There is said to be no reversal of anions when one changes from an acid to an alkaline solution. 2 Fibrine behaves very much like gelatine though the amount of swelling is different and the order of the anions is not absolutely identical. Although a great deal of work has been done on the swelling of gelatine and fibrine 3 and some work on the adsorption by fibrine 4 and by gelatine, 5 very little has been accomplished along the line of developing a satisfactory theory. If we pos- tulate that an adsorbed anion is the important factor, as seems to be the case with the salts, then the amounts of any two anions adsorbed must reverse as one changes from an acid to an alkaline solution. If an important factor is the adsorption of an undis- sociated salt, sodium and potassium salts should behave quite differently. It is probable that part of the difficulty is in the way that the experiments have been carried out. With fibrine, for instance, there is a maximum swelling at less than N/10 for hydrochloric acid, nitric acid, and sulphuric acid, whereas with acetic acid there is no sign of a maximum even with normal acid. If we compare dilute acids, say 0.02N, nitric acid and hydrochloric acid cause more swelling than acetic acid; but with normal acid the swelling is much greater in the case of acetic acid. 6 When sodium chloride is added to 0.015N solutions of 1 This must refer to a slightly acid gelatine. In none of the work has the hydrogen ion concentration been controlled sufficiently carefully. Loeb has shown the importance of this; but his work, so far, has been with such dilute salt solutions that he did not detect any adsorption of the salt cation in acid solutions or of the salt anion in alkaline solutions. 2 FISCHER: Das Oedem, 37 (1910). 3 FISCHER: Jour. Am. Chem. Soc., 40, 272, 292, 303 (1918). 4 TOLMAN and STEARN: Ibid., 40, 264 (1918). 5 DUMANSKI: Zeit. phys. Chem., 60, 553 (1907). 6 TOLMAN and STEARN: Jour. Am. Chem. Soc., 40, 264 (1918). 254 APPLIED COLLOID CHEMISTRY hydrochloric acid and acetic acid respectively until the salt con- centration is normal, the swelling is about the same in the two cases; but the decrease in swelling is very much greater with the hydrochloric acid than with the acetic acid. Until some of these factors have been taken into account more thoroughly than has yet been done, it will be impossible to discuss the swelling of gelatine with any hope of success. It must be remembered that gelatine may cause hydrolysis of a salt with simultaneous but separate adsorption of base and acid. This certainly takes place with chromic sulphate solutions and may occur in other cases. One must also keep in mind the possibility of metathetical reactions. Bechhold 1 cites an experiment in which 0.78 grams gelatine swelled in 100 cc of 0.05 HC1 until it weighed 14.61 grams. The weight was only about 7 grams when the solution was also half-molecular with respect to potassium chloride and 2.84 grams when half -molecular with respect to potassium citrate. On the face of it, the potassium citrate has cut down the swelling a great deal more than the potassium chloride ; but it is hardly fair to attribute this to the citrate ion exclusively, because the solution really had a much lower content of hydrogen ion than before and much of the decrease in swelling may have been due to that fact. The experiments of Fischer and Hooker 2 do not bear on this, because they kept the concentration of phosphate constant and varied the acidity and alkalinity. They found the greatest swelling at the acid and alkali ends with consequently a minimum at some intermediate point. Jacques Loeb 3 considers that gelatine forms salts such as gelatine chloride or hydrochloride and sodium gelatinate, which he assumes forms true solutions. There seems to be no possi- bility at present of getting people to agree on a definition of true solution. It is therefore going to be necessary for somebody with an adequate knowledge of colloid chemistry to study the behavior of gelatine in detail and to see whether a better explanation can be given from that point of view. One set of experiments that Loeb made on the swelling of gelatine is very interesting. 4 1 Colloids in Pathology and Medicine, 69 (1919). 2 Jour. Am. Chem. Soc., 40, 272 (1918). 3 Jour. Gen. Physiology, 1, 39 (1918). 4 LOEB: Jour. Biol. Chem., 31, 343 (1917). GELATINOUS PRECIPITATES AND JELLIES 255 Cooper's commercial gelatine, when powdered, swells to a certain extent in water and to a slightly greater amount in M/8 sodium chloride. If the gelatine powder is treated for a short time with the salt solution and then with water it swells more than twice as much as in the salt solution or in water alone. Similar results were obtained with purified pig's bladder. Loeb considers that the first effect of the salt solution is to cause a chemical reaction between the colloid and the salt which would give rise to a great swelling if it were not checked by the salt solution. Replacing the salt solution by water makes possible the swelling due to the unspecified reaction between colloid and salt. This in itself does not mean anything. If only the colloid is changed, there is no chemical reaction between the colloid and the salt. If there is a chemical reaction, then all the salt will not be washed out and one would like to know what and how much was left in the gela- tine. It seems probable that there was some hydrolysis, and that washing out the sodium chloride left the gelatine slightly more alkaline than before, in which case an increased swelling is what ought to take place. This hypothesis receives some con- firmation from the fact that a preliminary treatment of pig's bladder with M/16 disodium tartrate caused nearly twice as much swelling as the preliminary treatment with M/8 sodium chloride. Since tartrate ions are presumably adsorbed much less by gelatine than chlorine ions, one would expect the gelatine to retain more alkali when treated with sodium tartrate than when treated with sodium chloride. The in reased swelling is there- fore what might be expected. Before one can accept this expla- nation it is necessary to show experimentally that alkali is adsorbed in both cases and that more alkali is taken up from the tartrate solution than from the chloride solution. It has also been suggested that the preliminary treatment with the salt solution washed out some of the lime salts in the gelatine. Loeb 1 has done a number of brilliant experiments which bring out clearly the importance of considering the hydrogen ion concentration in all work with gelatine. He finds that the isoelectric point for gelatine lies at a hydrogen ion concentration of 2 X 10~ 7 , or pH 4.7 in Sorensen's logarithmic notation. Under the conditions of the experiments Loeb found that on the 1 Jour. Gen. Physiol. 1, 39, 237, 353, 483, 559 (1918-19). 256 APPLIED COLLOID CHEMISTRY acid side of the isoelectric point only anions of neutral salts are taken up and on the alkaline side of the isoelectric point only cations. Since the Hofmeister series calls for an effect due to both ions of a neutral salt on the swelling of gelatine, Loeb concludes that the Hofmeister series is a delusion and a snare. This does not follow at all. Loeb is working at such extreme dilutions that the specific effects of all ions but hydrogen and hydroxyl ions are practically negligible. In acid solutions only anions are taken up and in alkaline solutions only cations. Loeb recognizes the specific effect of iodine ions over chlorine ions in causing the liquefaction of gelatine; but he considers that liquefaction stands in no necessary relation to swelling, an assumption which will be shared by few. With higher salt concentrations Loeb will undoubtedly get entirely different re- sults. His conclusions as to the existence of definite compounds depend on the assumption that he is dealing with true solutions and will fall with that assumption. While it is not possible to accept his theoretical conclusions, his experiments with silver nitrate and with potassium ferrocyanide, to take two striking cases, are marvellous as an example of technique. The so-called osmotic pressure measurements are apparently meas- urements of swelling and not of osmotic pressure. Since the swelling and shrinking are practically reversible, the gelatine jellies do not become opaque at some point as was the case with the silica gels. Katz 1 has made some experiments on the vapor pressures of gelatine solutions. He obtained a smooth pressure-concentration curve with a point of inflection. With an 83 percent jelly the partial pressure of the water vapor was about half that of saturated water vapor and for a 90 percent jelly the partial pressure was about one-tenth that of saturated water vapor. The behavior of a gelatine j elly in liquid water and in saturated vapor has already been discussed elsewhere. 2 If the gelatine jelly is hardened somewhat by replacing the water with benzene, 3 it behaves in many respects like the silicic acid jellies. The pressures that can be exerted by a swelling colloid may be very large. We know how a growing vine will displace stones. In 1 Zeit. Elektrochemie, 18, 800 (1911). 2 See page 75. S BACHMANN: Zeit. Kolloidchemie, 23, 85 (1918). GELATINOUS PRECIPITATES AND JELLIES 257 some experiments on dried seaweeds, it was found that water was taken up against a pressure of forty atmospheres and Rode- wald states that starch will swell against a pressure of 2500 atmospheres. When gelatine dries in contact with glass, it may pull pieces of glass off the surface. Posnjak 1 has made some experiments on the amount of water with which gelatine is in equilibrium under different external pressures and on the cor- responding behavior of raw Para rubber in different organic liquids. The data are given in Tables XLVIII-XLIX, the pressures being centimeters of mercury and the concentrations being grams of TABLE XLVIII. SWELLING OF GELATINE IN WATER P = external pressure in cm Hg. C = grams water per gram gelatine. P 38.3 82 4 156 240 303 377 C 2.56 2.06 1.46 1.28 1.10 0.92 TABLE XLIX. SWELLING OF RUBBER IN ORGANIC LIQUIDS P = external pressure in cm. Hg. C = grams liquid per gram rubber. Liquid P = 82.4 LC P = 156 C P = 240 C P = 377 C Ether 2 40 1 79 1 49 Benzene 4.41 3.28 2.72 2.09 Thiophene 5.32 4 00 3.32 2.71 Acetylene dichloride. . . Chloroform 7.35 9.31 5.22 7.07 3.42 4.58 Carbon tetra chloride . . 11.06 8.47 6.69 liquid per gram of gelatine or rubber. As was to be expected, the liquid content decreases with increasing pressure. It would not be safe to extrapolate from these data to the pressures neces- sary to squeeze out the last drops of liquid; but they would evi- dently be very high. The specific effects of the different liquids are quite marked in the case of rubber, nearly five times as much carbon tetrachloride being taken up as ether. Owing to the differences in the densities of the two liquids, this is of course i Kolloidchemische Beihefte, 3, 417 (1912). 17 258 APPLIED COLLOID CHEMISTRY not a measure of the amount of swelling. With an external pres- sure of 156 cm Hg, Posnjak found that the amount of rubber in grams per liter was 283 with ether and 150 with carbon tetrachloride. The taking up of liquid may be accompanied by a marked evolution of heat. In Table L are some figures by Rodewald for the heat evolved when starch swells in water. TABLE L. HEAT EVOLUTION WITH STARCH AND WATER %H 2 O Calories per gram starch 0.23 28.11 3.23 20.97 8.16 12.43 12.97 7.37 19.52 2.91 We do not know at all what the solubility is of gelatine in water or of water in gelatine, except that it must be small. Experi- ments have been made to determine these values by pressing the gelatine and analyzing the supernatant liquid when a jelly con- tracts; but these data mean nothing because there is no proof that the so-called solution does not contain the substance in a colloidal form. DIFFUSION In dilute gelatine solutions the rate of diffusion of electrolytes is very nearly the same as in pure water; but this does not hold as the jelly is made more concentrated. The electrical conduct- ance is also less in gelatine solutions than in water. 1 Although diffusion takes place readily in gelatine jellies, some curious phenomena have been observed under special conditions. Liesegang 2 coated a glass plate with gelatine containing a small amount of potassium or ammonium bichromate. When a drop of concentrated silver nitrate solution is placed on the gelatine, the silver nitrate diffuses into the gelatine and reacts with the bichromate. 3 Instead of forming a continuous precipitate, the silver chromate comes down in concentric rings which are called 1 DUMANSKI: Zeit. phys. Chem., 60, 553 (1907). 2 Chemische Reaktionen in Gallerten, 1898. 3 BECHHOLD: Colloids in Biology and Medicine, 261 (1919). GELATINOUS PRECIPITATES AND JELLIES 259 Liesegang's rings, or rhythmic banding. A similar experiment can be carried out in a test-tube and it is then possible to get as many as twenty or more parallel membranes. Ostwald 1 at- tempted to account for the phenomenon by supposing that the silver chromate remained in supersaturated solution, diffusing along with the silver nitrate until the metastabile limit is reached, when it precipitates. This ingenious hypothesis has not proved satisfactory, and there is as yet no entirely satisfactory explana- tion of the phenomenon. Microscopic examination shows that the apparently clear spaces between the concentric rings are not absolutely free from crystals of the insoluble compound, but contain very fine crystals. The phenomenon varies with the salt used, sodium chloride not giving rings, and with the nature of the jelly. The reaction between silver nitrate and potassium bichromate gives rings or layers in gelatine but not in agar, while lead nitrate and potassium chromate give layers in agar and not in gelatine. Neither of these reactions gives rise to layers in silicic acid, 2 though other reactions do. Bradford 3 believes that one of the reacting solutes is adsorbed by the layer of precipitate, thus giving a zone practically free from it. There is of course no objection to the assumption; but, as Hatschek points out, this explanation and that of Ostwald leave the specific effect of the jelly entirely out of account. Rhythmic banding is important, because it is probable that something of this sort is the cause of the banded structure ob- served in agates and elsewhere. It is worth noting that Bech- hold claims that the phenomenon does not necessarily have anything to do with diffusion because similar structures can be produced in jellies by crystallization of sodium phosphate or by freezing water. Holmes 4 has made a large number of crystalliza- tions in silicic acid jelly. He has also put forward a theory of rhythmic banding, 5 which unfortunately does not account for the fact, recorded in the same paper, that colloidal gold gives three colored bands, red, purple and blue, before repeating. 1 Lehrbuch allgem. Chemie, 2d Ed., 2, 778. 2 HATSCHEK: Second Report on Colloid Chemistry, 21 (1919). 3 Biochem. Jour., 10, 169 (1916); 11, 14 (1917). * Jour. Phys. Chem., 21, 709 (1917). 5 HOLMES: Jour. Am. Chem. Soc., 40, 1187 (1918). CHAPTER IX EMULSIONS AND FOAMS EMULSIONS When considering the adsorption of solids by liquids, it was pointed out that certain solids will pass into the interface between two liquids and will be held there quite firmly. This will also happen if the solid is present in so finely divided a state that it forms a colloidal solution in one or both of the liquid phases, as was found experimentally by Winkelblech 1 who, however, was very much puzzled by the phenomenon. He shook a dilute solution of gela-tine with benzene and found a film of gelatine at the dineric interface. The method was so sensitive that 0.06 mg gelatine could be detected when working with 10 cc lots. Win- kelblech proposed to use this as a method of analysis; but this is not satisfactory because the sensitiveness varies with changes in the gelatine or in the acidity of the solution. The more readily the substance is peptized by one of the liquids, the less readily it will go into the interface. We have already seen that solid gelatine is not peptized readily by water. Similar results were obtained with albumin, water-soluble starch, soap, colophonium, the proteins in beer, and with tannic acid, though not with gallic acid. As precipitating liquids could be used petroleum, liquid paraffin, benzene, carbon tetrachloride, and chloroform, but not ether. Winkelblech did not find out why ether did not work; but it is interesting to note that Miller and McPherson 2 found that arsenic sulphide forms a colloidal solution both in ether and in water, and that on shaking a colloidal solution of arsenic sulphide in water with ether, the colloid distributes itself through the two phases. Winkelblech also succeeded in the partial extrac- tion of colloids from organic liquids by shaking colloidal solutions of fat,, with water, the fats forming a film at the interface. 1 Zeit. angew. Chem., 18, 1953 (1906). 2 Jour. Phys. Chem., 12, 706 (1908). 260 EMULSIONS AND FOAMS 261 Biltz and Krohnke 1 made some experiments to see whether the colloidal organic matter in sewage could not be extracted by shaking with organic liquids but they found that only about 30 percent went into the interface, so the method never came into technical use. 2 In these experiments of Winkelblech the amount of organic liquid is large and the drops formed on shaking are not very small. The tendency is to form a coarse emulsion but this is not stable and cracks, the colloid then passing into the interface. An emulsion consists of drops of one liquid suspended in another liquid. When these drops are stabilized by an electrical charge they behave like any other colloidal solution of the same type. 3 In most cases there is an actual film around the globules and this keeps them from coalescing. It is evident that with any pair of non-miscible liquids there may be two sets of emulsions, one with drops of oil suspended in water and one with drops of water suspended in oil, the word oil being used here to mean the other liquid. The water-in-oil emulsion may be important in certain cases. Wa. Ostwald 4 has pointed out that if one treats wood with creosote emulsified in water, the water comes in direct contact with the wood, which is not what one wants. If one takes an emulsion of water in creosote, the creosote comes in direct contact with the fibers of the wood, which is much more satisfactory. It seems probable that in the case of a kerosene spray, one type is better than the other, though that point seems not to have been considered. The necessary conditions for forming a stable emul- sion are that the drops shall be so small that they will stay sus- pended and that there shall be a sufficiently viscous film round each drop to keep the drops from coalescing. An emulsifying agent is therefore a substance which goes into the interface and produces a film having satisfactory physical properties. If the adsorption of the emulsifying agent lowers the surface tension on the water side of the interface more than it does on the oil side, the interface will tend to curve so as to be convex on the 1 Zeit. angew. Chem., 20, 883 (1907). 2 Cf. BRIGGS: Jour. Phys. Chem., 19, 210 (1915). 3 ELLIS: Zeit. phys. Chem., 80, 597 (1912); Powis: Ibid., 89, 9\ 179 (1914). 4 Zeit. Kolloidchemie, 6, 103 (1910). 262 APPLIED COLLOID CHEMISTRY water side, and we shall have a tendency to emulsify oil in water. If the adsorption of the emulsifying agent lowers the surface tension on the oil side of the interface more than it does on the water side, the interface will tend to curve so as to be concave on the water side, and we shall have a tendency to emulsify water in oil. The simplest way then to emulsify oil in water is to add a water-soluble colloid which is adsorbed strongly at the interface and the simplest way to emulsify water in oil is to add an oil-soluble colloid which is adsorbed strongly in the interface. As a matter of fact this is the way in which almost all emulsions are made. Since most emulsifying agents will not peptize the liquid by themselves, it is usually necessary to break up the liquid into fine drops by some form of mechanical agitation. Sodium and potassium oleates are water-soluble colloids and they are excellent for emulsifying oils in water. The gums are also water-soluble colloids and certain of them are much used in pharmaceutical work for emulsifying oils in water. Though gelatine is a water-soluble colloid, it is not used ordinarily as an emulsifying agent probably because there is too much danger of putrefaction. By adding salt to a soap solution, the soap film can be made to come out so brittle that it will not form a satisfactory emulsion. Pickering 1 made emulsions of kerosene in water with small amounts of soap as the emulsifying agent. With 70-80 percent of kerosene, the mass was viscous and with 99 percent oil it was as stiff as blanc-mange, and could be cut into cubes which would stand alone. When a lump of this solidified kerosene was left standing in dry air, it liquefied because the water evaporated from the soap film which then cracked, letting the kerosene run out. The solidified alcohol which is so much used nowadays is not an emulsion, though it does contain some water. It is es- sentially an alcoholic soap jelly and presumably does not con- tain alcohol dispersed as drops. Calcium and magnesium oleates form colloidal solutions in oil and can therefore be used to emulsify water in oil. Rosin and the resinates act in the same way and the water in ready-mixed paints is an emulsion of water as drops in linseed oil containing rosin. It is a simple matter to add fifty percent of water to a 1 Jour. Chem. Soc., 91, 2002 (1902). EMULSIONS AND FOAMS 263 ready-mixed white lead without changing the look of the mixture much, because the oil is the external phase. If water is stirred into a heavy lubricating oil containing a lime soap, the water will be emulsified in the oil and will change it to a grease, instead of decreasing the viscosity as might perhaps have been expected. Lanolin, or purified wool fat, will emulsify water and can easily be made to take up eighty percent of it. Nobody has deter- mined what the emulsifying agent is in this case. Friedrich 1 pro- poses making printing inks with about half the normal content of linseed oil, turpentine, drying oils, etc. by diluting with water emulsified in wool fat and adding a little more coloring matter. The degras which is used in dubbing leather is apparently an emulsion of water in oil because Fahrion 2 states that normal degras should contain 15-20 percent water but that English sod- oil may run 30-40 percent water and the white degras from the German chamois leather works may go as high as 40-50 percent water. The water in butter may be emulsified by means of a calcium soap; but more probably the water is merely dispersed as drops in a viscous medium without any true emulsion being formed. Milk is an emulsion with butter fat as drops in a water phase. It is possible to make imitation fresh milk by emulsifying renovated butter fat in skimmed milk. It is claimed that some day the milk problem of the large cities will be solved by shipping butter and a suitable milk powder to the cities and adding the water there, since there is a very large difference in cost between sending milk powder by slow 'freight and water by what is practically express, to say nothing of the difference in keeping qualities. Mayonnaise is an emulsion of oil in water with egg as the emulsi- fying agent. Cod-liver oil is of course an emulsion of oil in water. It is not always easy to tell by inspection, whether water is external or internal phase in a given emulsion. A good way, devised by Briggs, is to examine the emulsion under the micro- scope, 3 stirring in gently a little water or a little oil. The one that is external phase will mix readily with the emulsion, while the other will not. 1 Jour. Soc. Chem. Ind., 36, 296 (1917). 2 Ibid., 30,501 (1911). 3 NEWMAN: Jour. Phys. Chem., 18, 34 (1914). 264 APPLIED COLLOID CHEMISTRY The emulsion jellies are of course all of the honeycomb type and it is a pity that none of them have yet been examined iii the ultramicroscope. If they had been shown to be quite different in appearance from the gelatine and the silica jellies, for instance, this would have been a strong argument in favor of the view that the latter jellies are of a different type. If no marked difference could be discovered, as would undoubtedly be the case, 1 it would have been recognized long ago that the evidence of the ultramicroscope is worthless in matters of this sort. In making emulsions of oil in water by means of sodium oleate, it must be remembered that the soap does not go entirely into the interface. There is always a relation between the concen- tration of the soap in the interface and the amount still peptized in the water. 2 It is a case of the distribution of soap between the water and the interface. Anything that cuts down the peptiz- ing action of the water on the soap will increase the amount going into the interface and anything increasing the peptizing action of the water will decrease the amount of soap in the interface. It is not a question of osmotic pressure or of solubility. While most emulsions are made with gelatinous colloids as emulsifying agents, this is not necessary theoretically. Any- thing that will go into the interface and make it sufficiently viscous will give the same result. If we can put enough of a fine powder into the interface, we shall have a plastic mass there, which will stabilize the emulsion. This has been done by Pickering 3 who obtained emulsions of oil in water using powdered, basic copper or iron sulphate as the emulsifying agents. The emulsions were not very good ones but they were emulsions. Sheppard 4 has obtained similar results, emulsifying nitrobenzene in aqueous sulphuric acid by means of lead sulphate. When mayonnaise is made with the powdered yolk of a hard-boiled egg, the resulting emulsion comes under this head. Schlaepfer 5 and Moore 6 have emulsified water in kerosene by means of carbon black. 1 BACHMANN: Zeit. Kolloidchemie, 23, 89 (1918). 2 BBIGGS: Jour. Phys. Chem., 19, 210 (1915). 3 Jour. Chem. Soc., 91, 2010 (1907); Zeit. Kolloidchemie, 7, 14 (1910). * Jour. Phys. Chem., 23, 634 (1919). 5 Jour. Chem. Soc., 113, 522 (1918). 6 Jour. Am. Chem. Soc., 41, 940 (1919). EMULSIONS AND FOAMS 265 The pharmacists make great use of emulsions of oil in water and the books on pharmacy are filled with elaborate details as to the method of making emulsions. There are two general me- thods in vogue and both give good results, although they appear mutually contradictory. In the American method the gum, which is used as an emulsifying agent, is peptized and the resulting colloidal solution is placed in a large mortar. The oil is added a little at a time or in four separate portions with continuous stir- ring. If one starts stirring to the right, one must continue stirring to the right or no emulsion will be formed. Some books go so far as to say that a left-handed man cannot make an emulsion, but that seems a bit absurd. In the Continental method, the gum is ground in a mortar with the oil, in which it does not dissolve, and enough water to emulsify all the gum is then added at once. Since the essential condition for making an emulsion is to break the oil into sufficiently fine drops and coat them with the gum solution, there is no necessary reason for adding the oil in separate portions in the American method. When the constitu- ents are mixed in a bottle and shaken in a shaking machine, it is easy to make an emulsion if the amount of oil is not too large. With a mechanical shaker 1 giving 400 shakes per minute and a one percent solution of sodium oleate, 80 percent by volume of benzene could be emulsified in fifteen minutes but it took two hours to emulsify 96 percent of benzene, and 99 percent of benzene was not completely emulsified after eight hours of continuous shaking. If the shaking is done intermittently by hand so that the unemulsified benzene can separate at intervals, the number of shakes can be cut down enormously, from 48,000 to 80 in the case of the 96 percent emulsion and from 6000 to 7 in the case of the 80 percent one. Continuous shaking in a bottle disintegrates both benzene and soap solution into drops, the latter of which is disadvantageous. Since the soap solution is the continuous phase in these emulsions, any treatment which breaks up the soap solution retards emulsification. When ben- zene is dispersed into drops in the soap solution, the soap coats the drops and prevents them from coalescing, while there is nothing to prevent the rapid coalescing of the water drops in the benzene. The first few shakes cause the most disintegration of 1 BRIGGS: Jour. Phys. Chem., 24, 120 (1920). 26G APPLIED COLLOID CHEMISTRY the benzene and the least disintegration of the soap solution ; con- sequently intermittent shaking gives the best results. The pharmacist was doing intermittent stirring unconsciously when he added the oil in separate portions. The reason for using a mortar that is apparently much too large is to get a large surface and thus to facilitate emulsification. A similar result can be obtained by putting the mixture in a cylindrical glass bottle, with or without glass beads, and rolling the bottle. These experiments throw some light on the making of mayon- naise. Since mayonnaise is essentially an emulsion of oil in water (vinegar) with egg as the emulsifying agent, it ought to behave like any other emulsion and so it does for the experts in the Departments of Home Economics. They can add the in- gredients in any order, all at once or in separate portions, hot or cold, and the mayonnaise always comes. On the other hand, these same experts do not train their pupils so that these latter can make mayonnaise every time. It seems certain that the experts do something or other unconsciously which they conse- quently do not tell to their pupils. Probably the expert is so sure of the result that she works leisurely without being hurried or flurried, and is practically doing intermittent stirring. The person who is not an expert and who is uncertain of the outcome probably goes at her task so vigorously as to defeat her object in many cases. While this explanation has not been tested, one expert said that she had found that if the materials were beaten well together and then allowed to stand for a moment or two, a couple of swishes would make the mayonnaise. I have been told that Bearnaise sauce is an emulsion of melted butter in tarragon vinegar without any egg and that it is very easy to make. If this is so, tarragon vinegar must contain a good deal more of some emulsifying material, probably a tannin, than ordinary vinegar. The Continental method of making emulsions has also been studied by Briggs 1 who finds that the object of grinding the oil with the solid gum is merely to increase the interface between oil and water by suspending in the oil finely divided solids which are wetted preferentially by water. The gum acts both as finely divided solid and as emulsifying agent. It is possible to 1 Jour. Phys. Chem., 24, 147 (1920). EMULSIONS AND FOAMS 267 get the same results by grinding the oil with sand and adding a colloidal solution of gum, but the resulting emulsion then contains the sand. Since sodium oleate emulsifies oil in water and calcium oleate emulsifies water in oil, a mixture of the two oteates will behave differently, depending on the relative amounts. There will also be some ratio of calcium to sodium at which the two oleates will practically balance each other and the slightest relative change will change the type of the emulsion. This has been proved experimentally by Clowes 1 who worked with a sodium soap and calcium chloride. This antagonistic action of sodium and cal- cium salts with reference to emulsions throws light on some per- plexing physiological problems. Jacques Loeb showed that certain marine organisms died in sodium chloride or calcium chloride solutions which were isotonic with sea-water; but flour- ished when there was a definite ratio of sodium to calcium, a result which could not be explained on the basis of osmotic pressure. If we -consider protoplasm as consisting of lipoids (oil) and water, we shall have an emulsion of oil in water in presence of sodium salts and of water in oil in presence of calcium salts. When the sodium and calcium salts are present in a definite ratio, there will be a balancing between these two types of emulsion, and it may well be that this critical state is the one which is conducive to life and growth. As a matter of fact, Clowes found that the ratio of sodium and calcium salts necessary to produce a balancing between the two types of emul- sions when working with oil, water and soap, was practically the same as that found in sea-water. This indicates a very close connection between the two sets of phenomena. Osterhout 2 at Harvard has shown that the specific electric conductivity of certain seaweeds is increased by addition of sodium salts and decreased by the addition of calcium salts. If the hypothetical emulsion changed to one of oil in water, the conductivity should increase, and it should decrease if the emulsion changed to one of water in oil. Clowes 3 has succeeded in duplicating Oster- hout's results by impregnating filter paper with an emulsion 1 Jour. Phys. Chem., 29, 407 (1916). 2 Science, (2) 35, 112; 36, 350 (1912). 3 Proc. Soc. Exp. Biol. Med. 15, 108 (1918). 268 APPLIED COLLOID CHEMISTRY of oil, water, and soap, to which he afterwards added sodium and calcium salts. When a badly made emulsion is allowed to stand, it decom- poses progressively. This seems to be due to the fact that when the free oil comes in contact with the film around the oil drops, it tends to make it bend in the opposite direction, thereby crack- ing the emulsion. Temperature changes probably play a part in causing the ageing and cracking of the films; but there has been no careful study of this point. All we know is that anything which destroys the film destroys the emulsion. That is why acid destroys a soap emulsion. In the case of oil emulsions in water, we get cataphoresis and the emulsions can be broken down by the action of a high-tension direct current. 1 In the oil industry emulsions of one kind and another are a common occurrence and their removal is quite an important matter. Very little that is accurate has been published in regard to this. One interesting case in the technical application of emulsions is the splitting of fats by superheated steam in the presence of lime. Caustic soda is not effective because it is necessary to use so much; but one percent of lime is sufficient to cause complete hydrolysis. No explanation of this has been tested out; but it seems certain that the calcium soap emulsifies the water in the fat. This may give a sufficiently intimate mixture to hydrolyze the fats or the water globules may pass into steam, breaking down the emulsion which then forms again. Whatever the secondary details may be, it is clear that the formation of an emulsion of water in oil is the essential thing because the oxides of magnesium and of zinc act in a similar way. FOAMS In an emulsion of oil in water we have drops of oil coated by a more or less viscous film containing water, with water outside. If the emulsion is not sufficiently fine, the emulsified oil will rise to the surface and we get a sort of cream, which consists almost entirely of the oil drops with their viscous coating and with little or no free water. If we consider the globules of oil replaced by air, we shall have a mass of froth, the only difference that the 1 FANTO and STRITAB: Jour, prakt. Chem. (2) 81, 564 (1910). EMULSIONS AND FOAMS 269 bubbles would be much finer than in the ordinary froth. The conditions for forming froth should then be much the same as for forming an emulsion and that is the case, though there are of course some differences. It has already been shown that colloidal solutions of peptone form a film of solid peptone over the surface 1 and that the surface viscosity of a saponin solution is 60 times that of the mass of the liquid. The formation of a film at the surface depends primarily on the adsorption of the colloid at the interface and not on the presence of air. On the other hand the adsorption of air by the colloid will cause the latter to rise more rapidly and will also help to keep it there. This effect might be so strong as to give rise to a film at the surface in certain cases where none would be formed otherwise. Stark 2 has called especial attention to the adsorption of air by colloidal particles. We know that gum arabic adsorbs air strongly and sodium palmitate likewise. Findlay 3 has shown that carbon dioxide is taken up under some conditions by starch solutions. Freundlich 4 considers that no pure liquid can foam and in this he is apparently right. He then deduces that a low vapor pres- sure, a low surface tension, and a high surface viscosity are essential to foam. The low vapor pressure is put in avowedly because a foam with a volatile liquid like ether would evaporate very rapidly. That is not necessary because we can work in closed vessels. Freundlich has put in the low surface tension because all the foaming solutions which he knew about had lower surface tensions than the pure liquid; but this criterion is not sound because salt solutions will foam. Since salts may increase the fluidity of water, a high surface viscosity is not essential. All three of Freundlich's criteria thus go by the board, which shows the danger of the empirical generalization. On the other hand Freundlich's criteria do hold in practically all cases if one con- fines oneself to fairly permanent foams, which was what he was really considering. To get a foam the only essential is that there shall be a distinct 1 See page 189. 2 Wied. Ann., 68, 117, 618 U899). 3 Jour. Chem. Soc., 101, 1459 (1912); 103, 636, 1170 (1913). 4 Kapillarchemie, 302 (1909). 270 APPLIED COLLOID CHEMISTRY surface film, in other words, that the concentration in the surface layer shall differ perceptibly from that in the mass of the liquid. All true solutions will therefore foam if there is a marked change of surface tension with concentration, regardless whethert he surface tension increases or decreases. All colloidal solutions will foam if the colloid concentrates in the interface or if it is driven away from the interface. To get a fairly permanent foam the surface film must either be sufficiently viscous in itself or must be stabilized in some way. This can be done by introducing a solid powder into the interface. Solutions of aqueous alcohol, acetic acid, sodium chloride, and sulphuric acid all foam when shaken; but the foam is instable. Soap solutions foam when shaken and the foam is, or may be, quite stable owing to the viscosity of the soap film. With sapo- nin the surface film is even more stable. If we add to aqueous alcohol some substance like lycopodium powder which goes into the interface, we get a stabilized foam. We can do the same thing with aqueous acetic acid by adding lamp-black. The pres- ence of enough of a finely divided solid in the interface will make the film so viscous that the foam will be quite stable. Grease will help stabilize a foam in some cases and it has been claimed erroneously that the foaming of sulphuric acid solutions is due to grease. 1 In 1857 Gladstone 2 pointed out that aqueous solutions of organic substances are apt to froth and he cited beer as a then familiar instance. He did not realize however that it was the col- loidal matter in the beer which caused the frothing. It has been shown by Zeidler and Nauck 3 that removing the albumoses from beer destroys the foaming. Gladstone showed, however, that the dissolved air was not essential to the frothing. If this were pumped out in a vacuum, the liquid frothed freely when shaken with air. Gladstone states that aqueous solutions of the acetates of iron, copper, lead, and other metals froth readily, especially the ferric acetate solution. As we know, this is the one which hydrolyzes most readily and the frothing is due undoubtedly to the combined effect of the acetic acid and the hydrous ferric oxide. 1 LANG: Ber. deutsch. chem. Ges., 18, 1391 (1885). 2 Phil. Mag. (4) 14, 314 (1857). 3 Jour. Soc. Chem. Ind., 20, 260 (1901). EMULSIONS AND FOAMS 271 Ramsden 1 obtained precipitation of solids by shaking solutions of albumin, proteins, gelatine, cupric acetate, ferric acetate, mastic, etc. This is essentially the same in principle as Winkel- blech's later experiments with a second liquid phase. The colloid goes into the surface, forming a solid film which is so brittle, how- ever, that it does not give a good emulsion or foam. When the emulsion or foam cracks, the colloid does not go back at once into colloidal solution; but remains as a precipitate. An interesting case of this is to be found in the behavior of enzymes. Many enzyme solutions can be rendered inactive merely by shaking. 2 A rennet solution need be shaken violently in a test tube for only two minutes to destroy most of its power to coagulate milk. The inactivation was found to be a surface phenomenon, the inactivation increasing with the length of time and the violence of the shaking. The volume of air, the concentration of the enzyme and the temperature are all influencing factors. The en- zyme concentrates in the foam and on the walls of the vessel. Vanino and Hart I 3 shook up a red colloidal solution of gold and found that the gold first turned blue and then precipitated as brown gold. The irreversible formation of a film on agitation with air is an important thing in making whipped cream. It seems probable though not proved that we might get tiny air bubbles emulsified in a liquid, in which case the air in the bubbles would be given off relatively slowly when the pressure was relieved. If we have two colloids present, one of which goes into the interface more readily than the other, the first may keep the second out more or less completely. Herty and Willard 4 found that resene cuts down soap lather while it has been shown that shaking does not inactivate a rennet solution to which saponin has been added because the latter keeps the rennet from going into the interface. Under adsorption of solid by liquid, reference was made 5 to the Elmore bulk oil process of ore flotation where the sulphide i Proc. Roy. Soc., 72, 156 (1903); Zeit. phys. Chem., 47, 336 (1904). *BECHHOLD: Colloids in Biology and Medicine, 189 (1919). 3 Ber. deutsch. chem. Ges., 37, 3623 (1904). 4 Jour. Ind. Eng. Chem., 6, 895 (1914) 5 See page 89. 272 APPLIED COLLOID CHEMISTRY ore went chiefly into the interface between oil and water while the siliceous gangue settled to the bottom. This was an expen- sive process technically because it called for about one ton of oil per ton of ore. The Minerals Separation Company holds a patent for ore separation using only a fraction of one percent of oil per ton of ore, a froth being formed which carries up the values. If no ore is present, bubbles of air are formed coated with a thin film of oil, around which is water. The oil film is not viscous enough to form a stable froth. If we introduce particles of sulphide ore into the oil-water interface or into the oil itself, we get a stabilization of the film because we then have practically an armor-plated bubble. The resulting froth can be scraped or shovelled off. It is said that about sixty million tons of ore are handled by this process every year in America alone. In most cases acidification of the solution and rise of temperature increases the effectiveness of the separation. A blended oil seems often to give better results than a straight oil. One oil may spread better over the surface and be a more effective frothing oil, while a second oil may have a greater effect in carry- ing the sulphide particles into the interface. Although a good deal of experimenting has been done with this process, the theo- retical results are not what they should be, apparently because people have insisted upon looking upon the whole thing as very mysterious. This is one of the many industries involving colloid chemistry where the art is far ahead of the science. In Australia eucalyptus oil has proved very successful. In this country so- called pine oil has been used a great deal with or without admix- ture of mineral oils. While sulphide ores are the ones which have been worked with the most, the principle is a general one, and the Calumet and Hecla Mining Company are reported to be using the process for removing native copper from the tailings. If the amount of air is insufficient, the oil may cause the ore particles to agglomerate and sink. This was patented by Catter- mole but is not technically so successful as the frothing process. If the amount of air is excessive relatively to the ore, the number of bubbles will be so great that the ore particles cannot coat them sufficiently to armor-plate them ; the bubbles will therefore be in- stable. This principle is involved in the Callow process. Air is introduced in fine bubbles at the bottom of the cell These EMULSIONS AND FOAMS 273 bubbles float up the ore particles but break as soon as they reach the surface. The ore particles must be removed before they settle back. We can therefore have clotted particles, armor-plated bubbles, or fragile bubbles, depending upon the conditions. The flotation process will not work with large lumps because the force of gravity would carry them down. One of the ad- vantages of the method is that it will work with finely ground material. On the other hand there is a limit. It is customary to say that the ore goes into the interface and the gangue does not; but that is not to be taken too literally. Some gangue always goes into the interface and the percentage increases the finer the gangue. The ideal condition therefore is one in which the gangue is coarse relatively to the ore particles. That the essential thing is the adsorption of the particles at the oil-water interface is shown by the fact that the introduction of saponin kills the flotation of the ore under ordinary conditions although it increases the frothing tremendously. A carbon dioxide froth, made from aluminum sulphate and sodium bicarbonate, is used for putting out fires. It is not suc- cessful against burning aluminum powder for two reasons. The aluminum powder is floated up by the froth and consequently not blanketed by it. Also hot aluminum powder reduces carbon dioxide to carbon monoxide which then burns in the air. Fortu- nately few fires are due to aluminum powder. The alumina froth is said to be stabilized by the addition of other colloids such as licorice; but no details have been published. 18 CHAPTER X NON-AQUEOUS COLLOIDAL SOLUTIONS Very little systematic work has been done on sols and gels with liquids other than water. Graham replaced the water in a silicic acid jelly by alcohol, sulphuric acid, nitric acid, and formic acid. Van Bemmelen substituted acetone for water, while Bachmann 1 put in benzene. Butschli replaced the water in gelatine and agar jellies by alcohol, for which could then be substituted chloroform, turpentine or xylene. We know that the calcium and magnesium oleates form col- loidal solutions in organic liquids and that the so-called resinates do the same thing. It is very probable, though not proved, that stearin exists in colloidal solution in olive oil. The whole in- dustry of the cellulose nitrates and cellulose acetates depends upon the formation of colloidal solutions in organic solvents. Pieroni and Girardi 2 have studied some colloidal solutions in pyridine. Water can be emulsified in lanolin and in linseed oil. Svedberg 3 has prepared colloidal solutions of metals in ether, pentane, isobutyl alcohol, etc. Schneider 4 has made alcoholic solutions of. colloidal silver and finds that they are coagulated at once by isopropyl alcohol, normal and secondary butyl alcohol, trimethyl carbinol, and heptyl alcohol, but not by propyl or isobutyl alcohol. Colloidal sulphur may be formed when sul- phur chloride reacts with ethylene. 5 Michael 6 obtained col- loidal sodium chloride without knowing it by the interaction of chloracetic ester and sodium malonic ester, while Paal 7 showed 1 Zeit. anorg. Chem., 03, 165 (1911); ANDERSON: Zeit. phys. Chem., 77, 191 (1914). 2 Zeit. Kolloidchemie, 15, 159 (1914). 3 Ber. deutsch. chem. Ges., 38, 3616 (1905); 39, 1705 (1906). 4 Ibid., 26, 1281 U892). 5 CONANT: Jour. Am. Chem. Soc., 42, 585 (1920). 6 Ber. deutsch. chem. Ges., 38, 3217 (1905). 7 Ibid., 39, 1436, 2859 (1906); 41, 51 U908); 42, 277 U909). 274 NON-AQUEOUS COLLOIDAL SOLUTIONS 275 that petroleum ether precipitated the sodium chloride, which would go again into apparent solution on treatment with benzene. Copper oleate 1 stabilizes a colloidal solution of cupric chloride in benzene. It seems probable that trinitroxylene (TNX) is peptized by trinitrotoluene (TNT) for Marshall 2 says that when 30-50 parts of TNX were suspended in 70-50 parts of TNT, the mass was suf- ficiently fluid to be cast without the use of any extruding mech- anism. The cooled castings appeared amorphous and resembled paraffin in texture. The castings showed no segregation of either constituent, were free from air-holes, were of uniform density, were non-hygroscopic, and showed no tendency to become oily or leaky on storage. They could be detonated completely with smaller primers of tetryl than could castings of TNT alone. Bechhold 3 has been able to separate some of the asphalt out of a crude oil by ultrafiltration through a toluene-glacial acetic acid collodion filter. Amberger has made colloidal solutions of several metals in lanolin. Richardson 4 finds that asphalt from certain sources contains solid particles dispersed in the asphalt. Lorenz 5 has shown that the fused baths, from which we obtain sodium, aluminum, and other metals, electrolytically, tend to peptize the molten metal. The amount of peptization or production of metallic fog increases very rapidly with rising temperature, which is one reason why it is very important to carry on the electrolysis at as low a temperature as possible. In the glass industry col- loidal solutions are very common, with stannic oxide, gold, copper, etc., as the dispersed phase. During the war a colloidal fuel was developed, consisting of coal in fuel oil. Since the mixture is denser than water, a fire can be put out by spraying with water, so that the fire risk is claimed to be less than with oil fuel. Shoe blacking is another case of a colloidal solution with a non-aqueous solvent. Acheson oildag is a third variant of the same general type. 1 VON WEIMARN: Kolloidchemische Beihefte, 1, 409 (1910). 2 Jour. Ind. Eng. Chem. 12, 248 (1920). 'Colloids in Biology and Medicine, 11 (1910). 4 Jour. Phys. Chem., 19, 241 (1915); ROSINGER: Zeit. Kolloidchemie, 15, 177 (1914). 6 Van Bemmelen Gedenkboek, 395 (1910). CHAPTER XI FOG AND SMOKE FOG A fog consists of drops of liquid suspended in a gas, usually air, while a smoke has solid particles. To obtain a fog a vapor must be saturated suddenly, so that condensation takes place more or less uniformly throughout the mass. On a laboratory scale this cooling can best be brought about by an adiabatic expansion or by letting a jet of steam emerge into a cooler atmosphere. The first method is the more satisfactory for quantitative measure- ments and the second for qualitative determinations. If the supersaturation is excessive, the liquid comes down as a dense fog without presenting any specially interesting characteristics. If the supersaturation is slight, it is found that no fog forms unless " nuclei" are present in the air. This matter was studied with great thoroughness by Coulier 1 who showed that there is almost always dust in the air and that condensation takes place on the dust particles. The dust can be removed by filtering the air through cotton. Ozonized air is active in causing condensation and this activity is not destroyed by filtering the air through cot- ton. Oxygen, hydrogen, and nitrogen become active when heated and this activity is not destroyed by filtration. There are therefore other types of nuclei besides solid particles. It was found by R. v. Helmholtz 2 that the vapors of sulphuric acid, hydrochloric acid, and sulphurous acid act as nuclei. In a later paper R. v. Helmholtz and Richarz 3 advance the view that condensation occurs frequently during chemical reactions even though the reacting substances and reaction products do not act as nuclei. They point out that neither ozone nor oxygen acts 1 Jour. Chem. Soc., 29, 186 (1876); AITKEN: Trans. Roy. Soc. Edin., 30, 337 (1883). 2 Wied. Ann., 27, 508 U886). s Ibid., 40, 161 (1890). 276 FOG AND SMOKE 277 as a nucleus but that nuclei are present while ozone is decompos- ing. A similar phenomenon was observed in the case of a num- ber of slow oxidations and reductions. Hydrochloric acid and ammonia are said not to act singly nor does ammonium chloride; but ions are formed during the reaction. This statement in regard to hydrochloric acid contradicts the previous statement as to this substance but there is a difference of concentration and the authors are also willing to admit that some ammonia in the air may have reacted with the acid in the earlier experiments. On this statement of facts the authors conclude that the reactions pass through an intermediate stage in which ions are formed, and that these ions act as nuclei. This extremely interesting hypothesis is put on the defensive by the experiments of Engler and Wild 1 who claim that solid substances are the real nuclei in all these cases, iodine pentoxide in the oxidation of iodine, an oxide of phosphorus in the oxidation of phosphorus, ammonium chloride, etc. Helmholtz and Richarz made a good deal of the fact that the nuclei could not be suspended solids because they were not removed when the gases were bubbled through water. Engler and Wild point out that the solids in the centre of the gas bubbles do not come in contact with the water at all and that it is a simple matter to carry suspended particles of phosphorus pentoxide through several wash-bottles in spite of its tremend- ously hygroscopic nature. The difficulty of removing suspended solids is shown by the special precautions necessary to stop such toxic smokes as diphenylchlorarsine. Pringal 2 considers that the nuclei attributed to ozone are really oxides of nitrogen and believes that ozone oxidizes nitrogen to some extent. Bieber 3 concedes that oxides of nitrogen will act as nuclei; but shows that they are not formed by the action of ozone on nitrogen, and that the oxides of nitrogen do not form the stable, blue cloud which is observed so often when working with ozone. He attributes this stable, blue cloud to hydrogen peroxide, showing that the cloud is formed when hydrogen per- oxide is present and that it does not occur under conditions which preclude the formation of hj r drogen peroxide. Thus we 1 Ber. deutsch. chem. Ges., 29, 1929 (1896). 2 Drude's Ann., 26, 727 (1908). 3 Ibid., 39, 1313 (1912). 278 APPLIED COLLOID CHEMISTRY do not get it with water vapor alone or with water vapor plus a gas other than oxygen. Now hydrogen peroxide can be formed by the action of ultra-violet light on water, especially on liquid water, according to the equation H 4 O 2 = H 2 O 2 + H 2 . It may also be formed by the action of ultraviolet light on a mixture of hydrogen and oxygen, 1 according to the equation H 2 + 2 = H 2 O 2 . This second reaction of course plays no part in the case under consideration. Bieber rules out the first one because no hydrogen is formed in presence of ozone, and consequently he believes that the reaction is H 2 O + O = H 2 2 , where the oxygen atom is the third one in ozone or is set free from oxygen by ultra- violet light. While this may be true, it is not necessarily so. It is quite as simple to consider that the ozone acts as a depolarizer to the nascent hydrogen since it is very improbable that the action of ultraviolet light could give rise to the reaction H 4 O 2 + Oa = H 2 O 2 + H 2 O + O 2 , because we know that the intermediate product, electrically neutral monatomic hydrogen, reacts with oxygen. Haber and Just 2 have shown that when the alkali metals react with different gaseous substances such as water vapor, hydro- chloric acid, iodine, phosgene, oxygen, thionyl chloride, etc., negatively charged particles are set free in the reaction zone and can be removed from it by an electric field. When the pressure of the reacting gas is very low, the negative particles pass out from the reaction zone of their own accord and prove to be chiefly free electrons. In presence of light there is an enormous increase in the emission of electrons. Similar results were obtained with silver, copper, and aluminum when heated or when the pressure or the gas was very low. While these experiments of Haber and Just prove conclusively that electrons are emitted during some chemical reactions, they do not prove at all that there is always an intermediate formation of ions in every chemical reaction. On the other hand, we do know that gaseous ions, if present, may act as nuclei. As Donnan 3 points out, the very interesting experiments of C. T. R. Wilson 4 1 FISCHER and WOLF: Ber. deutsch. chem. Ges., 44, 2956 (1911). 2 Zeit. Elektrochemie, 16, 275 (1910); JUST: Ibid., 20, 483 (1914). 3 Phil. Mag. (6), 8, 306 (1902). 4 Phil. Trans., 189A, 265 (1897); 193A, 289 (1899). FOG AND SMOKE 279 on the condensation of water vapor in the presence of dust-free air and other gases, have rendered it extremely probable that this condensation occurs around electrical nuclei of some sort. It was found that an adiabatic expansion produced no condensation in dust-free air saturated with aqueous vapor until the expansion reached the value of 1.25. This expansion (ratio) produced a small shower of fine rain, and expansions ranging from the above value up to 1.38 produced small fine showers of not much greater density. At an expansion of 1.38, however, the phenomenon changes in character, and a more or less dense fog is produced. Now air or any other gas exposed to Rontgen rays becomes a conductor; and the experiments of J. J. Thomson and his fellow- investigators have made it practically certain that this conduction is of a convective nature, due to electrically charged nuclei in the gas. The importance of this lies in the fact that Wilson's experiments showed that air which was saturated with water vapor and exposed to Rontgen radiations gives, when expanded adiabatically, no condensation before the expansion 1.25, and a dense fog instead of a fine rain at that point. The natural con- clusion is that the rain-like condensation for expansions ranging from 1.25 to 1.38 is due to condensation on a comparatively small number of electrical nuclei either existing in the gas or produced by the sudden expansion. Donnan was especially interested in the fact that here is a "case of condensation of a vapor of a liquid of unique ionizing power round electrical nuclei. It becomes a matter of interest to compare with water vapor the vapors of other liquids whose ionizing powers are either much smaller or practically zero." He came to the conclusion that organic liquids condensed on ions less readily than does water. 1 When a gas is previously saturated with water vapor, the low- ering of the pressure necessary to produce condensation by adia- batic expansion 2 is about 15 cm for air, oxygen, and nitrogen, but about 21 cm for hydrogen. Condensation of water takes place more readily on negative ions than on positive ions; 3 the reverse is true for alcohol. 4 In dust-free and ion-free gases which do ' Cf. BARUS: Phil. Mag. (6), 4, 262 (1902). 2 BIEBER: Drude's Ann., 39, 1313 (1912). * WILSON: Phil. Trans., 193A, 289 (1890). 4 PRZIBRAM: Phys. Zeit. A., 8, 561 (1907). 280 APPLIED COLLOID CHEMISTRY not react with water vapor, the condensation takes place much less readily, at an eight-fold supersaturation according to Wilson. With water vapor the expansions 1.25, 1.28, 1.31, and 1.38 are the values: (a) to start catching the negative ions; (6) to catch all the negative ions; (c) to start catching the positive ions; (d) to catch the nuclei of cloudy condensation. 1 Since gaseous ions act as nuclei 2 it is not surprising that conden- sation may take place when inert air is exposed to cathode rays, Rontgen rays, Becquerel rays or ultra-violet light; also when a spark discharge or a silent discharge passes through the air. The ions given off by heated metals may also act as nuclei. Broadly speaking, the nuclei consist of very hygroscopic sub- stances, solids, and ions. 3 The hygroscopic substances form solutions having a very low vapor pressure and the water vapor is excessively supersaturated when they are present, consequently condensation takes place. Freundlich 4 accounts for the behavior of suspended solids by postulating that they have concave sur- faces on which water would have a lower vapor pressure than on a convex surface and consequently would precipitate more readily. This is very ingenious but carries with it the corollary that solid spheres would not act as nuclei. While this may be true, it is not probable. It seems much simpler to consider that water vapor is adsorbed by the suspended solid, forming the equivalent of a liquid surface on which the further excess of water vapor condenses. It has been shown by J. J. Thomson 5 that the effect of electrification of a drop of liquid is to decrease the vapor pres- sure of the liquid. Consequently water vapor will tend to pre- cipitate on an ion as a nucleus, which is what happens experimentally. While fog usually consists of condensed water, there are condi- tions under which special types occur. Thus Frankland 6 describes what he calls dry fog. Dense fogs have been noticed 1 Quoted from WILSON by OWEN and ROBERTS: Phil. Mag. (6), 23, 352 (1912). 2 FREUNDLICH: Kapillarchemie, 294 (1909). 3 For a recent general discussion, see ROTHMUND: Monatsheft fiir Chemie, 39, 571 (1918). 4 Kapillarchemie, 295 (1909). * Application of Dynamics to Physics and Chemistry, 166 (1888). 6 Proc. Roy. Soc., 28, 238 (1879). FOG AND SMOKE 281 around London when the degree of saturation was only 50-80 percent. Frankland believed that this was due to films of oil from coal smoke coating the drops of water and retarding the evaporation. While Frankland's explanation of the dry fogs is probably the correct one, there are two possibilities which are not covered by the experiments. He has not shown that the oily or tarry matters condense on the drops of water. It is certain that, under some conditions, the water vapor might condense on the oily drops as nuclei, in which case the water would be outside and not inside. He has also not considered the possibility of the " dry ness" being due to a lowering of the vapor pressure of the drops by the dissolving in them of sulphur dioxide, sulphuric acid, etc. While it is not probable that sufficiently concentrated solu- tions would be formed in this way to produce so great an apparent dryness, yet it is very probable that this is at least a contributing factor. At times a special state of things occurs at the seashore. If the wind blows very hard from the sea, a heavy surf springs up and salt spray is carried inland. Ordinarily this does no harm to vegetation because a wind of this sort is usually accompanied by rain which dilutes the spray sufficiently to make it harmless. One summer at Newport, R. L, the wind blew hard from the south for sixty hours without any rain. The result was that undiluted salt spray was carried inland in enormous quantities. Everything was coated with salt; and trees, shrubs, and plants were burned brown for nearly half a mile from the shore. At Cirencester in England the wind causes an average annual deposit of thirty-six pounds of sodium chloride per acre. 1 Supercooled fogs seem also to occur at times. 2 On Ben Nevis the top of the hill in ordinary weather is enveloped in drifting fog. When the temperature of the air and ground is below freez- ing, this fog deposits feathery crystals of ice on every surface that obstructs its passage. A round post will have an almost uniform crop of crystals on its windward half, pointing so accurately to windward that it is possible to trace changes in the direction of the wind from the successive layers of crystals lying at different 1 VON ENGELN: Am. Jour. Sci. (4) 40, 464 (1915). 2 POLLOCK and COLLIER: Nature, 31, 216; OSMOND: Ibid. 31, 532 (1885); MCCONNEL: Phil. Mag. (5) 29, 461 (1890). 282 APPLIED COLLOID CHEMISTRY angles. The rate of growth varies with the density of fog and the speed of the wind. A rough average is half an inch per hour and two inches per hour seems to be a maximum value. The phe- nomenon is what one would expect from liquid particles cooled below the freezing point. The same thing occurs on Mount Washington where the crystals are known as frost feathers, 1 though they are not considered as resulting from supercooled fogs. On Mount Washington the frost feathers are formed only when the wind is northerly. With a southerly wind, masses of greenish-blue ice are. formed, especially on the telegraph poles. These supercooled fogs are quite different from the so-called ice storms which occur once or more every winter and which sometimes do so much damage to trees. In the ice storms the moisture deposits as liquid water on the branches and twigs, freezing there to solid ice. These storms occur when the tem- perature is below freezing and a warm rain or mist comes down from above without getting chilled to the freezing point before it strikes the trees. The optical properties of suspended drops of water are very interesting, the best known case of this being the rainbow. 2 When the sun is shining on fog or mist, a fog bow 3 is often seen differing considerably in appearance from the ordinary rainbow far broader and, though quite bright, nearly colorless. The size of the bow depends on the size of the drops, the diameter of the bow being less with very small drops 4 than with large drops. Horizontal rainbows are often seen when the surface of a lake is covered with minute drops of liquid. 5 Under suitable circumstances we may see coronas surrounding the sun or the moon. 6 They are due to the interposition of small spherules of water, which act the part of diffracting obstacles. In order to have the formation of a well-defined corona, it is essential that the particles be exclusively or preponderatingly of one size. By measurements of coronas it is possible to infer the 1 Mount Washington in Winter, 285 (1871). 2 MOORE: Descriptive Meteorology, 249 (1910). 3 McCoNNEL: Phil. Mag. (5) 29, 453 (1890). 4 WOOD: Physical Optics, 346 (1911). 6 JuDAY: U. S. Monthly Weather Review, 44, 65 (1916); NAKAMURA: Jour. Meteorological Soc. Japan, 36, 1; OTOBE: Ibid. 36, 3, 7 (1917). 6 Encyclopaedia Britannica, llth Ed. 8, 244 (1910). FOG AND SMOKE 283 size of the particles to which they are due, an application of con- siderable interest in the case of natural coronas, the general rule being that the corona is larger the smaller the water particles. Young employed this method not only to determine the diam- eters of cloud particles (e.g., 1/1000 inch); but also those of fibrous materials for which the theory is analogous. Every one is familiar with the colored coronas surrounding bright objects, as seen through a pair of spectacles freshly bedewed with moisture. These colors are due to optical interference and are entirely analogous to those seen through a fog. . Such circles may be seen around the sun almost any day of the year if we view it by reflection from some poorly reflecting surface, like water, in order to eliminate the blinding glare of the sunshine. In the case of the largest corona, known as Bishop's Ring, the average diameter of the particles 1 of this ring by diffraction was calculated to be about 0.00015 of an inch (about 4/*). The production of coronas in the laboratory was observed by Coulier in the paper referred to and has been the subject of study for many years by Barus. 2 To get the best effects the drops should be uniform in size. Piazzi Smyth 3 reports that for several days in India and else- where the disc of the sun was blue or green in the morning and evening. He considers that this was due to a high concentration of water vapor which cut out the red end of the spectrum; but this does not seem to be an absolutely satisfactory explanation. The brilliant sunset and sunrise colors are due in part to the presence of suspended particles of water. 4 In the lower half of the atmosphere there are frequently larger particles, probably of aqueous vapor, dust, and smoke that scatter the blue and trans- mit the red, so that after sunset or before sunrise beams of reddish light permeate the atmosphere above our heads. While fog is due to the production of very small drops, rain is due to the formation of larger drops which fall with perceptible 1 These were solid particles; but formed a corona and not a halo. A corona is due to diffraction and is red on the outside. A halo is due to re- fraction and is red on the inside. 2 Smithsonian Inst., 29, No. 13 (1009); Carnegie Inst. Pub. No. 62 (1907); No. 96 U908). 'Trans. Roy. Soc. Edin., 32, 389 (1884). 4 MOORE: Descriptive Meteorology, 245, 255 (1910). 284 APPLIED COLLOID CHEMISTRY speed. There is no absolute dividing line between the two because most people would classify as rain what a Scotchman might call a mist. According to the Weather Bureau, rain appears 1 to begin when the drops reach a diameter of 0.04 mm. The maximum rate of fall of these drops is very small. Larger drops fall faster. According to Lenard 2 drops of 1.28 mm diam- eter have a final velocity of 4.8 meters per second, those with a diameter of 3.49 mm reach a velocity of 7.37 meters per second, while drops from 4.50-6.36 mm show a practically constant rate of about 8 meters per second. This limiting value is due to the drops becoming deformed, so that they become flattened out, instead of retaining the shape of spheres. They therefore offer an increased resistance to the air through which they fall. In consequence of this deformation large drops break up rapidly in the air into smaller drops. Lenard found that drops of 4 mm diameter were stable under all conditions, but that drops of 5.5 mm diameter and over could not exist for more than a few sec- onds after attaining their final velocity relatively to the air. This is in harmony with the results of other observers. Wiesner 3 concluded that natural rain drops cannot have a diameter larger than 7.2 mm, while Hitter found that the largest drops of natural rain did not exceed 6.6 mm diameter. An independent confirma- tion is found in Bentley's experiments on the size and frequency of drops during rains in northern Vermont. His data are given in Table LI. TABLE LI. MEASUREMENTS OF DIAMETERS AND FREQUENCIES OF RAIN DROPS Apparent size Diameter of Frequenc y of drops drops in mm No. observed Percent Very small Under 8 149 17 Small 0.8-1.5 288 34 Medium . 1 6-3 5 254 29 Large 3 4-5 141 16 Very large Over 5.0 1 MOORE: Descriptive Meteorology, 205 (1910). 2 Met. Zeit., 21, 249 (1904). 3 MOORE: Descriptive Meteorology, 208 (1910). FOG AND SMOKE 285 It is possible that the very smallest drops often escaped observation and that the percentage of very small drops should be more than 17. In any case the mean diameter is not over 1.5 mm and the instructive thing is to see the very small percentage of drops larger than 5.0 mm diameter. Simpson 1 has recently developed a very interesting theory in regard to thunderstorms. He considers the case of a rapidly ascending current of air which spreads out and slows up above a certain height. Large rain drops will fall through this air current until they reach a point where they are broken into smaller drops and are then carried to the upper and colder levels where they grow again and repeat the cycle. When the drops break up they become charged positively, while the negative ions are carried up more rapidly by the air and are finally caught by cloud parti- cles at some higher level. On this hypothesis positively charged rain should fall in bursts in the center of the thunderstorm while negatively charged rain should fall in the lulls and on the out- skirts of the thunderstorm. This actually happens. Simpson satisfied himself that the electricity generated by the breaking up of the falling drops might easily account for the gradient of 30,000 volts/cm necessary for lightning. No calculations were made by Simpson as to the possibility of duplicating these phenomena in the laboratory by a vertical blower. If it were feasible to have a synthetic thunderstorm, this would be of great value in the study of meteorological phenomena. It is interesting to note that while drops become charged posi- tively on splashing, water vapor condenses more readily on negative ions 2 than on positive ones and that liquid drops take up negative ions more readily than positive ones. 3 The precipitation of a fog or mist is an important problem; but can best be taken up in connection with the precipitation of smoke or dust because the technical methods are as yet the same in the two cases. Reference may, however, be made here to a process devised by Pelouze and Audouin, 4 who force the mist 1 Phil. Trans., 209A, 379 (1909); HUMPHREYS: Jour. Franklin Inst., 178 751 (1914); Phys. Rev. (2), 6, 516 (1915). 2 WILSON: Proc. Roy. Soc., 64, 127 (1898). 3 SCHMAUSS: Drude's Ann., 9, 224 (1902); SEELIGEE: Ibid., 31, 510; 33, 431 (1910); Lenhardt: Ibid., 42, 45 (1913). 4 MOORE: Descriptive Meteorology, 205 (1910). 286 APPLIED COLLOID CHEMISTRY through a series of small apertures and cause the streams to impinge on a flat surface placed opposite. They consider that by this means the drops are brought into contact with one another and therefore coalesce to larger drops, which latter adhere to the surface against which they are driven. The method has been used for the separation of tar, carbon bisulphide, etc., from coal gas. It is interesting as being the mechanical analogue of the Cottrell electrical method. SMOKE As an intermediate step between fog and smoke we have the precipitation of water vapor in solid form as snow, sleet or hail. When the dew point is decidedly below freezing, 1 water vapor con- denses directly to crystals of ice, ice needles, or snow flakes. When supercooled rain drops strike solid substances they cover these with a coating of transparent ice. When snow crystals and supercooled rain drops come in contact we get sleet, and sleet changes to hail when the supercooled drops precipitate in the form of sheaths of ice. In order that hail stones should grow to any appreciable size, they must be carried up and down a number of times. Even in midsummer the temperatures six kilometers above the earth's surface will be at least 26 and may easily be 40. The reason that we get hail chiefly in summer is because then we have the hot air of the earth's surface giving a sharp temperature gradient and therefore a rapid upward sweep. The warm air may contain a great deal of water vapor which permits the precipitation of large amounts of ice. In winter the vertical temperature gradient is less and also the amount of water vapor in the air. Consequently we get no real hail, especially far from the sea. Hail stones 2 can only form if they are supported during forma- tion by strong ascending currents. Further, the structure of a hail stone indicates that it is often carried up and down past the zero isothermal. Now a current of air sufficiently strong just to support a hail stone as big as a pea would be more than suffi- cient to carry up the water it condenses within itself; hence hail stones would always have a greater downward velocity relative 1 MOORE: Descriptive Meteorology, 205 (1910). 2 SIMPSON: Phil. Trans., 209A, 412 (1909). FOG AND SMOKE 287 to the ascending current than the water in the current and there would be a large amount of splashing between the two. There would be, consequently, a much greater amount of separation of electricity than would have taken place without the hail stones, and this might very well account for the great violence of the electrical discharges in hail storms. Tolman 1 and others have devised a form of ultramicroscope which they call a Tyndallmeter for the quantitative study of smokes and have determined the relation between the intensity of the Tyndall beam and the size of the particle, making use also of silica suspensions as standards. It has been shown by Lord Rayleigh on theoretical grounds that, when the particles are very small compared with the wave-length of light, the intensity of the Tyndall beam should be proportional to the number of par- ticles per unit volume and to the sixth power of their diameter. For larger particles Tolman considers that the intensity would be proportional to the area of reflecting surface, in other words to the square of the diameter. If we define the concentration, C, as the mass per unit volume irrespective of the number of par- ticles, we get two equations: T = RCd 3 for very small particles. , T = RiC/d for large particles. For very small particles, the Tyndallmeter reading, T, will in- crease with increasing diameter, whereas it will decrease for large particles. The results show that, in general, increased subdi- vision means an increased Tyndallmeter reading. The observa- tions of Mecklenburg 2 on sulphur suspensions agree well with Rayleigh's formula for diameters from 5-840/i/u; whereas some of the smokes show the opposite result for particles smaller than the largest of these. Mecklenburg's work will have to be re- peated before the discrepancy can be explained. Unless subjected to special treatment, the air always contains dust. The air of large cities invariably shows hundreds of thou- sands of dust-motes per cubic centimeter, 3 that of the village or town thousands, and that of the open country at least hundreds. 1 Jour. Am. Chem. Soc., 41, 297, 300, 304, 575 (1919). 2 Zeit. Kolloidchemie, 16, 97 (1915). 3 MOORE: Descriptive Meteorology, 30 (1910). 288 APPLIED COLLOID CHEMISTRY The dust may be soil 1 from the fields and roads, particles of animal or plant origin, including pollen, products of combustion, salt from the spray of the ocean, volcanic dust, or cosmic dust. The dust from the eruption of Krakatoa was shot high into the air and was carried entirely round the world, falling on the decks of ships and on various parts of the earth. Owing to the enor- mous distances to which volcanic dust may be carried, it is nec- essary to have some definite test before we can say that a given sample of dust is really cosmic dust coming from meteors which have burned up in passing through the earth's atmosphere. This matter has been studied by Hartley and Ramage, 2 who collected numerous samples of dust in the neighborhood of Dublin and ex- amined them spectroscopically as well as other samples furnished by friends. 3 "The specimens of soot required no preliminary treatment before being burnt, and the analysis of each is given in the tabular statement only; but the different kinds of volcanic dust and flue dust were dissolved and the silica removed, after which the bases were separated into groups, and the spectra of the groups photographed." The data for the bases are given in Table LII, italics being used when only a trace of the metal is present. The relative strength of the lines, as seen by comparing the different spectra, is, in some instances, indicated in the table by suffixes, the number 1 indicating the weakest line and 10 the strongest. Hartley and Ramage drew the following conclusions: (1) The presence of nickel, as shown by the examination of soot, is not positive evidence that the dust from the clouds comes from other than a terrestrial source. (2) The dust which feU on the 16th and 17th of November, 1897, with its regularity in composition and its similarity to meteorites, being magnetic, also its comparative freedom from extraneous matter, exhibits properties which are quite in favor of its cosmic origin. Moreover, its composition is totally unlike that of volcanic dust and flue dust from various chemical and metal- lurgical works. This dust for the most part fell on a perfectly !See EHRENBERG: Die Bodenkolloide, 209 (1915). 2 Proc. Roy. Soc., 68, 97 (1901). 3 Data on the composition of soot have been published recently by COHEN and HUSTON: Jour. Soc. Chem. Ind., 32, 1360 (1911). FOG AND SMOKE 289 . .D ,C ,5 ,Q ,O ,Q X!^X1J2^!^^J3 PH P-iPHpiP* P-, PH PH Pi PM PH PH PH PH Pi PH PH Pi PH U 1. t-, OOO C S C g C C C S C fi fl C C C C 53 0,0)0 0)0)4)4)4) H HHHHH HHH cj cs cs .3 oooe> a cicgg'cs cs os 03 cs e> oc^o o oo o OQ MCOM CO 02 ci c3 c3 rt c3 c3 o3 03 c3 c3 c3 oj c3 ( c3 o3 o3 o3 03 e3 c3c3o3o3 oooo ouoo oooooo ouoooo oooo ooooo oooo MMMM CsCsCsfcO CtMM MM MM M CeCsCsCsMMMM OOOO 33333 3 3333 333333 33333333 OOOOO O OOOO OOOOOO OOOOOOOO J2J2&AJ3 os ^5-0^242 ft! 05 PH ft^ PHPnPnPH MWWWM M MMMMWM WMMMMWMW tfl rt cj c3 S3 O3s3o3o3o3 o3 o3 c c c3 o3 IB IS i!-! 290 APPLIED COLLOID CHEMISTRY calm fine night, and there was no rain for twenty-four hours or more afterwards. "We beg to draw attention once more to the very wide dis- tribution of gallium in minute proportions; it occurs in all alu- minous minerals, flue dust of every kind, soot and atmospheric dust, also in a great variety of iron ores. Bauxite contains it in larger proportion than any other mineral, but the quantity even in this substance is very small. We have hopes of finding it concentrated in some mineral, as thallium, caesium, germanium, and indium are. Indium and thallium, the other members of the same group of elements, are found in blende and pyrites, and accordingly we might expect gallium to occur in a concentrated state in a sulphide, arsenite, or similar compound. Judging, however, from its analogy with aluminum, there does not seem to be much probability of this." Unfortunately Hartley and Ramage do not tell us exactly how to identify cosmic dust and it is not an easy matter to formulate a rule from the data. The first four samples of dust differ from the volcanic dust in not containing magnesia or strontium; they differ from the flue dust and the soot from chimneys in not con- taining rubidium and in containing more nickel. If we are to take only the dust which fell on November 16 and 17, 1897, as of cosmic origin, the only difference is the trace of chromium. The situation is complicated farther by the fact that the meteoric stones all contain magnesia and chromium. It is not clear at all why these should have disappeared in the alleged cosmic dust. So far as one can formulate a conclusion from this work, it seems to be that dust is of cosmic origin when it contains sodium, po- tassium, copper, silver, calcium, gallium, thallium, iron, nickel, manganese, chromium, and lead; but no rubidium, strontium or magnesia. The dust must also be magnetic, this last probably being the most important criterion of all though it does not appear in the table. Preliminary work on the precipitation of smoke and dust was done by Nahrwold 1 and by Aitken, 2 the latter of whom showed that tobacco smoke could be precipitated quantitatively by pass- ing it through a hot-cold tube, the smoke condensing on the cold 1 Wied. Ann., 6, 460 (1878). 2 Nature, 29, 322 (1884); Trans. Roy. Soc. Edin., 32, 239 (1884). FOG AND SMOKE 291 tube. The film of condensed air round the particles has a lower surface tension on the hot side than on the cold side and conse- quently flows round the particles from the cold side to the hot side, causing the particles to move in the opposie dirtection or along the gradient of falling temperature. The temperature gradient has to be pretty steep to produce a marked effect, say 100/cm, so that the method has never been developed as a practical means of removing smoke. A slight movement has been observed by Wells and Gerke 1 due to the heat from the source of light used in the ultramicroscopic examination of smokes. Lodge 2 tried the effect of electrification on a mass of smoke. With potentials of one hundred volts very little effect could be detected. When the potential rose to a few thousand volts and a brush discharge began to be possible, the smoke agglomerated and settled very rapidly. The theory of this has never been worked out. In ordinary smokes about 30 percent of the particles are charged electrically. It may be that the alternating current reverses the sign of the charge periodically and that the particles agglomerate when they are electrically neutral or it may be that it causes the charged particles to collide with the un- charged ones. In connection with this it would be interesting to determine the effect of a high-voltage alternating current on the stability of a suspension in a practically non-conducting liquid. The general principle of electrical precipitation of suspended particles was patented by A. O. Walker of the Dee Bank Lead Works. In 1885 an apparatus was installed consisting of a system of metallic points situated in the flue from the lead furnaces and excited from two Wimshurst machines with glass plates five feet in diameter, each machine being driven by a one-horse power steam engine. The process seems not to have been a commercial success. It is only through the work of Cottrell 3 that the elec- trical precipitation of suspended particles has been put on a com- mercial basis. Cottrell believed that in the case of the large volumes of rapidly moving gases in smelter flues the agglomerat- ing and settling process was too slow even though the flues were expanded into as large dust chambers as were commercially 1 Jour. Am. Chem. Soc., 41, 325 (1919). a Phil. Mag. (3), 17, 214 (1884); Jour. Soc. Chem. Ind., 6, 572 (1886). 3 Jour. Ind. Eng. Chem., 3, 542 (1911). 292 APPLIED COLLOID CHEMISTRY feasible. It is in such cases that direct current methods have been particularly important. If a needle point is connected to one side of a high potential direct current line opposite to a flat plate connected to the other side of the line the air space between becomes highly charged with electricity of the same sign as the needle point, irrespective of whether this is positive or negative, and any insulated body brought into this space instantly receives a charge of the same sign. If this body is free to move, as in the case of a floating particle, it will be attracted to the plate of opposite charge and will move at a rate proportional to its charge and the potential between the point and the plate. Even if there are no suspended particles the gas molecules themselves undergo this same process, as is evidenced by a strong wind from the point to the plate even in perfectly transparent gases. The old familiar experiment of blowing out a candle flame by presenting it to a charged point, is simply another illustration of the same phenomena. The procedure actually used to get direct current consisted in transforming the alternating current from an ordinary lighting or power circuit up to some 20,000 or 30,000 volts and then commutating this high potential current into an intermittent direct current by means of a special rotary contact maker driven by a synchronous motor. This direct current is applied to a system of electrodes in the flue carrying the gases to be treated. While any smooth conducting surface worked well for the plate electrodes, it was difficult to find anything suitable for the point electrodes when working on a commercial scale. The problem was solved by using asbestos or mica, the fine filaments of the one and the scales of the other supplying the discharge points or the edges of the excessive fineness required. These materials are twisted up with wires or otherwise fastened to suitable metallic supports to form the discharge electrodes in such wise that the current has to pass only a short distance by surface leakage over them, the slight deposit of moisture or acid fume, naturally set- tling on them, serving to effect the conduction. If the condition of the gases does not supply such coating sufficiently, a special treatment of the material before being placed in the flue is resorted to. The process was installed successfully at the contact sulphuric FOG AND SMOKE 293 acid plant of the Hercules Works of the E. I. du Pont de Nemours Powder Company at Pinole on San Francisco Bay; at the refining plant and at the pyrites and matte roasters of the Selby Smelting and Lead Company; and at the Balaklala smelter. In the case of smelting plants the process is not a universal panacea because it does not remove gases and the farmers object to sulphur dioxide about as much as they do to sulphuric acid, arsenic, etc. In cases where suspended solids are the only source of trouble, the process is capable of eliminating the whole trouble. This was shown at Riverside in California, where there is a cement plant in the neighborhood of orange groves and the dust of course settles on the leaves. 1 A rotary kiln 100 feet long and 7 feet in diameter, oil-fired, will yield about 50,000 cubic feet of stack gases per minute. The gases above the combustion zone in the stack have a temperature of 450C. and carry dust aggregating between four and five tons per twenty-four hours. With these dry gases it was possible to use steel construction without cooling the gases. The installation was placed at the top of the stacks 80 feet up. It is probable that a considerable portion of the needs of the country for potash can be met by utilizing the dust from the cement plants. In the Washoe reduction works of the Anaconda Copper Mining Company the Cottrell process has been introduced on so large a scale 2 that the point electrodes consist of one hundred and eleven miles of chains. The Cottrell process has proved a great success commercially and new uses for it are continually being found. A modification of the apparatus has recently been made for the study of toxic smokes in the laboratory. 3 The apparatus consisted essentially of a central wire cathode surrounded by a cylindrical anode of aluminum foil 0.001 inch in thickness. An alternating current running up to 15,000 volts was rectified by means of a General Electrical Company "Kenotron." Complete precipitation could be obtained even when smoke was drawn through the ap- paratus at a rate of about five liters per minute. The anode 1 SCHMIDT: Jour. Ind. Eng. Chem., 4, 719 (1912). 2 CHANDLER: Jour. Ind. Eng. Chem. 11, 138 (1919). 3 TOLMAN, REYERSON, BROOKS and SMYTH: Jour. Am. Chem., Soc., 41, 587 (1919). 294 APPLIED COLLOID CHEMISTRY could be removed and weighed, thus giving a satisfactory method of determining the concentration of any smoke. Examination of the precipitate microscopically makes it possible to tell whether one is really dealing with a smoke or with a fog. When dust is blown about by the wind, it becoms electrified. 1 In South Africa at Bloemfontein at an elevation of 4500 feet, the normal fine weather charge of the air is positive and seldom ex- ceeds a maximum value of 200 volts per meter. During a dust storm the sign of the charge changes and the value may exceed 500 volts per meter. A so-called dust devil or whirlwind which carries a column of fine sand up two or three hundred feet in the air will affect an instrument two miles away, reversing the charge. This is because the sand particles are charged positively. In England a cloud of dust increases the positive charge in the air instead of decreasing it. This is because the dust in England is usually calcareous and acquires a negative charge. It is also known that red lead acquires a positive charge when blown into the air and sulphur a negative one. From the experiments that he made, Rudge deduced four generalizations. 2 Non-metallic elements give positively charged clouds when the finely-divided solid material is blown into a cloud by a current of air. Metallic elements give negatively charged clouds when the finely divided material is blown into a cloud by a current of air. Solid acid-forming oxides give posi- tively charged clouds, and basic oxides negatively charged clouds. In the case of salts the charge apparently depends on the relative strength of adsorption of positive and negative ions. The whole subject of frictional electricity should be gone over from the view- point of adsorption of ions. If this were done, it would undoubt- edly bring some order out of the present chaos. Guttmann 3 gives some data on electrical disturbances in powder mills. I am indebted to Dr. Reid Hunt for information in regard to another difference between siliceous and calcareous dusts. The flint paricltes in South Africa are not rounded and the sharp edges of the dust wound the lungs and permit the tuberculosis bacteria to enter, whereas there is not the same danger in the 1 RUDGE: Phil. Mag. (6), 23, 852 (1912); 26, 481 (1913). 2 Cf. KNOBLAUCH: Zeit. phys. Chem., 39, 225 (1902). 3 Jour. Soc, Chem. Ind., 8, 60 (1889). FOG AND SMOKE 295 case of calcareous dusts. If dried rhubarb is ground up, the cal- cium oxalate forms a dust composed of needles exceedingly irritat- ing to the lungs. The optical phenomena due to suspended solids are quite as interesitng as those due to suspended drops of liquid. The Tyn- dall effect, or the scattering of a convergent or parallel beam by suspended particles, has been referred to in connection with the ultramicroscope; and in the discussion of the color of colloids it was pointed out that the blue color of skimmed milk is due to the fact that the blue rays are scattered more than the red. The blue of cigar smoke is a similar phenomenon and Lord Ray- leigh 1 has shown that the blue of the sky can be accounted for very well as due to scattering by particles of water, ice, or dust. He even claims that the molecules of oxygen and nitrogen are large enough in themselves to give rise to a blue sky even if the air contained no suspended matter in the ordinary sense of the word. From his formula he deduces the value of the Avogadro constant, the number of molecules per gram molecule. 2 Without being able to point to any error in Lord Rayleigh's mahtematics, it seems to me that his assumption in regard to the behavior of oxygen and nitrogen molecules is probably wrong. The molec- ular weight of oxygen is so low that practically all solutions and gaseous mixtures would be blue by reflecetd light and red by trans- mitted light if Rayleigh were right. Nichols 3 is by no means certain that one can account quanit- tatively for all the phenomena of the blue sky on the basis of solid or liquid particles. The distribution of intensities seems not to be exactly that required by Rayleigh's formula. The best explanation of the discrepancy is either that the air itself absorbs blue to some extent or that some variable constituent of the air does, such as ozone, for instance. Many of the color effects in the sky apart from the blue are due to the presence of solids. Moore 4 says that if it were not for dust-motes there would be a different and less brilliant twilight. The bending or refraction of light, as the sun's rays pass obliquely 1 Phil. Mag. (4) 41, 107, 274, 447 (1877); (5) 12, 81 (1881); 47, 375 (1899). 2 Cf. PERRIN: Die Atome, 129 (1914). 3 Phys. Rev., 26, 497 (1908). 4 Descriptive Meteorology, 32, 254, 245 (1910). 296 APPLIED COLLOID CHEMISTRY from the etjjer, at sunrise, or at sunset, into the optically denser medium of the air, displaces the apparent position of the sun, elevating it by an amount about equal to its apparent diameter, so that one may still see it and receive its light directly when geometrically it is entirely below the horizon. A little later in the evening and its rays fall upon the upper air too obliquely to be bent down to the earth by refraction, but darkness does not yet ensue, for the rays are scattered by the molecules of gases and the dust-motes and sent downward from particle to particle, resulting in a soft shimmering light that fades away almost imperceptibly, and which may last for hours in higher latitudes, because of the obliqueness there of the sun's path to the horizon. When the harmattan blows on the west coast of Africa, the sky has a chalky-white tint and the air is very dry. The whitish tint is due to the presence of innumerable fragments of micro- scopic diatoms or siliceous shells. The whitish color of the haze must be attributed to the reflection of light from their surfaces. A similar white haze occurs in air that is full of grains of pollen or . fine crystals of snow, or almost any other kind of small particles. "An unusual illustration of selective absorption was given on the occasion of the most violent volcanic eruption of historic times that of Krakatoa on August 27, 1883, when an immense volume of dust and aqueous vapor was thrown to a great height in the atmosphere above the Straits of Sunda. The antitrades spread this over the northern hemisphere, and winds higher up in the tropics carried it -westward round the world and formed a layer of from 5 to 15 miles above the earth's surface, consisting undoubtedly of minute particles. After sunset, and by virtue of the diffusion of red light by this layer of particles, the whole western sky, even to near the zenith, glared with a lurid red as though lighted up from some great and distant fire. These remarkable sunsets continued, diminishing slowly in brilliance, through the years 1884 and 1885 in temperate latitudes, while their northern limit advanced slowly toward the pole, showing that minute particles of solid matter may float for years in the high upper atmosphere, so long as slowly rising currents buoy them up. Isolated clouds of such material may always be floating in the upper atmosphere and may give rise to the nocti- lucent or night-shining clouds that are occasionally observed at FOG AND SMOKE 297 great altitudes when the observer looks northward in mid- summer." The tails of comets seem to consist almost entirely of colloidal particles. 1 The great comet of 1882 was invisible against the solar disc, a position which corresponds to attempted observation of colloidal particles in the ordinary microscope against a lumi- nous background. It became visible again after passing beyond the sun's disc, a position correspondng to successful observation of the same colloidal particles in the ultramicroscope against a dark background, the eye of the observer being protected from the source of illumination. The streaming of the tail of a comet away from the sun may be due to the ionization of the colloidal particles, and their consequent electrical repulsion, it may be a thermal effect, or it may be due to the pressure of light as was pointed out by Maxwell. The intensity of the action of the sun's rays on a particle depends on its surface which varies as the square of the diameter, while the force of gravity tending to carry the particle to the sun is proportional to the mass of the particle and this varies as the cube of the diameter. In the case of a particle having a density equal to water the repulsion should balance the gravitational effect 2 when the diameter is 1.5/u and when it is 70/xju. For particles of greater density a balance would be reached at a still greater degree of dispersity. In the very clear air of Egypt the contrast between lights and shadows is very marked because the shadows are illuminated but slightly by diffracted light and are therefore darker than in countries where there is more dust or moisture. If one is to get any detail in the shadows, one must expose a negative about twice as long 3 as one would in northern latitudes under equally favor- able conditions. Luckiesh 4 points out that a slight haze increases the brightness of the sky very much. Ait ken 5 has shown that water vapor in itself does not produce any effect of haze, that haze is due entirely to dust, and that the 1 ALEXANDER: Colloid Chemistry, 37 (1919). 2 Encyclopedia Britannica, llth Ed., 6, 761 (1910). 3 BECK: Jahrbuch der Photographic, 18, 159 (1904); Phot. Correspondenz, 1989, 45. 4 Jour. Franklin Inst., 187, 295 (1919). 5 Encyclopedia Britannica, llth Ed., 8, 715 (1910). 298 APPLIED COLLOID CHEMISTRY density of the haze increases with the number of the dust parti- cles in the air and also with the relative humidity. In Table LIII are given Aitken's results as to the number of particles of atmospheric dust in a column of air having a cross-section of one square centimeter required to produce a complete haze, that is, to make a distant object invisible. The figures are of course not dependent on the length of the column. TABLE LIII. PRODUCTION OF COMPLETE HAZE Wet-bulb depressions Number of particles to produce a com- plete haze 2 to 4 4 to 7 7 to 10 12,500,000,000 17,100,000,000 22,600,000,000 The whole effect of light on fog and haze may change in a re- markable way if one varies the ratio between polarized and unpol- arized light as has been brought out clearly by Tyndall. 1 The light of the sky is polarized and may be quenched in great part by a Nicol prism while the light from a cloud is not polarized and therefore cannot be extinguished. A cloud may appear to the naked eye dark against a bright sky and yet appear as a white cloud on a dark ground if the light from the sky is quenched by means of a Nicol prism. In the same way the effect of haze may be overcome and a mountain may be seen clearly when using a Nicol prism even though almost invisible to the naked eye. The elimination of polarized light in this way was made use of during the war to detect camouflaged ships and periscopes. Still better results were obtained with a Savart plate, the lines disappearing where the ship was. With ice crystals in the air, we sometimes get halos and mock suns. 2 The reflection and refraction of the sunlight by small ice crystals in the air give rise to a very complicated series of phe- nomena, which unlike the rainbow, can be seen at all altitudes of the sun. The ice crystals are all hexagonal, either flat plates or 1 Fragments of Science: The Sky of the Alps. 2 WOOD: Physical Optics, 347 (1911); MOORE: Descriptive Meteorology, 250 (1910). FOG AND SMOKE 299 needles (spiculse). In still air these crystals will fall slowly with the principal axis vertical in the case of the flat plates or with the principal axis horizontal in the case of spiculse, because bodies moving through a resisting medium set themselves in the position of greatest resistance. It is possible to show that the observer, looking through a filmy cloud of such crystals, would see in one part of the sky a halo, in another part an arc of light, and in other directions bright spots like the sun, all of them arranged symmetrically with regard to the sun and the observer's zenith. This resulting complex arrangement of circles and spots is known as a complete halo. It is rare that all possible com- bination are seen at any one time, but on several occasions solar halos have been recorded that corresponded very nearly to this ideal. The removal of dust from the air is an important matter in some industries. In photographic work dust particles in the film mean clear spots on the negative and black ones on the positive. In the manufacture of gas mantles, contamination by dust may cause pinholes in the product. Dust and soot in the air may be a very serious matter in the ventilating of elec- trical generators because 65,000 cubic feet of air may pass through a moderately large generator per minute or about 94,000,000 cubic feet per day. There are a number of firms which make apparatus for washing and cooling air. All of them remove the dust by spraying drops of water through the air instead of bub- bling air through the water, this latter method not being effective. 1 Everybody knows how effectively a rain clears the air and it is this principle which is employed commercially. The question of flames is one of colloid chemistry. A flame is a reacting gas and the luminosity is due either to incandescent solid particles or to chemical reactions. The luminosity of ordi- nary flames is due chiefly to incandescent particles of carbon, the presence of which can be shown by pressing a cooled dish down upon the flame, when a precipitate of carbon is formed. In the case of flames colored by salts, the luminescence is due chiefly to chemical reactions. 2 In the green cupric chloride 1 See however HURTER: Jour. Soc. Chem. Ind., 4, 639 (1885); 6, 707 (1887); 12, 227 (1893). 2 WEISER: Jour. Phys. Chem., 18, 281, 762 (1914); 19, 310 (1915). 300 APPLIED COLLOID CHEMISTRY flame, for instance, the cupric chloride dissociates temporarily into copper and chlorine, the light being due to the recombina- tion. The presence of free copper in the flame can be shown by placing in it a porcelain tube through which cold water flows. Metallic copper is precipitated as a metallic mirror on the cold portion of the tube. Nearly all the metals can be precipitated similarly from flames containing their salts, though in some cases the temperature of the oxyhydrogen flame is necessary for a satisfactory dissociation. If the smoke particles are concentrated enough to form a cloud which moves along, they will tend to carry with them the gases inside the cloud because the mingling with the air takes place relatively slowly by diffusion. At the eruption of Mont Pelee when St. Pierre was destroyed, the descending cloud carried down with it large amounts of heated steam. The mean density of the water vapor and the solid particles was greater than that of air and the cloud sank as a whole. In the late war the French always mixed some form of smoke with their gas clouds. That had several advantages. It enabled the observers to see what the gunners were doing, it increased the apparent density of the cloud, and it caused the cloud to hold together better than it would have otherwise. It is only fair to say, however, that there are no satisfactory data to show how real these last two advan- tages are; but one would expect them to be of less importance in gas warfare where the amount of smoke is small than in the case of a volcanic eruption where the total amount of solid matter in the cloud is relatively high. On the other hand a smoke cloud may hold together pretty well for a mile or two, which it could hardly do if the air in the cloud did not move with it to a great extent. Garcia de la Cruz 1 found that a hydrogen-filled balloon could be adjusted so that it would sink in pure air and would rise in air containing any kind of smoke. PRECIPITATION OF SOLID There seems to have been no systematic study of the effect of conditions on the form in which a solid precipitates from a gas 1 SPRING: Mem. Soc. beige Geologic, 17, 30 (1903). FOO AND SMOKE 301 or vapor, although there are a good many scattered observations. With carbon the matter is important, in its bearing on the pro- duction of lamp-black, charcoal, coke and metallized filaments. Monier 1 states that carbon made from pure sugar is very hard and will scratch topaz though not corundum. Schiitzenberger 2 reports that when cyanogen is passed over carbon soaked with cryolite and heated to a cherry-red heat, it decomposes into nitrogen and filamentous carbon. Constant and Pe*labon 3 have also made observations on filamentous carbon. Love 4 reports on the carbon deposits obtained in gas burners. Luzi 5 obtained a silvery graphite by decomposition of hydrocarbons at the surface of white-hot porcelain. Bone and Coward 6 believe that retort carbon is obtained essentially from methane because the carbon produced by cracking methane differs materially from that formed from the other hydrocarbons. In the early days of the war it was believed that a good adsorbent charcoal was to be obtained only by the decomposition of certain hydrocarbons; but that view did not last long. The work of Chaney of the National Carbon Company has thrown some light on portions of this field. Carbon obtained at low temperatures is adsorbent. If formed above 600 it is what Chaney calls gas-treated and is not adsorbent. The higher the temperature, the more the deposit is like graphite. Bone and Coward are wrong in thinking that carbon from methane is necessarily characteristic. The differ- ence is due to the fact that a higher temperature is necessary. While it is of course possible that carbon from different hydro- carbons may be different apart from temperature conditions, there is nothing at present to show this. Becker 7 claims that when vapors condense to solids, they pass direct to that form without any intermediate formation of liquid. Of course this cannot always be true and can only apply to the particular cases and conditions under consideration by Becker. 1 Jour. Chem. Soc., 27, 674 (1874). 2 Ibid., 60,265 (1891). 3 Comptes rendus, 137, 706 (1903). 4 Jour. Soc. Chem. Ind., 12, 433 (1893). 5 Jour. Chem. Soc., 62, 565 (1892). 6 Ibid., 93, 1211 (1908). 7 Zeit. phys. Chem., 78, 39 (1912). 302 APPLIED COLLOID CHEMISTRY Dewar 1 states that if a glass rod is cooled to the temperature of liquid air and is then brought into the air of the room, moisture will condense on it as a sheet of ice. If the glass rod is electrified by contact with a piece of silk, the ice forms as a forest of crystals and not as a sheet. The reason for this seems to be that the ice particles are themselves electrified and consequently precipitate as far from each other as possible. Platinum mirrors were obtained by Wright 2 by electrical dis- integration of platinum wires and by Edison 3 by volatilization of electrically heated platinum in hydrogen. Cottrell 4 modified the method of electrical disintegration so as to obtain platinum black. Any study of the form in which solids precipitate should extend to precipitation from solution, etc., and should include a study of the theory of the formation of mirrors, 5 of the way in which silver precipitates, 6 of mossy copper, 7 of filamentous potassium chloride, 8 of silver chloride growths with silver, sodium chloride, and gelatine, 9 etc. One particular case of precipitation is so puzzling as to call for special mention. If hydrogen is passed into a saturated aqueous solution of silver oxide containing an excess of the solid salt, silver precipitates to some extent as crystals and in part as col- loidal silver. 10 The form in which the metal comes down depends on the nature of the containing vessel. In a platinum vessel no hydrosol is formed and all the silver precipitates in a crystalline form on the walls of the vessel. In vessels of quartz and of ordinary glass, the colloidal silver is yellowish-brown by trans- mitted light, while it comes down red to blue in a flask made of Jena glass. The ratio of silver hydrosol to ordinary silver is Chem. News, 97, 5 (1908). Am. Jour. Sci. (3) 14, 169 (1877). Chem. News, 40, 152 (1879). Zeit. phys. Chem., 42, 385 (1903). Cf. WADSWORTH: Zeit. Instrumentenkunde, 15, 22 (1895); NEOGI: Zeit. anorg. Chem., 59, 213 (1906); CHATTAWAY: Proc. Roy. Soc., 80A, 88 (1908). 6 Cf. KOHLSCHUTTER: Zeit. Elektrochemie, 14, 49 (1908); 18, 373, 419 (1912); Liebig's Ann., 387, 86; 309, 340 (1912); 398, 47 (1913); PHILLIPS: Jour. Chem. Soc., 72 II, 32 (1897). 'HUTCHINGS: Jour. Chem. Soc., 32, 113 (1877). 8 WARINGTON: Jour. Chem. Soc., 8, 30 (1856). O LUPPO-CRAMER: Zeit. Kolloidchemie, 9, 116 (1911). 10 KOHLSCHUTTER: Zeit. Elektrochemie, 14, 49 (1908). FOG AND SMOKE 303 greater in the Jena glass vessel than in the other two. At first sight one would expect this difference in behavior to be due to differences in material dissolved from the walls; but this is not the case. Kohlschtitter allowed water to stand in an ordinary glass flask for a while and then poured it into the Jena glass flask. The reduction product was red. When water which had stood in a Jena glass flask was poured into an ordinary flask or into a quartz one, the silver came down yellowish-brown. The phenomenon is therefore connected with the presence and nature of the solid. The most plausible explanation is, as suggested by Kohlschlitter, that the reaction concentrations are highest at the surface of the platinum and lowest at the surface of quartz and the ordinary glass, so that the silver comes down coarsest and most crystalline in platinum vessels and finest in quartz. This could be checked experimentally by determining the adsorbing power of platinum, quartz, and Jena glass for silver oxide and for colloidal silver. It is probable that the adsorption is greatest with platinum and least with quartz. This behavior of the silver may be connected with the fact that it is easier to get a yellow stain 1 of silver on a potash-lime glass than on the hard glasses. Bredig 2 made use of the arc under water between wire termi- nals to give him colloidal solutions of platinum and other metals. Schoop 3 has developed a process for plating metals on all sorts of materials by blowing compressed air at about five atmospheres pressure through a pointed nozzle into an arc playing between two wire terminals. Smoo h well-adhering deposits of sprayed metal are obtained when aluminum or copper is sprayed on glass in this way, and the metal adheres so tenaciously that it is im- possible to remove the film without destroying the glass surface. It is claimed that a microscopic examination showed that the surface of the glass melts and that the metal is practically fused into the glass. This will happen only when the object to be coated is placed very close to the arc. By working at a longer range it is possible to coat combustible materials like paper and celluloid. It is possible to obtain deposits of brass by making one pole of copper and the other of zinc. 1 ROSENIIAIN: Glass Manufacture, 185 (1908). 2 Zeit. Elektrochemie, 4, 514 (1898); Zeit. phys. Chem., 31, 258 (1899). 3 Chem. Abs., 13, 2640 (1919). CHAPTER XII GASES AND SOLIDS IN SOLIDS GASES IN SOLIDS Reusch 1 points out that the volcanic ash erupted from a vol- cano often consists of an obsidian glass filled with innumerable gas bubbles and that consequently large pieces will float on water. If the expansion of the gases is too great a glass foam may be obtained which may break down into fragments of glass. Hannay 2 exposed glass at 200 to oxygen and to carbon dioxide at 200 atmospheres pressure, and allowed the glass to cool under pressure. So much gas was taken up that when the glass was heated quickly it passed into a foam. It would probably have been possible to have regulated the pressure and the rate of heat- ing so as to have obtained small but visible bubbles in the glass. In fact this has been done by others when making artificial pumice. 3 As has been pointed out previously, it is very probable that the blue color of many feathers is due to the presence of minute bubbles in the horny part of the feather. 4 When these are seen against a dark back-ground we get a blue analogous to the blue of the sky, due to suspended particles in a homogeneous medium. In the case of the sky the particles are solids or liquids suspended in a gas. In the case of the feathers, we have air bubbles in a solid medium. The floating soaps contain injected air bubbles and are there- fore suspensions of gas in a solid. Bread may also be included under this head. On the other hand a porous plate is really an interlacing system because the gas phase is, or should be, con- tinuous. Porous silica gels occur as opals, as hydrophane, and tabaschir, this last being found in bamboos. In so far as they 1 Jour. Chem. Soc., 46, 415 (1884). 2 Proc. Roy. Soc., 32, 407 (1881). 3 Jour. Franklin Inst., 174, 344 (1912). 4 HAECKER: Zool. Jahrbucher Abt. syst. Geog. Biol. Tiere, 16, 267 (1920). 304 GASES AND SOLIDS IN SOLIDS 305 are porous they are also interlacing systems. Meerschaum is a silicate mineral coming under the same head. A special case of gases and solids is a laminated system. Strictly speaking, the gas phase is continuous; but one cannot treat it as a case of a solid dispersed in a gas. Since the air film is very thin, it seems better to consider the theory of lamination here rather than elsewhere. This has been discussed in detail by Tyndall 1 who showed that a laminated structure is produced if an apparently homogeneous mass is subjected to pressure under such conditions that it can yield at right angles to the pressure. 2 He illustrated this by experiments on a mass of pure white wax. When chilled and compressed, it could then be split into laminae of extraordinary thinness. The same phenomenon occurs on a large scale in nature with many slates. Tyndall points out that we do the same thing in the kitchen when making puff-paste. "The cleavage of our hills is accidental cleavage, but this is cleavage with intention. The volition of the pastry cook has entered into its formation. It has been his aim to preserve a series of surfaces of structural weakness, along which the dough divides into layers. Puff-paste in preparation must not be handled too much; it ought, moreover, to be rolled on a cold slab to prevent the butter from melting and diffusing itself, thus rendering the paste more homogeneous and less liable to split. Puff-paste is, then, simply an exaggerated case of slaty cleavage." SOLIDS IN SOLIDS Adsorption of a solid by a solid is not properly to be considered as a case of a solid dispersed in a solid because the adsorption is a surface phenomenon. Consequently such things as the pigment lakes are excluded, though it is possible that after ignition they change their type. On the other hand the ultra-microscope shows the presence of a second phase in the case of ruby glass colored by gold. The ruby glass has some very interesting prop- erties. 3 Fused glass containing a small amount of gold is color- less and remains colorless if cooled quickly. If cooled slowly or 1 Fragments of Science: On Crystalline and Slaty Cleavage. 2 See, however, BECKER: Bull. U. S. Geol. Survey No. 241 (1904); LEITH; Ibid. No. 239 (1905). 3 ZSIGMONDY: Colloids and the Ultra-microscope, 165 (1909). 20 306 APPLIED COLLOID CHEMISTRY if re-heated to a temperature well below the melting-point, it becomes red owing to the separation of particles of gold of the size to transmit red light. If held too long at a high temperature the red color changes to a violet and then to a blue. The changes are exactly those which we get in aqueous solutions and for the same reason, the increase in the size of the particles. It is not quite clear, however, why the chilled glass should be colorless. This cannot be due to the formation of a gold salt because the same phenomenon is observed in a reducing atmosphere. It must either be that the gold is in true solution in the glass or that it has been so peptized by the glass that the colloidal parti- cles are too small to have any effect upon the beam of light. Zsigmondy believes that both things occur though he assumes without any apparently adequate reason that most of the gold is in true solution. It would probably be easy to settle this defi- nitely if experiments were made with a glass richer in gold. With stannic oxide in the glass there is less tendency for he gold to turn blue on prolonged heating and it is therefore easier to match colors. 1 The tinctorial power of the gold is very high, 0.01 per- cent giving a pink color and 0.1 percent a red. The number of colloidal particles is enormous, billions per cubic millimeter according to Zsigmondy. The cheaper form of ruby glass owes its color to colloidal copper 2 and the red in some glazes is due to copper in some form. The Chinese ox-blood is believed by Lauth and Dutailly to be metallic copper adsorbed by tin oxide. On rapid cooling a color- less glaze is produced, which turns red on slow heating. Seger 3 does not agree with the view that the red color is due to a segrega- tion of metallic copper. His experiments showed that alternately reducing and oxidizing conditions of the fire gases are necessary to produce the red color. He considers that the color is due to cuprous oxide or cuprous silicate. This opinion is also shared by Stokes 4 who believes that the color transmitted by colloidal copper is blue, whereas this is the color of the coarser colloidal copper. Bellamy: Jour. Am. Ceramic Soc., 2, 313 (1919). 2 WILLIAMS: Trans. Am. Ceramic Soc., 16, 284 (1914). 3 Collected Writings, 2, 734 (1902). 4 Mathematical and Physical Papers, 4, 245 (1904). GASES AND SOLIDS IN SOLIDS 307 Rosenhain 1 seems to be quite uncertain whether the color of a copper ruby glass is due to metallic copper or to cuprous oxide. In Aventurine glass, the second phase comes out in spangles, presumably of metallic copper. Zulkowski 2 believes that copper ruby glass owes its color to cuprous oxide but the general opinion is against him, though the evidence is not very conclusive. 3 The truth of the matter probably is that we may have metallic cop- per in one glass and cuprous oxide in another. At present there is no method of differentiating these two cases. On the other hand, it ought to be a simple matter to tell whether one were dealing with cuprous oxide or cuprous silicate because a cuprous silicate glass should be optically empty whereas a cuprous oxide glass should scatter light. It is not easy to see why a cuprous silicate glass should become colorless when fused. Berry 4 describes a red slag from a copper refinery which contained some metallic copper, though not much. When treated with acids about 27 percent Cu 2 dissolved, leaving a pale yellow slag containing 8.6 percent Cu20, 11.7 percent FeO, and 49.3 per- cent SiO 2 , the percentages referring to the original composition. Berry considers that the pale yellow is the resultant of green due to the iron and red due to the silver. Silver yields a beautiful yellow color 5 when applied to glass as a surface stain. This is the color of colloidal silver solutions having the finest particles. The best yellow stain is obtained with potash-lime glasses as the silver tends to come down metallic on the hard glasses. In the case of enamels, silver is added to the mass of the glaze 6 and not as a surface stain. Addition of silver carbonate together with bismuth oxide gives rise to an intense blue. One function of the bismuth oxide is to hold the silver to the body; but it must also cause a partial agglomeration of the silver because the lustres are blue to green instead of yellow to brown. A yellow stain of silver on a cobalt blue enamel gives a green. This is a ceramic analogue of the tree frog. 1 Glass Manufacture, 182, 184 (1908). 2 Jour. Soc. Chem. Ind., 16, 441 (1897). 3 THORPE: Dictionary of Applied Chemistry, 2, 721 (1918). 4 Am. Chem. Jour., 8, 429 (1887). 6 ROSENHAIN: Glass Manufacture, 185 (1908). 6 FBANCHETI Ann. Chim. Phys. (8) 9, 37 (1906). 308 APPLIED COLLOID CHEMISTRY Platinum and iridium oxide can be used to give grays, especi- ally in enamels, but they are expensive. Glass can be colored yellow by suspended carbon, the color of thin films of so-called carbon being yellow. Carbon cannot be used in lead glasses because of its reducing action. Glasses containing no heavy metals are colored yellow to yellowish brown by sulphur; but this is apparently due to forma- tion of polysulphides because these glasses are optically empty. 1 In calcium glasses addition of sulphur gives a blue color, ap- parently due to the presence of colloidal sulphur. We do not know at all why colloidal sulphur should ever be blue but ap- parently it may be. Wolfgang Ostwald 2 states that blue and green colloidal solutions of sulphur can be obtained by introducing sulphur into molten sodium chloride, into a borax bead, into liquid ammonia or into hot organic liquids like glycerine. He considers that this blue colloidal sulphur is the cause of the blue color of ultramarine. 3 In strongly alkaline glasses selenium gives a chestnut brown color and the glasses are then optically empty. 4 In moderately alkaline glasses the color is red to pink and submicrons can be seen. Potash glasses give a purer red than soda glasses. If the red glasses are heated to 620, the color becomes a pale brownish yellow. Sodium borate glasses containing selenium are often yellow, but turn red when annealed. Colloidal selenium solu- tions can easily be prepared which are red by transmitted light, 5 almost exactly the color of arterial blood. In low concentrations tellurium colors glass coral to purple-red and in higher concentrations a steel blue. It is possible to make a blue colloidal solution of tellurium by reducing a dilute boiling mixture of sodium tellurate and sodium protalbinate with aqueous hydroxylamine hydrochloride. It would probably be easy to make a coral-red solution of colloidal tellurium though this appears never to have been done. 6 I FENABOLI: Zeit. Kolloidchemie, 16, 53 (1915). 2 Theoretical and Applied Colloid Chemistry, 187 (1917). 3 See also MORGAN: Jour. Oil and Colour Chemists' Ass., 2, 120 (1919). 4 FENAEOLi: Zeit. kolloidchemie, 16, 53 (1915). 6 SCHULZE: Jour, prakt. Chem. (2), 32, 390 (1885). 6 BEBZELius: Ann. Phys. (2) 32, 1 (1834). GASES AND SOLIDS IN SOLIDS 309 Enamels' are " vitreous compositions capable of adhering by fusion to the surface of metal or of pottery. The base of the enamel is generally a colorless glass in which are suspended parti- cles of an opaque metallic oxide or salt, generally stannic oxide. Up to about 900 tin dioxide exists finely suspended in the enamel, but ta a higher temperature combination or solution may occur and opacity is diminished. Titanic oxide has been used in place of stannic oxide, and antimony compounds have been employed. Opacity may also be obtained by the use of arsenious oxide, calcium phosphate, cryolite, or fluorspar. A preparation known as artificial cryolite sometimes replaces the mineral. When bone-ash is used, a little nitre may be added to remove the color." The opacity produced by these added white compounds is due to the difference in refractive index. Natural tin oxide, for instance, has a much higher refractive index for yellow light than glass. While titania has been used as a substitute for stannic oxide, it is not a very good one because it gives a yellow tint. Haber 2 states that six million kilos of stannic acid a year were used for enamelling iron and that this amount was halved when people stopped mixing and melting the oxide with the other con- stituents of the glaze and merely added it to the enamel just before baking. In the old process a great deal of the stannic oxide was actually dissolved in the glass and consequently did not increase the opacity. Impure zirconia is now being used as a substitute for tin oxide in producing opacity. If the zirconia goes into solution as oxide or as silicate, it does not cause opacity. 3 Of five crystalline titania glazes which were examined, 4 four contained rutile and one probably asgenite. Almost all the titania had crystallized. In a crystalline zinc glaze the crystals resembled willemite. An opaque yellow lead glass is obtained with antimony oxide, the second phase being apparently lead antimoniate, sometimes called Naples Yellow. The black lustre of the Greek and Roman pottery has been reproduced 5 in an 1 THORPE: Dictionary of Applied Chemistry, 2, 333 (1918). 2 Jour. Soc. Chem. Ind., 33, 49 (1914). 3 GRUNWALD: Jour. Soc. Chem. Ind., 30, 210 (1911). 4 ENDELL: Jour, Soc. Chem. Ind., 30, 210, 489 (1911). 6 FRANCHET: Comptes rendus, 162, 1097 (1911). 310 APPLIED COLLOID CHEMISTRY oxidizing fire at 850 by fluxing 55 parts of quartz sand with 45 of sodium carbonate and grinding the frit with an equal weight of magnetite. Artificial ferroso-ferric oxide does not yield the black color which is probably enhanced by traces of manganese in the natural ore. The opacity of the enamel is caused by the presence of more stain than will dissolve in the flux. Franchet states that the black enamel was known to the Egyptians and consequently was not discovered by the Greeks. Foster 1 con- siders that ferrous iron, presumably as ferrous silicate, is the cause of the black color; but Franchet 's experiments seem pretty conclusive. Many of the natural obsidians 2 which appear black consist of a colorless matrix swarming with specks of magnetite. If there is complete oxidation to ferric oxide the color is red. Carnelian is chalcedony containing globules of hematite. If the amount of dispersed material is sufficient to make the mass opaque, we have jasper, 3 which is red with hematite, yellow with limonite and green with saponite or chlorite. Jasper may contain up to twenty percent suspended material. Blue quartz is not uncommon in the Southern Appalachians and the color is apparently due to the presence of hair-like inclusions of rutile. 4 Since the index of refraction of rutile is very different from that of quartz, this is apparently similar to blue feathers, except that we have here solid particles instead of minute bubbles. Iddings 5 states that the blue color of the opalescent quartz from Llano County, Texas, "is undoubtedly due to reflection of blue light from the minute colorless prisms whose width is a fraction of the length of light waves." Egyptian blue 6 is crystallized CaO.CuO.4SiO 2 . It is formed between 800 and 900, above 900 a green glass being obtained. The reaction is reversible, for blue crystals are formed if the green glass is held at 850. Glass can be colored yellow with colloidal cadmium sulphide i Jour. Am. Chem. Soc., 32, 1259 (1910). 2 PiRSsoN: Am. Jour. Sci. (4) 30, 104 (1910). 3 GOOD CHILD: Precious Stones, 174 (1908). 4 WATSON and BEARD: Proc. Nat. Museum, 53, 561 (1917). 6 Jour. Geology, 12, 227 (1904). 6 LAURIE: Proc. Roy. Soc., 89A, 418 (1914). GASES AND SOLIDS IN SOLIDS 311 and Homer 1 has produced a yellow of this sort in a boric acid bead. It really would be a good thing if somebody were to take up the whole question of colors in the borax bead so as to deter- mine which were due to colloids and which to true solutions; and also to find out definitely what chemical compounds cause the color in each particular case. Rock salt has been found having a blue color which is due apparently to colloidal particles of metallic sodium. The same result can be obtained by heating salt with sodium vapor 2 or by exposing salt to cathode rays. Radium emanation colors sodium glasses blue to violet and lead glasses brown. The action of heat, ultra-violet light, and radium on the colors of gems is very interesting 3 and the theory of it has not been worked out satisfactorily. Pale amethysts become darker when exposed to radium, while ultra-violet light has no effect. When heated to redness in hydrogen or oxygen, the pale amethysts become colorless while they turn yellow if heated in ammonia. The decolorized amethysts regain their color when treated with radium. Rose quartz is made colorless by ultra-violet light and blackish-brown by radium. It is not changed when heated in ammonia. Smoky quartz loses its color when heated and radium brings it back, while hydrogen peroxide tends to make the color yellower. Colorless topaz is made yellow to orange by radium and is decolorized when heated. Ultra-violet light tends to change the orange produced by radium to lilac. Kun- zite changes from lilac to green under the influence of radium and is changed back by ultra-violet light. It becomes colorless when heated to 400; but exposure to radium brings back the blue-green color. Corundum occurs as blue, green, violet, yellow and white sapphires and as ruby. The Oriental sapphire is said by Verneuil 4 to be colored by iron and titanium while the clear sapphire is colored by iron only. Blue sapphires are changed to yellow or yellowish-brown by radium, the blue-green sapphires to green and the white sapphires to yellow. Violet sapphires become pure red and natural rubies lose any violet 1 Chem. News, 29, 66 (1874). 2 SiEDENTOPF: Verb, deutsch. phys. Ges., 7, 268 (1905). 3 DOELTER: Das Radium und die Farben (1910). 4 Comptes rendus, 161, 1053 (,1910). 312 APPLIED COLLOID CHEMISTRY tinge . Artificial rubies and sapphires are no t changed by exposure to radium, but their coloring matter is chromium or cobalt. Ultra-violet light makes yellow sapphires blue and violet ones more violet. Heating sapphires in air makes htem colorless. The general result seems to be in all cases that heating makest he gems more nearly colorless 1 and that the action of radium and of ultra-violet light is antagonistic. The only possible explanation seems to be that radium increases the dispersity of the colloidal particles while ultra-violet light decreases it or vice-versa. We know that /3 rays increase the agglomeration of sulphur and that they change a selenium hyrosol into crystalline selenium. It should be possible to test this explanation on synthetic ma- terials, using perhaps borate glasses. For instance, radium produces no change in pure chromic oxide but turns it brown when the chromic oxide is dissolved in borax. Ultra-violet light changes the brown to yellow and when a chromium oxide borax glass is heated in ammonia it becomes pale. Alumina is not changed by radium but hydrous aluminum oxide sol is turned blue by it. If cases of this sort should be studied care- fully it would probably give us the necessary data to straighten out the question of the colors of gems without any difficulty. It is to be noticed that copper oxide adsorbed by alumina is apparently blue, 2 and it is quite possible that magnetite may be. The presence of surfaces of discontinuity may give rise to color effects without any pigment being present. In mother-of- pearl we have alternate layers of calcium carbonate and of organic matter called conchiolin. 3 On the outside each layer projects slightly beyond the edge of the preceding so that the outside of the nacreous portion shows innumerable steps or ridges running parallel to the growing edge of the shell, thus producing the effect of a diffraction grating. The color of mother-of-pearl is due to the diffraction of light from this grating-like structure and to the interference of light by reflection from numerous parallel laminae of sensibly equal thickness. Some crystals of potassium chlorate show brilliant colors and 1 Cf. HAMLIN: The Tourmaline, 87 (1873). 2 SCHENCK: Jour. Phys. Chem., 23, 283 (1919). 3 PFUND: Jour. Franklin Inst., 183, 453 (1917); SCHADE: Kolloidchemische Beihefte, 1, 385 (1910). GASES AND SOLIDS IN SOLIDS 313 this was found by Stokes to be due to the existence of planes within the crystals at which a periodic twinning had occurred and where there was probably a surface of discontinuity. 1 The same phe- nomenon is shown by the opal though it is not crystalline. Lord Rayleigh 2 assumes that the reflection takes place at a number of thin laminse sensibly equidistant, the distance between any two being of the order of magnitude of the light-wave. Crookes 3 says : " When a good fiery opal is examined in daylight, sunlight or artificial light, it appears to emit vivid flashes of crimson, green, or blue light, according to the angle at which the incident light falls, and the relative position of the opal and the observer; for the direction of the path of the emitted beam bears no uniform proportion to the angle of the incident light. Examined more closely, the flashes of light are seen to proceed from planes or surfaces of irregular dimensions inside the stone, at different depths from the surface and at all angles to each other. Occasion- ally a plane emitting light of one color overlaps a plane emitting light of another color, the two colors becoming visible alternately upon slight variations of the angle of the stone ; and sometimes a plane will be observed which emits crimson light at one end, changing to orange, yellow, green, etc. until the other end of the plane shines with a blue light, the whole forming a wonderfully beautiful solar spectrum in miniature. I need scarcely say that the colors are not due to the presence of any pigment but are interference colors caused by minute strisB or fissures lying in different planes. " 1 WOOD: Physical Optics, 161 (1911). 2 Phil. Mag., (5) 26, 256 (1888). 3 Proc. Roy. Soc., 17, 448 (1869). CHAPTER XIII THICKNESS OF SURFACE FILMS The preceding chapters have brought out the importance of films in colloid chemistry and it will perhaps be worth while to get some definite idea of the thickness of the films 1 with which we have been dealing. Oberbeck 2 made experiments to determine what thickness of metal deposited electrolytically on platinum gave the same electromotive force as the massive metal. He found values of 2-3/Aju for zinc, 1-2/i/j for cadmium, and 1/^u for copper. Since the metal was probably porous and was not burn- ished, these figures are too high rather than too low. It is interest- ing to note that the lowest value is obtained with copper, the metal which gives the smallest crystals of the three. Pring 3 determined the minimum thickness of metal which would give the maximum over-voltage. His values vary enormously, being 40MM for platinum, 5.5/z for zinc, 34/-1/* for copper, 0.16 mm for nickel, and 6.6 mm for lead. Vincent 4 determined the resistance of thin layers of silver precipitated chemically on glass and found that the specific conductance decreased rapidly after the film became less than 5(W in thickness. These figures are not very significant because Grimm 5 has shown that the resistance de- creases with time and also with pressure. The metal firts pre- cipitated is colloidal silver to some extent and this naturally throws the results out. Patterson 6 did similar experiments with platinum films pro- duced in a vacuum by cathodic disintegration and obtained con- stant values for a thickness of 7/i/i. These values could probably have been decreased considerably by burnishing the platinum. 1 FRBUNDLICH: Kapillarchemie, 265 (1909). 2 Wied. Ann., 31, 337 (1887). 3 Zeit. Elektrochemie, 19, 255 (1913). 4 Ann. Chim. Phys. (7) 19, 421 (1900). 5 Drude's Ann., 6, 448 (1901). 6 Phil. Mag. (6) 4, 652 (1902). 314 THICKNESS OF SURFACE FILMS 315 Wiener 1 prepared thin silver mirrors and determined the change in phase when light was reflected from the two surfaces. The change of phase was constant after a thickness of 12ju^ was reached, so this is the point at which a silver film behaves like massive metal under these conditions of test. The smallest thickness of metal which could be detected optically by this method was 0.2^. Konigsberger and Bender 2 found values of about 5(W for gold and platinum. The least thickness of lead peroxide which can be detected on platinum by optical methods is 0.84ju/u. Bredig and Weinmayr 3 found that a film of mercury thinner than 0.3-1.5/A/i would not cause the catalytic decomposition of hydrogen peroxide. Devaux 4 determined the smallest thickness of albumin film which could be considered as solid and obtained values between 0.4 and 8/i/i. Metcalf 5 found about Sjuju as the corresponding value for peptone films. Unfortunately he gives no data as to the maximum thickness of peptone films which can be obtained with a peptone solution and yet this is rather an important matter. If the film goes on growing in thickness until practically all the peptone is in the surface, the liquid- vapor inter- face cannot be important except perhaps just at the start. If the film stops growing in thickness while there is still peptone dispersed through the solution, what is the deciding factor? Shorter 6 has made some experiments on the thickness of saponin films but he does not give any actual thicknesses, merely saying that they are 10-15 molecules thick. Nagel 7 shows that solid surface films are formed on solutions of NiClj, Ni(NO 3 ) 2 , CoCli, CoSO 4 , FeCl 3 , FeSO 4 , K 4 Fe(CN) 6 , MnSO 4 , MnCl 2 , Pb(NO 3 ) 2 , AgNO 3 , and KMn0 4 when the water contains traces of ammonium carbonate, as is usually the case. The thinnest film which he recognized with a cobalt salt had a thickness of about 12^. In many cases the film was 55-270/iju in thickness. With colloidal silver a film at least 300/i/z in thickness was formed. When metals are polished, the surface becomes 1 Wied. Ann., 31, 629 (1887). 2 Drude's Ann., 26, 763 (1908). 3 Boltzmann Festschrift, 839 U904). 4 Jour. Phys. (4) 3, 450 (1904). 6 Zeit. phys. Chem., 62, 1 (1905). 6 Phil. Mag. (6), 17, 560 (1909). 7 Drude's Ann., 29, 1029 (1909). 316 APPLIED COLLOID CHEMISTRY more finely crystalline and thus constitutes a film which differs from the mass of the metal. 1 In the more easily flowed metals the surface films formed by polishing them may vary from 1-5/x in thickness, while in the less easily flowed metals they may be 0.5/* or less. At surfaces which have moved over each other very slightly, films of only a comparatively few molecules in thickness might be formed. Bodenstein and Fink 2 estimated the thickness of the adsorbed film of sulphur trioxide on platinum at 248 as 1.2/x/x on the ex- plicit assumption that the density of the film was 2. Gurwitsch 3 considers that the thickness of the adsorbed liquid on silver foil is 35, 33 and 28/i/^ for carbon tetrachloride, carbon bisulphide and acetone respectively. Ihmori 4 determined the amount of moisture on different sur- faces gravimetrically and estimated the thickness of the water film to be 3-KV/A on metals and 22juju on quartz. Langmuir 5 questions the accuracy of these results. Ihmori found no measur- able adsorption with polished or ignited platinum. With the other metals the amount of apparently adsorbed moisture be- came quite appreciable if the metal were allowed to stand in the air for a few weeks and Ihmori states expressly that the small amount of water adsorbed by the polished metals is perhaps due to the presence of a thin film of oxide. If the film of oxide is porous, as it probably would be, its surface is vastly different from that postulated by Ihmori in his calculations. Langmuir points out that with glass there is evidence that water may pene- trate quite a way. If a 40 watt incandescent lamp bulb, which has been exhausted and dried at room temperature, is heated to 500 for several hours, it loses enough gas to form a layer of water 55 molecules deep (0.00225cc/cm 2 ) of carbon dioxide 4.8 molecules deep (0.00015 cc/cm 2 ) ; and of nitrogen 0.9 molecules deep (0.000015cc/cm 2 ) all over the surface. 6 Since there can- not possibly have been this amount of gas on the surface of the 1 BEILBY: Phil. Mag. (6) 8, 258 (1904). 2 Zeit. phys. Chem., 60, 1 (1907). a Jour. Russ. Phys. Chem. Soc., 47, 805 (1915). 4 Wied. Ann., 31, 1006 (1887). 5 Jour. Am. Chem. Soc., 38, 2284 (1916). 6 LANGMUIR: Trans. Am. Inst. Electr. Eng., 32, 1921 (1913). THICKNESS OF SURFACE FILMS 317 bulb, most of it must have come from the mass of the glass. Katz 1 found that, with 70 percent humidity, powdered quartz takes up 1.3 X 10~ 3 mg of water and anorthite 6.2 X 10~ 3 mg per square centimeter. This corresponds to layers of water 43 and 205 molecules deep respectively. Langmuir believes that there either was penetration of water into the solids or that Katz was measuring moisture condensed as liquid in the capillary spaces between the fine grains. Sherwood 2 finds that glass which has been annealed above 200 adsorbs gases to give a layer about one molecule thick. From their experiments on the heats of adsorption of vapors by charcoal, Lamb and Coolidge 3 conclude that the adsorbed liquids are present in films many molecules deep. With carbon bisulphide the thickness of the layer was forty molecules, as- suming 0.1 HP as the thickness of a molecular layer. In the ex- periments with ethyl alcohol, the same calculation gives a layer about one molecule thick. Pettijohn 4 found about 128/zAi for the maximum thickness of a water film on some glass pearls made in Germany. With another lot of glass of different origin the corresponding value was about 68^At. With river sand the estimated thickness of the film varied from 285/x/i with 10-mesh sand to 114^/x with 60-mesh sand. Experiments with nitrobenzene, aniline, dime- thylaniline, phenyl iodide, toluene, turpentine, and pyridine indicated that the thickess of the film is independent of the liquid used, a result which can hardly be strictly true. Parks 5 finds that when glass wool at 12 has a surface film of water 134/i^c in thickness, further addition of water causes no evolution of heat. From data by Barus 6 on the formation of fog he calculates that, under the conditions of the experiments, the thickness of the water film is 0.1-0.8ju when the nucleus has a diameter of 2.6ju and is 0.05-0.5/u when the nucleus has a diame- ter of 3.6/,i. According to J. J. Thomson 7 the mean radius of 1 Proc. Amsterdam Acad., 16, 445 (1912). 2 Jour. Am. Chem. Soc., 40, 1645; Phys. Rev. (2) 12, 448 (1918). 3 Jour. Am. Chem. Soc. 42, 1168 (1920). 4 Ibid., 41, 477 (1919). * Phil. Mag. (6) 6, 517 (1903). 6 Ibid. (6) 4, 24, 262 (1902). 7 The Discharge of Electricity through Gases (1900). 318 APPLIED COLLOID CHEMISTRY drops formed on negatively electrified oxygen is about 0.8/* and on positively electrified oxygen is about 0.7/*. Sohncke 1 believed that films of oil on water were instable when less than lOO/^ in thickness. This does not mean that thinner films cannot be obtained; but merely that this thickness is necessary to give a stable film having the properties of oil in mass. Fischer 2 worked with films of oil on mercury less than 5/iyu in thickness; but these do break up into drops in time. Lord Rayleigh 3 found that a layer of oil 1/i/x in thickness causes a perceptible change in the surface tension of water and that a layer of 1.6/j/A in thickness contaminates the water sufficiently to affect the motion of a piece of camphor over it. Even with a layer of castor oil 8/x/x in thickness, the surface tension was still dropping. Langmuir 4 considers that when a drop of oil is allowed to spread over a large surface of water, it will do so until the layer is one molecule thick. This is very possibly true; but Langmuir gives himself a certain amount of leeway because he admits the possibility of the molecules of oil being packed on the surface in different ways. He is not interested at all in the maximum thickness of the film. Hatschek 5 estimates the thickness of the adsorbed water film on particles of colloidal sulphur at 0.87/x/i. In the case of cata- phoresis, Cotton and Mouton 6 believe that the effect of the walls is manifest to a distance of 25juM- The black film in a soap-bubble is apparently about 17ju/i in thickness 7 when there is no sodium nitrate in solution and about 12/xju in thickness 8 when there is. If this difference is real it is because the film is less hydrous in presence of sodium nitrate. At higher temperatures the thickness 9 may drop to SMA*. I have not been able to find any statement as to the maximum thickness of a soap-bubble film; but it is certainly not less than 1.4ju. 1 Wied. Ann., 40, 345 (1890). 2 Ibid., 68, 414 (1899). 3 Phil. Mag. (5) 48, 331 (1899). 4 Jour. Am. Chem. Soc., 39, 1848 (1917). * Zeit. Kolloidchemie, 11, 280 (1912). 6 NoBDENSON: Kolloidchemische Beihefte, 7, 108 (1915). 7 DRUDE: Wied. Ann., 43, 158 (1881). 8 REINOLD and RUCKER: Phil. Trans. 174, 645 (1883). 9 JOHONNOT: Phil. Mag. (5) 47, 501 (1899); (6) 11, 746 (1906). THICKNESS OF SURFACE FILMS 319 When studying reaction velocity in heterogeneous systems, it has been found desirable in many cases to assume the existence of a film of saturated solution in contact with the solid, the rate of reaction depending on the rate of diffusion. The thickness of this film can be varied a great deal by varying the rate of stirring. Under the conditions in which Brunner 1 was working the thick- ness of these films was 20-50ju. From this outline, it is evident that the thinnest films which can be detected at present are about 0.2-1. Ojuju in thickness, which corresponds fairly well to a layer one molecule thick, although there is no apparent reason why one should not detect a film before it covers the surface completely and it is probable that one does in a good many cases. The thickness at which a film has the properties of matter in mass may vary from 1-3/i/* in the case of metals, up to much higher values in the case of oils. The maxi- mum thickness to which films can grow and still be classed as films has not been determined in any case, except perhaps for the metals; but it may very likely run up to 100/* in special cases. There are certain matters in regard to the regions of discontinu- ity in soap-bubble films which call for a good deal more study. Nobody seems to have considered in any suitable way the ques- tion of the varying hydrolysis with varying thickness of the film. Perrin 2 gives the following estimated values in /*ju for the diame- ters of some molecules: helium, 0.17; argon, 0.27; mercury, 0.28; hydrogen, 0.20; oxygen, 0.26; nitrogen, 0.27; chlorine, 0.4; ether 0.6. 1 Zeit. phys. Chem., 47, 56 (1904). 2 Kolloidchemische Beihefte, 1, 265 U910). AUTHOR INDEX Abegg, 177 Acheson, 173 Adams, 70, 155 Adkins, 46 Aganin, 134 Aitken, 81, 202, 276, 290, 297, 298 Alexander, 151, 231, 297 Alleman, 13 Allen, E. T., 152 Allen, H. S., 23, 24 Amann, 176 Amberger, 173, 275 Andersen, 205 Anderson, 249, 250, 274 Antony, 235 von Antropoff, 44 Appleyard, 105, 106 Archibald, 23 Arisz, 76, 194, 195, 251, 252 Arndt, 13 Arnold, 243 Askenasy, 77 Audouin, 285 Ayres, 186 Bachmann, 240, 241, 245, 249, 256, 264, 274 Bacon, A. A., 147 Bacon, N. T., 13 Bahntje, 210 Baker, C. J., 12 Baker, W., 205 Baly, 43 Barker, 130 Bartell, 109, 110 Barus, 88, 167, 279, 283, 317 Basseches, 234 Bayer, 163 Bayley, 92 Bayliss, 122, 187 Beadle, 28 21 Beard, 310 Bechhold, 184, 186, 227, 234, 240, 254, 258, 259, 271, 275 Beck, 297 Becker, F., 301 Becker, G. F., 305 Bedford, 126 Beijerinck, 233 Beilby, 83, 136, 204, 316 Bellacci, 126 Bellamy, 306 van Bemmelen, 34, 103, 106, 246, 274 Bender, 315 Bennett, C. W., 79, 122, 149 Bennett, H. S., 210 Bennewitz, 23 Bentley, W. A., 284 Bentley, W. B., 169, 245 Bergter, 16 Berkeley, 177 fieri, 48, 51 Berliner, 28, 47, 50 von Berneck, 123 Berry, 307 Berthelot, 74 Berzelius, 234, 301, 308 von Bezold, 196 Bieber, 277, 278, 279 Bigelow, 67, 109 Billitzer, 174, 206, 209, 228 Biltz, 104, 106, 108, 187, 226, 233, 235, 261 Bingham, 155, 156, 157, 190, 192 Bishop, 283 Bleininger, 210 Blucher, 105, 122, 246 Blum, 169 Blumtritt, 11 Bodaszewski, 142 Bodenstein, 48, 316 321 322 AUTHOR INDEX Bodlander, 48 Bogojawlenski, 122 Bondy, 188 Bone, 46, 62, 63, 301 Bonsdorff, 246 Bose, 35 Bottger, 90, 176 Bowman, 164 Boyle, 16, 29 Boys, 66 Bradford, 259 Bragg, 37 Brand, 68 Bredig, 44, 123, 125, 175, 179, 180, 186, 205, 303, 315 Breuer, 149 Brewer, 137 Briggs, L. J., 19, 25, 38, 97 Briggs, T. R., 79, 87, 207, 208, 210, 224, 261, 263, 264, 265, 266 Brillouin, 139 de Broglie, 144 Brooks, 293 Brown, F. E., 67, 131 Brown, R., 137 Browne, 34 Brunner, 319 de Bruyn, Lobry, 168, 183, 186 Buchbock, 43 Buchner, 76 Budgett, 75, 82 Bunce, 244 Bunsen, 18, 48 Burnham, 79, 122 Burton, 140,141, 142, 182, 190, 212, 223, 288 Butschli, 274 Buxton, 227 Byk, 165 Cabot, 21 van Calcar, 186 Callow, 272 Calvert, 11, 39 . Cameron, 105, 121 Campbell, A. A., 70 Campbell, F. H., 65 Cartledge, 251 Cattermole, 272 Chamberlain, 201 Chamot, 167 Chance, 53, 54 Chandler, 293 Chaney, 8, 34, 301 Chappuis, 26, 34 Chatard, 177 Chattaway, 302 Chaudesaigues, 142 Christiansen, 197 Church, 170 Chwolson, 70, 77, 187 Clark, 67 Claude, 14 Claus, 54 Clermont, 151 Clowes, 164, 225, 267 Coehn, 207 Coggesha ' 21 Cohen, 288 Cohnstaedt, 19 Collier, 281 Conant, 274 Constant, 301 Coolidge, 27, 317 Cooper, 255 Corelli, 126 Cotton, 318 Cottrell, 286, 291, 293, 302 des Coudres, 83 Coulier, 276, 283 Courtepe"e, 179 Coward, 301 Craig, 11 Crookes, 18, 19, 313 Crum, 245 de la Cruz, Garcia, 300 Culverwell, 142 Cunningham, 136 Cushman, 21, 160 Dalladay, 150 van Dam, 233 Damoiseau, 39 Davies, 245 AUTHOR INDEX 323 Davis, N. B., 38, 160 Davis, O. C. M., 34, 104, 112, 113 Davison, 90, 119 Day, 152 Deacon, 48, 53 Deane, 193 Debus, 42 Defoe. 99 Delachanal, 235 Delafontaine, 244 De la Rue, 79 Denham, 124, 175 Desch, 79, 179 Devaux, 315 Dewar, 3, 7, 9, 10, 11, 13, 14, 15, 20, 24, 25, 26, 29, 302 Diamond, 205 Dietl, 100 Ditte, 152 Doelter, 166, 177, 223, 311 Donau, 174 Donnan, 130,161, 278, 279 Donnini, 177 Doyle, 166, 173 Dreaper, 164 Drucker, 19, 176 Drude, 318 Dudley, 114 Duff, 154 Duin, 217 Dumanski, 186, 253, 258 Dutailly, 306 Ebell, 137 Edgar, 35, 36 Edison, 302 Edson, 78 Ehrenberg, 22, 71, 72, 80, 137, 188, 222, 288 Ehrenhaft, 143, 144 Einstein, 138, 140, 141, 142, 144, 190, 191 Elissafoff, 209 Ellis, C., 60, 126 Ellis, R., 261 Elmore, 89, 271 Emden, 148 Emley, 156, 157 Endell, 153, 309 Engelder, 41, 48 Engels, 170 Engler, 20, 21, 277 Erdmann, 126 Espil, 126 Estrup, 115 Euler, 187 Evans, 107 Exner, 139, 141 Fahrion, 263 Fall, 154, 164, 243 Fanto, 268 Faraday, 47, 203, 208 Farmer, 223 Farnau, 105, 122, 246 Favre, 25 Felgate, 125 Fenaroli, 308 Ferguson, 135 Findlay, 269 Finch, 244 Fink, 48, 79, 80, 316 Firth, 35, 36 Fischer, F., 180, 278 Fischer, H. W., 222, 244 Fischer, K. T., 318 Fischer, M. H., 253, 254 Fischer, W., 169 Flade, 240 Fleischmann, 186 Foote, 78 Fortuyn, 121 Foster, 310 Foulk, 165 Franchet, 307, 309, 310 Frankland, 280, 281 Franklin, 186 Free, 162, 193 Freudenberger, 186 Freundlich, 16, 20, 100, 107, 108, 110, 111, 112, 114, 125, 129, 132, 135, 162, 167, 195, 208, 210, 216, 217, 226, 227, 269, 280, 314 324 AUTHOR INDEX Friedel, 151 Friedemann, 227, 228 Friedrich, 263 Friedenthal, 186 Frischauer, 223 Gadow, 199 Galecki, 176 Galeotti, 167, 176 Galpin, 166 Garbowski, 205 Gatin-Gruzewska, 187 Gaudechon, 27 Geddes, 29, 32, 34, 36 Gerke, 144, 291 Gengou, 172 von Georgievics, 100, 104, 106 Gerum, 20, 35 Ghosh, 23 Giampalmo, 167 Gibbs, 130, 131, 135, 189 Gilbert, 87, 107, 166 Giolitti, 186, 237 Girardi, 274 Gladstone, 187, 270 Glaser, 126 Goodchild, 198, 310 Gordon, 92 Gossage, 53 Goiiy, 137, 138 Graham, 1, 73, 113, 162, 168, 169, 177, 187, 228, 274 Grimaux, 168, 169, 172, 244 Grimm, 314 Groschuff, 238 Grout, 160 Grunwald, 309 Guignet, 169 Gurwitsch, 34, 123, 316 Gutbier, 166, 173, 205 Guthrie, 69 Guttmann, 294 Gyrlai, 245 Haber, 23, 37, 55, 56, 162, 174, 180, 181, 223, 278, 309 Haecker, 202, 304 Hallwachs, 23 Halske, 153 Hamlin, 312 Hannay, 304 Hanriot, 153 Hantzsch, 169, 171 Harbeck, 47 Hardy, 82, 162, 164, 212, 223, 241 Hargreaves, 52 Harkins, 67, 127, 131 Harned, 48 Harrison, 191 Hart, 244 Hartl, 271 Hartley, E. G. J., 177 Hartley, W. N., 288, 290 Hartwagner, 176 Hatschek, 90, 131, 134, 136, 150, 159, 190, 191, 192, 259, 318 Haussmann, 172 Hautefeuille, 148 Havard, 153 von Helmholtz, H., 207, 208 von Helmholtz, R., 276, 277 Hempel, 4, 7, 16, 40 de Hcmptinne, 47 Henderson, 43, 44 Henry, 46, 48 Herman, 147 Herty, 271 Herz, 169 Hilgard, 100 Himmelbaur, 174 Hitchcock, 170 Hober, 162, 223 van't Hoff, 140 Hofmann, A. W., 61, 113 Hofmann, F. B., 68, 72, 83, 84, 85, 86 Hofmann, K. A., 61 Hofmeister, 256 Hohenegger, 20 Hoitsema, 36 Holmes, 154, 164, 165, 243, 250, 259 Holt, 35, 36 AUTHOR INDEX 325 Homfray, 29, 34 Hooker, 254 Hopkins, 164, 199 Horner, 311 Horsin-Deon, 244 Hostetter, 154, J64 Houlehan, 34 Hulett, 13, 74, 148 Humphreys, 285 Hunt, 294 Hunter, F. W., 67 Hunter, J., 3, 6, 7, 16 Hurter, 44, 53, 299 Hutchings, 302 Iddings, 310 Ihmori, 18, 316 Ikeda, 123 Iljin, 139 Ingraham, 169 Ipatieff, 41, 48 Ives, 21, 64 Jablczynski, 125 Jacquemin, 121 Jamin, 77 Jannetz, 151 Jellinek, 56 Jobling, 52, 53, 54, "56, 57, 58, 60, 62, 64, 126 Johnson, 20, 41 Johnston, 70, 155, 164 Johonnot, 318 Jones, Chapman, 205 Jones, M., 148 Jones, W. J., 136 von Jonstorff, 148 Jordis, 167 Jorissen, 223 Jowitschitsch, 178 Juday, 66, 282 Jullion, 49 Just, 278 Kaempfer, 114, 125 Kahlbaum, 154 Kahlenberg, 109 Kaiser, 66 Karl, 126 Karrer, 21, 64 Kastle, 49, 124 Katz, 256, 317 Kaufmann, 250 Keane, 205 Kellner, 94 Kelvin, 146 Kingsbury, 21, 64 Kirchman, 69, 150 Klein, 169 Klemp, 245 Knietsch, 22, 49, 51 Knoblauch, 294 Kogonossow, 183 Kohlrausch, 208 Kohlschutter, 152, 302, 303 Kolb, Ellsworth, 137 Kolb, Emery, 137 Koller, 73, 77, 151 Konigsberger, 315 Koppen, 48 Kratz, 177 Kressman, 77 Krohnke, 261 Krukenberg, 199 Kriiss, 235 Krutwig, 52 Kruyt, 217 Kugelmann, 57 Kundt, 18 Kuriloff, 244 Kurzmann, 225 Kiister, 104, 106 Kutscheroff, 180 Lachaud, 170 Lachs, 103, 106, 114, 115 Lagergren, 34, 108 Lake, 73, 111, 113, 115, 118 Lamb, A. B., 8, 27, 34, 317 Lamb, H., 208 Landis, 8 Lang, 270 Langmuir, 37, 38, 39, 40, 41, 47, 316, 317, 318 326 AUTHOR INDEX Laplace, 134, 146 Larguier des Bancels, 167 Larsen, 194 Lassar-Cohn, 122 Laufer, 154 Laurie, 310 Lauth, 306 lea, 98, 173, 203, 205 Lear, 70 LeBlanc, 61, 184 LeChatelier, 23, 236, 237 Lecoq, 174 Lefebure, 20 Lefort, 170, 172 Lehmann, 142 Leighton, 18, 77, 92, 93, 94 Leith, 305 Lenard, 23, 134, 135, 284 Lenhardt, 285 Leonhardt, 216 von Lerch, 127 Levites, 167 Lewis, D. S., 10 Lewis, W. C. McC., 41 Liesegang, 258, 259 Linder, 168, 189, 206, 221 Liversidge, 174 Ljubavin, 222 Lloyd, 92, 97 Lockemann, 91 Lodge, 291 Loeb, J. 232, 253, 254, 255, 256, 267 Loeb, L. B., 28 Loevenhart, 49, 124, 244 Loew, 169 Long, 180 Lorenz, 275 Lottermoser, 106, 167, 168, 176, 187, 222 Love, 301 Lowenstein, 250 Lucas, 152 Lucchesi, 235 Lucke, 62, 63 Luckiesh, 200, 297 Lueger, 158 Lumiere, 228 Lunge, 47 Luppo-Cramer, 170, 205, 302 Luzi, 301 McBain, 29, 34, 100, 210, 225 McConnel, 281, 28 McDonald, 150 Mclntosh, 78 McKay, 194 McPherson, 260 Maglia, 7, 29, 34 Magnusson, 19 Mahin, 169 Mahla, 49 Mailhe, 58 Malcolm, 196 Malfitano, 184 Mannich, 56 Manville, 13 Mar, 165 Marc, 99, 100 Marek, 175 Margules, 23 Marshall, J., 235 Marshall, J., 275 Martin, 183 Martine, 26 Masius, 114 Masson, A., 179 Masson, O., 25, 28, 29, 34, 76 Mathers, 73 Mathieu, 109 Mat win, 17 Maxted, 47 Maxwell, 297 Mecklenburg, 167, 221, 287 Megigrecianu, 176 Meigen, 126 Meissner, 78 Melloni, 179 Melsens, 9, 26 Mendeleeff, 158 Mermet, 235 Merton, 20 Merwin, 154, 246 Metcalf, 189, 315 Meunier, 168 AUTHOR INDEX 327 Meyer, A., 19,5, 240, 241, 242, 243 Meyer, G., 133 Michael, 274 Michaelis, 97, 103, 106, 114, 115 Middleton, 218 Miller, 260 Millikan, 136, 144 Mills, 121 Mines, 219 Mitscherlich, 24, 38 Moissan, 152 Mond, 35, 36 Monier, 301 Moore, R. L., 164 Moore, W. C., 264 Moore, W. L., 282, 283, 284, 285, 286, 287, 295, 298 Morgan, 308 Morley, 19 Morris-Airey, 180 Morsi, 16 Mott, 205, 206 Mouret, 166 Mouton, 318 Miilfarth, 18 Mulhauser, 137 Muller, A., 92 Muller, A., 168, 169, 176 Muller, E., 180, 210 Muller, M., 235 Murray, 107, 108 Mylius, 238 Nagel, C. F., 169, 170, 233, 244 Nagel, G., 315 Nahrwold, 290 Nakamura, 66, 282 Namias, 229 Nauck, 270 Navassart, 223 Neel, 151 Neisser, 227, 228 Neogi, 302 Neumann, 111 Newbigin, 199 Newton, 197 Newman, 263 Nicholas, 250 Nichols, 295 Nipher, 65 Nobili, 197 Nordenson, 223, 318 Northcote, 170 Oberbeck, 314 Oden, 173, 191, 213, 215, 237 Ogg, 216 Oordt, 162 Osaka, 96, 103, 108, 112, 114 Osborne, J. L., 165 Osborne, T. B., 220 Osmond, 281 Osterhout, 267 Ostwald, W., 57, 58, 90, 110, 147, 259 Ostwald, Wa., 261 Ostwald, Wo., 1, 210, 308 Otobe, 282 Oudemanns, 167 Owen, 280 Owens, 188, 193 Oxland, 52 Paal, 20, 35, 126, 274 Pappada, 215, 216, 217 Parker, F., 223' Parker, F. L., 208 Parker, R. G., 150 Parks, 26, 78, 317 Parsons, 58 Partington, 148 . Paternd, 176 Patrick, 40, 127 Patten, 105 Patterson, 314 Paucke, 91 Pawlow, 148 Pease, 56 P&abon, 301 Pelet-Jolivet, 115, 116, 118- P&igot, 189 Pelouze, 285 Perley, 58 Perrey, 148 328 AUTHOR INDEX Perrin, 138, 139, 140, 144, 208, 209, 295, 319 Pettijohn, 317 Pfund, 198, 132 Phillips, F. C., 302 Phillips, P., 49 Pickering, 245, 262, 264 Picton, 168, 189, 206, 221 Pieroni, 176, 274 Pierson, 210 Pingree, 224 Pirsson, 310 Piutti, 7, 29, 34 Plashke, 61 Playfair, 123 Pluddemann, 34, 111 Pockels, 72 Pollock, 281 Poppe, 160 Porlezza, 16 Posnjak, 246, 257, 258 Potts, 161 Pouillet, 26 Powis, 261 Prat, 153 Pring, 314 Pringal, 277 Prost, 168 Prudhomme, 170 Przibram, 279 Publow, 77, 220 Quincke, 160, 207, 240 Ramage, 288, 290 Rammelsberg, 154 Ramsay, 13, 23, 35, 36, 245 Ramsden, 271 Randau, 75 Raoult, 154 Raschig, 43 Rasenfosse, 174 Rayleigh, 65, 67, 72, 81, 82, 83, 88, 131, 132, 134, 151, 182, 197, 202, 287, 295, 313, 318 Recoura, 229 Reed, 180 Reedy, 149, 152 Reese, 50 Regnault-Betton, 204 Reid, E. E., 16 Reid, E. W., 179, 187 Reinders, 86, 87, 98, 165 Reinold, 318 Reinsch, 91 Reitlinger, 23 Reusch, 304 Reuss, 207 Reychler, 20, 36 Reyerson, 293 Reynolds, 66, 158, 159 Rhead, 12, 37 Rice, 75 Richards, 28, 76 Richardson, C., 123, 275 Richardson, L. B., 31, 32 Richarz, 276, 277 Rideal, 46, 49, 52, 54, 55, 56 Riffard, 168, 172 Rindfusz, 164 Ringer, 244 Rittenhausen, 173 Ritter, 284 Roberts, 280 Robinson, F., 52 Robinson, I., 24 Rodewald, 257, 258 Roessler, 175 Rohland, 20 Rona, 97, 114 Rontgen, 129, 136 Rose, G., 70 Rose, R. P., 169, 237, 245 Rose, T. K., 235 Rosenhain, 303, 307 Rosinger, 275 Rosnyak, 125 Rothe, 106 Rothmund, 23, 280 Rubenbauer, 169 Rucker, 318 Rudge, 294 Ruer, 23 Ruff, 153 AUTHOR INDEX 329 Russ, 163 Huston, 288 Rutherford, 15 Sabaneeff, 187 Sabatier, 41, 44, 57, 58, 59, 60, 126 Sabin, 21 Sack, 180 Sahlblom, 92 Saida, 126 Sakulka, 66 Salamon, 54 Sansonoff, 174 Satterly, 16 de Saussure, 7 Saxton, 78 Schade, 312 Schaposchnikoff, 122 Scheermesser, 20 Scheetz, 205 Schenck, 206, 312 Schiaparelli, 170 Schiff, 167 Schlaepfer, 264 Schloesing, 22, 223 Schmauss, 285 Schmidt,G. 15, 100, 102, 106, 110, 114 Schmidt, W. A., 293 Schmidt, Walter, 104 Schneider, E. A., 274 Schneider, J., 129 Schoep, 184 Schonbein, 47, 123 Schoop, 303 Schottlander, 173 Schraube, 13 von Schroeder, 75, 76, 110, 177 Schucht, 107, 217 Schultze, 71, 72, 80 Schulze, 112, 209, 213, 217, 218, 223, 228, 308 Schuster, 22 Schutzenberger, 301 Schwalbe, 76 Schwarz, 235 Schwerin, 210 Seager, 63 Seddig, 142 Seeliger, 285 Seger, 306 Selmi, 1 Senderens, 41, 57, 58 Senter, 123, 124 Sestini, 90 Seyewetz, 228 Shaw, 60 Shenstone, 40 Sheppard, 252, 264 Sherrick, 94, 99, 218 Sherwood, 18, 317 Shields, 35, 36 Shorter, 189, 315 Siedentopf, 182, 311 Siegmund, 126 Siemens, 153 Simcoe, 73 Simpson, 285, 286 Sisley, 105 Skey, 10, 91, 114, 177, 223 Slade, 169 Smith, Angus, 39 Smith, A. F., 73 Smith, C. R., 195, 224, 252 Smith, J. Cruickshank, 72, 88 Smith, J. Lawrence, 147 Smoluchowski, 140, 142 Smyth, H. D., 293 Smyth, Piazzi, 283 Snowdon, 149 Sohncke, 318 Sorensen, 255 Sosman, 154 Soxhlet, 170 Spencer, 23 Spengel, 205 Spezia, 154 Spring, 77, 151, 154, 168, 170, 178, 202, 300 Squibb, 44 Stark, 269 Stearn, 253 Stefan, 74 Steiner, 108 Steinmetz, 64, 69 330 AUTHOR INDEX Stepanoff, 23 Stevenson, 249 Stewart, 169 Stockle, 133, 134 Stokes, 136, 142, 211, 306, 313 Stoney, 21 Storey, 10 Stritar, 268 Strong, 175 Strutt, 42 Svedberg, 138, 141, 144, 169, 172, 173, 174, 175, 176, 179, 180, 223, 274 Swan, 127 Swinton, 18 Szilard, 168 Tait, 9 Tammann, 81, 151, 249 Tangl, 136 Taylor, F. W., 157 Taylor, H. S., 47, 48, 49, 52, 54, 55, 56 Taylor, W. E., 163 Teague, 227 Thibault, 170, 172 Thiel, 56 Thompson, 157 Thomson, J. J., 279, 280, 317 Thomson, W., 25, 146 Thorp, 68, 155 Thorpe, 52, 54, 55, 178, 307, 309 Thoulet, 91 Tiebackx, 232 Tinker, 109 Titoff, 29, 31, 34 Toch, 80 Tolman, 253, 287, 293 Tommasi, 122 Travers, 29, 30, 34 Traube, 219 Trouton, 29, 30, 99 Triimpler, 100 Tubandt, 169 Tumlirz, 74 Twomey, 127 Tyndall, 22, 179, 182, 195, 197, 200, 202, 287; 298, 305 Ullmann, 19 Upton, 158 Valentiner, 15 Vanino, 175, 271 Van Name, 181 Van Slyke, 77, 103, 106, 220, 244 Vater, 4, 7, 16 von Vegesack, 187 Verneuil, 311 Vignon, 76, 221 Vincent, 314 Vogel, 53 Von Engeln, 148, 281 van der Waals, 134 Wadsworth, 302 Wagner, 106, 107 Wald, 70 Waldbott, 97 Walden, 185 Waldie, 223 Walker, A. O., 291 Walker, J., 105, 106 Warburg, 18 Warington, 302 von Wartenberg, 23 Washburn, 76 Watson, G., 166 Watson, T. L., 310 Weber, 71 Wegelin, 178 von Weimarn, 163, 164, 165, 166, 167, 176, 183, 238, 275 Weingartner, 173 Weinmayr, 315 Weisberg, 168 Weiser, 94, 99, 165, 218, 299 Weiske, 92 Welker, 235 Wells, 144, 291 Wetherill, 155 Whatmough, 129 Wheeler, 12, 37 Wiedeburg, 77 Wiedemann, 207, 208 Wiegner, 105 AUTHOR INDEX 331 Wieland, 61 Wiener, 137, 315 Wiesner, 284 Wigand, 176 Wild, 21, 277 Willard, 271 Williams, A. E., 306 Williams, A. M v 32 Willows, 131, 134, 136, 150 Wilson, C. T. R., 278, 279, 280, 285 Wilson, G., 127, 128 Wilson, R. E., 8, 34 Winkelblech, 260, 261, 271 Winssinger, 168, 176 Wintgen, 188 Witzmann, 174 Wohler, F., 49 Wohler, L., 20, 34, 111, 153, 170, 174, 205, 235 Wohler, P., 34, 111 Wolf, 278 Wolff, 76 Wood, 197, 199, 205, 282, 298, 313 Worden, 88 Worley, 129, 130 Worthington, 66, 74 Woudstra, 223 Wright, A. W., 302 Wright, F. E., 152 Wright, L. T., 175, 177 Yorke, 92 Young, S. W., 168, 224 Young, T., 283 Zangger, 139 Zawrieff, 180 Zeidler, 270 Zeisberg, 51 Zlobicki, 189 Zsigmondy, 139, 182, 206, 217, 221, 230, 235, 239, 240, 249, 305, 306 Zulkowski, 307 SUBJECT INDEX Abnormal adsorption, 106-108 end-points, 49 Acid, hydrochloric, 4, 9, 20, 26, 44, 49, 50, 52, 53, 92-96, 107, 116-119,121,127,129,165, 175, 214, 216 lysalbinic, 173 nitric, 57, 58, 123, 129, 175 picric, and silk, 105 protalbinic, 173, 308 sulphuric, 22, 49-51, 61, 92-96, 115-119, 165, 180 Action, catalytic, of charcoal, 39, 40 mutual, of sols, 225-235 Addition agents, 165, 231 of agglomerating agent, rate of, 220 of peptizing agent, sols by, 177, 178 Adhesion of hot powders, 81 Adhesives, 74 Adsorbed gas, density of, 24 vapor pressures of, 25 Adsorbent, poisoning of, 47-53, 56- 59, 123-125 Adsorbing power of fuller's earth, 121 Adsorption, abnormal, 106-108 Adsorption and density, 70, 91 and hydrogen ion concentra- tion, 219 and hydrolysis, 120-122 and peptization, 166 and solvent, 112, 113 and surface tension, 135, 166, 167 and the triple point, 70 and valence, 112, 209, 213-219 and vapor density, 19 Adsorption by charcoal, 3-18, 24-40, 90, 91, 96-98, 101-106, 108, 110-114, 233 by crystals, 99 by filter paper, 92, 107, 108, 178 effect of pressure on, 3, 9, 16, 20, 29, 32, 36 of temperature on, 3-5, 9, 13, 29, 32, 113 from solution by liquid, 126-128 by solid, 90-126 gravimetric determination of, 93-97 heat of, 24^27, 78, 113, 258 in cathode ray tubes, 18, 19 isotherm, 29-37, 100-107 exponent in, 33, 100, 106 negative, 90, 108-110 and osmotic pressure, 109, 110 of dyes, 90, 104, 112, 115-120, 232 of strong electrolytes by char- coal, 96-98, 238 of essential oils, 73, 332 of gases by solids, 3-64 of liquid by fibers, 76, 77 by liquid, 81-83 by solid, 67-78, 140 of mixed gases, 10, 16 of rouge by metals, 79 of several solutes, 113-120 of solid by liquid, 83-89, 260 by solid, 78-81 of vapor by liquid, 65-67, 134, 135 of vapors by solids, 3-64 positive, 90 reversibility of, 110, 111 333 334 SUBJECT INDEX Adsorption, selective, catalysis and, 41 specificity of, 111-113 titration determination of, 93- 97 vacuum by, 9 Agar-agar, 74, 190, 233, 243, 274 Agate, 198 Agent, agglomerating, rate of addi- tion of, 220 sols by removal of, 177 Agents, addition, 165, 231 emulsifying, 261, 264 Agglomerating agent, 172, 173, 177, 220 rate of addition of, 220 sols by removal of, 177 Agglomeration, 111, 153, 161, 162, 170, 177, 192, 195, 212, 220, 221, 240 reversibility, 177, 221 Air films and mobility, 21 and voltage, 22, 23 washing of, 299 Albumin, 1, 91, 173, 185, 187, 188, 219, 221, 231, 260, 271, 315 Alcohol, solidified, 262 Alumina, 1, 18, X 20, 41, 44, 85, 91 120, 123, 169, 170, 172, 206, 226, 227, 245, 246, 312 Aluminum, 1, 19, 164, 273 Amalgam, sodium, 69 American method of making emul- sions, 265 Ammonia, 4, 6, 16, 19, 26, 31, 43, 55-57, 228, 277 process, 55-57 Amplitude, 138 Analysis of smokes, 293 Angle, contact, 67 Anions, 115, 213-220, 224 Antagonism of ions, 267 Aquadag, 156, 173 Argon, 4, 6, 13, 14, 26 Armor-plated bubbles, 272, 273 Arsenious oxide, 50, 51, 53, 56, 91, 175 Arsenious sulphide, 175, 185, 188, 226-228, 260 Asphalt, 275 Baking powders, phosphate, 80 Banding, rhythmic, 259 Barium salts, 1, 92, 120, 163-165, 172, 190, 209, 214-218, 220, 239 Bayer process, 163 Beaches, wells near, 99 Beta rays, 223 Bishop's ring, 283 Bitumen, 123 Blackness, 199 Blue eyes, 200-202 feathers, 201, 202, 304 powder, 79 quartz, 310 sky, 202, 295, 304 sun, 283 Body color, 203 Bombardments, molecular, 138 Borneol, dehydrogenation of, 60 Bronzing liquids, 87, 88 Brownian movements, 136-145, 183, 186, 187 and size of particles, 139 in gases, 143, 144 laws of, 138-142 Bubbles, 2, 66, 198, 199, 268-273, 304, 318 and drops, coalescence of, 65, 66, 286 armor-plated, 272, 273 Butter, 2, 241, 263, 266 Calcium soaps, 262, 263, 267, 268, 274 sulphate, filtration of, 159 Callow process, 272 Cape Cod lighter, 68 Capillarity, 25, 67, 69, 70, 146, 247 Carbon black, 21, 72, 157, 178, 264, 270 dioxide, 4-6, 11-13, 20, 24, 26, 32, 37, 39, 133 SUBJECT INDEX 335 Carbon monoxide, 4, 5, 11-13, 26, 36, 37, 39, 46-48 Caseine, 2, 77, 103, 163, 167, 169, 173, 185, 191, 219, 220, 231-233 Cassius, purple of, 234, 235 Catalysis and degree of dispersion, 125 and fields of force, 43 and selective adsorption, 41 and viscosity, 43 contact, 40-64, 122 of hydrogen peroxide, 44, 123- 125 theories of, 40-14 Catalytic action of charcoal, 39, 40 Cataphoresis, 206, 268, 318 Cathode ray tubes, adsorption in, 18, 19 Cations, 115, 213-220 Cattermole process, 272 Cellulose, 35, 76, 161 nitrates, 88, 167, 274 Cement dust, 293 Portland, and gypsum, 80 Centrifuging of sols, 186 Ceria, 64, 226, 227 Chance-Glaus process, 53 Charcoal, adsorption by, 3-18, 24- 40, 90, 91, 96-98, 101-106, 108, 110-114, 233 adsorption of strong elec- trolytes by, 96-98, 233 catalytic action of, 39, 40 cocoanut, 3, 4, 67 from different woods, 6, 8 gas-treated, 301 oxidation of, 10-13 structure of, 7 synthetic, 5 Charge on diaphragms, 207-209 reversal of, 219 Chlorine process, Deacon, 52 Chloropicrin, 8, 16, 48 Christiansen effect, 197 Chromic oxide, 111, 123, 168, 170, 184, 226-230, 233, 234, 246, 311 Chromite and magnetite, 69 Clay, plasticity of, 160 Clouding of gels, 247-250, 256 Coalescence of drops and bubbles, 65, 66, 286 of liquids, 146, 147, 161, 261, 265 of solids, 147-154, 161 Coarse and fine powders, 79 Cocoa, 190 Cocoanut charcoal, 3, 4, 6, 7 Coherer, 23 Collodion membranes, 184, 240 pores in, 185, 240 Colloid, definition of, 1 peptization by, 167, 170 stabilization by, 212, 225 Colloidal fuel, 275 ice, melting of, 78 solutions, true and, 144, 183 Colloids, color of, 196-206 protecting, 170-173, 180 Color, body, 203 of coUoids, 196-206 of gems, 224, 311, 312 of metal sols, 203-205, 308 of powders, 198 of rouge, 205 of water, 202, 203 pigment, 196 structural, 196 surface, 203 Colors, sunset, 283 Combustion, surface, 62 Comets, 297 Composition of dust, 288-290 Compounds, indefinite, 39 intermediate, 44 stabilization of, 122 Concentration and surface tension, 130, 131 surface, 130-132 Condensation methods, 172-176 Conductance of emulsions, 267 of sols, 210 without solutes, 211 Consolidation settling, 193 336 SUBJECT INDEX Contact angle, 67 catalysis, 40-64, 122 sulphuric acid, 49-51 Continental method of making emul- sions, 265, 266 Copper, 20, 59, 60, 79, 150, 173, 174, 180, 272, 275, 278, 306, 312, 314 ferrocyanide membranes, 110 oxide, 52, 107, 122, 168-170, 172, 173, 206, 246 Coronas, 282, 283 Cotton, 27-30, 76, 77, 92-96, 108, 116, 118-120, 147 Cottrell process, 286, 291-293 Cracking of emulsions, 268 Cream, 186, 194, 268, 271 Crystallization of gels, 107, 166, 223 Crystals, adsorption by, 99 filamentous, 302 size of, 164 Crystal size and melting-point, 148 and solubility, 148 Cubical piling, 157, 159 Curdy precipitates, 162, 163, 231 Daguerreotype, 73 Deacon chlorine process, 52 Decomposition of esters, 45, 46 of vermilion, 79 photochemical, sols by, 176 Definition of colloid, 1 Dehydrogenation, 60 of borneol, 60, 61 Density and adsorption, 70, 91 of adsorbed gas, 24 of smokes, 300 of sols, 188 vapor, adsorption and, 19 Detection of strychnine, 113 Determination, gravimetric, of ad- sorption, 93-97 of type of emulsion, 263 titration, of adsorption, 93-97 Diagram for silica, 248 Dialysis, 188, 243 Diameter of molecules, 319 Diaphragms, charge on, 207-209 flow through, 206-210 Diffusion, 164, 187, 258 Dilatancy, 158, 159 Dineric interface, 83, 260 Discharge, oscillatory, 180 Discontinuity, surface tension and, 134 Disintegration, electrical, sols by, 179 electrochemical, sols by, 180 mechanical, sols by, 161, 178 Dispersed phase, 1, 2 Dispersing phase, 1, 2 Dispersion, degree of, catalysis and, 125 methods, 172, 177-180 Drops, 2, 65, 66, 71, 146, 148, 161, 163, 242, 261, 265, 282 and bubbles, coalescence of, 65, 66, 286 electrification of, 65 optical properties of, 282, 283 rain, 1, 283-285 rain, size of, 284-286 rolling, 66, 71, 282 Dry fog, 280, 281 Drying, electrical, 210 Dust, 287-299 cement, 293 composition of, 288-290 electrification of, 294 from Krakatoa, 288, 296 meteoric, 288-290 optical properties of, 295-297 Dyes, adsorption of, 90, 104, 112, 115, 120, 232 Dynamic and static surface tensions, 131-134 Effect, Christiansen, 197 of gravity on sols, 139, 186 of ions on syneresis, 244 of pressure on adsorption, 3, 9, 16, 20, 29, 32, 36 of temperature on adsorption, 3-5, 9, 13, 29, 32, 113 SUBJECT INDEX 337 Electrical disintegration, sols by, 179 drying, 210 endosmose, 206-210 properties of sols, 206-212 tanning, 210 Electrification of drops, 65 of dust, 294 of ice, 302 Electrochemical disintegration, sols by, 180 Electrolysis, sols by reduction with- out, 173 Electrolytes, adsorption of, by char- coal, 96-98, 233 Electrolytic reduction, sols by, 174 Electromotive force measurements with sols, 206, 211, 212 Electrons, emission of, 278 Elmore bulk oil process, 89, 271 Elutriation, 136 Emanation, 15, 16, 114, 311 Emission of electrons, 278 Emulsifying agents, 261, 264 Emulsion jellies, 240, 264 Emulsions, 2, 161, 240, 241, 260- 268, 275 conductance of, 267 cracking of, 268 intermittent shaking of, 265 making, American method of, 265 Continental method of, 265, 266 photographic, 231 type of, determination of, 263 types of, 261 Enamels, 307, 309 Endosmose, electrical, 206-210 End-points, abnormal, 49 Enzymes, 48, 49 Essential oils, adsorption of, 73, 232 Esters, decomposition of, 45, 46 Ethylene, 17, 41, 45 Exponent in adsorption isotherm, 33, 100, 106 External phase, 1, 2, 166, 187, 263 Eyes, blue, 201, 202 22 Fabrics, shower-proof, 71 Fatigue, photo-electric, 23 Fats, hydrolysis of, 268 Feathers, blue, 201, 202, 304 frost, 282 Ferric oxide, 51, 52, 79, 87, 91, 111, 168, 172, 175, 185, 187- 189, 205, 217, 218, 221, 226-228, 232-234, 238, 245, 270, 310 Ferrocyanide, copper, membranes, 110 Fibers, adsorption of liquid by, 76, 77 Fields of force, catalysis and, 43 Filamentous crystals, 302 Filaments, 2, 37, 42, 47, 240, 245, 302 Films, 2, 20-23, 25, 38, 65-67, 81, 82, 89, 128, 151, 160, 189- 191, 195, 260, 269, 271, 314-319 air, and mobility, 27 and voltage, 22, 23 liquid, 67, 81, 128, 160, 191, 269, 315-319 solid, 89, 189, 190, 195, 260, 271, 314-316 thickness of, 25, 38, 82, 160, 189, 191, 314-319 Filter paper, adsorption by, 92, 107, 108, 178 Filtration of calcium sulphate, 159 Fine and coarse powders, 79 Flames, 174, 299, 300 Flamingoes, 199 Flocculation, 105, 106 Flocculent precipitates, 162, 163 Floridin, 123 Flotation, 2, 271-273 Flow through diaphragms, 206-210 Fluidity, 154, 210 zero, 156, 157, 190, 192, 240 Foam, 268-273, 304 stabilization of, 270, 272 Fog, 1, 275-286, 298 -bows, 282 338 SUBJECT INDEX Fog, dry, 280, 281 precipitation of, 285 supercooled, 281 Fogs, metallic, 275 Force, fields of, and catalysis, 43 Formaldehyde, 61, 184, 225 Freezing of sols, 78 Fritting, 151 Frost feathers, 282 Froth, 268, 269, 272 Fuel, colloidal, 275 Fuller's earth, adsorbing power of, 121 Gamboge, 139-142 Gas, adsorbed, density of, 24 illuminating, purification of, 17 mantles, 54 vapor pressures of adsorbed, 25 Gases, adsorption of, by solids, 3-64 Brownian movements in, 143, 144 in solids, 304, 305 mixed, adsorption of, 10, 16 Gas-treated charcoal, 301 Gelatine, 1, 73-76, 161, 165, 168, 172, 174, 178, 184, 185, 188, 189, 194, 195, 204, 224, 225, 228-233, 251-260, 274 jellies, vapor pressures of, 75, 256 strains with, 73 Gelatinous precipitates, 162, 163, 177, 236-240 Gels, 162 clouding of, 247-250, 256 crystallization of, 107, 166, 223 structure of, 236-245 vapor pressures of, 236, 245-250 Gems, color of, 224, 311, 312 Glass, ruby, 2, 305-307 welding of, .149, 150 Gold leaf, granulation of, 83 number, 230, 231 Grains, 2, 83, 148, 161 Granulation of gold leaf, 83 Graphite, 2, 44, 156, 190, 192, 240 Gravimetric determination of ad- sorption, 93-97 Gravity, effect of, on sols, 139, 186 welding by, 151 Gum arabic, 68, 170, 172, 179, 231, 232, 265, 266, 269 Gypsum and Portland cement, 80 Haemase, poisoning of, 123, 124 Hail, 286, 287 Halos, 283, 298 Harmattan, 296 Haze, 297, 298 Heat of adsorption, 24-27, 78, 113, 258 Helium, 4, 5, 9, 13, 14, 24 Hematite, 205 Hexagonal piling, 157, 159 Honeycomb structure, 163, 239, 264 Hot powders, adhesion of, 81 Hydrochloric acid, 4, 9, 20, 26, 44, 49, 50, 52, 53, 92-96, 107, 116-119,121,127,129,165, 175, 214, 216 Hydrogels, 162 Hydrogen, 4, 5, 9, 13, 14, 17, 23-26, 28-31, 35, 37, 40, 41, 45-48, 58-61, 126, 133 Hydrogenation, 58-60, 126 of oils, 60, 126 Hydrogen ion concentration, adsorp- tion and, 219 peroxide, catalysis of, 44, 123- 125 Hydrolysis, adsorption and, 120-122 of fats, 268 sols by, 175 Hydrophane, 304 Hydrosols, 162 Ice, colloidal, melting of, 78 cream, 231 electrification of, 302 Illuminating gas, purification of, 17 Indefinite compounds, 39 Index, refractive, 188, 199 Indigo, 21, 123, 203 SUBJECT INDEX 339 Inhomogcneity of sols, 182 Ink, 1, 92, 178 Interface, dineric, 83, 260 Leaf, gold, granulation of, 83 Leaking of tents, 71 Liesegang's rings, 258, 259 Interlacing systems, 2, 239, 242, 243, Light, ultra-violet, 223, 278, 280, 311, 312 Liquid, adsorption from solution by, 126-128 adsorption of, by fibers, 76, 77 by solid, 67-78, 140 liquid by, 81-83 solid by, 83-89, 260 vapor by, 65-67, 134, 135 films, 67, 81, 128, 160, 191, 269, 315-319 peptization by, 167, 224, 230 Liquids, bronzing, 87, 88 coalescence of, 146, 147, 161, 261, 265 in sieves, 70, 72 tensile strength of, 74 304 Intermediate compounds, 44 Intermittent shaking of emulsions, 265 Internal phase, 1, 2, 166, 263 Iodides in sea-water, 90 Iodine, 90, 98, 104, 111, 147, 277, 278 starch, 104, 111 Ion, peptization by, 167, 168 Ions, antagonism of, 267 effect of, on syneresis, 244 stabilization by, 212-223 Iron, 36, 56, 72, 122, 154, 180 parkerized, 72 passive, 122 Irregular series, 227 Isotherm, adsorption, 29-37, 100- Lloyd's reagent, 97 107 Lycopodium powder, 68, 80, 270 adsorption, exponent in, 33, Lysalbinic acid, 173 100, 106 Itacolumite, 155 Jasper, 310 Jelly, 2, 163, 236, 237, 240-259, 262, 274 emulsion, 240, 264 gelatine, vapor pressures of, 75 256 stabilization of, 244 swelling of, 243, 251-256 Kenotron, 293 Kerosene, 68, 72, 84, 85, 87, 190, 261, 262, 264 Krakatoa, dust from, 288, 296 Krypton, 15 Lamination, 305, 312 Law, Schulze's, 112, 209, 213-219 Stokes', 136, 142, 211 Laws of Brownian movements, 138- 142 Leaching of whiskey, 114 Magenta, 121, 203 Magnesium soaps, 262, 268, 274 Magnetite and chromite, 69 Making emulsions, American method of, 265 Continental method of, 265, 266 Mantles, gas, 64 Welsbach, 64 Marksmen, 201 Mayonnaise, 264, 266 Mechanical disintegration, sols by, 178 Meerschaum, 20, 305 Melting of colloidal ice, 78 Melting-point, crystal size and, 148 Membranes, collodion, 184, 240 pores in, 185, 240 copper f err ocyanide, 110 semipermeable, 109 Mercury, 24, 44, 68-70, 72, 73, 89, 128, 133-135 Metallic fogs, 275 340 SUBJECT INDEX Metal sols, color of, 203-205, 308 Metals, adsorption of rouge by, 79 Metathetical reactions, sols by, 175 Meteoric dust, 288-290 Methane, 5, 17, 39, 45, 46 Method, American, of making emul- sions, 265 Continental, of making emul- sions, 265, 266 Methods, condensation, 172-176 dispersion, 172, 177-180 Mica, 74, 151 Micelles, 211 Migration velocities, 206, 210 Mirrors, 302 Mixed gases, adsorption of, 10, 16 paints, 80, 156, 262 solvents, peptization by, 167 Mobility, air films and, 21 Molecular bombardments, 138 weights, 141, 145, 187 Molecules, diameter of, 319 Monoxide, carbon, 4, 5, 11-13, 26, 36, 37, 39, 46-48 Mordant, 90, 112, 122 Mother-of-pearl, 198, 311 Movements, Brownian, 136-145, 183, 186, 187 and size of particles, 139 in gases, 143, 144 laws of, 138-142 Mutual action of sols, 225-235 Negative adsorption, 90, 108-110 and osmotic pressure, 109, 110 Neon, 13, 14 Newton's rings, 197 Nickel, 41, 43, 45, 58-61, 79, 125, 126, 149 Nitric acid, 57, 58, 123, 129, 175 process, 57 Nitrogen, 4-6, 10, 11, 13-16, 24, 26, 39, 42, 43, 55-58, 133 Nobili's rings, 197 Non-aqueous sols, 230, 274, 275 Non-electrolyte, peptization by, 167, 168 Nuclei, 276-280 Number, gold, 230, 231 Oat-meal, 25, 147 Obsidian, 310 Oil, 2, 48, 81, 161, 189, 242, 260, 268, 274 Oil-in-water, 2, 262, 264, 268 Oils, essential, adsorption of, 73, 232 hydrogenation of, 60, 126 spreading of, 82 Opacity of paper, 80, 199 Opals, 304, 313 Optical properties of drops, 282, 283 of dust, 295-297 of sols, 182, 183, 188, 193-205 Order of ultra-filtration, 185 Oscillatory discharge, 180 Osmium, 56 tetroxide, 61 Osmotic pressure, negative adsorp- tion and, 109, 110 pressures, 109, 110, 141, 186, 187, 189 Over-voltage, 23 Oxidation of charcoal, 10-13 preferential, 46 sols by, 174 Oxide, arsenious, 50, 51, 53, 56, 91, 175 chromic, 111, 123, 168, 170, 184, 226-230, 233, 234, 246, 311 copper, 52, 107, 122, 168-170, 172, 173, 206, 246 ferric, 51, 52, 79, 87, 91, 111, 168, 172, 185, 187-189, 205, 217, 218, 221, 226-228, 232-234, 238, 245, 270, 310 Oxygen, 4-6, 10-16, 23, 24, 26, 36, 37, 39, 40, 42, 44, 47-54, 57, 58, 62-64, 133 Paints, mixed, 80, 156, 262 Palladium, 20, 25, 28, 35, 37, 47, 73. 126, 172 SUBJECT INDEX 341 Paper, filter, adsorption by, 92, 107, 108, 178 opacity of, 80, 199 Parkerized iron, 72 Particles, size of, Brownian move- ment and, 139 Passive iron, 122 Pectization, 162 Peptization, 162, 166-171, 212, 221, 222, 233, 242, 264, 275, 306 and adsorption, 166 by colloid, 167, 170 by ion, 167, 168 by liquid, 167, 224, 230 by mixed solvents, 167 by non-electrolyte, 167, 168 by salt, 167, 170 Peptizing agent, sols by addition of, 177, 178 Peroxide, hydrogen, catalysis of, 44, 125 Phase, dispersed, 1, 2, 161 dispersing, 1, 2 external, 1, 2, 166, 187, 263 internal, 1, 2, 166, 263 Phosphate baking powders, 80 Photochemical decomposition, sols by, 176 Photo-electric fatigue, 23 Photographic emulsion, 231 Picric acid and silk, 105 Pigment colors, 196 Piling, cubical, 157, 159 hexagonal, 157, 159 Platinum, 20, 23, 25, 28, 35, 36, 46- 51,57-59,94,111,123-126, 179, 185, 187, 205, 223, 308 Plasticity, 154-160, 191, 264 of clay, 160 Poisoning of adsorbent, 47-53, 5&- 59, 123-125 of haemase, 123, 124 Pores in collodion membranes, 185, 240 Portland cement and gypsum, 80 Positive adsorption, 90 Potash, recovery of, 293 Potassium soaps, 262 Powder, lycopodium, 68, 80, 270 Powders, phosphate baking, 80 adhesion of hot, 81 coarse and fine, 79 colors of, 198 welding of, 151 Power, adsorbing, of fuller's earth, 121 Precipitates, curdy, 162, 163, 231 flocculent, 162, 163 gelatinous, 162, 163, 177, 236- 240 sandy, 162, 163, 236 types of, 161-163 Precipitation of fog, 285 of smoke, 290-293 of solid, 300-303 Preferential oxidation, 46 Preparation of sols, 172-181 Pressure, effect of, on adsorption, 3, 9, 16, 20, 29, 32, 36 osmotic, negative adsorption and, 109, 110 welding by, 151 Pressures, osmotic, 109, 110. 141, 186, 187, 189 swelling, 257 unequal, 154 vapor, of adsorbed gas, 25 of gelatine jellies, 75, 256 of gels, 236, 245-250 wetting, 77 Promoters, 56 Properties, electrical, of sols, 206- 212 of sols, 182-235 optical, of drops, 282, 283 of dust, 295-297 of sols, 182, 183, 188, 193- 205 Protalbinic acid, 173, 308 Protecting colloids, 170-173, 180 Protoplasm, 308 Puff-paste, 309 Purification of illuminating gas, 17 Purple of Cassius, 234, 235 342 SUBJECT INDEX Quartz, 25, 72, 97, 179, 302 blue, 310 Quicksands, 159 Radium, 15, 223, 311, 312 Rain, 1, 283-286 Rainbow, 197, 282, 298 Rain drops, 1, 283-285 size of, 284-286 Rate of addition of agglomerating agent, 220 Reactants, 41 Reactions, metathetical, sols by, 175 Reagent, Lloyd's, 97 Recovery of potash, 293 Reduction, electrolytic, sols by, 174 sols by, without electrolysis, 173 Reflection, selective, 203 Refractive index, 188, 199 Reinsch's test, 91 Resonance, 203, 204 Resultants, 41 Reversal of charge, 219 of stabilization, 219 Reversibility of adsorption, 110, 111 of agglomeration, 221 Rhubarb, 295 Rhythmic banding. 259 Ring, Bishop's, 283 Rings, Liesegang's, 258, 259 Newton's, 197 Nobili's, 197 water, 92 Rolling drops, 66, 71, 282 Rouge, 79, 178, 205 adsorption of, by metals, 79 color of, 205 Rubber, 2, 305, 307 Ruby glass, 2, 305-307 Salt-cake process, 52 Salt, peptization by, 167, 170 spray, 281 Salts, silver, 85, 98, 107, 111, 165, 168, 172-174, 176, 177, 183, 215, 216, 221, 231, 233 Salts, thorium, 16, 43, 45, 64, 80, 114, 169, 226, 227 Sandy precipitates, 162, 163, 236 Saponin, 131, 132, 170, 172, 189, 196, 269-271, 273, 315 Savart plate, 298 Schoop process, 303 Schroeder-Grillo contact mass, 49 Schulze's law, 112, 209, 213-219 Sea-water, iodides in, 90 Selective adsorption and catalysis, 41 reflection, 203 wetting, 83, 85, 260 Semipermeable membranes, 109 Series, irregular, 227 Settling, consolidation, 193 subsidence, 193 Shaking of emulsions, intermittent, 265 Shower-proof fabrics, 71 Sieves, liquids in, 70, 72 Silica, 18, 20, 40. 160, 177, 185, 187, 188, 220, 228, 236-238, 240- 251, 274, 288, 304 diagram for, 248 Silk, 105, 115, 116, 121 picric acid and, 105 Silver, 62, 166, 173, 198, 199, 203- 205, 215, 223, 230, 274, 278, 302, 303, 307 salts, 85, 98, 107, 111, 165, 168, 172-174, 176, 177, 183, 215, 216, 221, 231, 233 Sintering, 151-153, 200, 230 Size, crystal, and melting-point, 148 and solubility, 148 of crystals, 164 of particles, Brownian move- ments and, 139 of rain drops, 284-286 Sky, blue, 202, 295, 304 Sleet, 286 Smoke, 1, 2, 22, 184, 286-300 precipitation, 290-293 Smokes, analysis of, 293 density of, 300 wetting of, 22 SUBJECT INDEX 343 Snow, 286 Soaps, 2, 131, 132, 170, 178, 189, 225, 231, 240, 260, 262-265, 267-271, 274, 304 calcium, 262, 263, 267, 268, 274 magnesium, 262, 268, 274 ' potassium, 262 sodium, 131, 132, 225, 231, 240, 260, 262, 264, 265, 267, 269 Sodium amalgam, 69 soaps, 131, 132, 225, 231, 240, 260, 262, 264, 265, 267, 269 Solid, adsorption from solution by, 90-126 adsorption of, by liquid, 83-89. 260 liquid by, 67-78, 140 by solid, 78-81 films, 89, 189, 190, 195, 260, 271, 314-316 precipitation of, 300-303 solution, water in, 249, 250 Solidified alcohol, 262 Solids, adsorption of gases by, 3-64 vapors by, 3-34 coalescence of, 147-154, 161 gases in, 304, 305 in solids, 305-313 t surface tension of, 136 Sols, 162 by addition of peptizing agent, 177, 178 by electrical disintegration, 179 by electrochemical disintegra- tion, 180 by electrolytic reduction, 174 by hydrolysis, 175 by mechanical disintegration, 178 by metathetical reactions, 175 by oxidation, 174 by photochemical decomposi- tion, 176 by reduction without electroly- sis, 173 by removal of agglomerating agent, 177 Sols, centrifuging of, 186 conductance of, 210 density of, 188 electrical properties of, 206-212 electromotive force measure- ments with, 206, 211, 212 effect of gravity on, 139, 186 freezing of, 78 inhomogeneity of, 182 metal color of, 203-205, 308 mutual action of, 225-235 non-aqueous, 230, 274, 275 optical properties of, 182, 183, 188, 193, 205 preparation of, 172-181 properties of, 182-235 stability of, 212-235 surface tension of, 131 Solubility and crystal size, 148 Solutes, conductance without, 211 several, adsorption of, 113-120 Solution, adsorption from, by liquid, 126-128 by solid, 90-126 solid, water in, 249, 250 Solutions, true and colloidal, 144, 183 Solvent, adsorption and, 112, 113 Solvents, mixed, peptization by, 167 Specificity of adsorption, 111-113 Sphagnum moss, 77 Splashes, 66 Sponge structure, 239, 240, 241 Spray, salt, 281 Spreading of oil, 82 Stability and temperature, 221, 222 of sols, 212-235 Stabilization by colloid, 212, 225 by ions, 212-223 by water, 224 of compounds, 122 of foam, 270, 272 of jellies, 244 reversal of, 219 Starch, 1, 2, 104, 111, 188, 191, 231, 233, 240-242, 260 iodine, 104, 111 344 SUBJECT INDEX Static and dynamic surface tensions, 131-134 Stokes' law, 136, 142, 211 Strains with gelatine, 73 Strength, tensile, of liquids, 74 Structural colors, 196 Structure, honeycomb, 163, 239, 264 of charcoal, 7 of gels, 236-245 sponge, 239, 240, 241 Strychnine, detection of, 113 Subsidence settling, 193 Sugar, 1, 161, 168, 183, 225, 230 Sulphate, calcium, nitration of, 159 Sulphide, arsenious, 175, 185, 188, 226-228, 260 Sulphur, 17, 47, 174, 176, 191, 213- 215, 223, 237, 238, 274, 287, 308 Sulphuric acid, 22, 49-51, 61, 92- 96, 115-119, 165, 180 Sun, blue, 283 Sunset colors, 283 Supercooled fog, 281 Surface color, 203 combustion, 62 concentration, 130-132 tension, 127, 129-136, 166, 167, 186, 187, 189, 261, 262, 239 adsorption and, 135, 166, 167 and discontinuity, 132 concentration and, 130, 131 of solids, 136 of sols, 131 tensions, static and dynamic, 131-134 Swelling of jellies, 243, 251-256 pressures, 257 Syneresis, 236 effect of ions on, 244 Synthetic charcoal, 5 Tabaschir, 304 Tannin, 1, 167, 173, 188, 210, 224, 232, 260, 266 Tanning, electrical, 210 Tarragon vinegar, 266 Temperature and stability, 221, 222 effect of, on adsorption, 35, 9, 13, 29, 32, 113 Tensile strength of liquids, 74 Tension, surface, 127, 129-136, 166, 167, 186, 187, 261, 262, 269 adsorption and, 135, 166, 167 and discontinuity, 132 concentration and, 130, 131 of solids, 136 of sols, 131 Tensions, surface, static and dyna- mic, 131-134 Tents, leaking of, 71 Theories of catalysis, 40-44 Thickness of films, 25, 38, 82, 160, 189, 191, 314-319 Thorium salts, 16, 43, 45, 64, 80, 114, 169, 226, 227 Thunderstorms, 285 Titmice, 202 Titration determination of adsorp- tion, 93-97 Tree-frog, 201, 307 Trinitrotoluene, 275 Triple point, adsorption and the, 70 True and colloidal solutions, 144, , 183 Tuberculosis, 294 Tungsten, 42, 45, 153 Tyndall blue, 182 Tyndallmeter, 287 Tyndall phenomenon, 182, 195 Type of emulsion, determination of, 263 Types of emulsions, 261 of precipitates, 161-163 Ultra-filtration, 184, 185, 225, 234, 240 order of, 185 Ultramarine, 137, 308 Ultra-microscope, 182, 183, 264, 305 Ultra-violet light, 223, 278, 280, 311, 312 Unequal pressures, 154 Uranium carbide, 56 SUBJECT INDEX 345 Vacuum by adsorption, 9 Valence and adsorption, 112, 209, 213-219 Vapor, adsorption of, by liquid, 65- 67, 134, 135 density, adsorption and, 19 pressures of adsorbed gas, 25 of gelatine jellies, 75, 256 of gels, 236, 245-250 Vapors, adsorption of, by solids, 3-64 Velocities, migration, 206, 210 Vermilion, decomposition of, 79 Vinegar, Tarragon, 266 Viscosity, 140, 150, 190-193, 268 and catalysis, 43 Vitrification, 137 Voids, 156-158, 192 Voltage, air films and, 22, 23 Washing of air, 299 Water, color of, 202, 203 -in-oil, 241, 261-264 in solid solution, 249, 250 Water rings, 92 wings, 71 Weights, molecular, 141, 145, 187 Welding by gravity, 151 by pressure, 151 of glass. 149, 150 of powders, 151 Wells near beaches, 99 Welsbach mantles, 64 Wetting of smokes, 22 of solids, 23, 67, 69 pressures, 77 selective, 83, 85, 260 Whiskey, leaching of, 114 White lead, 72, 88, 204, 263 Whiteness, 198 Willemite, 15 Wings, water, 71 Woods, different, charcoal from, 6, 8 Wool, 18, 29, 30, 115-120 Xenon, 15 Zero fluidity, 156, 157, 190, 192, 240 14 DAY USE RETURN TO DESK FROM WHICH BORROWED BIOLOGY LIBRARY TEL. NO. 642-2531 This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall. LD 21A-15m-2,'69 (J6057slO)476 A-32 General Library University of California Berkeley UNIVERSITY OF CALIFORNIA LIBRARY