$ v\ a ..The... Quantitative Determination of Boric Acid in Tourmaline THESIS Presented to the Faculty of the Department of Philosophy of the University of Pennsylvania, for the Degree of Doctor of Philosophy BY GEORGE WILLIAM SARGENT BELLWOOD, PA. PHILADELPHIA AVIL PRINTING COMPANY 1898 This investigation was undertaken at the suggestion of Professor Edgar F. Smith, and I take this opportunity of sincerely thanking him for his encouragement and ready advice throughout the work. INTRODUCTION. Nothing of the constitution of this mineral is understood. In fact, exceedingly little is known concerning the structure of much less complex minerals. Structural formulas have been proposed, but these are conjectures only. The problem that confronts the chemist undertaking to throw light upon the constitution of any of the products of nature’s laboratory, is a very difficult one. No mode ol attacking the problem has ever been formed. Substitutions, syntheses and decompositions cannot be made as with organic bodies. There is no means of estimating the size of the molecule. The question of determining which ele- ments are linked together in many instances seems unan- swerable. Again, many elements are capable of existing in two conditions ; this complicates the problem still further. Before attempting to answer these questions, however, the composition of the body must be known. In a careful review of the literature on mineralogy, it will be noticed that in many cases some constituent of a mineral has been determined by difference, owing probably to the fact that no good direct method of estimating it existed at that time. Every chemist feels the necessity of directly determining any constituent. It is especially desirable that this be done in the analyses of complex substances. It may happen that many years have elapsed since the analysis ot the mineral was made. Since then the old methods prob- ably have been shown to be inaccurate, hence a redetermi- nation is desirable. Owing to the unreliability of estimating boric acid, Rammelsberg, in numerous instances, in the analyses of tourmaline, obtained it by difference. p 15744 4 Boric acid constitutes from 6 per cent to 12 per cent ot this mineral. Silica and alumina are its common associates from which it is separated with great difficulty. Since the time of Rammelsberg, many methods have appeared, but few have been found applicable to insoluble borates or borosilicates ; with tourmaline, they have proven tedious or- unreliable. A chronological review of the various methods for the determination of this acid may be interesting. At least, it will show the condition in which the quantitative determination of boric acid rests at the present time. | Historical Resume. Probably the first work that was done toward the quan- titative determination of boric acid, was that of Arfvedson (K. Vetensk. Acad. Handb. 1822 , 22 and Schweigg. Jour. 8, f). By heating a mixture of borax and calcium fluoride with sulphuric acid, he volatilized the boron as fluoride ; the sodium sulphate resulting from this treatment was dissolved out with water, evaporated to dryness, weighed and the boric acid calculated. Berzelius ( Lehrbuch 3, Aufl. 84; Poggendorff' s Annalen 1824 :, 2, 1 18) attempted the determination by precipitating boron as potassium borofluoride. Berzelius also used Arfvedson’s method with success, but in place of the calcium fluoride and sulphuric acid, he took hydrofluoric and sul- phuric acids. In 1828, Menil ( Jahresber . der Chemie undPhysik 2, j6J) endeavored to estimate boric acid by means of the silver salt, 3Ag 2 0,B 2 0 3 , but met with little success. Gay Lussac ( Annales de Chew, et de Physique 40, jg8) in the year 1830, called attention to the titration of borax by a sulphuric acid solution with tincture of litmus as the indi- cator. The reaction is : Na 2 B 4 0 7 -f-H 2 S 0 4 = Na 2 S 0 4 -f 2 B 2 0 3 . The liquid changes color only when free sulphuric acid is present. This was the first volumetric method proposed. 5 Rose {Poggendorff' s Annalen 80, 261) was the first to volatilize boric acid as the ethyl ester and obtain its amount by difference. This was in 1850, but as early as 1732, Claude Geoffrey knew that boric acid imparted a character- istic green color to the alcohol flame and in 1818, Stro- meyer {Poggendorff' s Annalen 61, 179) mentioned the vol- atility in an alcoholic solution and the characteristic green color given to the burning liquid. Rose, in his paper, gave a second method, in which a weighed amount of sodium carbonate was added to the borate, evaporated to dryness and ignited until no further evolution of carbon dioxide. The residue of sodium carbonate and boron trioxide was weighed and from this the boric acid calculated. Berzelius’ method gave poor results, owing to the solu- bility of the precipitate of potassium borofluoride in alcohol,, with which it was washed to free it from potassium fluoride. W eber {Poggendorff ' s Annalen , 80, 276), 1850, tried modi- fying the method of Berzelius, in order to avoid the formation of the large quantity of fluoride, but did not succeed. Rammelsberg {Poggendorff ' s Annalen , 1850, 80, 4.66) found that if the alcohol used for washing the precipitate was free from water, potassium fluoride contaminated the precipitate, and that alcohol containing water dissolved some of the potassium borofluoride. Schweitzer {Pharm. Centralbl. Nr. 24, jpo, and Fresenius' Anleituug zur quantit. Analyse. 6 Aud. y 1873, 1, 424) esti- mated the base and determined the boric acid by difference, by evaporating the base with hydrochloric acid until all the acid was expelled and estimating the chlorine as silver chloride. In 1856, Kraut {Hennebergs Jour, fur Landwirtsch. 4, 1 12 ; abs. by Zeit. fur analyt. Chemie 1863 , 73) showed that a borax solution boiled with ammonium chloride, reacted according to the following equation : Na 2 B 4 0 7 T 2NH,CI"= 6 2NaCl+2B 2 0 3 -f 2NH 3 , and the ammonia could be caught in a standard acid solution, and thus the boric acid determined. This same year, the method of Stromeyer (Liebig* s Anna- len , ioo, 82) appeared. He added to the solution of the borate two equivalents of potassium hydrate to one equiva- lent of boron trioxide contained in the borate, then an excess of hydrofluoric acid and evaporated to dryness. The residue he dissolved in a 20 per cent solution of potassium acetate, and filtered through a weighed filter, using a gutta percha funnel. The precipitate was washed first with a solution of potassium acetate to remove potassium fluoride, then with alcohol, dried at ioo° C. and weighed. Upon applying this method to a silicate, Stromeyer obtained 97.5 per cent of all the boron trioxide. Where the boric acid content was approximately known, Schafifgotsch [Poggendorff' s Anua/en,i8y^ t 107, 42 f) added to the solution a weighed amount of sodium carbonate, in which not less than one and not more than two equivalents of sodium oxide, Na 2 0,to two equivalents of boron trioxide were contained. After evaporation to dryness and ignition, the residue was sodium oxide and boron trioxide, the weight of which less the weight of the sodium oxide taken, gave the boron trioxide. This method is essentially the same as that of Rose. Marignac (Zeit. fur analyt. Chemie 1862 , 1 , 405) in a solution containing alkali salts only, estimated the boric acid thus : After the addition of magnesium chloride, am- monium chloride and ammonia, he evaporated the solution to dryness in a platinum dish. The residue after ignition was taken up with water and the insoluble portion removed : the filtrate was made ammoniacal, evaporated to dryness, ignited, the residue taken up with water and the insoluble portion filtered off as before. This was repeated three times. The insoluble portions were ignited and weighed as magnesium pyroborate ; the magnesia was determined by dissolving 7 the pyroborate in^nitric acid and precipitating as magnesium ammonium-phosphate, and from this the boric acid was obtained by difference. Insoluble borates were fused with sodium carbonate, the melt extracted with water and the solution acidified with hydrochloric acid; ammonia was then added and the whole concentrated to a small volume. Any precipitate that formed was filtered off : any lime present was removed as oxalate. The solution was then supposed to contain the borate together with alkali salts and was ready for the treatment with magnesium chloride solution. In regard to this method, Marignac himself says : “ If good results are obtained, it is due to a balancing of errors.” 'R.osz^Handb. der analyt. Chemic 6 Aufl. 2, 721), 1871, substituted sodium metaborate for the sodium carbonate used in his earlier method. By fusing boric acid with calcium chloride in the pres- ence of sodium and potassium chlorides, Ditte ( Comptes rendus , 1873, 80, 490, a?id 561) obtained the salt Ca 0 ,B 2 0 3 which crystallized from the fusion in needles and was insolu- ble in cold water. He utilized this salt for the estimation of the acid. In 1877, Berg (Zeit. fur analyt. Chemie 22, 25) precipi- tated from an alcoholic solution Ba 0 ,B 2 0 3 4 H 2 0 . The precipitation was fairly complete, enabling him to determine boron trioxide in borax and obtain results within .1 per cent to .3 per cent of the theoretical amounts. Smith {Am. Chew. Jour. 1882-83 . 2 79 ) precipitated manganese borate, MnO, 2 B 2 0 3 , from a borax solution in the presence of alcohol and determined the excess of man- ganous sulphate added, by evaporating the filtrate from the manganese borate to dryness, taking up with water and titrating the manganese according to Volhard. In applying this method to a silicate, the latter was fused with sodium carbonate, the melt extracted with water 8 and the solution digested with ammonium sulphate to throw out silica and alumina. After removing these, a known amount of manganous sulphate solution with an equal volume of alcohol was added ; the precipitate was removed and the manganese in the filtrate determined as above. In a tourmaline, the percentage of boric acid found by Marignac’s method was io per cent, while that found by Smith’s method was 9.7 per cent. Bodewig (Zeit fur anaylt. Chemie 1884., 23, /^p) tried the method of Smith upon a silicate, but obtained no result. With some success, Bodewig used Stromeyer’s method in the determination of the free acid, but upon applying it to a silicate, he experienced considerable difficulty in obtaining a solution of the borate free from silica. To accomplish this, the procedure of Berzelius (use of an ammoniacal zinc oxide solution) was resorted to with a fair degree of success. The methods of Gooch and Rosenbladt appeared sim- ultaneously in 1 887. Gooch (Am. Chem. Jour. 9, 2j ) evaporated the solution of free boric acid to dryness with a weighed amount of lime and ignited to constant weight. The increase in weight represented boron trioxide. Rosen- bladt (Zeit. fur analyt. Chemie 26, 21) used magnesia in place of lime. Both Gooch and Rosenbladt showed that boric acid could be completely expelled from the concentrated acid solution of a borate as the methyl ester. To accomplish this, Gooch used an apparatus consisting of a large pipette bent at one end of the bulb at a right angle and at the other like a goose neck ; the end of the pipette extended into a con- denser. Insoluble borates were fused with sodium carbon- ate; the melt was extracted with water and the solution evaporated to dryness in the bulb of the pipette, which was immersed in a paraffin bath. The residue was made acid with nitric or acetic acid and the distillation with methyl 9 alcohol begun ; the methyl alcohol was added in portions of io c. c. The distillate was poured into a platinum dish, containing a weighed amount of lime, evaporated to dry- ness and ignited to constant weight ; the increase in weight represented boron trioxide. The apparatus used by Rosenbladt was more simple than that used by Gooch. It consisted essentially of an Erlen- meyer flask connected with a condenser. Rosenbladt’s method differed from Gooch’s only in that the ester was caught in an ammonium carbonate solution then poured into a dish containing a weighed amount of magnesia and evaporated to dryness. Morse and Burton ( Am . Chem. Jour. 1888, 10, 154) to a concentrated borax solution of small volume, added sul- phuric acid to acid reaction using tropaeoline 00 as the indicator. After the addition of anhydrous copper sulphate, the resulting mass was extracted with absolute alcohol ; this alcoholic solution was run into an excess of standard barium hydrate. The excess of baryta was changed to carbonate, the whole evaporated to dryness and weighed as barium carbonate and barium metaborate, from which the boric acid was calculated. Insoluble borates were decom- posed by fusion with sodium hydrate in a nickel crucible. The extract of the fusion was evaporated to a small bulk and treated as the borax solution. By this procedure, Morse and Burton obtained from a tourmaline the follow- ing percentages of boron trioxide : 10.03 P er cent, 10.08 percent and 10. n per cent. Dr. Riggs obtained from the same tourmaline by a different method 10.15 per cent, 10.00 per cent and 10.31 per cent. By means of a solution of baryta, Will (Archiv. der Pharmacie 225, iioi> abs. by Zeit.fur analyt . Chemie i88g , 28, 100) titrated boric acid. Kriiss and Moraht {Liebig's Annalen 18 go, 259, 184) in the analysis of beryllium borate, gave the method of Stromeyer preference. IO To the salt of the alkali metal, Parmentier ( Comptes rendus 1891, 113, 41) added an excess of hydrochloric acid or sulphuric acid. After dividing the solution into two equal parts, he added to one methyl orange and titrated the excess of hydrochloric or sulphuric acid : to the other, litmus was added and the total acidity determined. The difference gave the boric acid. Reischle {Zeit. fiir anorg . Chemie 189J , 4, ///) stated, that the color reaction of the litmus was so indistinct that he could obtain no results. In 1893, Thompson {Jour. Soc. them. Ind . 12, 432) noticed that a boric acid solution containing 30 per cent of glycerol could be titrated with sodium hydrate with phe- nolphthalein as the indicator. Starch, glucose and cane sugar could be substituted for glycerol. In borax the acid was set free by hydrochloric acid using methyl orange as the indicator and the boric acid titrated by a solution of sodium hydrate with glycerol and phenolphthalein. Thompson applied this method to boracite. As the commercial method of estimating boron trioxide in boronatro calcites, Le Roy {Jour. Soc. Chem . Ind. 1893 , 867) gave the following: Decompose the mineral with dilute sulphuric acid and hydrochloric acid in a flask con- nected with a reflux condenser: remove the insoluble residue and precipitate the iron and aluminium as hydrates by caus- tic soda. After the removal of these, the solution is made acid with hydrochloric acid and boiled to expel carbon dioxide ; upon cooling, it is made up to a known volume. Twenty cubic centimeters of this solution are titrated with standard alkali and Porrier orange ill, until the shade produced corresponds to the yellowish red caused by the same amount of orange 1 1 1 in 20 c.c. of water. Another volume of 20 c.c. is now titrated with alkali using orange 1 1 as the indicator, until the color is a dark red. The dif- ference between the two titrations indicates the boric acid present. Schwarz ( Pharm . Zeitung 1894., 32, 362) suggested Congo Red as an indicator to be used with hydrochloric acid. Hefelmann (Pharm. Centralhalle (N. P) 1894, 9, 116) gave litmus the preference. Hefelmann recommends the expulsion of the boric acid by heating with ammonium fluoride and the estimation of it by difference. Barthe (Jour. Pharm. Chem. 1894 , 29, 163) used the method of Thompson with success as did also Honig and Spitz (Zeit. fiir angewandte Chemie 1896 , jjo). Insoluble silicates were fused by the latter with sodium and potassium carbonates. The melt was dissolved in water and as much ammonium chloride added as equaled the carbonates used in the fusion. The solution was then boiled and the pre- cipitated silica removed. The last traces of silica were removed by Berzelius’ method. After concentration to a small volume, the liquid was made acid with hydro- chloric acid, boiled a few minutes to expel carbon dioxide and titrated. By this procedure, the percentage of boron trioxide found in a “ Gasgliihlicht-cylinder ” was 5.12 per cent, while that found by difference was 5.34 per cent. Honig and Spitz also applied the reaction made use of by Kraut — Na 2 B 4 0 7 -f 2 NH 4 C 1 = 2NH 3 -f- 2NaCl -f- 2 B 2 0 3 — to “ Boraxkalk.” The results obtained differed .10 per cent to .30 per cent from those got by their first method. In 1897, Kraut (Zeit. fur analyt. Chemie 36, 165) em- ployed Gooch’s method successfully for the analysis of colemanite and pandermite. The apparatus used in the distillation of the ester was, however, more simple than that of Gooch. Schneider and Gaab (Pharm. Centralhalle 1897, 37, 672) distilled the boric acid with alcohol and evaporated the dis- tillate with a weighed amount of sodium carbonate. 12 By shaking with ether, Beliocq (Rev. Int. falsific. 9, 119 , abs. by Zeit. fur anorg. Chemie 1897, 380 ) removed boric acid from other salts, and upon the evaporation of the ethereal extract obtained the trioxide. Thaddeeff (Zeit. fur analyt. Chemie 1897, 36, 368) applied the Berzelius-Stromeyer method, somewhat modified, to borax. His modification consisted in adding to the potas- sium acetate solution containing the potassium borofluo- ride, 100 c.c. of alcohol .805 sp. gr. and allowing it to stand twelve to fourteen hours before filtering. The pre- cipitate was washed with alcohol of the same specific gravity. By this means, results varying from .08 per cent to .40 per cent of the theoretical were obtained. Thad- deeff also volatilized boric acid as the methyl ester which was caught in a solution cf potassium hydrate. From this potassium borofluoride was precipitated. To facilitate the volatilization of the ester, Thaddeeff used a current of air. By this procedure, he obtained results varying from .01 per cent to .40 per cent of the theoretical. INVESTIGATION. From this review, it is apparent that, while boric acid in borax, soluble borates and minerals not requiring fusion with alkalies for their decomposition, is estimated with a fair degree of success, this acid, where fusion is required, is not determined with the desired accuracy. Minerals containing silica and alumina usually require fusion with sodium carbonate or caustic soda for their decomposition. The extract from this fusion is supposed to contain all the boric acid together with silicate and aluminate of soda. Digestion with ammonium chloride or sulphate is generally resorted to for the removal of the latter two. The solution is then ready for treatment according to any method applicable to borax. 13 The methods of Gooch, Stromeyer and Marignac have been universally used. If Marignac’s method is adopted, results are obtained about which there is more or less uncertainty. If Stromeyer’s method is used, the potassium borofluoride weighs more than it should, owing to fluosili- cate ( Fresenius ’ Quant. Chem . Anal. 424). To avoid this silica, Wohler [Handb.der Mineral Analyse , under Datholite) recommends evaporating the hydrochloric acid solution of the fusion to dryness, in a flask connected with a condenser, adding the distillate to the residue and filtering off the silica. By this means, silica is entirely removed, but alumina and other bases are yet to be separated, and in the removal of the alumina as hydrate, there is a tendency on the part of the precipitate to retain boric acid ( Wohler , Ann. der Chem. und Pharm. 141, 268). In the use of Gooch’s method the trouble is met in weighing the lime. It is evident from the foregoing, that the quantitative determination of this acid, where it exists in such combina- tion as it does in tourmaline, is not attended with the most desirable results. With the hope of obtaining a more accurate and more rapid method and of casting some light, perhaps, upon the constitution of this mineral, this investi- gation was undertaken. The isolation of the acid is evidently necessary for its successful determination. Its separation from these two associates, silica and alumina, is difficult, and alkalies apparently increase this difficulty; hence it was thought, if the tourmaline could be decomposed by heating with metallic magnesium or fusion with carbonates other than those of the alkalies, the isolation of the boric acid would be more readily accomplished ; or if the tourmaline was fused with alkali carbonates, possibly it could be completely removed from these objectionable associates in an acid solution. Since the volumetric method of Thompson seemed to offer the most rapid and accurate means, I decided to investigate it, with the view of utilizing it for the determi- nation of the isolated boric acid. Determination of Boric Acid in Borax by the Method of Thompson. One-tenth normal hydrochloric acid solution was pre- pared and standardized by precipitating the chlorine with silver nitrate. No. i, 10 c.c. HC 1 sol. gave .1430 gram AgCl = .035358 gram Cl. No. 2, 10 c.c. HC 1 sol. gave .1440 gram AgCl = .035605 gram Cl. Average = .035481 gram Cl. Theory = .03545 “ “ Difference = .000031 “ “ From this hydrochloric acid solution a tenth normal caustic soda solution was prepared. A borax solution, 10 c.c. of which contained 1 gram of Na 2 B 4 0 7 , ioH 2 0, or .03659 gram of B 2 0 3 , was prepared from recrystallized borax. As nearly all glycerol is likely to be acid owing to fatty acids, it was found best to add a very small amount of water to the glycerol, then phenolphthalein, shake well, and introduce sodium hydrate until a faint pink tinge appeared. This neutralized glycerol was kept in a well stoppered bottle. For the estimation of boric acid in the latter, a number of c.c. of the borax solution were introduced into an Erlenmeyer flask, a few drops of methyl orange added and tenth normal hydrochloric acid run in from a burette until 15 all the boric acid was free : 50 c.c. of the glycerol were added, together with a few drops of phenolphthalein, and the titration with tenth normal caustic soda begun. Accord- ing to the reaction : 2 B 2 0 3 -f- 4NaOH = 4 NaB 0 2 + 2 H 2 0 , 1 c.c. of tenth normal alkali is equivalent to .0035 gram of B 2 0 3 . The following table shows the value of this method : Taken. Found. Borax sol. B0O3. N — NaOH B2O3. No. C.C. Grams, C.C. Grams. Per Cent. I ... 10 .03659 IO.4 •0364 36.40 2 ... 10 .03659 10.5 .03675 36.75 3 ... 10 .03659 IO.4 .3640 36.40 4 ... 20 .07318 21.0 •0735 36.75 5 ... 20 .07318 20.9 •07315 36.57 6 ... 25 .O9I47 26.1 •09135 36.54 7 ... 25 .09147 26.1 •09135 36.54 8 ... 15 .O5488 15*7 •05495 36.63 9 ... 15 .O5488 157 •05495 3663 10 ... 15 .05488 15.6 .0546 36.40 The atomic weights used were : Oxygen — 16.00 Boron = 10.95 Silver = 107.97 Chlorine = 35.45 Sodium = 23.05 The theoretical percentages of the constituents of crys- tallized borax are : Na 2 0 = 16.26 per cent. BA = 36.59 “ H 3 0 = 47-15 “ Having proved the volumetric method of Thompson to give reliable results with borax, the problem of isolating the boric acid in tourmaline was taken up. i6 Decomposition of Tourmaline by Heating with Metallic Magnesium. Two-tenths of a gram of brown tourmaline (No. 2) was heated to expel the water, mixed with half a gram of mag- nesium powder, the whole placed in a porcelain crucible and covered with a thick layer of thoroughly dried salt. It was then heated for thirty minutes and allowed to cool. The contents, which had the appearance of amorphous sili- con, together with any parts of the crucible to which the con- tents adhered, were powdered, placed in a porcelain boat and chlorine passed over the same. The chlorine and any volatile portion were caught in two U tubes containing water ; later, heat was applied and gradually raised to a dull red, where it was held until the contents of the boat became quiescent. The heat was then gradually withdrawn. After four days, most of the chlorine had gone from the U tubes, leaving the contents exceedingly acid. Sodium carbonate suffi- cient to neutralize the acid was added, and the solution evaporated almost to dryness. After making acid and expelling the carbon dioxide by drawing air through the solution for fifteen minutes, the excess of acid was neutral- ized by sodium hydrate with methyl orange as the indicator, and the boric acid titrated with the following results : Weight of Mineral. — Na OH. B 2 0 3 . PerCent. 10 No. Grams. C. C. Grams. 1 2000 5.0 .0175 8.75 2 2000 2.5 .00875 4-357 3 2000 1.2 .0042 2.10 4 2000 20 .007 3.50 5 .... . .2000 1.5 .00525 2.625 Number four was allowed to stsnd two days before treat- ing the contents of the U tubes. Very little acid was present. Numbers two, three and five were allowed to stand but a i7 few hours before treating. Little acid was present and the chlorine was removed in each instance before the addition of the sodium carbonate, by drawing air through the tubes. It might be added, that the time of heating the mineral with the magnesium powder was varied from fifteen minutes to an hour, and in every case the contents of the crucible had the appearance of amorphous silicon. This procedure is evidently not a success; just why, I do not know, but am inclined to believe that the boron chloride was not completely broken up when it came in contact with the water saturated with chlorine, but was carried on out with the escaping gas. It may have been that the mineral was not completely decomposed by the magnesium, but from all appearances the magnesium had entirely reduced it. At any rate, this procedure was abandoned, and the fusion of the mineral with the alkali-earth carbonates under- taken. Decomposition of Tourmaline by Fusion with Calcium Carbonate. Some of the very finely powdered mineral was fused with a mixture of eight times its weight of precipitated calcium carbonate and once its weight of ammonium chloride, at a low red heat for one hour, as in the J. Lawrence Smith fusion for the alkalies ; when cool the fusion was powdered, placed in a platinum dish, covered with water and an amount of sodium carbonate introduced slightly in excess of that required. After digesting for half an hour the solu- tion was allowed to cool and the insoluble portion was removed. To the filtrate sulphuric acid to almost neutral reaction was added. The precipitate which formed was fil- tered off and the solution made slightly acid. For fifteen minutes air was drawn through the solution to remove the carbon dioxide, after which the excess of acid was i8 neutralized with caustic soda, with methyl orange as the indicator, and the boric acid titrated. In this instance half-normal sodium hydrate was used in the titration of the boric acid. One cubic centimeter of a half-normal alkali solution is equivalent to .0175 gram of boron trioxide. The results are shown in this table : Weight of Mineral (No. 2)- — NaOH. B 2 0 6 . 2 No. Grams. C. C. Grams. Per Cent. 1 3OOO 2.4 .0420 I4.OO 2 3OCO 2.8 .O49O 16.33 3 3000 3.7 .06475 21.58 The decomposition of the mineral was complete, because all the portion insoluble in water, except silica, dissolved in hydrochloric acid. Upon allowing the titrated solutions to stand, in every case a precipitate of alumina separated. Thinking that during the boiling of the fusion with soda some alumina was dis- solved, I used barium carbonate in the place of the soda and decomposed the fusion by boiling with sodium sulphate, with the hope of avoiding the solution of alumina. That this was not a success is shown by the following : Weight of Mineral (No. 2). — NaOH. B 2 0 3 2 No. Grams. C.C. Grams. Per Cent. 1 3000 1.5 .02625 8.7 5 2 3000 3.3 .05775 I9.25 3 3000 4.3 -07525 25.08 4 3000 2.6 -0455 *5.16 5 3000 2.1 .03675 12.25 6 3000 1.85 .3237 10.79 In every instance, after titration, a copious precipitate of alumina came down upon boiling with hydrochloric acid and adding an excess of ammonia. It is very likely that in the digestion of the fusion with sodium sulphate, barium carbonate becomes barium 19 sulphate and sodium carbonate is formed, this then dis- solves the alumina. I thought that possibly this alumina might be eliminated completely by adding a small quantity of ammonium sul- phate to the sodium sulphate solution free from the insolu- ble portion, and digesting with a reflux condenser; the ammonia could easily be expelled by boiling with an excess of alkali. This was done and the following results obtained : N Weight of Mineral (No. 2 ). — NaOH. b 2 0 3 . No. Grams. C. C. Grams. Per Cent. 1 5000 2.6 .0455 9.10 2 5000 2.25 .O3937 7.87 3 5000 2.00 .0350 7.00 Upon testing after the titration, alumina was found. The insoluble portions of some of these fusions were fused with sodium carbonate and the boric acid therein, determined by distillation with methyl alcohol — a method which will be given later — and the following amounts found : N Weight of Mineral (No. 2 ). 2 B 2 0 3 . No. Grams. C. C. Grams. Per Cent. 2 3000 1. 1 .OI925 6.41 3 3000 0.6 .0105 3.50 1 5000 1.0 .0175 3.50 2 5000 1. 1 .01925 3.85 3 5000 1.4 .0245 4.90 From this it appears that the decomposition of tourma- line by fusing the mineral with calcium or barium carbonate does not succeed ; nor is it a success with datolite, as the following show : N Weight of Mineral. 2 B 2 0 3 . No. Grams. C. C Grams. PerCent. 1 5000 4.6 .0805 l6.IO 2 30C0 2.8 .O49O 16.33 20 The decompositions were complete, but some boric acid had remained with the insoluble portions. Tourmaline to be completely decomposed must be fused with alkalies. Then if boric acid is to be cleanly removed from the alumina, it must be done in an acid solution. The methods of Gooch and Rosenbladt are the only means of accomplishing this and directly esti- mating the boric acid ; but the process as practiced by Gooch or Rosenbladt is rather slow. Thaddeeff lessened the time necessary for the complete removal of the boric acid as the methyl ester, by using a current of air. By utilizing the discovery of Gooch and Rosenbladt and the volumetric method of Thompson, I thought a rapid and accurate method could be devised. The Volatilization of the Boric Acid as the Methyl Ester and Subsequent Titration. To diminish the time required for the complete volatilization, this apparatus, the plan of which is here shown, was constructed. A weighed amount of fused borax was placed in the bulb, moistened with sulphuric acid, io c. c. of commercial wood alcohol added and the distillation begun. The distillate was caught in the flask B, which contained 25 c. c. of tenth normal caustic soda, into which the condenser tube extended about one inch. When nearly all the alcohol had distilled over, air was drawn through the whole apparatus for a few minutes, by attaching a suction pump to A and opening the stopcock at D sufficiently far to let the air 21 gently bubble through the solution in B. This operation was repeated until 50 c. c. of wood alcohol had been added and distilled off into the sodium hydrate. The flask was disconnected, 25 c. c. of tenth normal hydrochloric acid were introduced, then 100 c. c. of glycerol with a few drops of phenolphthalein and the titration of the boric acid with tenth normal alkali begun. By this procedure the following results were obtained : No. Taken Weight of Borax. Grams. b 2 o 3 . Grams . Found N — NaOH. 10 C. C. b 2 o 3 . Grams. Per Cent. I . . . . .1000 .O6924 23.OO .0805 80.50 2 . . . . .1000 .O6924 22.80 .0794 79.80 The theoretical percentages of the constituents of fused borax are : Na 2 0 .... 30.76. B 2 0 3 .... 69.24. The amount of alcohol in each case being the same and the. results obtained agreeing so closely, I concluded that the alcohol was the cause of the high percentages. Upon evaporating the distillate obtained as above and igniting, a black charred mass was left, which, however, was easily burned. This showed either that some organic compound had been formed in the distillation of the alcohol from the sulphuric acid or existed as an impurity in the alcohol and was carried over during the distillation. If the substance were acetic acid or formic acid, the sodium salt would be formed, which when treated with hydrochloric acid would regenerate the acid and thus increase the amount of sodium hydrate required. To ascertain whether a formate was produced, some pure methyl alcohol was distilled with sulphuric acid. The dis- tillate which was caught in caustic soda, was concentrated, the alkali neutralized with nitric acid and silver nitrate added. The odor of formic acid was distinctly perceptible 22 upon neutralizing the sodium hydrate, and upon adding the silver nitrate and boiling, silver separated. When the ester comes in contact with the alkali, the probable reaction is : B(0CH 3 ) 3 +Na0H + H 2 0 =B 00 Na-l-CH 30 H. The heat- ing of methyl alcohol with sulphuric acid in the air seems to react thus : CH 3 OH+ 0 2 =H 2 0 + HC00H. To destroy this organic acid, the distillate was evaporated to a small volume, carbon dioxide rapidly run through to insure the formation of borax, the evaporation then con- tinued to dryness and the residue ignited until all the car- bon was consumed. When cool, hydrochloric acid was added in slight excess and the carbon dioxide removed by drawing air through the solution for a quarter of an hour. The excess of hydrochloric acid was neutralized and the boric acid titrated with half normal sodium hydrate. Taken. Found. Weight of Borax. b 2 o 3 . E.Na 0 H. 10 b 2 o 3 No. Grams. Grams. c.c. Grams. Per Cent. I . . . .2000 .13848 9.0 •1575 78.7s 2 . . . .2000 .13848 8.2 •1435 71-75 3 * . . .1000 .06924 4.0 .0700 70.00 These results are too high and variable, so pure methyl alcohol was substituted for the commercial article and about 60 c.c. were used each time. The boric acid was titrated in this instance with tenth normal caustic soda. It will be noticed that sometimes tenth normal alkali was used in the titration, and at other times half normal. Either gave good results in the titration of boric acid in borax, but with half- normal sodium hydrate, the end reaction was more pro- nounced. Taken. Found. Weight of Borax. b 2 o 3 . E NaOH. IO b 2 o 3 . No. Grams. Grams . c.c. Grams. Per Cent. I . . . .5000 .3462 97-5 •34125 68.25 2 . . . .2000 .13848 39-5 00 eo 69.13 3 • . . .3OOO .20772 58.8 .2058 68.60 23 According to these reactions : 4B(OH) 3 -}-4NaOH= 4 B 0 , 0 Na-|- 8 H 2 0 and 4BO,OH+4NaOH=4BO,ONa+ 3H2O, the treatment with carbon dioxide was unnecessary and was dispensed with in the following determinations : Taken. Found. Weight of Borax. B2O3. NaOH. 10 B2O3. No. Grams. Grams. C.C. Grams. Per Cent. I . . . .2000 .13848 39 - 8 •1393 69.65 2 . . . .3000 .20772 594 .2079 69.30 3 * . . .4000 .27696 79-3 .2775 69.36 4 • . . .2000 .I3848 39 - 6 .1386 6930 The theoretical per cent of acid in fused borax is 69.24. This shows that this method is a good one, applied to borax. The next thing was to apply it to tourmaline. A brown tourmaline, marked (No. 2), from McAfifee, New Jersey, on analysis gave the following : Silica Per Cent. 33-72 Boric acid .... Alumina Ferric oxide . . . 1.82 Ferrous oxide . . 323 Lime Magnesia .... 14.07 Loss on ignition . . Potassium oxide . . 20 Sodium oxide . . . In this instance, the boric acid was estimated according to Marignac’s method. This tourmaline was fused with potassium and sodium carbonates : the fusion was taken up with water, evaporated almost to dryness, then transferred to the distilling bulb together with the insoluble oxides. The distilling bulb 24 instead of resting on an asbestos pad, was immersed to half the depth of the bulb, in a glycerine bath. The tem- perature of the bath was raised to 135? C. and the contents of the flask rapidly evaporated to dryness by the aid of a current of air. This required about an hour. The glyce- rine bath was allowed to cool down to 50° C. and a flask containing sodium hydrate was substituted for the one pre- viously used to catch the distillate. The contents of the bulb were made decidedly acid and the distillation with methyl alcohol commenced. The remainder of the opera- tion was the same as with borax. The following table shows the results oi this method : Weight of Mineral (No. 2). N — NaOH. 10 B2O3. >. Grams. C.C. Grams. Per Cent. Ol O O O 16.0 .0560 ( I 1.20) 5000 16.0 .O56O (ll.20) 31.0 VO 00 O (10.85) I. OOOO 22.4 .0784 (7-84) I. OOOO 27.4 .0959 9-59 5000 I3.6 .0476 9.52 5000 137 •04795 9-59 In numbers 5, 6 and 7, the fusion was allowed to cool completely and by rolling the crucible between the fingers with gentle pressure, the fusion dropped out. It was placed in the distilling flask, the crucible rinsed with sul- phuric acid (1 : 1) and the washings added. Sulphuric acid and water sufficient to break the mass up into a mushy state, were poured over it and the distillation with methyl alcohol conducted as with the others. By the procedure used in these latter determinations, the entire time consumed was about five hours. A black tourmaline which gave according to Marignac’s method 9.87 per cent of boric acid, treated as the above gave the following : 25 Weight of Mineral. — NaOH. 2 B 2 0 3 . No. Grams. c.c. Grams. Per Cent. i 5coo 2.9 .05075 10.15 o o o 1-0 2.9 .05075 10.15 From datolite, the following percentages were obtained : Weight of Mineral. F NaOH 2 B2O3. No. Grams. iC.C Grams. Per Cent. I 5000 5.50 .O9625 19.25 2 5000 5*55 .097125 1943 Another sample of a brown tourmaline marked (No. 4) gave these results : Weight of Mineral. ? NaOH 2 B2O3. No. Grams. c.c. Grams. Per Cent. I . . . .5000 2.80 .O49O 9.80 2 . . . .5OOO 2.90 .05075 10.15 3 * • • -5000 2.85 .04988 9976 4 . . . 1. 0000 5.70 .09975 9-975 In each of the preceding determinations the residue in the distilling bulb was tested for boric acid, but none was found. These results show this to be a good method of estimating boric acid wherever it is associated with silica and alumina. The time consumed is considerably less than that required by other methods, and the complete isolation of the boric acid is accomplished. The construction of the apparatus required by this method is an objection, a very slight one, however. It occurred to me that possibly a more simple method might be found in that proposed by Smith in the Am . Chew. Jour., Vol.^,p.2yg. With this in view I took up the method of Smith for investigation. 26 Estimation of Boric Acid by the Method of Smith. A borax solution, I c. c. of which contained .004 gram of Na 2 B 4 0 7 , a potassium permanganate solution containing zinc sulphate, 1 c.c. of which was equivalent to .0018705 gram of manganese, and a manganous sulphate solution containing .005228 gram of manganese in each c. c., were prepared. To a volume of the manganous sulphate solu- tion, a known portion of the borax solution, with an equal quantity of alcohol, was added. As soon as the precipitate settled it was filtered off by means of a Gooch crucible and a suction pump. The filtrate containing the excess of man- ganous sulphate was evaporated to dryness, the residue gen- tly ignited and when cool dissolved in water. A few drops of a saturated solution of sulphur dioxide were added to aid in dissolving the residue. The liquid was then transferred to a flask and brought to boiling. The titration of the manganese according to Volhard’s method was now made, and from this the boric acid calculated by the reaction MnS 0 4 + Na 2 B 4 0 7 = MnB 4 0 7 + Na 2 S 0 4 . Borax Sol. Grm. of MnS 0 4 Sol. Grm. of C. C. of Excess Per cent Vol. of Al. No. C. C. Borax . c. c. Mn. KM11O4. of Mn. b 2 o 3 . C. C I . . . IO .04 5 .02614 8.0 .OI4964 71.03 15 2 . . . IO .04 5 .92614 8.0 .OI4964 71-03 15 3 • . . IO .04 5 .02614 8.2 •015338 68.33 IO 4 . . . IO .04 10 .05228 22.1 •041338 68.9I IO 5 • . . 15 .06 10 .05228 18.9 •035354 71.71 15 6 . . . 15 .06 10 .05228 19.2 • 0359 H 69.38 15 7 • . . 25 .10 15 .O7842 27-3 .05 IO65 69.54 20 8 . . . 25 .10 15 .O7842 27-3 .051065 69.54 20 9 • . . 25 .10 25 .13071 55-3 .10344 69.30 25 10 . . . 20 .08 20 .10456 43-5 . . . (67.68) 20 The precipitate of Mn 0 , 2 B 2 0 3 was dissolved off the filter by a warm aqueous solution of sulphur dioxide, transferred to a flask, brought to boiling and the manganese then 27 titrated with permanganate of potash according to the method of Volhard. This was done with each precipitate with the following results : No. KM11O4. c. c. Grams Mn with 2B2O3. Per Cent b 2 o 3 . Total Mn determined. Difference from amount taken. I . . . 6.0 .011223 7 *- .026187 .000046 plus. 2 . . . 6.0 .011223 7 I - 3 I .026l87 .000046 plus. 3 • • . 5<8 .OIO849 68.94 .026187 .000046 plus. 4 • • . 5.9 .OIIO36 70.1 1 .052374 .00005 2 plus. 5 • • . 9.0 .OI6834 7 i. 3 i .052188 .000094 minus. 6 . . . 9.0 .OI6834 7 i. 3 i .052188 .000466 plus. 7 • • . 14.5 .027122 68.94 .O78187 .000236 minus. 8 . . • 147 .O27496 69.89 .O78561 .000138 plus. 9 • • . 14.6 .027310 69.41 0 <-n 0 .000045 plus. 10 . . . 15.0 .028057 7 I- 3 1 . . . . . . This shows that boric acid in borax may be successfully estimated either by the titration of the excess of manganous sulphate, which is Smith’s method, or by titration of the manganese combined with B 2 0 3 . For the estimation of boric acid in a tourmaline, Smith digested the extract of the sodium carbonate fusion with an amount of ammonium sulphate equivalent to the car- bonate. The precipitate that formed was removed and the filtrate concentrated to 25 c. c., after which it was treated as the borax solution. This I did, but upon adding, the manganous sulphate obtained no precipitate, and upon test- ing with litmus, found the liquid acid. This corroborates Bodewig’s statement in the Zeit.fur analyt. Chemie } Vol. 23, p. 14.3. Another fusion was made and the extract evapor- ated to a volume of 20 c. c., then transferred to a flask connected with a reflux condenser, and digested with am- monium sulphate. The precipitate was removed and the filtrate and washings concentrated to a volume of 25 c. c. To the solution, which was faintly ammoniacal, a known volume of manganese sulphate, with an equal volume of 28 alcohol, was added. As soon as the precipitate had settled it was filtered off and washed with alcohol and water (one to one). The filtrate was evaporated to dryness and the residue ignited ; when cool, it was taken up with water and a drop of aqueous sulphur dioxide added. The liquid was brought to boiling and the manganese titrated with potas- sium permanganate. Two determinations gave respectively 23 per cent and 28.12 per cent of boric acid. There is no doubt but that in the ammoniacal solution manganic hydrate is formed and carried down with the manganese borate. I dissolved this precipitate in aqueous sulphur dioxide, and attempted to remove the manganese from the boron trioxide, in order to titrate the B 2 0 3 with sodium hydrate, but could find no means of separating the two without volatilizing some boric acid or introducing harmful reagents. My attention was next turned towards the method used by Honig and Spitz (Zeit. fur angewandte Chemie 1896,551), for the determination of boric acid in a “ Gasgliihlichtcyl- inder,” with the hope of obtaining a simple method appli- cable to tourmaline. The Honig and Spitz Method of Determining Boric Acid in Insoluble Silicates. The extract of the fusion of a brown tourmaline was digested with an amount of ammonium chloride, equivalent to the carbonate used in the fusion, and the precipitated alumina and silica filtered off. An ammoniacal zinc oxide solution was added, and the whole boiled until all the am- monia was expelled. The zinc oxide, with any silica it carried, was removed and the filtrate concentrated to a small volume. After being made slightly acid with hydrochloric acid, the solution was boiled for fifteen minutes in a flask connected with a reflux condenser ; when cooled the 29 condenser was washed out with water (the washings going into the flask), and after neutralizing the excess of hydro- chloric acid, the titration for boric acid was made. The results obtained were very high, and in every in- stance after titration, a white precipitate which proved to be alumina and zinc oxide, separated. This further shows that boric acid cannot be separated from alumina by digestion with ammonium salts and that the removal of zinc oxide by this means is a difficult task. The following course was then pursued. After the addi- tion of ammonium chloride, the extract of the fusion was evaporated to dryness, the residue gently ignited to expel the ammonium chloride and when cool, taken up with water. Any insoluble matter was removed, a small amount of sodium carbonate added and the whole boiled for a few minutes. When cool, the carbon dioxide was expelled by adding hydrochloric acid and boiling, and the boric acid titrated after neutralizing the excess of hydrochloric acid. Tenth normal sodium hydrate was used with the following results : Weight of Mineral (No. 2). ^NaOH IO b 2 o 3 Grams. c.c. Grams. Per Cent. . . .2000 4-5 .01575 7-875 . . .5000 10.5 •03675 7-35 . . .5000 1 1.2 .0392 7.84 These results are lower than those previously obtained in the same sample (page 24), and are due to the volatilization of some of the boric acid during the ignition necessary to completely expel the ammonium salts, and it may be that some of the acid is lost in the boiling with the ammonium salt. In commenting on the method of Smith, Bodewig (Zeit.fur analyt. Chentie 1884, 14.3) says that some of the boric acid resulting from the decomposition of the borax by the ammonium salt, would be volatilized in the boiling; 30 upon testing the escaping vapor, he found it alkaline, while the solution at the same time was acid. To determine just how much boric acid was volatilized in this way, I decomposed a solution 200 c.c. in volume, containing 1 gram of Na 2 B 4 0 7 , by distilling with ammonium chloride; the solution was distilled to one-fourth its vol- ume. The distillate contained .02275 gram of B 2 0 3 , equiva- lent to 2.275 P er cent of the borax or 3.29 per cent of the total boric acid. This shows the necessity of avoiding the boiling of an ammoniacal boric acid solution. The digestion with ammonium chloride, using a reflux condenser to prevent the loss of boric acid, was tried, but it was impossible to get rid of the silica, alumina and ammo- nium salt in this manner. Evaporation to dryness and ignition seemed to be the only way to make the silica and alumina insoluble and destroy the ammonium salt. The Use of Lime to Prevent the Volatilization of Boric Acid During Evaporation. It occurred to me that the loss of boric acid during the evaporation to dryness and subsequent ignition, might be avoided by the addition of lime. Therefore, after digesting the extract of the fusion with ammonium chloride in a flask connected with a reflux condenser, and removing the pre- cipitate, I added lime obtained by the ignition of precipitated calcium carbonate, and evaporated the whole to dryness in a platinum dish and ignited strongly. Soda in excess of the amount required by the lime, and 25 c.c. of water were added and the whole digested for half an hour. When cool, the precipitate was removed and well washed with cold water. After concentrating the filtrate and allowing it to cool, sulphuric acid was carefully added to almost the neutral point ; the precipitate that formed was filtered out and the solution made just acid and the boric acid then titrated. The following table shows the results obtained 3i from the brown tourmaline (No. 4), which according to the previous method (page 25) contained 9.97 per cent of boric acid : Weight of Mineral (No. 4). N — NaOH. 2 Bg0 3 . No Grams. C.C. Grams. Per Cent. I • . . .5000 2.9 .05075 IO 15 2 . . .5000 30 .0525 10.50 3 • . . .3OOO 1.6 .0280 933 4 • .3000 i -75 .03062 10.21 In these determinations the insoluble oxides were refused and the extract of the fusion added to that of the first. To ascertain whether the CaO,B 2 0 3 , formed upon the addition of lime to the ammoniacal boric acid solution, was completely decomposed by the boiling with sodium car- bonate, a weighed amount of borax was subjected to the same treatment as the extract of the fusion. That the C l 0 ,B 2 0 3 was entirely decomposed by the boiling with soda, is proven by this table. Taken. Found. Vol. Weight of of Sol. Borax. b,o 3 . N - NaOH. 2 b 3 o 3 . No. C.C. Grams. Grams. C.C. Grams. I . . . 10 .01 .OO6924 04 0.007 2 . . . 10 .01 .OO6924 O.4 0007 3 • • . 20 .02 .013848 0.8 0.014 A black tourmaline, in which, as previously shown, the boric acid amounted to 10.15 P er cent, yielded 9.92 per cent when subjected to the above treatment. Datolite gave the following : Weight of Mineral. N — NaOH. 2 Bi0 3 . No. Grams. C.C. Grams. Per cent. 1 . . . . . .5000 54 .0945 1 8.90 2 . . . . . .3000 3-3 •05775 19.25 32 This same datolite, according to the method on page 25, contained 19.25 per cent of boric acid. By using barium hydrate in place of the lime and sodium sulphate instead of the carbonate, the following results were obtained from the brown tourmaline (No. 4) : Weight of Mineral. — NaOH. £ BgOj. ). Grams. C.C Grams, Per cent. 3000 1.3 .02275 7.58 3000 1.8 •0315 10.50 1.6 .0280 9-33 2770 1-5 .02625 949 5000 2.6 .0455 9 10 5000 2.4 .0420 8.40 In this series, the insoluble oxides were not refused, but were dissolved in sulphuric acid and subjected to distilla- tion with methyl alcohol, etc., whereupon the following percentages were obtained : ^NaOH. b s o 3 . 2 No. C.C, Grams. Per Cent. 1 0.2 .0035 1. 16 2 0.1 OOI75 O.58 3 0.1 .00175 0.58 4 005 .000875 0.29 5 o i .00175 0.58 Upon igniting the distillate of number 6, the charac- teristic green flame of boron was obtained. If these per- centages are added to those previously obtained, it will be noticed that it makes the boric acid found in numbers 3, 4 and 5 respectively 9.92 per cent, 9.80 per cent and 9.45 per cent. These numbers agree fairly closely with those obtained from the same tourmaline by the method described on page 25. From the foregoing, the necessity of refusing the insoluble oxides in order to entirely remove the boric acid is apparent. 33 This method, while not giving results as reliable as that in which the boric acid is volatilized as the methyl ester, yet answers fairly well and avoids the use of complicated apparatus : the time is lessened very little if at all. If the digestion with ammonium chloride could be avoided, that is if the alumina could be separated from the boric acid by some other means, the evaporation to dryness and ignition might be unnecessary and thus the time shortened. In looking for a means of accomplishing this object, I hap- pened upon the reaction 2A10 0 Na -f- C 0 2 -f- 3 H 2 0 = AI 2 (OH) 6 -f- Na 2 C 0 3 which is made use of in the manufac- ture of soda from cryolite. Hoping by this means to separate the alumina from the boric acid, I conducted carbon dioxide into the boiling solution of the sodium car- bonate fusion free from the insoluble oxides. The separa- tion, however, was not complete, as the following show : Weight of Mineral. *1 NaOH. P 2 0 3 . 2 No. Grams. C. C. Grams. Per Cent. 1 5000 12.0 .2100 42.00 2 3000 100 - 1/50 58.33 The solutions, after titration, were found to contain large amounts of alumina. By passing carbon dioxide through a cold solution of sodium aluminate, Day (Am. Chem. Jour., Vol. 19, p. 718) completely separated all the alumina from the soda. I then passed carbon dioxide through the cold fusion extract, as shown in this table : Time of Conducting C0 2 Volume through the Sol. of Sol. Hours. C. C. 2-5 200 a precipitate formed 1 . 2.0 300 << << a 1.0 350 »< «< << 2.0 500 no 7.5 total time. Temp, of sol. i°C. 34 The solution was then evaporated to a small volume, and after the expulsion of all the carbon dioxide, titrated for boric acid. N Weight of Mineral (No 4). — . NaOH. 2 e 2 o 3 . No. Gram. c. c. Gram. Per Cent. I . . 3000 1.95 •03413 n -34 After titration, upon boiling with hydrochloric acid and adding ammonia, a precipitate of alumina was obtained which weighed .0039 gram, equivalent to 1.3 per cent of the mineral. This procedure was then repeated with the following results : Time of Conducting C 0 2 Volume through the Sol. of Sol. Hours. C. C. 2.0 300 a precipitate 4.6 400 a a 1.0 ROO no 7.0 total time. Temp, of sol., i° C. The 'treatment was as in the former case and 12.77 per cent of boric acid was found. Upon testing the titrated solution for alumina, .0075 gram, equivalent to 1.5 per cent of the mineral was obtained. In all attempts to determine the boric acid in tourmalines containing much alumina great difficulty has been experi- enced in separating the alumina from the boric acid and alkali carbonate. In following out Marignac’s method, alumina is always found with the magnesium borate, mag- nesium chloride and silica. The tenacity with which boron oxide and alumina hold to one another in other than acid solutions is extremely great, as has already been demon- strated and is further shown by the following experi- ment : 35 .1733 gram of aluminium, corresponding to .32565 gram of A 1 2 0 3 , was dissolved in hydrochloric acid and the excess of acid expelled ; 2.019 grams of fused borax and 1.0 gram of sodium carbonate dissolved in water, were added to the aluminium chloride and the whole boiled one hour. These proportions were taken from the equation 3ll 2 0 + 3Na 2 B 4 0 7 -{- A 1 2 C 1 6 = Al 2 (OH) 6 + 6NaCi -f- 2 B 2 0 3 . The precipi- tated alumina was well washed, dissolved in hydrochloric acid, reprecipitated by ammonia, ignited and weighed. It equaled .2906 gram, or 89.23 per cent of all the alumina present in the solution. The washings, after boiling to expel the ammonia, were added to the filtrate. The volume of the solution then equaled 400 c. c., and when cooled to i° C. carbon dioxide was conducted through it for fifteen hours. The precipitate was removed and treated as the first ; it amounted to .0332 gram, or 11 plus per cent of all the alumina. This with the former percentage equals 10023 per cent. Carbon dioxide passed through two hours longer caused no further precipitation, yet upon acidifying, then adding ammonia, a precipitate was obtained which amounted to .005 1 gram of A 1 2 0 3 , or 1.56 per cent. The total amount of A 1 2 0 3 was .32565 gram, while that estimated equaled .3289 gram. This excess is due beyond a doubt to the contamination of the precipitates with boric acid, although in all except this last instance the precipitates were dissolved in hydrochloric acid and reprecipitated by ammonia. The obstinacy with which A 1 2 0 3 and B 2 0 3 cling together, it seems to me, would indicate the existence of a salt of aluminium and boron in this alkaline solution. This would also favor the view held by Clarke ( Bulletin 125, United States Geological Survey), which he expresses in his structural formula for tourmaline. Si0 4 = A1 / A'-Si0 4 =Al Si0 4 ■= A1-B0 2 Al-BOg = NaH s;o 4 = A1-B0 2 Al-Si0 4 = MgH \ Si0 4 = MgH boric acid, the alumina The alumina, as is seen, links the boric acid to the silica. This view, I think, is further strength- ened by the fact that calcium or barium carbonate fused with borax, then boiled with sodium carbonate or sulphate, gave up all the B 2 0 3 to the alkali ; while tourmaline, heated with calcium or barium carbonate, then boiled with car- bonate or sulphate of soda, yielded to the alkali but a portion of the most likely held the remainder. SUMMARY. In order to avoid the introduction of alkalies, which, as it has been shown, are objectionable, new methods for the decomposition of the mineral were attempted. The first of these was the heating of the mineral with metallic mag- nesium and subsequently volatilizing the boron by heating in a stream of chlorine. From all appearances, the decom- position of the mineral was complete, but in the treatment with chlorine, either the boron was partially retained by the other chlorides formed at the same time, or if it was entirely volatilized as the chloride, it was not decomposed when it came in contact with the water saturated with chlo- rine. By using a large volume of water this difficulty might be avoided, but if the chlorine is removed by draw- ing air through the solution, some boric acid would be carried out with it. Again, if alkali is added, hypochlorite of the alkali would be formed, and this, when neutralized, would free chlorine, which again would likely carry with it boric acid, or the hypochlorous acid formed would cause trouble in the titration. 37 By fusing the mineral with calcium carbonate, also with barium carbonate, a complete decomposition was obtained, but the separation of the boric acid was incomplete. Some of the boric acid was held by the alumina, which was refused to yield it to the alkali with which the fusion was digested. This must be the case, since calcium or barium carbonate fused with borax, then boiled with sodium carbonate or sulphate, gave up all the B 2 0 3 . This strengthens the idea that the boron, which is so constant a quantity in tourma- lines, is linked through the aluminium to the silica, prob- ably as an aluminium metaborate, as Clarke suggests. The mineral was then fused with alkali carbonates, and by volatilizing the boron from an acid solution as the methyl ester, the boron was successfully isolated and its estimation made by the volumetric method of Thompson. The use of a current of air with the apparatus described, shortens the time required for the volatilization of the ester, very materially. As has been shown, impure methyl alcohol prevents the complete volatilization of the boric acid. The results obtained by this method are reliable, and the method is capable of very wide application. The great affinity that alumina has for boric acid, has been pretty thoroughly demonstrated in the foregoing work, and the most successful means of overcoming this affinity and completely separating these two, is by volatilizing the boric acid from an acid solution. If an acid or ammoniacal borax solution is boiled, there is the likelihood of a very appreciable loss of boric acid : by this method there is no boiling of acid or ammo- niacal solutions. These facts, and the brief time required for its performance, recommend this method for the estima- tion of boric acid in all silicates. Could boric acid be separated from manganese, without volatilization, its estimation in silicates could be accom- plished through the salt MnO,2B 2 O s . As has already been proven, the method is successful when applied to borax. 38 If the attempt is made to remove the manganese with alkali, the precipitation is incomplete; if ammonium sul- phide is used, the difficulties encountered are well known, and if the manganese is precipitated from an acid solution, boric acid is volatilized. As has been shown, the volatility of boric acid in am- monia is very appreciable, hence any method in which an ammoniacal boric acid solution is boiled in the open, is unreliable, unless some substance is introduced to bind the acid. I have used lime for this purpose, and found it to fulfill its end. By refusing the insoluble oxides with alkali carbonate, it is possible to separate the boric acid. However, I do not believe the separation is complete even then, but that the amount of boric acid left after the first fusion is so diluted by the second, that what remains with the insoluble oxides may be neglected. By twice fusing and digesting the extract with ammonium chloride with a reflux condenser, then evaporating the solution to dryness with lime in a platinum dish and igniting the residue, the silica and alum- ina are made insoluble ; then digestion with sodium car- bonate gives borax in which the boric acid is readily titrated by the method of Thompson. This method works very well with alumina containing silicates ; with datolite, better results were obtained. It avoids the use of complicated apparatus and requires for its execution about four hours. With silicates decomposable by digestion with acids ( Wohler , Cliem. News , i86j } 255, and Handb. der Mineral, Analyse , under Datholite ), it seems to me, this would be the method. The obstinacy that is encountered in separating alumina from B 2 O s in an alkaline solution, calls to my mind a point which I think is worthy of some attention. In the com- plete analysis of minerals, such as tourmaline, it is cus- tomary to evaporate the hydrochloric acid solution of the 39 fusion to dryness, remove the silica and precipitate the iron and alumina from the filtrate as hydrates by ammonia. Although these hydrates have been dissolved in hydro- chloric acid and reprecipitated, yet considerable difficulty is experienced in obtaining two determinations from the same sample that agree. This trouble I believe to be due to boric acid, which does not go off in the evaporation to dryness with hydrochloric acid. If this be true, by the addition ot methyl alcohol during the evaporation with hydrochloric acid, this difficulty might be eliminated. The facts to which I have called attention in the latter part of this paper, argue strongly for the formula proposed by Clarke to represent the structure of tourmaline. But, as I have said, we know nothing of the magnitude of the mole- cule, hence such representations are surmises only. Until some means of ascertaining the size of the mineral molecule has been devised, such formulas will have to be looked upon with doubt. BIBLIOGRAPHY. Gilbert’s Annalen der Physik 61, 179. K. Vetensk. Acad. Handb. 1822, 22. Schweigg. Jour. 8, 7. Lehrbuch 3 Aufl. 84 ; Pogg. Annalen 2, 118. Jahresber. der Chem. und Physik 2, 364. Annales de Chem. et de Physique 40, 398. Pogg. Annalen 80, 262. Pogg. Annalen 80, 276. Pogg. Annalen 80, 466. Liebig’s Annalen 100, 82. Pharm. Centralbl. 24, 390. Pogg. Annalen 107, 427 and Fres. Anleit. z. quant. An. 6 Aufl. 1, 424. Liebig’s Annalen 98, 28. Henneberg’s Jour. f. Latidwirthsch 4, 1 12 and Zeit. f. analyt. 1863, 73. Zeit. fur analyt. Chemie 1, 405. Annales de Chem. und Pharm. 141, 268. Handb. der analyt. Chem. 6 Aufl. 2, 723. Comptes rendus 80, 490 and Zeit. f. analyt. Chem. 1875, 36. Zeit. fur analyt. Chemie 22, 25. Zeit. fur analyt. Chemie 23, 143. Zeit. fur analyt. Chemie 25, 202 Am. Chem. Jour. 4, 279. Menschutkin, Analyt. Chemie, Russische 5 Aufgabe 1883, 356- Am. Chem. Jour. 10, 154. Zeit. fur analyt. Chemie 26, 18. Am. Chem. Jour. 9, 23. Archiv. der Pharm. 225, 1101 and Z. f. an. Chem. 28, 100. Liebig’s Annalen 259, 1 84. Comptes rendus 113, 41. Zeit. fur anorgan. Chem. 4, in. Jour. Soc. Chem. Ind. 12, 432. Pharm. Centralhalle 9, 116. Pharm. Zeitung 32, 562. Jour. Pharm. Chem. (5) 29, 163. Jour. Am. Chem. Soc. 1894, 418. Jour. Am. Chem. Soc. 1897, 385. Pharm. Centralhalle 37, 672. Rev. Int. falsific 9, 119 and Z. f. anorg. Chem. 1897, 380. Zeit. fur angewan. Chem. 1896, 549. Zeit. fur analyt. Chem. 1897, 568. Zeit. fur analyt. Chem. 1897, 165. Annales de Chem. et de Physique 4, 6184. Chem. News 58, 175. Bull. 148, U. S. Geol. Survey. Chem. News 1867, 255. Chem. News 1893, 309. Soc. Chem. Ind. 12, 867.