LIBRARY 
 
 UNIVERSITY OF CALIFORNIA 
 DAVIS 
 
Digitized by the Internet Archive 
 
 in 2007 with funding from 
 
 IVIicrosoft Corporation 
 
 http://www.archive.org/details/electromotiveforOOrichrich 
 
p 
 
The Electromotive Force of Iron Under 
 Varying Conditions, and the Effect 
 
 of Occluded Hydrogen 
 
 BY 
 
 THEODORE WILLIAM RICHARDS 
 
 AND 
 
 GUSTAVUS EDWARD BEHR, JR. 
 
 PUBLISHED BY THE 
 
 CARNEGIE INSTITUTION OF WASHINGTON 
 
 1906 
 
 V \ 
 
 UNIVERSITY OF CALIFORNIA 
 DAVIS 
 
CARNEGIE INSTITUTION OF WASHINGTON, PUBLICATION No. 6i 
 
 CONTRIBUTIONS FROM THE CHEMICAL LABORATORY 
 OF HARVARD COLLEGE 
 
 THE KRIEDENWALD COMPANY 
 BALTIMORE, MD., U. 8. A- 
 
THE ELECTROMOTIVE FORCE OF IRON UNDER VARYING 
 
 CONDITIONS, AND THE EFFECT OF 
 
 OCCLUDED HYDROGEN. 
 
 The topics named in the title of this paper are of great practical import- 
 ance, as well as of great interest from the theoretical point of view. The 
 rate of rusting and the change of properties caused by the occlusion of 
 hydrogen are essential factors in determining the permanence and strength 
 of most modern buildings, bridges, and underground pipes. For this reason 
 the literature which has already accumulated upon the subject has grown 
 almost to the dimensions of a library in itself. The theoretical side of 
 the matter also is of great interest. The cause of the sudden diminution 
 of activity produced in iron under certain well-known circumstances has not 
 yet found its definitive explanation, and the mechanism of the occlusion of 
 hydrogen by a solid so tenacious and compact as iron is a theoretical ques- 
 tion of no small interest. 
 
 Phenomena relating to these topics are considered together in this paper, 
 because the method of investigation furnished evidence with regard to both. 
 The matter was considered rather from the theoretical than the practical 
 standpoint. The method was the determination of the electromotive force 
 of iron, after subjection to a great variety of changing conditions, on immer- 
 sion in a solution of one of its salts, in order by comparison of these different 
 values of electromotive force to draw inference concerning the state of the 
 metal and its occluded gas. 
 
 The object of the research, as originally planned, was to add to the experi- 
 mental evidence concerning the significance of changing atomic volume, and 
 it was proposed to discover whether or not changes in free energy were 
 associated with changes of density and of other properties of the solid. 
 During the progress of the work it was found that this matter could not be 
 satisfactorily studied without studying the properties of occluded hydrogen, 
 because, in spite of the great volume of literature upon the subject, certain 
 important questions seemed to have remained hitherto almost untouched. 
 Lack of time has as yet prevented the completion of the program ; but the 
 
 3 
 
4 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 present results, although only a part of the whole plan, are reasonably com- 
 plete in themselves, and therefore worthy of separate publication. The 
 subject will be taken up again in the near future. 
 
 For the sake of clearness and convenience, the present paper will be 
 divided into two parts, entitled : Part I. " The Electromotive Force of Pure 
 Iron under Varying Conditions ; " Part II. " The Electromotive Force of 
 Iron Containing Occluded Hydrogen." 
 
PART I. THE ELECTROMOTIVE FORCE OF PURE IRON 
 UNDER VARYING CONDITIONS. 
 
 When in a definite state of internal equilibrium, the electromotive force of 
 a metal immersed as a reversible electrode in a definite solution of one of its 
 salts must have a perfectly definite value under given conditions. A change 
 in the internal state must of necessity change this electromotive force, unless 
 the temperature happens to be exactly at the transition point. For this 
 reason it seemed not improbable that the measurement of the electromotive 
 force of iron which had been subjected to great stress would afford the most 
 convenient method of determining whether or not this stress had affected a 
 permanent alteration in the internal equilibrium of the metal, a " permanent 
 set," as it is sometimes called, or an extraordinary condition of metastable 
 equilibrium. 
 
 A solid may be subjected to a variety of stresses. The first which it 
 seemed desirable to study was direct pressure, in order to determine if this 
 pressure, by causing closer internal structure, diminishes the outward or 
 dissolving tendency as measured by electromotive force, or if, on the other 
 hand, the substance, when released from the pressure, returns at once to 
 essentially the state in which it was before the pressure was applied. Experi- 
 ment alone could decide between these alternatives. 
 
 It was desirable also to test the possibly opposite effect of a negative 
 pressure, a distending tendency, such as the effect of stretching a wire to 
 the breaking-point. This effect was also made the subject of experiments 
 recorded below. 
 
 During these tests it was found that the temperature used in the reduction 
 of the iron from its oxide caused a very important effect on the dissolving 
 tendency, and therefore this matter also had to be considered in a special 
 series of experiments. 
 
 PREPARATION OF MATERIAL. 
 
 In any research with iron it is of primary importance that the substance 
 experimented upon should be pure — free from carbon, silicon, sulphur, 
 phosphorus, manganese, and, as the following work testifies, from hydrogen 
 also. 
 
 5 
 
b ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 Most of the iron used in this research was prepared by a method which 
 had been found in an exhaustive investigation of the atomic weight of the 
 metal to yield good results.^ It is true that not quite all the steps were used 
 in the present work, but the more essential ones were retained, and a very 
 pure sample of the metal was prepared. " Piano " wire of good quality 
 served as the original substance ; it was carefully sandpapered and wiped, 
 and dissolved in dilute nitric acid. Concentrated acid was first used, but this 
 oxidized the silicon to silicic acid, and filtration from the silicon itself (or 
 silicide of iron, SiFcg)/ which the dilute acid leaves unchanged, was con- 
 sidered more safe, besides being more convenient. All iron carbide or 
 graphite is either completely oxidized or else is unchanged and filtered out. 
 The dilute acid also lessens the chances of any sulphur being oxidized to 
 sulphate. 
 
 The nitrate formed was then concentrated and recrystallized three or four 
 times. Recrystallization is here a peculiarly good method of purification, 
 because the ferric nitrate has an unusual crystal form ; therefore most im- 
 purities can only be held mechanically.' The recrystallized salt was then 
 heated in a platinum dish at a moderate temperature in an electric oven. 
 The nitric acid was partly driven off and the dry, dark-red basic nitrate was 
 then powdered in a Hempel steel mortar* and heated to full redness to 
 complete the denitration. It is important that this operation be carried out 
 in steps as described, because in this way a fine, spongy powder is obtained 
 which allows of perfect reduction. If denitrated too quickly by too high a 
 temperature the oxide obtained is exceedingly hard and the powder con- 
 sists of very compact particles, in which the complete reduction of the core 
 is more or less uncertain and decidedly less rapid. 
 
 The oxide so produced was reduced in a stream of electrolytic hydrogen ' 
 in an unglazed porcelain boat ° set in a hard glass combustion-tube when a 
 low-temperature reduction was desired, or in a large glazed porcelain tube ^ 
 in a gas-furnace when a higher temperature was used. 
 
 * Richards and Baxter, Z. f. anorg. ch., 23, 245 (1900) ; Proc. Amer. Acad., 35, 253 
 (1900). 
 
 ^Lebeau, Bull. Soc. Chim., Paris, (3) 27, 39-42 (1902); Ann. Chim. Phys. (7), 26, 
 5-31 (1902). 
 
 'T. W. Richards, Zeitschr. f. phys. Chem., 46, 189 (1903). 
 
 *W. Hempel, Z. f. ang. Chem. 14, 843 (1901). 
 
 "The generators used were zinc amalgam, dilute hydrochloric acid, platinum cells, 
 and were of the convenient form used by Cooke and Richards, Proc. Amer. Acad. 23, 
 168 (1887). The hydrogen was twice washed in strong caustic solution and dried 
 by passing through two caustic potash towers. 
 
 * Unglazed, because at the high temperature used the glaze frequently fused with 
 the iron. 
 
 * Hempel water-cooled stoppers were used in the porcelain tube. 
 
THE REDUCTION OF PURE IRON. 7 
 
 The iron from these reductions was of a gray color and of a porous struc- 
 ture. A piece of it could usually be broken with ease between the fingers 
 and rubbed to powder. This consistency was of course due to its mode of 
 preparation, the powdered oxide having been simply pressed into the boats 
 and the temperature having been only high enough to cause slight cohesion 
 between the particles, not to fuse the mass. 
 
 The cohesion of the particles varied considerably with the temperature at 
 which the reduction was carried on, a product reduced at 600° having no 
 cohesion at all, while one heated to about 1,100° for some time could be 
 broken only with some difficulty. The color also varied, owing to the differ- 
 ence in degree of subdivision, from almost black to a clear, silvery gray. 
 
 A convenient index of complete reduction was the absence of any trace 
 of black oxide in the cooler portions of the boat; for water formed in the 
 hotter places is decomposed in the somewhat cooler ones, and if the reduc- 
 tion is not complete water is of course present. Ordinarily, the reduction 
 was carried out by heating the substance in a rapid stream of hydrogen for 
 three or four hours with a Bunsen burner, and then for about half an hour 
 at the highest temperature of the blast lamp. Such iron showed not the 
 slightest trace of black oxide, even when finely powdered in a mortar. After 
 heating the iron remained in a continuous stream of dry hydrogen until 
 quite cold, and was then transferred for preservation to a good desiccator 
 containing finely divided fused caustic potash. No trace of oxidation was 
 ever visible on iron thus preserved; but, as will be explained later, the 
 electromotive force measurements gave clear evidence that an exceedingly 
 thin coating of oxide or adsorbed oxygen had been formed, and therefore, 
 for the more accurate experiments, small boat-loads of pieces about the size 
 of peas were reduced a second time and sealed in hydrogen until just before 
 using, thus avoiding the possibility of oxidation. Whenever this specially 
 sealed iron is used in the experiments below the fact is mentioned. The 
 sealing was done as follows: A hard glass tube was drawn out in several 
 places, having between each contracted place a boat containing porous iron. 
 Three or four such boats in series were laid in a combustion furnace, and 
 after reduction the boats were sealed off by drawing out the contracted places. 
 
 All this iron undoubtedly contained hydrogen in greater or less amounts. 
 Although, as Baxter' has shown, the weight of hydrogen retained by pure 
 iron ignited at a high temperature is inconsiderable, some of the other 
 properties of the occluded gas are more important, as will be seen. Most 
 of the hydrogen can be expelled by heating for a long time in vacuum, or 
 by standing for a long time in dry air; but in order to drive out the last 
 traces fusion in vacuum is necessary. 
 
 "Baxter, Am. Chem. Journ., 22, 363 (1899). 
 
ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 METHOD OF MEASUREMENT. 
 
 The solution-tension of the iron was measured as its electromotive force 
 in a normal solution of ferrous salt, in this case ferrous sulphate. The 
 sulphate ion is desirable here, since it is neutral as regards passivity phe- 
 nomena, having- neither passivizing nor activizing effect." The voltage of 
 the cell Fe, nFeSOi, n/ioKCl, HgCl, Hg, was measured by the well- 
 known compensation method of Poggendorff." 
 
 A d'Arsonval galvanometer served as zero instrument, being sensitive to 
 o.ooi volt with the resistance used. A Helmholtz cell was used as a standard 
 of comparison. Its exact value was 0.993 international volt, as determined 
 by a comparison with carefully prepared cadmium normal elements. These 
 agreed accurately among themselves, and were assumed to possess at 
 21.50° C. the value 1.0185 volts. They agreed also with the standards of 
 the Jefferson Physical Laboratory. 
 
 As a standard electrode the mercury-calomel-potassium chloride electrode 
 first suggested by Ostwald" was considered the best for the reasons he so 
 convincingly puts forward in his argument with Wilsmore." In the present 
 research tenth normal instead of normal potassium chloride was usually 
 used as the electrolyte. All values of electromotive force are given for the 
 whole cell, Fe, FeS04, KCl, HgCl, Hg, a definite measurable quantity, 
 instead of for the single potential difference Fe, FeSO^, because of the uncer- 
 tainty as to the absolute value of the decinormal electrode. If 0.56 volt is 
 supposed to be the value of the normal calomel electrode, the value 0.612 
 volt for the tenth-normal electrode should be used, as recently determined 
 by Sauer" and verified by us. Subtracting this from the values given, a 
 value comparable to the commonly accepted values of single potential differ- 
 ences will be obtained. 
 
 A special arrangement of the decinormal electrode was adopted, in order 
 to prevent its contamination with iron. To the connecting-tube, used for 
 immersion in the ferrous sulphate cell, was attached above a siphon-inlet 
 from a large bottle of decinormal potassium chloride on the shelf above. 
 The fact that it was possible to wash out the tube of the decinormal elec- 
 trode by means of this fresh solution removed the possibility of variation 
 through the backward diffusion of normal ferrous sulphate or other electro- 
 
 •Sackur, Z. f. Electrochem., 10, 843 (1904). 
 " Ostwald-Luther, Physico-Chemische Messungen, p. 367. 
 "Lehrbuch, 2, 945 (1893). 
 
 ^^Z. f. Physik., ch. 35, 333 (1900) ; ibid., 36, 91 (1901). 
 
 "Z. f. Physik., ch. 47, 184 (1904). That is, N/10 electrode = 0.612 volt if the N/i 
 electrode is called 0.560 volt. 
 
MEASUREMENT OF ELECTROMOTIVE FORCE. 9 
 
 lyte. By means of this arrangement one such electrode lasted through a 
 whole season and at the end of that time gave no test for iron in the cell 
 itself. Thus the uncertainty of this possible error is removed and the 
 nuisance of repeated renewal of the electrode done away with. The arrange- 
 ment is shown on the left-hand side of figure i. 
 
 The usual arrangement of the iron and of the cell was as shown in figure I, 
 and on a larger scale in figure 2. The sample, about half the size of a pea, 
 
 Fig. I. — The Cell and its Appurtenances. 
 
 A. Reservoir of decinormal potassic 
 chloride. 
 
 J3. Reservoir of normal ferrous sul- 
 phate containing pure iron 
 wires. 
 
 C. Calomel electrode (decinormal). 
 
 D. Connecting receptacle containing 
 
 ferrous sulphate. 
 
 E. Inverted U tube. 
 
 F. Iron electrode (enlarged in fig. 2). 
 
 G. Automatic hydrogen generator, pro- 
 
 tecting Reservoir B from air. 
 
 was inserted in a running noose of platinum wire, the end of which projected 
 through a fine hole in a cork stopper. The noose was then pulled tight, 
 drawing the sample against the cork, and the wire was fastened by winding 
 around the shank of an iron tack on the other side of the cork. The cork 
 was fitted into one end of a glass tube and carefully sealed with molten 
 paraffin. The tube was filled with normal ferrous sulphate solution by 
 
10 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 means of a siphon from the reservoir, and an inverted (J "tube and stopper 
 were inserted in the top to make connection with an open test-tube contain- 
 ing ferrous sulphate fastened alongside by means of a rubber band. In this 
 way the iron was kept for some time safe from oxidation, yet always open 
 to measurement. The arrangement is illustrated in figure 2. When the 
 end of the tube of the decinormal electrode (C in figure 2) was dipped into 
 this test-tube full of ferrous sulphate, the cell was complete and ready for 
 measurement. A large number of " iron electrodes " could be placed around 
 
 Ky 
 
 Q. 
 
 I \ 
 
 ^ 
 
 v-y \^ 
 
 Fig. 2. Fig. 3. 
 
 Receptacles for Iron Electrodes. 
 
 A 
 
 o 
 
 Fig. 4. 
 
 C. Tube leading to 
 
 trode. 
 F. Contains normal 
 
 solution. 
 O. Protecting layer of paraffin oil. 
 
 decinormal elec- 
 ferrous sulphate 
 
 P. Hard paraffin. 
 V. Iron wire. 
 
 W. Platinum wire and loop for holding 
 porous iron. 
 
 the sides of the thermostat, all attached to one wire, and the change of con- 
 nection from one cell to another was made by simply lifting the iron stand 
 to which the decinormal electrode was attached and dipping the end of the 
 standard electrode's tube into the next test-tube. Thus a large number of 
 measurements could be taken with a minimum expenditure of time and 
 trouble, a very necessary matter in an investigation involving thousands of 
 measurements. The thermostat, electrically regulated and heated, in which 
 the electrode was immersed, was kept at 20.0° C. The normal ferrous 
 
THE EFFECT OF PRESSURE. 
 
 II 
 
 sulphate used was prepared by dissolving almost the necessary amount of 
 pure crystals in a quantity of water, adding a small measured amount of 
 sulphuric acid, and making up to a liter. Then pieces of piano wire, care- 
 fully cleaned, were added. The acid acting on the iron thoroughly reduced 
 the solution, the amount of the acid having been so chosen that just the 
 necessary amount of iron dissolved to make the solution normal. Several 
 months were needed to remove the last traces of acid — an essential condition. 
 The flask was connected with an automatic hydrogen generator whose cock 
 was always open, and when liquid was withdrawn through the siphon 
 hydrogen took its place. This also is represented in figure i. Thus after 
 having once prepared the solution, it could be kept indefinitely, perfectly 
 constant and always at hand. 
 
 THE EFFECT OF PRESSURE. 
 
 In the following work the term " porous iron " will be used for the sake 
 of brevity to designate the iron which had been reduced from oxide in the 
 manner described on pages 6 and 7. 
 
 The results of a series of preliminary experiments with this porous iron 
 are given in table i. 
 
 The first two were made with samples which had merely suflfered exposure 
 to air. The next four had received further treatment, having been put 
 into a small iron mortar such as is used in powdering rocks, and beaten with 
 a heavy hammer. Under this treatment the friable, porous iron became 
 compact and gained a shining metallic surface. The electromotive forces 
 of these samples were then measured after immersion in normal ferrous 
 sulphate. The cracked and powdered edges of numbers 2 and 3 were 
 covered with shellac, so that only the smooth metallic surface was exposed 
 to the solution. 
 
 Table i. — Preliminary experiments — Electromotive force of iron against 
 decinormal electrode. 
 
 Time 
 
 elapsed 
 
 after 
 
 immersion. 
 
 Sample 1, 
 
 Sample 2, 
 
 Sample 3, 
 
 Sample 4, 
 
 Sample 5, 
 
 Sample 6, 
 
 porous 
 
 porous 
 
 beaten 
 
 beaten 
 
 beaten 
 
 beaten 
 
 iron. 
 
 iron. 
 
 iron. 
 
 iron. 
 
 iron. 
 
 iron. 
 
 Minute. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 1 
 
 
 
 
 0.704 
 
 0.655 
 
 0.655 
 
 2 
 
 
 
 0.674 
 
 
 
 
 0.685 
 
 4 
 
 0.754 
 
 0.755 
 
 
 
 
 
 
 20 
 
 0.780 
 
 0.774 
 
 0.724 
 
 0.720 
 
 0.674 
 
 0.722 
 
 55 
 
 
 
 0.737 
 
 0.728 
 
 0.674 
 
 0.736 
 
 240 
 
 6.801 
 
 0.799 
 
 
 
 
 
 
12 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 A glance at the table shows that the beating had made considerable dif- 
 ference in the electromotive force of the iron in every sample which had 
 been beaten. It will be shown later, however, that this difference was due 
 not to the compression, but to another effect of the treatment. 
 
 This effect was soon found to be closely connected with another relation 
 to be observed in the above figures; namely, the fact that after immersion 
 the electromotive force of each piece of iron begins at a low point and rises, 
 attaining constancy at a maximum only after some time. In order to study 
 this phenomenon more closely, six other small pieces were immersed in the 
 solution and readings of the electromotive forces taken at short intervals. 
 The numerical results are given in table 2. 
 
 Because it was evident that the electromotive force continued to rise slowly 
 for a very long time, four of these pieces were carefully sealed up with 
 paraffin immediately after immersion in order to prevent oxidation. This 
 was done in order to determine the electromotive force of the final equi- 
 librium. After three days No. 9 and No. 10 were opened and measured; 
 and after five days No. 11 and No. 12. 
 
 Table 2. 
 
 The increase of electromotive force with time — Irori electrode 
 against decinormal electrode. 
 
 Time elapsed 
 
 after 
 immersion. 
 
 Sample 7. 
 
 Sample 8. 
 
 Sample 9. 
 
 Sample 10. 
 
 Sample 11. 
 
 Sample 12. 
 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Oh 2m 
 
 0.590 
 
 0.690 
 
 
 
 
 
 
 3.75 
 
 0.645 
 
 0.745 
 
 
 
 
 
 
 6 
 
 0.685 
 
 0.764 
 
 
 
 
 
 18 
 
 0.714 
 
 0.774 
 
 
 
 
 
 50 
 
 0.749 
 
 0.785 
 
 
 
 
 
 73 
 
 
 
 
 0.789 
 
 0.793 
 
 
 
 78 
 
 
 
 0.791 
 
 0.794 
 
 
 
 119 
 
 
 
 
 
 0.782 
 
 0.784 
 
 120 
 
 
 
 
 
 0.784 
 
 0.786 
 
 This was later repeated with other samples, which gave similar results. 
 It should be noted that after several days a pale-green precipitate of ferrous 
 hydroxide was always found in the tube, but this seemed to exercise no 
 considerable effect on the potential until seven or eight days had passed and 
 the amount had become large. Even then by shaking the constant value of 
 the potential was usually regained. 
 
 Thus it was obvious that spongy iron which has been exposed to the air 
 always gives at first an abnormally low electromotive effect, which slowly 
 rises. The only reasonable hypothesis capable of explaining the phe- 
 
THE EFFECT OF HIGH PRESSURE. 1 3 
 
 nomenon seems to be the conclusion that the exposure to the air results in 
 the formation of an exceedingly thin coating of oxide or at least of adsorbed 
 oxygen, and that when this coating is removed, by solution or reduction, 
 the true electromotive force is obtained." 
 
 Accordingly, the iron to be used in the succeeding experiments was sealed 
 in hydrogen. The tubes were broken open just before the samples were 
 to be used, and their electromotive forces v*^ere measured as soon as possible 
 afterwards. In this way the exposure to air was as short as possible and 
 the theory was confirmed, for they were found to attain their maximum 
 values very promptly. 
 
 With these samples a more careful study of the effect of pressure was 
 begun. A number of preliminary experiments were made with pressures 
 produced by 30,000 and 60,000 pounds. These pressures were quite suffi- 
 cient to cause cold-welding of the particles of the porous iron, giving firm, 
 hard plates with bright metallic surfaces. The area of these surfaces was 
 seldom more than the half of a square centimeter, hence the pressures were 
 approximately 30,000 to 60,000 kilograms per square centimeter. The 
 compression was done between two smooth plates of steel in a large testing- 
 machine in the Engineering Department of Harvard University. The 
 pressure was great enough to press the soft iron into the steel plates, leaving 
 distinct indentations. The iron was handled only with iron pincers. Every- 
 thing with which it came in contact was kept scrupulously clean. The edges 
 of the compressed pieces being cracked and split, they were embedded in 
 soft paraffin before immersion in the ferrous sulphate solution, leaving 
 exposed for measurement only the center, which had certainly borne the full 
 effect of the pressure. 
 
 It appeared that iron thus treated gave results in no way different from 
 the spongy pieces before compression, except that the compact specimens 
 were slower in reaching their maxima. When reached, these maxima were 
 essentially identical with those of the porous specimens. 
 
 Nevertheless, it seemed worth while to carry the matter further. Through 
 the kindness of Lieutenant-Colonel J. G. Butler in charge, it was possible 
 to subject other portions of pure spongy iron to still greater pressure on the 
 magnificent testing machine at the United States Arsenal at Watertown, 
 Massachusetts. This machine is capable of not only exerting a weight-effect 
 of 1,000,000 pounds, but of measuring with great accuracy the magnitude 
 of the effect it' exerts. Two special blocks were made between which to 
 
 "This experience, and the conclusion derived from it, agrees with that of Mutti- 
 mann and Fraunberger (Sitzber. Akad. der Wiss. Miinchen, 34, 201) (1904). Their 
 investigation was published after this part of our research had been completed, and 
 was wholly independent. 
 
14 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 compress the iron. In order to resist such enormous pressures they were 
 made of the very finest tool steel, and after the lathe-work was finished they 
 were quenched to a condition of maximum hardness. They were frustra 
 of cones, the diameters of the bases being about i6 cm. and those of the 
 tops, between which the iron was compressed, about 3 cm. Their heights 
 were about 8 cm. The pressures exerted were 500,000 pounds and 700,000 
 pounds. Exactly how much of this pressure bore on the iron test-piece it 
 is impossible to say, for the soft iron was pressed so deeply into the hard 
 steel that there was some possibility of the whole top surface (7 sq. cm.) 
 receiving part of the pressure. On the other hand, if, as is probable, the 
 test-piece alone bore the stress, the maximum pressure was at least 350,000 
 kilograms per sq. cm., for the surface of the soft iron was never i sq. cm. 
 
 In order to make more striking any differences which might result from 
 these pressures, samples 13 and 14 had each been broken in two, and one 
 piece of each was kept in a spongy condition, in order to compare with the 
 other compressed one. Thus it was certain that all treatment previous to 
 measurement" had been precisely the same, except for this one matter of 
 compression. Other pieces of porous iron gave essentially the same results. 
 
 As in the previous experiments, the edges of the compressed iron were 
 protected with paraffin, so that the exposed surface should be that alone 
 which had been compressed. 
 
 Table 3. — The effect of previous compression on the electromotive force — 
 Iron electrode against decinormal electrode. 
 
 Time 
 elapsed 
 after 
 immer- 
 sion. 
 
 Sample 
 
 13. 
 porous. 
 
 Sample la 
 
 after 
 subjection 
 
 to 
 350,000 kg. 
 pressure. 
 
 Sample 
 
 14. 
 porous. 
 
 Sample U 
 
 after 
 subjection 
 
 360,000 kg. 
 pressure. 
 
 Sample 15 
 
 after 
 subjection 
 
 to 
 350,000 kg. 
 pressure. 
 
 Sample 16 
 
 after 
 subjection 
 
 to 
 350,000 kg. 
 pressure. 
 
 Oh llm 
 
 28 
 
 6 
 
 21 
 
 Volt. 
 
 0.775 
 
 0.781 
 0.780 
 
 Volt. 
 
 0.7.55 
 0.781 
 0.782 
 
 Volt. 
 
 *0.'786 
 0.787 
 0.787 
 
 Volt. 
 0.731 
 
 0.770 
 0.782 
 
 Volt. 
 
 0.744 
 0.783 
 788 
 
 Volt. 
 0.722 
 
 '6 '.750 
 0.751 
 
 The results with samples 13 to 15, inclusive, lead to the conclusion that no 
 permanent change in free energy had resulted from the compression. They 
 are exactly similar to those obtained with the lower pressures. The appar- 
 ently contradictory result of sample 16 will be shortly explained. 
 
 As a step toward this explanation, it is necessary to call attention once 
 
 "Including the temperature to which it had been exposed, the importance of which 
 will shortly be pointed out. 
 
THE PERFECT ELASTICITY OF PURE IRON. 15 
 
 more to the fact that the compressed pieces at first yielded too low an 
 electromotive force, although the value afterwards rose to equal that of the 
 uncompressed pieces. An explanation for these low values is to be found 
 in the idea already mentioned on page 13, there advanced to explain the lag 
 of porous iron which had been kept for some time in a desiccator. It is 
 that an exceedingly thin coating of oxide is formed on the iron by exposure 
 to the air and thus the true electromotive force is masked until this coating 
 is dissolved. Most of the samples of porous iron had been kept a whole 
 day in a desiccator after reduction before immersion, while sample 14 in 
 table 3 had been kept only three hours. The difference in the speed of 
 attaining the maximum value is very marked. This theory not only seems 
 to account for the effect of long exposure to air, but also offers a simple 
 explanation for the slightly lower initial potential of the compressed samples. 
 The length of time during which the compressed pieces were exposed to 
 the air before measurement was exactly the same as that of the parallel 
 porous ones, but the former were inevitably heated somewhat in the act of 
 compression, and thus an oxide coating might well have been formed which 
 would equal in effect one of long, slow formation in a desiccator. The 
 theory probably also offers the most satisfactory explanation of the exag- 
 gerated deficiency exhibited by No. 16 in table 3. The blocks between 
 which No. 16 was compressed had been three times subjected to such enor- 
 mous pressures that they were beginning to fail, and this fourth and last 
 time the blocks split with bright sparks and loud reports, and threw small 
 pieces in all directions. Under such treatment the blocks undoubtedly 
 became warm, and it is quite to be expected that this last sample should 
 possess the heaviest and firmest coating of oxide. 
 
 Thus it is fairly safe to conclude that even such enormous pressures as 
 300,000 kilograms per square centimeter produce no permanent change of 
 free energy in iron. 
 
 It is interesting to compare this constancy with the behavior of other 
 properties of metals on subjection to great stress. Since this work was 
 done. Dr. StuU and one of us has found, in the work already referred to, 
 that after subjection to pressures up to 500 atmospheres all of thirty-odd 
 elements examined regain precisely the original volume. The method used 
 is unique and allows of the detection of volume changes of 0.000,01 milliliter 
 in a sample of 10 milliliters. 
 
 After this work was done we became acquainted with two pertinent 
 papers by W. Spring," in which he proves the same fact by still another 
 method, the one, in fact, which would most naturally be adopted if that 
 were the main object of the research, namely, the determining of the spe- 
 
 "Bull. Acad. Roy. de Belg. (3), 6, 509 (1883) ; Ibid. (1903), p. 1066. 
 
l6 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 cific gravity before and after compression. He first submitted to pressures 
 up to 10,000 atmospheres more than eighty soHds of the most diverse chem- 
 ical nature, and found that powders weld together perfectly under pressure, 
 as we found with iron. In order to do this, he found it necessary to com- 
 press in a vacuum, because imprisoned air not only prevented contact of the 
 particles, but also tore them apart again when pressure was released. All 
 of his work was performed with the care which this precaution indicates. 
 The result of one investigation, entitled *' L'elasticite parfaite des corps 
 chimiquement definis," is best given in his own words: 
 
 La plupart des corps examines ont manifeste une petite augmentation permanente 
 de densite apres avoir subi une pression d'environ 20,000 atmospheres, mais, celle-ci 
 une fois realisee, ils ont resiste opionatrement a toute diminution permanente ulterieure 
 de leur volume. Pour tous la densite a atteint bientot un maximum. La faible aug- 
 mentation de densite permanente observee n'a cependant jamais eu pour cause une 
 contraction reelle de la matiere. Tou jours elle a ete due a I'ecrasement, par pression, 
 des cavites existant d'avance dans le corps examine, ou bien a la disparition de fissures 
 plus ou moins evidentes. En un mot, on s'est trouve en presence d'un fait accidentel, 
 pouvant facilement amener une erreur, mais non d'un phenomene physique essentiel. 
 
 Thus the present work on the electromotive force of iron subjected to 
 great stress confirms the very different work of Spring on densities. In so 
 far as these data show that when no change of specific gravity is effected by 
 a given application of energy, no change in the free energy of the system 
 results, they support the theory of compressible atoms. 
 
 TENSION EXPERIMENTS. 
 
 Although it thus appears to be impossible to produce permanent free energy 
 changes in pure iron through pressure alone, it appeared possible that 
 measurements of electromotive force made while the metal was under tensile 
 stress might show varying values, because tension produces marked physical 
 changes in the metal. These changes are as follows: (i) A strain which 
 immediately disappears on release of the tension, inside the elastic limit. 
 
 (2) Beyond the elastic limit a permanent strain, with increasing stress 
 needed to produce equal increment of strain, up to the ultimate resistance. 
 
 (3) Beyond the ultimate resistance, with ductile metals, there is a region 
 before the actual break occurs, where, for increase of strain, decreasing 
 stress is required, until failure takes place. It seemed reasonable to expect 
 that such decided and sudden changes in the cohesion would be accompanied 
 by free-energy changes. 
 
 While under stress, or after a " permanent set " has been produced by 
 strain beyond the elastic limit, the iron is in unstable equilibrium. Sufficient 
 proof that it is so within the elastic limit lies in the fact that it returns of 
 
THE WORK OF BARUS. I'J 
 
 itself to its original dimensions. Beyond the elastic limit, the custom in 
 wire manufacture, of returning the wire perhaps half a dozen times to the 
 annealing pots to be softened, speaks for itself. By heating to 300° to 400° 
 for several hours " the molecular tension in the wire caused by drawing " 
 is released." The molecules, given the opportunity through increased mo- 
 bility at higher temperatures, may be supposed to return to their normal 
 relations to each other and plasticity is restored to the iron. 
 
 Further evidence in this direction is furnished by a piece of work done by 
 Barus, entitled " The Energy Potentialized in Permanent Changes of Molec- 
 ular Configuration." " He pulled wires of various metals to the point of 
 failure, determined their rise in temperature by means of a thermopile, and 
 subtracting the heat evolved from the total work done on the wires, obtained 
 values for the amounts of energy which had become potential in the metals. 
 He says: 
 
 To summarize, it appears that as much as one-half of the work done in stretching 
 up to the limit of rupture may be stored up permanently; that the amount of work 
 thermally dissipated varies considerably with the metal acted upon, being very large, 
 for instance, in copper (75 per cent) and smaller in the case of iron (50 per cent) ; 
 that in the case of the same given metal the work is largely potentialized during in- 
 cipient stages of strain, and very largely dissipated during final stages of strain. When 
 stress of a given kind is applied to different metals, the total amount of energy which 
 can be stored per unit section, per unit of length up to the limits of rupture, may 
 therefore be looked upon as a molecular constant of the metal. 
 
 In a table he gives results showing that in an iron wire of 0.136 cm. 
 diameter, stretched almost to the limit of rupture, at least 2 megergs per 
 centimeter are potentialized, about the same amount having been dissipated as 
 heat. Knowing this value, it is possible to calculate the maximum rise in elec- 
 tromotive force theoretically required by such an increase in the free energy 
 of the metal, supposing that all this work were available as free energy. The 
 general formula for such a calculation is: 
 
 Atz = 
 
 96s8o;;2 
 m which 
 
 W = the work in joules done per centimeter of the wire ; 
 
 E = the electrochemical equivalent of the metal, in grams; 
 
 m = the weight of the metal per centimeter of wire ; and 
 
 W X E/m = the work done per gram equivalent of the metal. 
 
 Substituting values found by Barus, we have: 
 
 ^7r = 0.432 X 27.95/0. 1 1 14 X 1/96580 = o.ooii volt. 
 
 (The radius of the wire was 0.068 cm. and its density 7.68.) 
 
 " " Wire, its Manufacture and Uses," J. B. Smith, pp. 30, 54, and 56. 
 "Amer. Journ. Sci. (3), 38, 193 (1889); also U. S. Geological Survey Bulletin No. 
 94, p. loi (1892). 
 
1 8 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 Two Other calculations will be given to illustrate more fully the changes 
 to be expected. The data for these were found in a book" which is still 
 quoted as an authority by engineers, although published as long ago as 1879. 
 
 In the first of these (p. 24) the test-pieces were taken " from a bar of 
 remarkably pure, refined, and uniform iron." The average of nine tests, 
 very uniform in their results, gave : Breaking stress per square inch, 55,489 
 pounds ; elongation, 23.9 per cent. The diameters of the test-pieces were 
 0.974 inch. Calculating as before, we get 0.0034 volt as the change to be 
 expected. 
 
 In the second one the iron is described as " very soft and ductile." The 
 average of four tests gave : Stress per square inch, 45,873 pounds ; elonga- 
 tion, 29.7 per cent. The diameters were 1. 000 inch. These data give 0.0035 
 volt as the expected rise. 
 
 The results of all these calculations should be divided by 2, since Barus 
 showed that only 50 per cent of the work done becomes potential. These 
 calculated values, never exceeding 1.8 millivolts, represent the maximum 
 values for the increase of potential — values which may not be actually 
 attained, because it is by no means certain that all the energy thus stored is 
 available as free energy. Moreover, the distribution of the strain between 
 surface and interior is uncertain. This is indicated by the following state- 
 ment of Burr : "^ 
 
 It has been found by experiment that bars of wrought iron which are apparently 
 precisely alike in every respect, except in area of normal section, do not give the 
 same ultimate tensile resistance per square inch. Other things being the same, bars 
 of the smallest cross section give the greatest intensity of ultimate tensile resistance. 
 
 This is due to the fact that shearing strains increase with greater cross 
 section. And also,"^ 
 
 If the tensile stress is uniformly distributed over each end of the test-piece, it will 
 not be so distributed over any other normal section. Since lateral contraction takes 
 place, the exterior molecules of the piece must move toward the center; but if this 
 motion exists, the molecules in the vicinity of the center must be drawn farther apart, 
 or suffer greater strains, than those near the surface. Hence the stress will no longer 
 be uniformly distributed, but the greatest intensity will exist at the center and the least 
 at the surface of the piece. These effects will evidently increase, with a given form 
 of cross section, with the area. 
 
 Smith," in pointing out the strengthening effect of drawing wire has said : 
 " In the case of the wire drawn through three holes the tenacity of the 
 
 " " Experiments on the strength of wrought iron." Report of the Committee of 
 the United States board appointed to test iron, steel, and other metals. Commander 
 L. A. Beardslee, U. S. N. Abridged by William Kent. New York, 1879. 
 
 * Elasticity and resistance of the materials of engineering, p. 218. New York, 1903. 
 
 " Ibid., p. 206. 
 
 ""Wire, its manufacture and uses," p. 58. 
 
THE SMALL EFFECT OF TENSION. I9 
 
 metal was increased 95 per cent above that of the billet ; " and Burr says : ^ 
 " Wire is the strongest form in which iron can be used to resist tensile 
 stress." 
 
 It is evident, then, that by pulling a wire of small cross-sectional area it is 
 possible to exert the maximum of stress on the surface of a test-piece. 
 
 The apparatus (shown in figure 4, page 10) employed in our experiments 
 was exceedingly simple. The upper end of the wire was twisted around an 
 iron support and the pull was exerted by weights in a basket attached to the 
 lower end. In the middle the wire was narrowed at one place, and this was 
 surrounded by a tube of normal ferrous sulphate and connected with a nor- 
 mal calomel electrode. The results were at first highly irregular. The 
 electromotive force of the wires used (pure piano wire and pure soft iron 
 wire) varied over a range of 0.005 volt without apparent cause. Merely 
 tapping the stretched wire sometimes caused a momentary rise of from 0.003 
 to 0.004 volt. Such electrical disturbances from mechanical jarring have 
 been observed by von Helmholtz. In one case the addition of 5 kilograms 
 to the load caused a net rise of about 0.0007 volt and the next 5 brought it 
 down again by the same amount, while one more brought it up again by 
 0.0005. 
 
 Calculating the theoretical change of potential, we get 0.0017 volt for the 
 total work done, and half of this, or 0.0009 volt, as the change to be expected 
 according to Barus. The data for this calculation are: Tensile strength, 
 25 kilograms ; elongation, 9.3 per cent ; diameter of wire, 0.08 cm. 
 
 Hence the accidental disturbances exceeded the maximum theoretical effect 
 and entirely masked the latter. 
 
 Only one regularity was noticeable, namely, that at the instant at which 
 the load was increased there was a momentary drop in electromotive force. 
 Starting with no load at all and the potentiometer reading 0.7145 volt,** a 
 10 kg. load was applied, and at the same instant the galvanometer swung 
 18 to 20 divisions to the right, and then instantly swung back again very 
 rapidly to the permanent reading, 0.7125 volt. Exactly how much fall of 
 potential this swing to the right indicates could be determined only with a 
 ballistic galvanometer, but it may have been as much as 0.015 volt. The 
 addition of another kilogram had a similar effect, except that the swing was 
 only 12 to 15 divisions to the right and it returned more slowly to 0.7125. 
 The third 10 kg. caused a smaller swing still, and a much slower return to 
 0.7125. About that value was always reached after standing a few minutes. 
 
 ^ Elasticity and resistance of the materials of engineering, p. 241. 
 
 '*This reading, being taken with the help of a normal calomel electrode, must have 
 0.052 volt added to it in order to be comparable with the other potentials recorded in 
 the paper. It thus becomes 0.766, the same as before. The normal electrode has less 
 resistance and allows more sensitive readings than the decinormal. 
 
20 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 The weights were then added to 48 kg., the last ten still causing a slight 
 swing to the right. A 13 kg. weight was then added to the load, and the 
 wire lengthened rapidly and would have undoubtedly broken had not the 
 support stopped the basket. The galvanometer swung 8.5 divisions to the 
 left, indicating a rise of about 0.004 volt, and quickly swung back again. 
 
 These temporary changes and the change in sign which they undergo 
 are highly interesting. The most obvious explanation of the first effect of 
 diminished potential is to ascribe this to heating incidental to pulling, and 
 experiment gave the idea some support. Heating the wire to redness with 
 a small gas flame just above the surface of the liquid gave a somewhat 
 variable electromotive force, the mean value being about 0.015 volt below 
 the initial one — an effect about like the first effect noted above. When 
 cold the wire returned to the initial value. This is advanced as a suggestion 
 rather than a definitive explanation, however. 
 
 The decreasing temperature effect due to the successive additional stresses 
 may have been due to the larger percentage of storing as potential energy as 
 the elastic limit is approached — ^and the final excess of potential may have 
 been due to the sudden manifestation of this increased potential at the point 
 of fracture. 
 
 Nevertheless, these conclusions are highly hypothetical. The results do 
 not permit of conclusions broader or more definite than these. On the other 
 hand, they afford the opportunity of correcting two pieces of work which 
 have been widely published and whose incorrect results have been given 
 general credence. The subject is of such great importance to modern 
 engineering that, naturally enough, engineers were the ones attracted to its 
 investigation. M. P. Wood, in his book " Rustless Coatings ; Corrosion and 
 Electrolysis of Iron and Steel," ^ under the heading ^' Corrosion Increased 
 by Stress " (p. 348), quotes three men as entitled to a hearing on this subject. 
 
 The first of these, Thomas Andrews,^ carried out experiments far from 
 satisfactory from the electrochemical point of view, because he used a solu- 
 tion of common salt as an electrolyte, and prepared the two pieces of iron 
 for comparison in different ways. The observed difference in potential is 
 quite as probably to be ascribed to the oxidation effect already discussed as 
 to the effect of strain, although a part of the effect may have been due to 
 strain. 
 
 The second paper quoted by Wood is " An Experimental Study of the 
 Corrosion of Iron under Different Conditions," by Carl Hambuechen." Of 
 
 " Wiley & Sons, N. Y., 1904. 
 "Proc. Inst Civil. Engin. 118, 356-374 (1894). 
 
 '"Trans. Amer. Soc. Mechan. Engin., 22, 816-821 (1901) ; or Bull. Univ. Wisconsin, 
 Engin. Series, vol. 2, No. 8, (1900). 
 
ERRORS IN PREVIOUS INVESTIGATIONS. 
 
 21 
 
 this paper the section on " Corrosion of Strained Metal," which is the only 
 part to be discussed here, is no more conclusive than Andrews's work, 
 although Hambuechen states his conclusions in unqualified terms. The 
 work was very carefully performed in every particular except one, namely, 
 the fact that he chose ferric chloride solution as the electrolyte in which to 
 measure the electromotive force of the iron test-pieces. This unfortunate 
 choice completely invalidates all the results. The reason is very simple. 
 Ferric chloride is always hydrolyzed to a considerable extent — that is, it 
 contains acid. When such a solution comes in contact with iron the three 
 following reactions take place : 
 
 Fe + 2Fe+++ = 3Fe++ (or Fe + 2FeCl3 = sFeCl^), 
 
 Fe + 2H+ = Fe++ + H^ (or 2HCI + Fe = FeCl^ + H^). 
 
 H (nascent) + Fe+++ = Fe^"^ + H+ (or H + FeCla = FeCl^ + HCl) . 
 
 There exists, then, at the electrode a continually changing concentration of 
 ferric, ferrous, and hydrogen ions, and reliable work is out of the question. 
 The effect of all these influences combined would cause the electromotive 
 force to change steadily, quite independently of any stress and strain effects. 
 The rate of this change would depend on such indeterminate and chance rela- 
 tions as the area of the iron surface to the volume and the concentration of 
 the solution, and on the mechanical arrangement, allowing more or less 
 rapid diffusion, etc. The following experiments are enough to show this. 
 Wires of pure iron were cleaned with fine emery cloth, carefully wiped, and 
 simply immersed in a ferric chloride solution.** They were under no arti- 
 ficial strain of any kind when measured. 
 
 Table 4. — The electromotive force of iron in ferric chloride against the 
 decinormal electrode. 
 
 Piano wire (20). 
 
 Soft wire (31). 
 
 Elapsed time. 
 
 E. M. F. 
 
 Elapsed time. 
 
 E. M. F. 
 
 
 Volt. 
 
 
 Volt. 
 
 Oh 0.^^ 
 
 0.550 
 
 Oh im 
 
 0.595 
 
 5 
 
 0.5725 
 
 5 
 
 0.608 
 
 10 
 
 0.576 
 
 10 
 
 0.610 
 
 30 
 
 0.585 
 
 15 
 
 0.610 
 
 40 
 
 0.588 
 
 15 55 
 
 0.718 
 
 17 
 
 0.743 
 
 16 40 
 
 0.716 
 
 During the first day hydrogen was steadily evolved from both wires, but 
 the second day this side reaction was greatly diminished. As is seen in the 
 table, the electromotive forces rose in these cases respectively 0.192 and 0.123 
 
 Similar results were found by Finkelstein in 1902. (Z. phys. Cham., 39, 91). 
 
^2 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 volt. This rise would have been ascribed by Hambuechen to strain. Further 
 experiment could add nothing to the argument. It will be noticed that even 
 finally the potential was very far from the value in ferrous sulphate, 0.765. 
 
 Hambuechen's results are quite as irregular, as one would infer from these 
 considerations. For example, the change in electromotive force, caused 
 supposedly by tensile stress, varies in two cases from 0.004 to 0.056 volt, and 
 yet in the former case the load per square inch was actually 600 pounds 
 greater than in the latter. Hambuechen realized that some of his observed 
 potentials were much larger than were to be expected, and therefore ex- 
 plained that " One of the assumptions it was necessary to make is that the 
 stress equally affected the entire cross section, which is by no means the 
 case, for it is the outside layer which is affected the most and would there- 
 fore give higher values of electromotive force than the calculated amounts." 
 This is exactly the opposite assumption to that demanded by the reasonable 
 explanation of Burr already quoted (p. 18). 
 
 From the theoretical point of view, the matter appears, then, to be more 
 complicated than might be inferred from a hasty survey of it. One can not 
 safely conclude that the potential on the surface of a wire is an index of the 
 free-energy change in the interior of the wire ; and one can not assume that 
 the electrochemical behavior of wire while under stress is the same as when 
 the stress is removed. Still further, one can not but believe that the perma- 
 nent strain caused by a stress (that is to say, the energy potentialized by a 
 stress) will differ in the case of impure iron from that in the case of pure 
 iron. Moreover, it is by no means certain that energy stored in iron by 
 change of internal structure is available as free energy. 
 
 Taking all these considerations together, and weighing also the electro- 
 chemical defects in some of the papers just discussed, it is apparent that 
 none of these investigations furnish conclusive evidence concerning the 
 magnitude of the changes of potential to be noticed on stretching a rod or 
 wire. It is apparent, moreover, that this change is smaller than is usually 
 supposed, although it is probable that a slight change of electromotive force 
 is caused by a tensile strain of impure iron. 
 
 Qualitative evidence on this point is furnished by a paper entitled '^ The 
 Effect of Strain on the Rate of Solution of Steel," published by Barus in the 
 Bulletin 94 of the U. S. Geological Survey (p. 61), already quoted. He 
 experimented with soft steel, hard-drawn steel, and the latter annealed; he 
 determined the rate of solution by loss of weight, and he concluded his 
 report as follows : " Summarizing the above results as a whole, it follows 
 that the rate of solution of drawn steel is greater than the rate of the same 
 homogeneous metal similarly circumstanced." In applying the results it 
 
VARYING THE TEMPERATURE OF IGNITION. 2$ 
 
 should be noted that steel alone, not pure iron, was experimented upon ; that 
 any speed of reaction in which two phases are involved is complicated by 
 many disturbing factors ; and that the tendency of iron to occlude hydrogen, 
 discussed in a later section of this paper, must have added another 
 complication.'" 
 
 Because our results obtained by applying both compressing and distending 
 stresses upon pure iron had been chiefly negative, indicating very slight 
 changes of free energy for large stresses, attention was now directed to the 
 change of the other conditions of experiment, with the hope of tracing to 
 these other conditions the fairly large changes of electromotive force actually 
 to be observed in specimens of pure iron prepared in different ways. 
 
 THE EFFECT OF VARYING THE TEMPERATURE OF IGNITION. 
 
 It will be noticed on studying tables 2 and 3 that the electromotive forces 
 of cells made from different samples of iron which were chemically alike 
 varied between such wide limits as 0.780 to 0.794 volt, a range far beyond 
 the limits of experimental error. 
 
 In tracing these differences back to their fundamental cause, the first clue 
 was furnished by the different degrees of cohesion or compactness exhibited 
 by the several specimens. When the four pieces of porous iron, the results 
 from which are given in table 3, were removed from the cells at the comple- 
 tion of the electrochemical experiments, they were examined with regard to 
 their cohesion by the simple method of resistance to pressure and to fracture 
 between the fingers. It was thus found that No. 2 iron felt softest. No. 3 
 the hardest and firmest, and Nos. i and 4 somewhere between. Reference 
 to the table shows that No. 2 had the highest electromotive force and No. 3 
 the lowest; hence a more powdery structure evidently went with greater 
 tendency to dissolve. Repetition with other samples confirmed this view. 
 
 What, now, could be the cause of this difference in compactness of the dif- 
 ferent samples? It is well known that many substances, on being heated, 
 begin to cohere at temperatures far below their melting-points, exhibiting 
 the shrinkage commonly called " sintering." It is also well known that this 
 effect becomes more and more noticeable as the melting-point is approached. 
 Probably, then, these different samples had been subjected to different tem- 
 peratures during their reduction. This relationship was easily confirmed 
 by further experiment. Iron reduced at high temperatures was found to be 
 far firmer in structure than that reduced at low temperatures. In order thus 
 to find the highest potential attainable in this way, iron should obviously be 
 
 *• In this connection attention is called to the interesting work of C. S. Burgess on 
 the effect of impurities on the rate of solution, read at the recent meeting of the 
 American Electrochemical Society, May, 1906. 
 
24 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 reduced at the lowest possible temperature. Accordingly two samples were 
 prepared, the first being very slowly reduced at a temperature well below 
 800°. In the second of these reductions the temperature was measured by a 
 Le Chatelier thermopile laid under the boat containing the iron, and the 
 highest temperature it indicated through a reduction lasting three days was 
 615°. The iron in boat No. i was of a clear gray color in those parts where 
 the temperature became highest and dark blackish gray on the surface. All 
 the iron in boat No. 2 was like the top of boat No. i. Even the firmest iron 
 in boat No. i could easily be crushed between the fingers. 
 
 Two samples of the light-gray material from boat No. i gave the follow- 
 ing measurements : 
 
 Table 5. — The electromotive force of porous iron reduced at low temperature. 
 
 Sample 22. 
 
 Sample 23. 
 
 Elapsed time. 
 
 E. M. F. 
 
 Elapsed time. 
 
 E. M. F. 
 
 Oh ^m 
 
 30 
 
 1 
 24 
 70 10 
 
 Volt. 
 0.796 
 0.800 
 0.798 
 0.799 
 0.796 
 
 Oh 14m 
 
 1 
 
 24 5 
 
 70 40 
 
 92 
 
 Volt. 
 0.797 
 0.797 
 0.799 
 0.800 
 0.800 
 
 These values are somewhat higher (0.002 and 0.006 volt) than any ob- 
 tained in previous work, and are further evidence in favor of believing that 
 the low temperature of reduction really causes a product having a higher 
 potential. The contents of boat No. 2 were found to contain much hydrogen, 
 and will be discussed later. 
 
 Having thus shown that iron reduced at low temperature may give in 
 combination with the decinormal electrode a potential as high as 0.800 volt, 
 and that increasing temperature of reduction causes a steady decrease in this 
 value until the metal prepared at perhaps 1,200° gives a value as low as 0.776, 
 it became a matter of interest to discover if iron subjected to a still higher 
 temperature might not give a still lower value. The obvious subsequent 
 step was therefore to test iron which had been fused, and accordingly several 
 specimens of the purest commercial iron (Swedish bar iron and pure piano 
 wire) were tested in preliminary experiments. The Swedish bar iron was 
 very pure, and soon reached a constant value. The piano wire was less 
 regular, and hence was examined more thoroughly. The wire was " hard 
 drawn," and its purity was indicated by the fact that when heated to 1,000° 
 in hydrogen and quenched in ice-water it was as soft and pliable as the 
 purest iron. Two pieces of this wire were scraped bright and clean with a 
 
COMPACT IRON PREPARED AT HIGH TEMPERATURE. 
 
 25 
 
 dull knife and measured as usual in a cell shown in figure 3, page 10. The 
 potential of these cells began at very low value, and gradually rose to a con- 
 stancy at about 0.76 volt. It was evident that the end-point is reached only 
 after a long time; and accordingly six wires were sealed up in ferrous 
 sulphate solution. Three of these had been scraped as before to clean them 
 and the other three were rubbed with fine emery cloth, all of them being 
 carefully wiped with a cloth before immersion. Below are given the final 
 constant values attained by these samples : 
 
 Table 6. — The electromotive force of commercial iron of moderate purity. 
 
 Elapsed 
 time. 
 
 Swedish 
 bar iron. 
 
 Piano wire (hard drawn). 
 
 Sandpapered. 
 
 Scraped. 
 
 No. 24. 
 
 No. 25. 
 
 No. 26. 
 
 No. 27. 
 
 No. 28. 
 
 No. 29. 
 
 No. 30. 
 
 3 hours 
 
 1 day 
 
 3 days 
 
 4 days 
 
 7 days 
 
 Volt. 
 0.765 
 0.770 
 0.770 
 
 Volt. 
 
 0.762 
 0.762 
 0.762 
 
 Volt. 
 
 0.762 
 0.762 
 0.763 
 
 Volt. 
 
 ' 'o*.762 
 0.762 
 0.763 
 
 Volt. 
 
 " 0.767 
 0.763 
 0.765 
 
 Volt. 
 
 'o!769 
 0.767 
 0.767 
 
 Volt. 
 
 0.765 
 0.765 
 0.763 
 
 Thus all these samples yield a result lower than the most compact of the 
 sintered samples. The results show excellent agreement, and the number 
 0.765 volt may be taken as the true electromotive force of the cell made 
 from this wire. The Swedish iron gave a slightly higher value, about 0.770. 
 A third sample of commercial iron, the softest and -purest wire obtainable, 
 was next examined. It was cleaned with fine emery cloth and wiped with a 
 clean cotton cloth. Two specimens after two days gave respectively 0.760 
 and 0.771 volt, or in mean 0.766, essentially the same as before. 
 
 Thus all the commercial compact samples of iron showed a distinctly lower 
 solution tension than any of the spongy pure samples. Before this differ- 
 ence could be certainly ascribed to any peculiarity in the freshly reduced 
 spongy iron, proof must be afforded that it was not due to impurities in the 
 commercial samples. This possibility was most conveniently subjected to 
 trial by fusing a sample of pure reduced iron, and comparing the product 
 with the commercial sample. At first an attempt was made to prepare pure 
 fused iron by fusing a small quantity in a small lime. crucible in an oxyhy- 
 drogen flame. The hydrogen used was very pure, having been made by 
 electrolysis, for impurities would be taken from the gas by the fused iron. 
 The flame was so directed as to cover the iron completely and was provided 
 with excess of hydrogen, but nevertheless all the iron oxidized. The oxide 
 unfortunately did not act as a protection to the iron underneath, because it 
 fused and was absorbed by the porous lime. No acceptable variations of 
 
26 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 apparatus or manipulation greatly bettered the result, which could have been 
 easily accomplished with larger quantities of material. Therefore another 
 method was tried, namely, the melting of the iron in a vacuum by means of 
 the heat generated by its own resistance to a great electric current. Be- 
 sides an apparatus suitable for exhaustion, the essentials for success were 
 iron electrodes, a suitable crucible, and a sufficiently high current. Bundles 
 of piano wire provided the electrodes, and a nest of powdered iron, sup- 
 ported on a crucible of lime made from the purest marble, was used to 
 contain the melted particles. In order to increase the resistance of the 
 metal it was powdered in the Hempel steel mortar, and the current was 
 passed through this powder. The arrangement of the apparatus and its 
 description are given in figure 5. 
 
 The operation of the experiment was as follows: After everything was 
 arranged as shown in the figure, the Sprengel vacuum-pump was started 
 and the stoppers thus firmly forced in. This, however, disarranged the elec- 
 trodes and air was again admitted, so that they could be carefully placed at 
 the desired places in the lime boat. The apparatus was now evacuated a 
 second time and the current turned on. For some time nothing happened, 
 but finally the metal fused with the help of vigorous jarring. More and 
 more powder was added from above as the fusion progressed. Tliere were 
 brilliant flashes of violet light with each addition of powder, followed by 
 bright-red glowing. By feeding powder into the open spaces left by the 
 fused material it was possible to maintain the fusion almost continually until 
 the supply was exhausted. The product consisted mainly of small pellets 
 about the size of the head of a pin or a little larger, resting in the nests of 
 pure powdered iron. Only one was large enough to use conveniently in a 
 cell. Nevertheless, this gave a result agreeing so well with that from the 
 commercial materials that a duplicate seemed unnecessary. The sample was 
 covered with soft paraffin on those parts which did not have a smooth fused 
 surface. 
 
 Table 7. — Pure iron fused in vacuum. (33). 
 
 Elapsed time. 
 
 E. M. P. 
 
 Elapsed time. 
 
 Oh 7m 
 9 
 20 
 40 
 
 Volt. 
 0.745 
 0.750 
 0.757 
 0.760 
 
 40h 40m 
 
 41 30 
 72 10 
 90 25 
 
 E. M. F. 
 
 Volt. 
 0.761 
 0.759 
 0.763 
 0.761 
 
 Thus this sample of exceedingly pure iron fused in vacuum gave essen- 
 tially the same potential as good quality commercial material (0.765). 
 
PURE IRON HEATED IN VACUUM. 
 
 n 
 
 This iron must have been wholly free 
 from carbon, from any other metal, and 
 indeed even from all but the merest 
 trace of hydrogen. It is hence clear that 
 the impurities in the commercial iron 
 could not have been enough to affect its 
 electromotive force, and that pure, com- 
 pact iron really gives a lower value than 
 the same iron reduced by hydrogen. To 
 what cause, then, is this difference to be 
 ascribed? Is the higher solution tension 
 of the spongy iron due to the traces of 
 hydrogen within it, or to difference of 
 internal structure, or to the disposal of its 
 external surface? 
 
 The first of these three possible causes 
 of an extraordinarily high potential does 
 not seem to be a very probable one; 
 nevertheless, it seemed worthy of test, 
 especially in view of the results on oc- 
 cluded hydrogen to be recorded later. 
 Accordingly, several samples of spongy 
 iron were ignited for varying lengths of 
 time at varying temperatures in a 
 vacuum, while inclosed in a stout porce- 
 lain tube. It was thus found that while 
 short ignition or ignition at a low tem- 
 perature left much of the gas in the 
 metal, treatment at a higher temperature 
 for a longer time seemed to remove 
 nearly all of it, while causing the iron to 
 sinter into a compact mass. Iron thus 
 treated at i,i6o° gave in conjunction 
 with the decinormal electrode the usual 
 potential 0.787 in a few hours, a value 
 which remained unchanged 
 even in the last decimal place 
 after ninety hours. Even this 
 iron, however, undoubtedly 
 contained traces of hydrogen, although 
 much less than at first. 
 
 oh 
 
 ^^i:=r| 
 
 ^ 
 
 
 \ 
 
 
 0. 
 
 
 .a 
 
 
 
 
 .ti 43 
 
 
 ^ .2 
 
 
 CO 
 
 
 ^''O -3 
 
 
 4; jLJ g 
 
 
 
 
 .S « 
 
 s 
 
 j^«S^ 
 
 
 ^ .2!% 
 
 p 
 
 
 lA U >yA 
 
 < 
 
 > 
 
 
 i 
 
 Porcelain 
 Lime cru 
 Iron pow 
 Feed, gl 
 operate 
 e about o 
 
 »-H 
 
 cqG»~^'fe: s 
 
 
 
 C/2 
 
 iz; 
 
 
 l-H 
 
 
 in 
 
 
 £ 
 
 
 M 
 
 
 g 
 
 11 
 
 CO 
 
 
 ^ 
 
 o'S 
 
 H 
 
 \* ctf 
 
 i 
 
 0. 
 
 < 
 1 
 
 . !i 
 
 1. 
 
 0. w S 
 
 tn 
 
 i s^i 
 
 2 
 
 ^d's^s 
 
 fe 
 
 >>3 rt H «J 
 
 
 « 3 Cii O.'S 
 
 
 c &L(u a2 
 
 
 chim 
 to va 
 
 to d 
 ■wire 
 r sto 
 
 sevei 
 d). 
 
 
 %^m 
 
 
 i-JHHia^« 
 
 
 '^^lli^'c^i 
 
 
 
 
 4ttfco 
 
26 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 The permanence in the value in the potential is in itself the best evidence 
 that the hydrogen, even if present, is so shielded from the electrolyte as not 
 to affect the measurement of electromotive force. It is improbable that a 
 trace of occluded gas could exercise a lasting effect in maintaining the 
 potential above that of the reversible electrode of iron in its sulphate. This 
 matter will be discussed in detail in the second section of the paper. 
 
 There are two other hypotheses capable of explaining the abnormal poten- 
 tial of the spongy iron. It is certain that the " sintering ' is increased and 
 the surface of the iron diminished in this way ; and it is possible that change 
 in the degree of polymerization of the iron, resulting in the formation of a 
 more stable modification, is likewise effected.^" Either of these changes 
 might affect the electromotive force. A method of deciding between these 
 two would demand either the formation of compact iron at low temperature, 
 or else of loose spongy iron at high temperature — and neither is capable of 
 certain preparation. The former was nearly obtained in the experiment first 
 recounted, in which the spongy iron was rendered compact by great pressure ; 
 the latter can hardly be obtained by mechanical subdivision. 
 
 It is to be noticed that of the first three specimens used in the experiments 
 recorded in table 3, the untouched porous pieces gave an average potential 
 of 0.783, while the same pieces after cold-welding at very high pressure gave 
 the essentially identical value 0.782. The cold-welded pieces were very hard 
 and tenacious, and appeared qualitatively in every way similar to iron which 
 had been fused (which gave only 0.765 volt) ; but it is not certain that some 
 particles on the surface were not as loosely coherent as in the sponge, and 
 hence capable of giving as high a potential. It is true that one fact mili- 
 tates against this possibility — the fact, namely, that the compressed pieces 
 steadily rose to the maximum potential and staid there a long time, which 
 could have hardly been the case if the high potential had depended upon a 
 few particles. 
 
 Because of the experimental difficulties it is impossible now to decide defi- 
 nitely between these possible causes of difference between the spongy and 
 compact metal. It is possible that all three are superposed ; but of the three. 
 
 ""The allotropic theory of iron need not be discussed here. It has satisfactorily 
 explained so many phenomena and has had such wide usefulness that it is universally 
 accepted by scientific metallurgists. Its theoretical side from the standpoint of the 
 phase rule is thoroughly treated by Roozeboom, Z. f. Phys., ch., 34, 436 (1900). The 
 metallurgist's use of it is shown by Baron von Juptner in his work " Grundziige der 
 Siderologie," vol. 2, 1901. See also Berichte der d. Chem. Gesell, 39, 2381 (1906). 
 
 A short explanation of the terminology may not be out of place. Pure iron has at 
 least two well-defined transition points. Iron stable below 766° is known as alpha 
 iron, between 766° and 895° as beta iron; and the phase formed above 895° is called 
 gamma iron. 
 
THE SUDDEN COOLING OF PURE IRON. 29 
 
 the writers are rather inclined to believe that the chief cause is probably the 
 state of subdivision — the very finely divided particles giving a higher electro- 
 motive force than the fused mass." If this is true, the case is one of great 
 interest because of the magnitude of the effect, corresponding as it does to 
 a tenfold greater solubility of the finest particles. 
 
 In this connection it was deemed worth while to test the electromotive 
 force of iron which had been suddenly cooled or quenched. In this way it 
 might be found whether or not pure iron, when suddenly cooled, could be 
 caught in a condition of internal structure less stable than when cooled 
 slowly. 
 
 Preliminary experiments, to be recounted later in another connection, 
 showed that in order to obtain significant results for this purpose the heat- 
 ing and sudden cooling must occur in the absence of any foreign substance 
 capable of reacting upon the iron ; therefore the metal must be quenched in 
 the absence of any gas or water. 
 
 The apparatus needed was not complex. A stout tube of Berlin porcelain 
 was clamped in a vertical position, closed at both ends by Hempel water- 
 cooled stoppers, and was heated in the middle by an encompassing perforated 
 Fletcher furnace. The temperature was registered by a platinum-rhodium 
 thermopile. The porcelain plate on which the test-pieces of iron were laid 
 was suspended in the middle of the tube by an iron wire on one side and a 
 small silver wire on the other. On the lower Hempel stopper lay a smooth, 
 thick iron plate, and ice-cold water was circulated through the stopper, 
 while the whole lower end of the tube was packed in ice. The apparatus 
 was now evacuated with a Sprengel mercury pump, and when the pressure 
 had fallen to less than i mm. the tube was heated. For ten minutes the 
 temperature was kept above i,ooo° ; the silver wire was melted and one side 
 of the porcelain plate was dropped. The iron of course fell from this region 
 of high temperature to the cold iron plate. The apparatus was left to cool 
 over night, without the admission of a trace of air. The two pieces of iron 
 when put in had been blackish-gray and very loosely knit together; when 
 taken out they were quite cohesive and of a clear-gray color, exactly 
 resembling iron which had been reduced in hydrogen above 700°. This 
 illustrates the effect of sintering and proves that the. original dark color of 
 the iron powder was not due to incomplete reduction, but solely to its state 
 of division. One surface of one of the two pieces was found lying flat on the 
 cold iron plate and this one was immersed in ferrous sulphate and measured. 
 It gave in eight minutes a potential of 0.788, which slowly rose to 0.805 in 
 an hour, and then settled down to constancy at the value of 0.795. 
 
 Ostwald, Zeitschr. fur Phys. Chem., 34, 45 (1900) ; Hulett, ibid., 37, 385 (1901). 
 
30 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 In repeating the experiment two improvements were introduced. For 
 fear that some silver might have volatilized and contaminated the iron, the 
 silver-wire support was replaced by a simple mechanical device, one side of 
 the porcelain plate supporting the iron being held in place by a pipe-stem, 
 while the other side was attached to an iron wire running through a lubri- 
 cated hole in the stopper. By pulling the wire the iron could be dropped. 
 Furthermore, quick and effectual cooling was obtained by surrounding the 
 outside of the base of the porcelain tube, containing the iron cooling-plate, 
 by a mixture of chloroform and solid carbon dioxide. Time was allowed for 
 the thorough cooling of the inclosed iron plate. The temperature of porous 
 iron was thus almost instantaneously lowered from i,ioo° to — 75° or there- 
 abouts. Great pains was taken to exclude the vapor of water, which would 
 have condensed in the cold part of the tube and vitiated the experiment. 
 This was accomplished by running a stream of pure dry hydrogen through 
 the tube for many hours. 
 
 As soon as possible after the fall of the porous iron, which took place 
 this time in an atmosphere of hydrogen, the pieces were taken out, immersed 
 in ferrous sulphate, and measured in the usual way. In six minutes after 
 immersion the potential was 0.790, in twenty minutes 0.796, and in an hour 
 0.798. After a day it settled down to perfect constancy at 0.793 volt, essen- 
 tially identical with the result of the previous experiment. This is not 
 enough above the usual value (0.787) obtained from iron slowly cooled from 
 the same high temperature to allow the conclusion that the sudden cooling 
 had made any considerable difference in the potential or difference in internal 
 structure. 
 
 It becomes now an interesting question as to which of the many values 
 for the gradually changing potential of the cell Fe, ftFeSO^, V^oKCl, HgCl, 
 Hg correspond to the true values of the solution tension of the different 
 forms of iron. Consideration shows that the final high values are probably 
 to be taken in each case, for several reasons. In the first place, these values 
 are approached asymtotically, and are then maintained at a constancy for a 
 long time. For example, one sample of iron which in 16 hours had given 
 in the cell a potential of 0.793 volt, after 150 hours more was still practically 
 unchanged at 0.792 volt. In the next place, the different kinds of iron, 
 although sometimes beginning at impossibly low values, all finally rise to 
 nearly the same point. Again, it is not conceivable that the iron should 
 gradually raise itself to a potential above its true value and maintain itself 
 there in a reversible reaction with a large excess of both iron and ferrous 
 sulphate at hand, while it is easy to see that a coating of oxygen would lower 
 the potential at first. The conclusion is, moreover, reinforced by the fact 
 that the same potential is attained from the opposite direction by the gradual 
 
THE SINGLE POTENTIAL OF THE PURE IRON ELECTRODE. 3 1 
 
 falling off of a higher potential artificially and temporarily created in the 
 metallic electrode with the help of hydrogen conveyed to the system from 
 without, as will be explained in the succeeding section of the paper. Hence 
 there can be little doubt that the potential of porous iron reduced at about 
 1,000° and immersed in normal ferrous sulphate is really as high as 
 0.79 — 0.612 := 0.18 volt, if the decinormal calomel electrode is taken as 0.612, 
 the normal calomel electrode being taken as 0.56. Pure iron which has been 
 fused in a vacuum gives a value nearly 0.03 volt lower, or about 0.15 volt. 
 This latter value, corresponding to the flat surface of a coherent mass, is the 
 one to be chosen ordinarily as the normal value. Even this is nearly twice 
 as large as that usually ascribed to it. The errors of the work of others were 
 probably insufficient time of immersion and the presence of acid in the electro- 
 lyte, both of these tending to lower the observed value, as is shown above. 
 
PART II. THE ELECTROMOTIVE FORCE OF IRON CON- 
 TAINING OCCLUDED HYDROGEN. 
 
 The fact that hydrogen freely penetrates iron at red heat and that iron 
 retains some of this hydrogen on cooHng was discovered as long ago as 1866 
 by Graham ; but the more detailed study of this and related phenomena has 
 been undertaken only in recent years, after its importance in the metallurgy 
 of the metal had been established. Many papers have been published on this 
 topic, but the field has been by no means thoroughly treated. 
 
 Our attention during this research was first directed to the well-known 
 affinity of iron for hydrogen by observing the copious evolution of this gas 
 from metal reduced at a low temperature.^ Iron obtained from the oxide at 
 570° is, as well known, a fine powder, which oxidizes easily, and hence after 
 even a short exposure to the air gives a low electromotive force. In order 
 to test properly the free energy of such iron, it must be cooled in pure hydro- 
 gen and tested instantly after opening the reduction-tube. In this way it 
 was found that the potential is not essentially different from iron reduced 
 at 700°, being about 0.795 volt in conjunction with the decinormal electrode. 
 
 Nevertheless, such powder, upon immersion in ferrous sulphate evolved 
 copious bubbles of gas, which were proved to be hydrogen. That this hydro- 
 gen was not in any extraordinary condition is shown by the fact that it 
 neither altered the potential of the iron nor caused the iron containing it to 
 produce any unusual reducing effect on mildly oxidizing solutions. It is 
 probable that the hydrogen thus held by the fine powder was merely adsorbed, 
 as the gas is adsorbed by charcoal. 
 
 That the gas could exist in the metal in a radically different condition we 
 did not at first suspect ; but this conclusion was forced upon us by the sur- 
 prising and quite unexpected results of several other experiments. 
 
 In the first place, several pieces of porous iron which had been reduced at 
 high temperature, slightly oxidized by long standing in the air, and then 
 reheated in hydrogen at 575° gave at first an extraordinarily high value for 
 the usual cell couple, namely 0.825. In seven hours this had decreased to 
 0.810, and in five days became constant at 0.798, essentially the normal value. 
 
 ^ Baxter has shown that pure iron reduced at a high temperature contains but little 
 occluded hydrogen. Am. Chem. Journ., 22, 363 (1899). 
 
 32 
 
THE QUENCHING OF IRON. 33 
 
 A slight indication of an excessive value of this kind had already appeared 
 in other previous samples, many of which after a short immersion in ferrous 
 sulphate showed a slight maximum. This was noticeable even in samples 
 which had been ignited in a vacuum. In the cases mentioned at the begin- 
 ning of this paragraph, after standing some time, bubbles of hydrogen 
 appeared around the metal, as the electromotive force decreased. This 
 seemed to show that in some way the excessive electromotive force was con- 
 nected with hydrogen. But hydrogen gas has a lower, not a higher potential 
 than iron. Therefore, it is clear that if the abnormality is produced by 
 hydrogen, this impurity must exist in the metal in a different form — no 
 longer as merely adsorbed gas on the surface of the fine powder, but in some 
 new state, inclosed in the less open structure of the sintered iron. 
 
 More light was thrown upon all these matters by interesting series of 
 experiments which showed that the dissolved active hydrogen could be more 
 easily introduced into the iron in other ways. There follows a brief descrip- 
 tion of these experiments. 
 
 The first series furnishing this further light was a set of experiments 
 originally begun as a preliminary attempt to cause a change in the internal 
 structure of iron by quick cooling. In it iron was plunged while hot into 
 water. 
 
 In order to quench iron suddenly from a high temperature without coating 
 the metal with a hardly soluble film of oxide, it was clearly necessary to 
 conduct in an atmosphere free from oxygen both the heating of the metal 
 and the transference to the cooling agent. 
 
 An atmosphere of hydrogen was first used for this purpose. As before, 
 a stout tube of Berlin porcelain was erected in a vertical position, the middle 
 portion being heated by a Fletcher furnace. Pure electrolytic hydrogen, 
 thoroughly washed and dried, was supplied at the top of the tube through a 
 cooled Hempel stopper, which was further protected against the rapid con- 
 vection of hot hydrogen by a pipe-bowl suspended at the upper limit of the 
 heated zone by means of an iron wire. The iron was heated at first in an 
 unglazed basket of the best porcelain, and in the later experiments on a 
 shelf or disk of the same material. Simple mechanical devices as before 
 enabled the iron to be plunged or dropped quickly into the cooling agent at 
 the base of the tube. The temperature of heating. was determined by a 
 Le Chatelier platinum-rhodium thermoelectric junction. Cold boiled water 
 was used as the cooling agent at the base of the tube, the operation being, 
 therefore, merely the quenching of pure iron without exposure to oxygen gas. 
 
 Even the first experiments gave interesting results. Porous iron reduced 
 at 800° was quenched from 1,000*^. Twenty minutes afterwards the super- 
 
34 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 ficially oxidized sample was immersed in ferrous sulphate and half an hour 
 later it gave the value 0.89 volt against the decinormal electrode, instead of 
 only 0.79 as before. This extraordinary value was only temporary, the 
 electromotive force falling rapidly, at the end of one hour being 0.825 volt 
 and of five hours, 0.803. Three other pieces quenched at the same time gave 
 similar results. Because the electromotive force fell rapidly it appeared pos- 
 sible that an immediate reading might have given even higher results. A 
 repetition of the experiment with frequent readings was evidently very 
 desirable. 
 
 The second quenching was a repetition of the first. To remove the 
 adhering water after the operation, the samples of iron were rinsed for a 
 minute or two in ferrous sulphate solution before putting into the cells, thus 
 preventing dilution of the normal solution in which they were measured. 
 The results verified those of the first quenching, but added to them, in that 
 the later ones showed the speed of decrease in potential to be much less in 
 the first half hour than afterwards. The potential started at about 0.9 volt, 
 and after remaining almost constant for thirty minutes began to fall off 
 rapidly. A number of other similar quenchings were made, most of which 
 gave similar results, only a few showing no excess of potential. Typical 
 examples are given in the following table, under the heading. Sample 44. 
 
 Further experiments were made to determine if the temperature before 
 quenching caused any considerable effect on the results. 
 
 The apparatus used was the same as the one which served in the previous 
 experiments. One sample was quenched from 695° and another from 600°. 
 In both cases the quenched iron rose to a maximum electromotive force less 
 strikingly above the normal value, the former attaining 0.87 volt and the 
 latter 0.82 volt, which potentials, as usual, settled down to the normal values 
 of 0.795 in the course of half a day. These experiments showed that the 
 excessive value of the electromotive force was directly dependent upon the 
 temperature before quenching, a higher temperature giving a greater excess. 
 They added to the evidence already given that the " gamma " or high-tem- 
 perature phase of iron could have nothing to do with the excessive values, 
 because the latest quenching occurred from a temperature below the recog- 
 nized transition point. It will be remembered in this connection that pure 
 iron quenched from 1,100° on a cold iron plate gave normal values for its 
 electromotive force. 
 
 The experiment was next made of reheating the quenched iron in hydro- 
 gen in order to discover whether or not it would return to its normal condi- 
 tion. Two pieces of this same quenched iron were thoroughly dried in 
 alcohol and ether and three days later were reheated in hydrogen for one 
 hour (the highest temperature being faint redness) and then slowly cooled. 
 
THE SOURCE OF THE OCCLUDED HYDROGEN. 
 
 35 
 
 The results, together with those of the previously described quenched iron, 
 are given in the following table : 
 
 Table 8. — The effect of reheating quenched iron. 
 
 Time elapsed 
 
 after 
 
 inijnersion. 
 
 Sample 44, 
 quenched, not reheated. 
 
 Sample 46, 
 
 reheated 
 
 iron. 
 
 Sample 46, 
 
 reheated 
 
 iron. 
 
 Fresh. 
 
 Exposed to air. 
 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Volt. 
 
 Oh 3m 
 
 0.880 
 
 0.79 
 
 0.796 
 
 0.794 
 
 11 
 
 0.910 
 
 0.80 
 
 0.800 
 
 0,801 
 
 20 
 
 0.906 
 
 0.90 
 
 0.800 
 
 0.804 
 
 45 
 
 0.890 
 
 0.89 
 
 0.797 
 
 0.804 
 
 2 
 
 0.810 
 
 0.81 
 
 0.797 
 
 0.803 
 
 4 
 
 0.797 
 
 0.80 
 
 0.797 
 
 0.802 
 
 It is clear at once that samples 45 and 46 had once more acquired the 
 same condition they possessed before they were quenched — as was to be 
 expected. Thus there can be no doubt that the reheated iron returns essen- 
 tially to its normal condition. 
 
 From these experiments it was clear that something, either chemical or 
 physical, had happened to the iron during its heating and plunge into water 
 which caused it to assume an excessive potential. Two possible causes 
 might underlie these interesting results. Either the sudden cooling might 
 have preserved the iron in an unstable allotropic form yielding a higher 
 electromotive force, or else occluded hydrogen might in some way be respon- 
 sible for the differences. The former of these explanations was disproved 
 by the already described experiments on cooling iron in a vacuum on a cold 
 iron plate. Hence only the latter explanation remained, and the phenomenon 
 was obviously to be classed with the occlusion of hydrogen from the hot gas. 
 
 It now became an interesting matter to find whether this hydrogen came 
 from the hydrogen gas or from the water during the instant of quenching. 
 This was easily decided by an experiment in which the metal was heated in 
 and quenched from an atmosphere of nitrogen. The quenching apparatus 
 was exactly the same as that used in all the previous experiments. The 
 nitrogen was prepared by Wanklyn's method of blowing air through aqua 
 ammonia, passing this ammoniacal air over hot copper gauze, and purifying 
 the resulting nitrogen in the usual way. All rubber used in the apparatus 
 had been boiled in caustic solution and was thickly coated on the inside with 
 the semi-solid paraffin. In this apparatus a typical sample of porous iron 
 was heated for twelve minutes in nitrogen at a temperature of 1,040° and 
 was then quenched in ice-water. Of the two pieces thus quenched, one was 
 
36 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 immersed in the ferrous sulphate solution five minutes after quenching, 
 while the other was thoroughly dried in alcohol and ether and put into the 
 cell to be measured on the following day. 
 
 The results agreed in a general way with those already obtained by 
 quenching from an atmosphere of hydrogen. They differed in two respects. 
 The maximum electromotive force was 0.03 volt higher than any other thus 
 far obtained, rising to 0.94 volt, and this high electromotive force was 
 retained for a much longer period of time. These are differences in degree 
 only, and since the other results differed among themselves, although to a 
 less extent, the higher value is to be ascribed merely to more favorable 
 experimental conditions. 
 
 Could the formation of iron nitride in any way have produced the high 
 potential in this case? Reflection promptly decided this question in the 
 negative. Iron nitride is formed only at low temperatures and is decom- 
 posed even in a stream of ammonia at higher temperatures ; " free nitrogen 
 is not able to unite with iron, even when the latter is reduced in nitrogen, 
 and the formation of nitride is dependent on the coming in contact of nascent 
 nitrogen with iron." ^ Moreover, the nitride begins to decompose in a 
 stream of nitrogen at 600°, and the nitride FcgN is an exothermic compound. 
 Hence probably it has a lower electromotive force than iron,^" not a higher 
 one, and the nitride is evidently out of the question. 
 
 Upon standing in ferrous solution the quenched iron evolved an appre- 
 ciable volume of gas, which was shown to be hydrogen by explosion with 
 oxygen in a micro-eudiometer. 
 
 In view of this fact, and because iron quenched from nitrogen gave the 
 same high potential as iron quenched from hydrogen, there can be no doubt 
 that the bulk of the occluded hydrogen must have come from the water at 
 the moment of quenching. Some of the iron must have been oxidized, and 
 the resulting nascent hydrogen must have dissolved in the iron, in its active 
 state. No other explanation seemed to be compatible with the extraordinary 
 potential observed. 
 
 Other workers on this subject (for example Heyn,^ who found that 
 hydrogen exists in iron quenched from above 730° in an atmosphere of 
 hydrogen) have usually supposed that the occluded impurity was taken 
 from the gas and not from the water, but our experiments show con- 
 clusively that the water is the source of the greater part of the hydrogen. 
 
 «« Stahlschmidt, Fogg. Ann., 125, 37 (1865). 
 
 ** Assuming total energy change to be an approximate guide to free energy change 
 in reactions of this kind. 
 
 '"Stahl und Eisen, 20, 837 (1900). See also Roberts-Austen, Fifth report of the 
 Alloys Research Committee, Proc. Inst. Mech. Eng., 1899, p. 35. 
 
THE ELECTROMOTIVE FORCE OF OCCLUDED HYDROGEN. 3/ 
 
 The discovery that in this case nascent hydrogen may be absorbed more 
 freely than the gaseous material by iron at once reminded us of the well- 
 known absorption of the light element by metals used as a cathode in acid 
 solutions. Cailletet*' and Johnson*^ showed in 1875 that iron will take up 
 at ordinary temperatures electrolytic hydrogen deposited upon it, while 
 ordinary gaseous hydrogen has no effect. Johnson's work was particularly 
 interesting. His iron was in the form of soft wire and his simple method of 
 testing for hydrogen was to bend the wire to test its brittleness, since he had 
 discovered the interesting fact that hydrogen imparted this property to iron 
 previously tough. Immersing the wire in acids which react on it to form 
 hydrogen uniformly made it brittle. Also when the fractured surface was 
 wetted with water it was seen to froth. Making the wire the cathode in 
 neutral, acid, or alkaline solutions had the same effect, but iron anodes were 
 unaffected even in the acid solution. Other wires were placed in a bottle 
 full of water and hydrogen was made to bubble violently through the water, 
 without any trace of absorption. 
 
 Another interesting research yielding similar results was conducted by 
 Bellati and Lussana.^ A barometer was closed at the top by an iron plate, 
 and by cementing a glass ring on this an electrolytic cell was made in which 
 hydrogen was generated at the iron plate ; the mercury at once fell by dif- 
 fusion of the hydrogen through the plate into the barometric vacuum. This 
 work was later confirmed and amplified by Shields.'' 
 
 In view of these phenomena, it became a very interesting point to dis- 
 cover if the active hydrogen introduced into cathode iron from solution is 
 capable of raising the potential of a normal cell to the value of 0.93, that 
 observed with spongy iron quenched from 1,100° in nitrogen. Shields had 
 already observed that this occluded hydrogen raised the electromotive force 
 of the iron above the normal value, but the maximum had not been 
 determined. 
 
 This point was easily tested. In order to obtain satisfactory evidence it 
 is of course necessary that iron alone should be immersed as cathode; the 
 platinum wire supporting it must be above the liquid. Preliminary experi- 
 ments showed that a large excess of potential is as a matter of fact attained, 
 but that it falls off with very great rapidity. Spongy iron which had been 
 
 ^"Comptes rendus, 80, 319 (1875). 
 
 ''Proc. Roy Soc. (London), 23, 168 (1875). 
 
 ^^ Bellati and Lussana, Z. Phys. Chem., 7, 229 (1891). 
 
 '"Shields, Chem. News, 65, 195 (1892). Further discussion of this matter will be 
 found in papers mentioned later; see also Hoitsema, Z. Phys. Chem., 17, i (1895); 
 Winkelmann, Drude's Ann., 8, 388 (1902) ; Richardson, Nicholl and Parnell, Phil. 
 Mag. (6), 8, I (1904). St. Schmidt considers the assumption of a split of the hydro- 
 gen molecule to be unnecessary and unwarranted. (Drude's Ann., 13, 747 (1904).) 
 
38 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 made the cathode of a dense current in an acid sohitlon for an hour was 
 quickly washed with neutral ferrous sulphate, and was measured as soon as 
 possible. In half a minute after immersion it had sunk to 0.822 volt, in four 
 minutes later 0.800, and in an hour to constancy at 0.790. Evidently the 
 hydrogen is only very superficially deposited, and most of it is gone before 
 measurement is possible. Longer exposure as cathode might have more 
 effect. Much more striking results were obtained from electrolytic iron 
 obtained from a neutral solution with the help of a current so dense as to 
 deposit hydrogen with the iron. Such a specimen (sample 48), measured 
 about three seconds after the breaking of the current, gave a potential of 0.93, 
 equal to that observed in the quenched iron. In a minute this had sunk to 
 0.85 volt ; but even after an hour it remained as high as 0.83 volt, not sinking 
 to about the normal value 0.798 for several days. 
 
 Thus it is clear that as far as potential measurements can show, the 
 hydrogen driven into iron with the help of electrical energy is essentially 
 similar to that absorbed in the act of quenching, although the latter operation 
 seems to be especially favorable to the occlusion. Probably the action of 
 the red-hot core upon the steam produced by the exterior of the porous metal 
 supplies the nascent hydrogen to the iron at a temperature especially suitable 
 for occlusion. 
 
 It is interesting to note that even without outside electrical assistance 
 hydrogen in this active form may be taken up by iron from acid solutions. 
 Upon immersing the metal in an acid solution of ferrous sulphate, hydrogen 
 is of course evolved, and the potential of the iron while still in the acid is 
 lowered as much as two decivolts — evidently by the coating of gas, for 
 hydrogen gas has a potential much lower than iron. But besides this super- 
 ficial eflfect, a more deep-seated one is occurring, as is easily shown. 
 
 Upon removing this iron quickly to a neutral solution of ferrous sulphate 
 and measuring at once, its potential is found to be above the normal value, 
 the usual cell often reading as high as 0.83 instead of 0.78 or 0.79. In a 
 few hours it settles down as usual to its normal level, evolving in the process 
 bubbles of hydrogen gas. The iron alone could not of course raise the active 
 hydrogen which it absorbs to a concentration above that capable of giving 
 an electromotive force equal to its own (0.79 with the decinormal electrode), 
 but with the help of the osmotic pressure of the ionized hydrogen in the 
 acid more hydrogen is driven in. This excess manifests its potential and 
 changes to hydrogen gas when the surrounding acid is removed. As a 
 change of only tenfold in the concentration of the ionized hydrogen would 
 be expected to produce a change of potential of nearly 0.06 volt, the observed 
 effect is by no means excessively large. Stated in another way, it may be 
 
THE DISCHARGE OF THE EXCESSIVE POTENTIAL. 39 
 
 said that the hydrogen taken in by iron from an acid solution is in equilibrium 
 with nascent hydrogen, and therefore possesses a high chemical potential. 
 
 In the course of the quenching experiments it was found that iron 
 charged with active hydrogen lost its impurity much more quickly when 
 immersed in ferrous sulphate than when immersed in water or kept in air. 
 This interesting fact seemed worthy of more careful study, because it might 
 be capable of throwing light on the singular occlusion. Accordingly the 
 following series of experiments was instituted upon six pieces of iron, all 
 quenched at once under such conditions that all were exactly alike : 
 
 Piece No. 51 was immersed in ferrous sulphate and measured immediately 
 after quenching. Its electromotive force began at about 0.8 volt, gave the 
 maximum value 0.91 volt after a few minutes, and then fell off as before, 
 reaching 0.90 volt in half an hour and 0.793 volt in four hours. 
 
 No. 52 was left in the water in which it was quenched for an hour and 
 twenty minutes before measurement. When the electromotive force had 
 reached its maximum of 0.90 volt after 26 minutes subsequent immersion 
 in ferrous sulphate. No. 51 had passed that point over an hour and a half 
 before. 
 
 No. 53 was also left in water and was first measured after two hours and 
 forty minutes had elapsed. By the time it gave its highest electromotive 
 force, 0.91 volt. No. i had dropped to 0.803 volt. 
 
 No. 54 was left in water for 26 hours and then immersed in ferrous 
 sulphate. It behaved but little differently from No. 53, but its maximum was 
 lower, being only 0.85 volt. Before its immersion all the preceding had 
 reached their constant values 0.795. 
 
 No. 55 was thoroughly dried in alcohol and ether immediately after 
 quenching and kept in a desiccator an equal length of time (26 hours) 
 before measurement. This specimen began at 0.79 volt, rose regularly to 
 0.88 volt in two hours, and reached 0.80 volt again in twenty-four hours. 
 
 No. 56 also was immediately thoroughly dried and was measured after 
 73 hours, giving a similar but less marked maximum (0.85 volt). 
 
 The whole series of experiments was then repeated with essentially identi- 
 cal results. 
 
 The comparison of the results of these series of experiments shows that 
 mere exposure to dry air slowly lowers the high -electromotive force due to 
 quenching, that pure water hastens this lowering a very little more than 
 exposure to the air, but that immersion in ferrous sulphate quickly estab- 
 lishes equilibrium. The more typical of these experiments, with some of the 
 previous ones, are plotted in figure 6. 
 
 It became now an interesting matter to discover if any other electrolyte 
 
40 
 
 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 would effect the adjustment of equilibrium as quickly as the ferrous sul- 
 phate. In an easily carried out variation on the preceding series, a normal 
 solution of potassic suphate was used as a typical electrolyte. Brief im- 
 mersion in this solution seemed to show no more effect in discharging the 
 hydrogen than in water, but long immersion showed a slight difference in 
 favor of the electrolyte. The rate was nevertheless so much slower than 
 that observed in the case of ferrous sulphate as to indicate entirely another 
 mechanism of reaction. 
 
 0.9. 
 
 0.8 
 
 0.7 
 
 0.6 
 
 as 
 
 ■ 
 
 
 
 
 1 '^^^*s4L.-_bt 
 
 
 : : ■ 
 
 4S . 
 
 33 lis 
 
 
 
 ■"^' 7 ' 
 
 r 
 
 
 
 ^'33 ■ 
 
 
 
 
 
 ..i— I i..,-i . 
 
 II 1 1 
 
 1 III 
 
 1 t I.I, 
 
 2 3 4. 
 
 8 
 
 10 II 12 13 14 15 16 17 18 19 20 
 
 Fig. 6. — ^The Changing Potential of the Iron Electrode. 
 
 In the direction of ordinates are plotted the potentials of the iron electrodes in normal sulphate, in 
 combination with a decinormal calomel electrode. Time in hours is plotted in the direction of 
 abscissae. The curves are typical, No. 7 of porous iron which has been exposed to air, No. 33 of 
 pure fused iron, No. 48 of porous iron which has been made a cathode. No. 51 of porous iron 
 quenched in water, and No. 53 of the same after exposure to air for 2% hours. 
 
 The progress of the evolution of the gas in ferrous sulphate may be sup- 
 posed to be as follows : The active hydrogen, having a higher potential than 
 the iron, causes metal to deposit from the ferrous sulphate, and the acid 
 thus formed immediately attacks the deposit, or a neighboring portion of 
 metal. Without the interposition of the iron as carrier, the hydrogen seems 
 to find it difficult to go from its occluded to its gaseous state. When an 
 electrolyte which contains no easily deionized cation is substituted for the 
 ferrous sulphate, the reaction involving deposition can not take place, and the 
 hydrogen remains in its occluded condition. 
 
THE POSSIBLE EXPLANATION. 4I 
 
 These considerations show further that the final potential which the iron 
 attains after long immersion in ferrous sulphate is probably the true poten- 
 tial, because, since both iron and ferrous sulphate are greatly in excess, the 
 equilibrium finally attained must correspond to these two alone. Even if 
 some dissolved hydrogen is retained by the iron it can not possess a perma- 
 nent potential above that of the iron. That this is true is shown by the 
 fact that quenched iron finally attains a potential equal to that finally 
 attained by a similar sample which has been long exposed to the air — one 
 approaching this value from above, the other from below. 
 
 The attempt to explain the nature and condition of the active hydrogen in 
 iron is assisted by a suggestive paragraph which occurs in a paper by Thos. 
 Graham *" '^ On the relation of hydrogen to palladium and on hydro- 
 genium " : 
 
 The chemical properties of hydrogenium *^ also distinguish it from ordinary hydro- 
 gen. The palladium alloy precipitates mercury and calomel from a solution of the 
 chloride of mercury without any disengagement of hydrogen; that is, hydrogenium 
 decomposes chloride of mercury, while hydrogen does not. . . . Hydrogenium 
 (associated with palladium) unites with chlorine and iodine in the dark, converts red 
 prussiate of potash into yellow prussiate, and has considerable deoxidizing powers. It 
 appears to be the active form of hydrogen as ozone is of oxygen. 
 
 Of these chemical tests one was easily extended to hydrogen contained 
 in iron. It was found that carefully cleaned pure iron wire, and also porous 
 iron which had been kept for a year in the pure air of a desiccator over 
 potash, had no appreciable tendency to reduce neutral potassic ferricyanide. 
 Even after the iron had stood in a solution of this salt for five days the 
 solution gave no precipitate with ferric chloride. Such iron was evidently 
 practically free from active hydrogen. On the other hand, iron which had 
 been freshly made showed a marked reducing tendency, and this tendency 
 was even greater in quenched iron and iron which had been used as a 
 cathode (or had been simply immersed in acid) and thoroughly washed. 
 These results are quite in accord with the potential measurements already 
 recorded, and furnish additional evidences of active hydrogen in freshly 
 reduced or quenched iron, but point to its absence on the surface of iron 
 long exposed to the air. 
 
 In what form may this hydrogen be supposed to exist? Ramsay was per- 
 haps the first to suggest that hydrogen undergoes dissociation when passing 
 through a hot metal, or through one made a cathode in acid solution.** 
 Our experience supports this hypothesis, and seems to be explicable in no 
 
 **Proc. Roy. Soc. (London), 17, 219 (1869); Collected Papers, p. 290-299; Pogg. 
 Ann., 138, 49 (1869). 
 
 *^ Hydrogenium = hydrogen in a metal. 
 
 "Sir William Ramsay, Phil. Mag., 5, 38, 206 (1894). 
 
42 ELECTROMOTIVE FORCE OF IRON AND OCCLUDED HYDROGEN. 
 
 Other way. In one respect only does Ramsay's opinion appear question- 
 able, namely, in his assumption that the dissociated gas is ionized/' In 
 recent years the word ion has been used in so many senses that an unfortu- 
 nate vagueness has crept into its definition ; but the idea of ionization seems 
 always at least to be associated with electric charges. Now, in this case 
 there seems to be no need of assumption of an electric charge on the 
 occluded hydrogen atom ; indeed it is hard to see how much a charge could 
 be held in the midst of so good a conductor as iron. To the writers it 
 appears much more probable that the hydrogen is rather in the condition of 
 nascent hydrogen, set free from the positive charges which had caused it ta 
 ionize in the aqueous solution, but not yet consolidated into the form of 
 hydrogen gas. It would appear that iron is permeated with minute cavities 
 into which only this dissociated form of hydrogen is able to enter, and that 
 upon all occasions when the nascent element is liberated in the presence of 
 the iron, the opportunity of entrance is at once seized. In the wording of 
 the atomic hypothesis, the active hydrogen occluded by iron seems to be 
 atomic but not ionized hydrogen, very different in its properties from the 
 molecular hydrogen which is adsorbed by the fine powder reduuced at low 
 temperatures. 
 
 In conclusion, it is a pleasure to express our gratitude to the Carnegie 
 Institution of Washington for generous pecuniary assistance in this 
 investigation. 
 
 SUMMARY OF PART FIRST. 
 
 ( 1 ) A method, both rapid and convenient, is given for the preparation of 
 iron containing no impurity but hydrogen. 
 
 (2) The potential of this spongy iron in ferrous sulphate was measured 
 and found to be at first greatly affected by previous exposure to the air. 
 After long immersion in ferrous sulphate solution a constant and trustworthy 
 value was reached. 
 
 (3) Even the enormous pressure of about 350,000 kilograms per square 
 centimeter did not produce any appreciable permanent change in this value, 
 although the masses were effectively cold- welded. Taken in conjunction 
 with the results of Spring, this fact is shown to be consistent with the 
 hypothesis of compressible atoms. 
 
 (4) Measurements of the free-energy change in iron during a pull upon 
 a wire great enough to cause rupture gave results showing that this change 
 must be very small, and called attention to regrettable errors in previous 
 work on this subject. 
 
 *' Ramsay, "Modern Chemistry," II, 31 (London, Dent, 1904). 
 
SUMMARY AND CONCLUSION. 43 
 
 (5) Iron reduced at a low temperature was found to have an electro- 
 motive force higher by at least 0.02 volt than iron which had been fused. 
 If the normal calomel electrode is taken as having a single potential differ- 
 ence of 0.56, pure compact iron has a single potential difference of 0.15, 
 and spongy or porous iron about 0.17 to 0.18 volt. The former is to be 
 chosen as the normal value. It is pointed out that the much lower results 
 of others are probably to be referred to the acidity of their solutions and 
 to insufficient waiting for equilibrium. 
 
 (6) No important change in these values was caused by sudden cooling 
 from a high temperature. 
 
 (7) Speculations concerning the relations of these facts to the structure 
 and internal pressure and solubility of iron are tentatively recorded. The 
 difference of potential of the different forms of iron is probably but not 
 certainly to be referred to the different sizes of their separate particles. 
 
 SUMMARY OF PART SECOND. 
 
 (8) It was found that hydrogen could be taken up by finely powdered 
 iron reduced at low temperatures without affecting the metal's electromotive 
 force. When the metal is wholly coated with hydrogen the electromotive 
 force is diminished. It is probable that hydrogen thus held is merely 
 adsorbed or held as molecular hydrogen. 
 
 (9) It was found further that by quenching in water, whether from 
 hydrogen or from nitrogen gas, that iron takes up hydrogen in an active 
 form, raising the single potential by as much as 0.15 volt. This hydrogen is 
 quickly expelled in ferrous sulphate solution, and very slowly in water or 
 potassic sulphate solution, the potential returning to the normal value. The 
 gas thus evolved was proved to be hydrogen gas. 
 
 (10) A small amount of hydrogen in the same active form may be taken 
 in from hot hydrogen gas. 
 
 (11) Active hydrogen thus occluded by iron seems to be in every way 
 similar to that occluded by iron in the presence of nascent hydrogen, whether 
 this is chemically or electrolytically produced. 
 
 (12) It is pointed out that the most reasonable explanation of these facts 
 is to suppose that the active dissolved hydrogen is dissociated but not ionized. 
 
 The investigation will be continued in the near future. 
 
p 
 
p 
 
^l^IJ™*'^*^****'^^^^ 
 
mi 
 
 Richards, T.W, 
 
 The electromotive force 
 of iron under varying 
 conditions, and the effect 
 of occluded hydrogen. 
 
 PHYSICAL 
 SCIENCES 
 LIBRARY 
 
 Mmi 
 
 MMm 
 
 LIBRARY 
 
 UNIVERSITY OF CALIFORNIA 
 
 DAVIS 
 
 WW 
 
 »^»»/F