EXCHANGE 
 

 The Formation of Addition Compounds Between Formic 
 
 Acid and Metallic Formates. A Discussion of 
 
 the Factors Affecting the Stability of 
 
 These Compounds. 
 
 by 
 
 HOWARD ADLER, B.S., M.A. 
 
 DISSERTATION ONIV 
 
 Submitted in Partial Fulfillment of the Requirements 
 
 for the Degree of Doctor of Philosophy in 
 
 the Faculty of Pure Science of 
 
 Columbia University. 
 
 NEW YORK CITY 
 1920 
 
The Formation of Addition Compounds Between Formic 
 
 Acid and Metallic Formates. A Discussion of 
 
 the Factors Affecting the Stability of 
 
 These Compounds. 
 
 by 
 
 HOWARD ADLER, B,S., M.A. 
 
 tt 
 
 DISSERTATION 
 
 Submitted in Partial Fulfillment of the Requirements 
 
 for the Degree of Doctor of Philosophy in 
 
 the Faculty of Pure Science of 
 
 Columbia University. 
 
 NEW YORK CITY 
 1920 
 

ACKNOWLEDGMENT 
 
 The author wishes to express to Professor James 
 Kendall his sincere gratitude for the suggestion of 
 the problem, and for his helpful advice throughout the 
 course of the investigation. 
 
 The author also wishes to acknowledge with 
 thanks the co-operation of the other members of the 
 Chemistry Department of Columbia University. 
 
 451724 
 
ABSTRACT OF DISSERTATION 
 
 1. What was attempted? 
 
 2. In how far were the attempts successful? 
 
 3. What contribution actually new to the science of chemistry 
 
 has been made ? 
 
 1. The attempt was made to demonstrate the applicability 
 of rules previously formulated governing the variation of ad- 
 dition compound formation, to systems of the type HX-RX. 
 The particular systems studied were the formic acid metallic 
 formate series. A few cases of the series acetic acid metallic 
 acetate were also studied. 
 
 2. (a) It has been shown that the extent of combination 
 between formic acid and the metallic formates varies uniformly 
 with the position of the metal (R) in the electromotive series. 
 As the position of R changes from potassium through the 
 series, the extent of compound formation decreases to a mini- 
 mum (in the neighborhood of hydrogen). A similar variation 
 occurs in the acetate systems, with a slight increase in the ex- 
 tent of combination when the position of R is considerably be- 
 low hydrogen. 
 
 (b) It has also been shown that the variation in the extent 
 cf compound formation is parallel to the change in conductivity. 
 
 (c) The change in solubility follows regularly the change 
 in compound formation. 
 
 3. (a) The initial steps in the formulation of a general- 
 ized theory of ionization have received additional experimental 
 confirmation. 
 
 (b) Further development of methods for determining the 
 existence and extent of compound formation in solution has 
 been indicated. 
 
 (c) Evidence has been obtained of causal relationships 
 between compound formation, conductivity, solubility, and di- 
 versity of components in systems of the general type HX RX. 
 
 (d) In the course of the investigation, five new acid 
 formates have been isolated. 
 
THE FORMATION OF ADDITION COMPOUNDS BE- 
 TWEEN FORMIC ACID AND METALLIC FOR- 
 MATES. A DISCUSSION OF THE FACTORS 
 AFFECTING THE STABILITY OF 
 THESE COMPOUNDS. 
 
 INTRODUCTION 
 
 In the preceding articles of this series 1 , it has been con- 
 clusively demonstrated that the extent of compound formation 
 in a solution depends essentially upon the difference in charac- 
 ter of the two components. On the basis of the generalization 2 
 that compounds increase in stability uniformly with increas- 
 ing divergence in the basic or acidic nature of the two compo- 
 nents, it has been possible to predict the relative extents of 
 combination in various systems, as well as the relative stability 
 of the complexes formed. Agreement of experimental results 
 with these predictions has been extremely satisfactory 3 . 
 
 The study of aqueous systems of the type HX-HOH, and 
 to a less extent, of the type ROH-HOH 4 , revealed the fact that 
 the generalization developed for non-aqueous solvents could be 
 applied, with equally good results, to solutions in water. Prac- 
 tically complete predetermination of the extent of the reaction 
 HX + HOH ^ HX-HOH from left to right was possible. 
 
 The results of the investigation mentioned above led to the 
 correlation of ionization and compound formation 5 . It was 
 shown that for systems of the type HX-HOH, and ROH-HOH, 
 the extent of ionization varied directly with the extent of com- 
 bination between the two components of the system. The 
 causal relationship between the two phenomena, which this 
 variation suggested, has been verified by Gross . The conduc- 
 
 1 Kendall and Booge, J. A. C. S. 38, 1712 (1916). For resume see Ken- 
 
 dall, Booge and Andrews, ibid, 39, 2304 (1917). 
 
 2 Lewis, System of Physical Chemistry, V. I, pg. 417. 
 
 3 The exceptional behavior of phenols has been noted. See Kendall, 
 
 Booge and Andrews- loc. cit. p. 2306; also Kendall J. A. C. S. 38. 
 1317, 1322 (1916) ; 36, 1240 (1914). 
 
 4 Kendall, Booge and Andrews, loc. cit. 
 
 5 Kendall and Booge, J. A. C. S. 39, 2323 (1917). - 
 
 6 Gross, Columbia University Dissertation, 1919. 
 
tivity measurements made by this investigator confirm the 
 hypothesis that ionization in solution is preceded by compound 
 formation between solvent and solute. 
 
 The generalization concerning compound formation was 
 postulated for aqueous systems of the type ROH-HOH in the 
 following form 1 : 
 
 (a) No indication of hydrate formation with extremely 
 weak bases. 
 
 (b) A regular increase in the extent of combination in 
 the liquid state as the strength of the base is increased. 
 
 (c) Extensive compound formation with transition and 
 strong bases. 
 
 (d) Increase in complexity 2 as well as stability of hydrates 
 with the strength of the base. 
 
 Experimental verification of these statements for solutions 
 of the type water-base (as afforded by the work of previous 
 investigators) was presented at the time of their proposal. It 
 is to an extension of this topic that the present paper is de- 
 voted. Additional data upon this problem are essential to a 
 more general formulation of the ionic theory, since, as has 
 already been mentioned 3 , more complete knowledge of the com- 
 paratively simple systems RX-HX and ROH-RX is necessary 
 before any quantitative knowledge of the complex system 
 RX-H 2 O can be gained. 
 
 Solvent-Base-Systems. With a generalization of the 
 theory of solutions will come, of necessity, a broadening of 
 definitions, so as to remove the restrictions now imposed by 
 the limited range of applicability of the accepted theory. In 
 terms of the broader conceptions, a base is defined as a binary 
 compound, which on solution yields the same negative ion as 
 the negative radical of the solvent 4 . With this in mind, it is 
 
 1 Kendall, Booge and Andrews, loc. cit. p. 2320. 
 
 2 i. e. compounds of the type (ROH) a (HOH)b can be isolated when 
 there is considerable divergence between the components. 
 
 3 Kendall, Booge and Andrews, loc. cit, p. 2322. 
 
 4 This terminology has been used by Schlesinger and Calvert, T. A. C. S. 
 
 33, 1933 (1911"). 
 
obvious that the non-aqueous system HX-RX is identical in 
 nature with the system ROH-HOH. Hence it is of interest to 
 see whether the postulates proposed for the latter type will 
 holdi equally well for the former. This condition is essential 
 to the evolution of a more comprehensive theory of solutions. 
 
 Criterion of Diversity. Since diversity of the two compo- 
 nents is postulated as necessary for combination between them, 
 ir becomes necessary to arrive at some criterion of the extent 
 of divergence in systems of the type HX-RX. The X radical 
 being common to both components, the "difference" between 
 H and R must afford a measure of the tendency to complex 
 formation, i. e. of the tendency of the reaction HX + RX ^ 
 HX-RX to go forward. The best available criterion of this 
 divergence is the relative positions of the metal (R) and hydro- 
 gen in the electromotive series. This follows from the rela- 
 tionship which exists between the electromotive series (or 
 electrode potential series, since these are identical in order) and 
 the chemical activity of the metals 1 . The relative activities of 
 the metals are given by the order of the electromotive series, 
 the. activity decreasing in order from K down the series to 
 the noble metals. 
 
 It is to be expected, then, that the higher the position of 
 the metal (R) in the series, the stronger will be the resulting 
 base (RX) 2 . As R is varied, and approaches hydrogen, the 
 strength of the resulting base, as evidenced by the extent of 
 compound formation and, consequently 3 , of ionization, should 
 diminish. In the immediate neighborhood of hydrogen, the 
 base should be very weak. On continuing the variation be- 
 yond hydrogen, the difference between the components be- 
 comes more pronounced, the lower the position of the metal in 
 the series. This divergence should result in the formation of 
 
 1 For a full discussion of chemical affinity and its measurement by 
 
 e. m. f., see particularly Lewis, System Physical Chemistry, V. 2, 
 Ch. XII. Also Lehfeldt, Electrochem., pp. 181, 182, 194. Mellor, 
 Modern Inorganic Chem., 361-376; LeBlanc, Electrochem., 267 et seq. 
 
 2 This ought to be equally true, irrespective of the nature of X i.e., 
 
 whether it be NO 3 , SO 4 , OH etc. 
 
 3 Kendall and Booge, loc. cit., p. 2324. 
 
 8 
 
bases stronger than those of the metals located near hydrogen 
 in the series 1 . 
 
 It is seen that on the basis of the above assumption the 
 strength of the base RX should diminish to a minimum and 
 then increase again, as R is varied from one extreme of the 
 electromotive series to the other. Those properties which are 
 usually associated with the term "strength" extent of ioniza- 
 tion (compound formation), and, as will be seen from the 
 sequel, solubility, should undergo concomitant variation. The 
 experimental work to be described, will seek to establish the 
 validity of this argument. 
 
 SYSTEMS HX RX. ACID SALTS. 
 
 There are in the literature numerous references to com- 
 pounds of the type (RX) a (HX) b , i. e., acid salts. There have 
 been, however, very few systematic investigations of such salts, 
 with a view to the correlation of the fact of their existence 
 with theory. 
 
 It might be worth considering briefly those compounds 
 mentioned in the literature, in order to see how far they are in 
 agreement with the requirement of the theoretical basis de- 
 veloped above. A complete survey would not be of any value, 
 because such acid salts as bisulfites, bicarbonates, and others, 
 could never lend themselves to systematic study, since the acids 
 exist only in solution. The following review will be limited 
 to those cases where the compounds can be obtained from the 
 pure acid and base. 
 
 1 The position of the metals in the electromotive series (or rather, elec- 
 tro-affinity) as a factor in determining the properties, of metallic 
 compounds, such as the chlorides, has been used by Bodlander, Z. 
 Phys. Chem. 27, 55 (1898), and Abegg and Bodlander, Z. Anorg. 
 Chem. 20, 453 (1899). They attempted to show for example, that 
 the solubility of the chlorides increased as the position of the metal 
 varied from bottom to top of the series, i. e. from Ag to K. This is 
 notoriously not the case ; which fact led to the use of additional 
 hypotheses to maintain the original thesis. The procedure was not 
 altogether warranted. The failure to establish the validity of the 
 propositions advanced is probably due to the total neglect of the 
 influence of the solvent. Furthermore, the systems examined were 
 of the complex type RY-HX or RY-H O, and the complications 
 introduced by the fourth radical prevented any real connection be- 
 tween the electromotive series and properties such as solubility from 
 being discovered. 
 
The only acid whose acid salts have been completely ex- 
 amined is sulfuric acid 1 . The results of the entire investigation 
 are to be published shortly, so that a complete discussion of 
 this system is not now advisable. The results in general are, 
 however, in highly satisfactory agreement with the theory. 
 There has been no other complete series studied, with any 
 theoretical objective. The compounds mentioned in the liter- 
 ature, as conditioned above, are given in Table I, which follows : 
 
 TABLE I. 
 
 Solvent Base Compounds 
 
 HN0 3 KN0 3 KN0 3 -2HN0 3 ; KNO 3 -HNO 3 2 
 
 NH 4 NO 3 NH 4 NO 3 -2HNO 3 ; NH 4 NO 3 -HNO 3 2 
 HAc KAc KAc 2HAc 3 ; KAc-HAc 4 
 
 NaAc NaAc-2HAc; NaAc-H.V 
 
 NH 4 Ac NH 4 Ac-HAc 6 
 Li Ac LiAc-HAc 7 
 
 TIAc TIAc-HAc 7 
 
 HF KF KF-3HF 8 ; KF-HF 9 
 
 NaF NaF-HF 10 
 
 LiF LiF-HF 11 
 
 NH 4 F NH 4 F-HF 12 
 AgF AgF-3HF 13 ; AgF-HF 13 
 
 The existence of all of these compounds, as well as that 
 of the hydrates of bases which have already been enumerated 14 , 
 is in agreement with the requirements of the theory. All 
 
 J Landon, J. A. C. S., 42, 2131 (1920). Completed by Davidson, as yet 
 
 unpublished. 
 2 Groschuff, Ber. _37, 1488 (1904). The same author's work on formic 
 
 acid will be referred to in the following section. 
 sLescoeur, Ann. Chim. Phys. (6), 28, 245 (1893). 
 *Melsens, Compt. Rend. 19, 611 (1844). 
 5 Lescoeur, loc. cit., pg. 241. 
 6 Reik, Monatshefte f. Chemie, 23, 1033 (1902). 
 7 Lescoeur, Bull. Soc. Chim., 24, 517 (1875). 
 s Moissan, Compt. Rend., 106, 547 (1888). 
 9 Abegg, Hand. Anorg. Chem., 2-1, 343. 
 10 Abegg, ibid., 220-221. 
 
 11 Ibid., pg. 120. 
 
 12 Marignac, Ann. Min. (5) 15,_221 (1859). 
 
 Guntz, Bull. Soc. Chim. (3) U 114 (1895). 
 
 14 Kendall,' Booge and Andrews, loc. cit., p. 2320. 
 
 10 
 
metals whose "bases" form addition compounds of the acid- 
 salt type, are either strongly electropositive or strongly elec- 
 tronegative. It is especially noteworthy that silver fluoride is 
 soluble in hydrofluoric acid, and is extensively solvated. 
 
 It is evident that the data existing are not sufficient to 
 supply a rigorous test of the validity of the argument. The 
 varying reliability of results from scattered sources, the lack 
 of completeness in all the series examined, as well as the ab- 
 sence of investigations of freezing-points so as to permit the 
 determination of relative extents of combination throughout 
 the series, all tend to diminish the value of any conclusion which 
 may be drawn. For these reasons, it was deemed necessary to 
 determine as completely as possible the freezing-point curves 
 for the series formate-formic acid. In this series the only part 
 which was varied was that which corresponds to R in the type 
 system RX HX. After the effect of this variation of R has 
 been determined, the role of X can be more exactly examined, 
 by a study of several series similar to the one in question. 
 
 FORMIC ACID AS SOLVENT 
 
 It has already been determined by Schlesinger and col- 
 laborators 1 , that solutions of the formates in formic acid are 
 excellent conductors. The alkali formates are highly ionized, 
 and are entirely analogous to the alkali hydroxides in water. 
 It has also been shown 2 that the conductivity of a solution de- 
 pends upon two factors (a) the extent of compound forma- 
 tion and (b) the extent of dissociation of the complexes into 
 ions of opposite charge. Since the metallic formates form high 
 ly conducting solutions in formic acid, it follows that they are 
 highly solvated in solution. 
 
 The formates in formic acid should give rise to compound 
 formation, varying in extent with the position of the metal in 
 the electromotive series 3 . The variation in compound forma- 
 
 1 With Calvert J. A. C. S. 33, 1924 (1911) ; Martin, ibid, 36, 15S9 (1914) ; 
 
 Coleman, ibid, 38, 271 (1916) ; Mullinix, ibid, 41, 72 (1919) ; Reed, 41_, 
 1921 (1919). 
 
 2 Kendall and Booge, loc. cit, p. 2324 (1917). Gross, loc. cit, pg. 7. 
 
 3 The order of increasing conductivity is given as that of increasing 
 
 electrolytic solution tension, Schlesinger and Coleman, loc. cit., 
 p. 278. 
 
 11 
 
tion should parallel the change in conductivity. To test the 
 validity of these conclusions, a representative series of the for- 
 mates was examined, namely, K, Na, Li, NH 4 , Ba, Ca, 
 Mg Zn, Ni, Pb, Cu and Ag. The solubilities of these for- 
 mates were determined, using the freezing-point method, as 
 described below. 
 
 Due to unavoidable complications, inherent in the nature 
 cf the solvent (i. e. of X), it was ndt possible to work with 
 silver. In order that the increase of compound formation as 
 R is varied below H might be demonstrated, several members 
 of the acetate-acetic acid series were studied. Na, Zn, Ni, 
 Fe (ic) and Ag were taken as representing the different por- 
 tions of the electromotive series. 
 
 The agreement between the deductions from the theoreti- 
 cal considerations discussed, and the results of these experi- 
 ments ought to furnish sufficient evidence to demonstrate the 
 applicability of the generalization given above to systems of 
 the type HX RX. 
 
 EXPERIMENTAL 
 
 EXPERIMENTAL PROCEDURE 
 
 Freezing-point curves for mixtures of formate and formic 
 acid were determined in the usual manner 1 . Points on the 
 curves were taken at intervals of from 2 to 3 molecular % ; 
 at points of change of phase, the intervals were small enough 
 to fix accurately the different branches of the curve. Each 
 point was determined at least twice. 
 
 Those mixtures which gave a melting-point 2 below sixty 
 degrees (60C) were investigated in freezing point tubes by 
 
 1 See Kendall and Booge, J. A. C. S. 38, 1718 (1916), and Landon, loc. 
 
 cit, for discussion of method. In some cases, i. e., at low tempera- 
 tures, considerable supercooling was encountered. The mixtures 
 were then cooled in CO 2 acetone paste and allowed to warm up 
 slowly, with stirring, to induce crystallization. 
 
 2 The temperatures which follow refer to the point at which a negli- 
 
 gibly small amount of the solid phase is in equilibrium with the 
 solution. 
 
 12 
 
the usual method. Every precaution was exercised to insure, 
 as far as possible, anhydrous conditions. The addition of 
 formate to the acid was done with the aid of a specially de- 
 signed weighing bottle 1 , thus reducing exposure to a minimum. 
 The stirrer was connected to the stopper by means of rubber 
 tubing, the system being in this way entirely closed. .The com- 
 position given for any of these solutions! is accurate within 
 less than 0.05%. Above about 60C, recourse was had to 
 sealed bulbs 2 , because of the increasing vapor pressure of the 
 formic acid. These bulbs were so blown as to reduce the air 
 space to a minimum, thus decreasing the amount of solvent 
 present as vapor to a negligible magnitude. The composition 
 given for solutions whose freezing-points were measured in 
 bulbs, may be taken as accurate within 0.1 molecular per cent. 
 
 The bath in which the tube or bulb was placed during the 
 determination of the, melting point, varied with the tempera- 
 ture range in which the point lay. Those baths used, and the 
 temperature interval of their use, were : 
 
 Acetone + CO 2 (solid) Up to 25C. 
 
 HNO 3 + ice 25 
 
 NaCl + ice -15 
 
 Water 100. 
 
 70 mol. % H 2 SO 4 ; 30 mol. %(NH 4 ) 2 SO 4 above 100. 
 
 Considerable attention was paid to the factors affecting 
 thermal equilibrium between the tube or bulb and the bath. 
 This resulted in the following precautions being observed, to 
 avoid any appreciable error from this cause : 
 
 1. The tube and bath were stirred constantly. 
 
 1 See Landon, Columbia University Dissertation (1920), pg. 9, for de- 
 
 tailed description and cut of this bottle. 
 
 2 Points in bulbs were, of course, determined under excess pressure, 
 
 i. e., the vapor pressure of the system plus the pressure of the en- 
 closed air. Since the limiting temperature was 160C, and the effect 
 of pressure on the freezing points so very small (probably <C0.01 
 per atmosphere), there \vould be no advantage in attempting, if it 
 . were possible, to reduce all freezing-points to atmospheric pressure. 
 
 13 
 
2. The temperature was changed slowly enough so as 
 to maintain 1 , as nearly as practicable, thermal equilibrium 
 throughout the heating process. 
 
 3. The bulbs were sufficiently thin to prevent lag when 
 the above precautions were observed 2 . 
 
 The effects of draughts, and radiation at higher tempera- 
 tures, were excluded by the use of an asbestos shield sur- 
 rounding 1 the bath. This shield had glass windows to permit 
 observation of the bath. 
 
 TEMPERATURE MEASUREMENT 
 
 Temperatures were measured by means of three mercury 
 thermometers, graduated in tenths of a degree (C.), and having 
 the respective ranges, 35 + 25 ; 100 ; 100 200. 
 These were calibrated at and 100, and the two with lower 
 range were compared with a certified thermometer at inter- 
 mediate points. The 100 200 thermometer was tested at the 
 boiling points of pure monobrombenzene and aniline (Kahl- 
 baum), giving results in agreement with the literature. Hence 
 it was considered as correct within the limits of experimental 
 error (as discussed below). 
 
 The correction for exposed stem was determined experi- 
 mentally 3 by measuring constant temperatures 4 with the ther- 
 mometer exposed, and then repeating with the thread entirely 
 immersed. The length of thread exposed was the same as 
 would be left outside the bath in a determination of a melting 
 point. The corrections obtained were plotted against tempera- 
 
 1 App. 0.2 per minute was the average rate of heating. This was varied 
 
 slightly, according as the slope of the curve changed. Where the 
 rate of change of composition with temperature was high, the rate 
 of heating was diminished. The rate was increased slightly under 
 the reverse circumstances. 
 
 2 This is proven by concordant results when points on the same portion 
 
 of a curve were determined by either method. 
 
 3 The apparatus used for this determination consisted of a glass tube, 
 
 resembling the outer jacket of a Victor Meyer apparatus, to the top 
 of which a Liebig condenser was attached. Its use was suggested 
 by Dr. P. M. Gross. 
 
 4 Boiling liquids of good quality, not necessarily ultra pure, give tem- 
 
 peratures constant to 0.1 for a sufficient length of time to permit 
 both measurements being taken. . 
 
 14 
 
tures, and from the graph so made, the stem correction for any 
 intermediate point could be read 1 . 
 
 X: Precision of Measurements The freezing-point of a mix- 
 ture prepared by the above method and determined with one 
 of the thermometers just described, possesses a definite pre- 
 cision value. This depends not only upon the temperature 
 interval in which the point lies, but also upon the nature of the 
 curve. This is due to the fact that it is easier to determine, 
 with any desired precision, a point which lies on a flat curve 
 than one which is on a steep curve 2 . 
 
 The fact that both of these factors, temperature and slope, 
 have to be considered makes it difficult to give any definite 
 probable values. Those which follow are to be taken as ap- 
 proximations, true for the average type of curve only 3 : 
 
 Temperature Interval Possible Error 
 
 _35 to _IQO o.2 to 0.5 
 
 -10 to + 100 0.1 to 0.2 
 
 100 to 200 0.2 to 0.5 
 
 ANALYSIS OF COMPOUNDS 
 
 The composition of the solid phase separating, in every 
 case where it was not evident from the curve, was determined 
 by analysis. A mixture of suitable composition was prepared, 
 and the substances to be analyzed frozen out. The compound 
 was then collected in a Gooch crucible, the solution being drawn 
 through by suction. The filtration was carried out under anhy- 
 drous conditions 4 . The solvent adhering to the crystals was re- 
 moved by sucking air dried by CaCl 2 through for sufficient 
 time to guarantee complete removal. 
 
 1 While it is evident that these corrections hold exactly only when the 
 
 temperature surrounding the stem is the same as when the correc- 
 tions were determined, a change of*3 or 4 in room temperature 
 produces no appreciable effect upon the values. 
 
 2 Flat and steep refer to the slopes of the curves the change of tem- 
 
 perature with respect to a slight change in composition is small 
 and large on the respective curves. 
 
 3 As Booges, Kendall and Booge, loc. cit. To be exact, distinction must 
 
 be made between points determined in bulbs, and in open tubes. 
 
 4 Water decomposed the compounds. The apparatus was so set up that 
 
 when it was necessary to maintain a low temperature during filtra- 
 tion, the funnel could be surrounded by a freezing mixture, 
 
 15 
 
The composition was calculated from the volume of stand- 
 ard alkali required to neutralize a weighed amount of the com- 
 pound. Check determinations were run to preclude the possi- 
 bility of unremoved acid giving erroneous and misleading re- 
 sults. 
 
 FORMATE SYSTEMS 
 Formic Acid. 
 
 The formic acid used was prepared from Baker and Adam- 
 son c. p. acid by treatment with boron trioxide to remove the 
 water it contained. The mixture was distilled in vacuum, 
 moisture being excluded 1 . The acid which was used froze be- 
 tween 8.35 8.5, and was generally better than 8.4 2 . 
 
 SYSTEM FORMIC ACID POTASSIUM FORMATE 
 
 This system was investigated partially, by GroschufP, 
 who determined the solubility of potassium formate in formic 
 acid between and 100C. He used a method similar to that 
 employed in this work, and succeeded in isolating an acid salt 
 KCHO 2 -H 2 CO 2 , which, according to the investigator, under- 
 went transition before it melted. In view of the incompleteness 
 of the work, it was thought advisable to repeat the part already 
 done, in addition to completing that part left undone. As will 
 be seen from the data, the course adopted was justified. Gros- 
 chufPs work was not only incomplete, but was erroneous as 
 well. 
 
 The potassium salt was prepared by dissolving a pure sam- 
 ple of potassium carbonate in 90% formic acid. After ex- 
 pelling the CO 2 , the hydrate of potassium formate 4 was crys- 
 tallized from the solution. This salt was dehydrated and dried 
 as completely as possible by prolonged heating just below the 
 
 1 Berichte H 1709 (1881). See particularly, Schlesinger and Martin, 
 
 loc. cit, whose apparatus and procedure were followed. 
 
 2 Varying values are given in the literature, 8.5 being probably correct. 
 
 8. 43, Peterson, Ber. U 1191 (1880). 8.5 Walden, Trans. Far. Soc._6_ 
 71 (1910). 8.52, Novak (by extrapolation), Phil. Mag. (5) 44, 828. 
 8.6, Schlesinger and Calvert, loc. cit. 
 
 3 Berichte 36, 1783 (1903). 
 
 4 Groschuff, loc. cit., who also discusses the hygroscopicity of the salt, 
 
 which is very great. 
 
 16 
 
25 
 
 75 
 
 100 
 
 50 
 Mol. % Base 
 
 Fig. I. 
 
 (A) Potassium Formate Formic Acid, (subtract 30 from the tem- 
 perature readings). 
 (B) Ammonium Formate Formic Acid. 
 
 17 
 
melting point. The last trace of water was removed by recrys- 
 tallization from absolute ethyl alcohol, and dessication over 
 99% H 2 SO 4 in vacuum. 
 
 The salt obtained melted at 167.5 Q.5C. 1 
 
 The data for the system follow ; see also Fig I A. 
 
 (a) Solid phase H 2 CO 2 . 
 
 % Base 0.97 3.02 4.73 6.36 8.43 
 
 Temp. 8.35 7.4 4.9 2.2 0.9 5.7 
 
 % Base 10.74 12.63 13.88 15.57 
 
 Temp. -12.6 -18.7 23.8 31.5 
 
 (b) Solid phase KHCO 2 -3H 2 CO 2 (?) 
 
 % Base 16.32 16.52 17.44 18.10 18.88 
 
 Temp. 29.0 -27.3 23.5 21.7 -19.5 
 
 (c) Solid phase KHCO 2 -2H 2 CO 2 . 
 
 % Base 19.48 19.91 21.21 22.79 
 
 Temp. 19.0 16.0 8.0 0.6 
 
 (d) Solid phase KHCO 2 -H 2 CO 2 . 
 
 % Base 23.04 24.14 25.98 28.56 30.41 31.97 33.74 
 Temp. 10.1 7.3 29.9 53.0 65.1 72.9 80.6 
 
 % Base 37.29 38.34 41.68 42.63 42.91 45,95 50.25 
 
 Temp. 93.0 96.1 103.2 104.3 104.6 107.5 108.6 
 
 % Base 51.49 54.24 58.47 61.14 63.14 
 
 Temp. 108.2 107.2 103.4 100.6 98.7 
 
 (e) Solid phase KHCO 2 
 
 % Base 66.45 68.71 71.24 75.18 77.75 82.41 86.68 
 
 Temp. 108.1 114.5 122.3 130.7 135.8 143.6 150.0 
 
 % Base 91.24 100.00 
 
 Temp. 157.3 167.5 
 
 1 The salt used by Groschuff melted at 157C. 
 
 18 
 
Analyses: 
 
 (b) It was not practicable to carry out this analysis, be- 
 cause of the low temperature at which compound undergoes 
 transition ( 17.5). The slope of curve indicates the probable 
 correctness of the composition given. 
 
 (c) Analysis was carried out at 4C. 0.3838 gm. salt 
 contained 0.2050 gm. acid. Calculates to 67.9 mol. % acid. 
 Theoretical 2 : 1 is 66.7%. 
 
 (d) 1 : 1 compound melts at 108.6 0.1. Analysis was 
 unnecessary. 
 
 SYSTEM SODIUM FORMATE FORMIC ACID 
 
 This system had also been partially, and as the work herein 
 reported shows, erroneously examined by Groschuff 1 . The 
 salt was obtained by recrystallizing a pure commercial sample 
 twice from water, and dehydrating at 140C to constant weight. 
 The salt gave a melting point of 255 I . 2 
 
 The curve does not extend beyond approximately 160C, 
 because of the decomposition of the acid at this point 3 . This 
 places an automatic limit upon all the curves. 
 
 The data for the system follow. The curve is given in 
 Fig. II A. 
 
 (a) Solid phase H 2 CO 2 . 
 
 % Base 0.0 1.04 2.92 4.61 6.35 8.72 
 
 Temp. 8.4 7.5 5.3 3.0 0.4 3.8 
 
 % Base 10.50 12.58 
 
 Temp. 7.6 12.8 
 
 1 Loc. cit. 
 
 2 Groschuff obtained 253 as the m. pt. of sodium formate. 
 
 3 Richter, Org. Chem. 1,266 Lorin, Jahresbericht uber Fortschritte 
 
 Chemie,28, 515 (1876). 
 
 19 
 
200 
 
 150 
 
 100 
 
 50 
 
 50 
 
 / 
 
 12.5 
 
 25 
 Mol. % Base 
 
 Fig. II. 
 
 (A) Sodium Formate Formic Acid. 
 (B) Sodium Acetate Acetic Acid. 
 
 37.5 
 
 50 
 
 20 
 
(b) Solid phase NaHCO 2 -2H 2 CO 2 . 
 
 % Base 14.52 16.45 18.18 20.69 21.13 21.86 23.23 
 
 Temp. 17.4 0.3 10.7 22.9 24.4 26.3 30.5 
 
 % Base 24.43 24.91 
 
 Temp. 33.1 34.5 
 
 (c) Solid phase NaHCO 2 -H 2 CO 2 . 
 
 % Base 25.18 25.90 27.00 27.10 28.39 29.80 31.15 
 
 Temp. 31.0 37.3 45.2 45.6 52.2 59.0 65.1 
 
 % Base 31.65 32.14 
 
 Temp. 67.5 69.6 
 
 .(d) Solid phase NaHCO 2 . 
 
 % Base 29.80 31.65 33.40 35.71 39.10 42.79 43.47 
 
 Temp. 45.1 63.6 81.2 99.3 118.6 135.2 137.7 
 
 Analyses: 
 
 (b) 0.2842 gm. compound contained 0.1615 gm. acid; 
 equivalent to 0.1674 mole acid to 0.0888 mole salt or 66.06 mol. 
 % acid. Theoretical 2 : 1 is 66.7 r /< . 
 
 (c) 0.2013 gm. compound contained 0.0814 gm.- acid; 
 equivalent to 0.1770 mole acid to 0.1763 mole salt ; or 50.1 mol. 
 % acid. Theoretical 1 : 1 is 50.0%. 
 
 SYSTEM LITHIUM FORMATE FORMIC ACID 
 
 Groschuff 1 also worked with this system, but did not do the 
 part of the curve of chief theoretical interest. His work was 
 extended to complete the curve as far as possible. 
 
 Lithium formate was prepared from a pure sample of the 
 carbonate and 90% B. and A. acid. The hydrate was crystal- 
 lized from the solution, dehydrated at 100C 2 , and the dry salt 
 
 1 Groschuff, loc. cit. 
 
 2 It decomposes to water and anhydrous salt at 94C Groschuff, loc. cit. 
 
 21 
 
recrystallized from alcohol, and dessicated over 99% H 2 SO 4 
 in vacuum. 
 
 The data for the system are given below : 
 
 (a) Solid phase H 2 CO 2 . 
 
 % Base 0.0 1.58 3.47 5.33 7.09 8.93 10.75 
 
 Temp. 8.4 7.0 5.2 3.2 +1.1 1.3 3.5 
 
 % Base 12.23 13.99 18.19 19.56 21.15 22.24 23.49 
 
 Temp. 5.6 8.2 14.6 17.1 19.8 21.7 23.5 
 
 % Base 24.33 
 
 Temp. 25.0 
 
 (b) Solid phase LiHCO 2 . 
 
 % Base 23.49 23.93 25.31 25.91 26.38 27.71 
 
 Temp. 18.0 34.0 80.0 90.5 97.9 113.1 
 
 % Base 29.87 31.98 33.04 35.01 36.13 
 
 Temp. 131.2 145.1 150.4 159.1 163.5 
 
 Analysis proved (b) to be the neutral salt. 
 
 SYSTEM AMMONIUM FORMATE FORMIC ACID 
 
 The existence of an acid salt of ammonium (equimolecular) 
 was shown by Groschuff 1 , after Reik 2 had failed to find one. 
 No complete examination of the system has been previously 
 undertaken. 
 
 The ammonium salt was prepared by passing ammonia 
 into 90% acid. The acid was cooled in ice until it was almost 
 saturated ; it was then allowed to warm up sufficiently to insure 
 the separation of the neutral salt. This salt was collected on 
 
 1 Groschuff, Berichte, 36, 4351 (1903). 
 
 2 Reik, Monatschefte,23, 1033 (1902). 
 
 22 
 
a Biichner funnel, and recrystallized from absolute alcohol ; it 
 was dessicated over 99% H 2 SO 4 in vacuum. The salt gave a 
 melting point of 117.3 0.2, that reported by Groschuff being 
 116. 
 
 The data for the system are given below and in Fig. I B : 
 
 (a) Solid phase H 2 CO 2 . 
 
 % Base 0.0 1.53 3.70 6.28 8.19 10.14 12.43 
 
 Temp. 8.47 7.0 4.5 0.6 2.8 6.9 12.6 
 
 % Base 14.86 17.11 18.95 
 
 Temp. 19.8 26.9" 33.8 
 
 (b) Solid phase NH 4 HCO 2 -3H 2 CO 2 (?). 
 
 % Base 20.43 21.73 23.33 
 
 Temp. 31.3 30.0 29.3 
 
 (c) Solid phase NH 2 HCO 2 -H 2 CO 2 , two crystalline modifi- 
 cations, less soluble (stable) needles ; more soluble (unsta- 
 ble) prisms. 
 
 % Base 23.33 24.83' 25.41 26.64 27.90 
 
 Temp.* 32.5 26.0 23.5 18.7 14.0 
 Temp.f 13.7 8.7 
 
 % Base 30.11 33.25 36.29 39.30 41.90 42.94 44.38 
 
 Temp.* 1.3 7.4 11.3 13.8 14.3 15.0 
 
 Temp.f +0.7 10.4 17.3 22.2 24.9 25.8 
 
 (d) Solid phase NH.HCCV 
 
 % Base 
 Temp. 
 
 44.38 
 20.4 
 
 46.06 
 29.3 
 
 47.87 
 
 37.5 
 
 51.88 
 53.1 
 
 59.98 
 74.3 
 
 68.23 
 89.5 
 
 74.32 
 96.5 
 
 % Base 
 Temp. 
 
 76.20 
 98.5 
 
 83.46 
 103.7 
 
 88.41 
 108.5 
 
 91.64 
 111.7 
 
 100.00 
 117.3 
 
 
 
 * Unstable. f Stable. 
 
 1 At the higher points, upon keeping the solutions in molten condition 
 for a while, there appeared to be a tendency for the melting points to 
 be slightly lowered, making it difficult to check points. This must 
 be due to decomposition in the liquid state, possibly to formamide, 
 although this generally takes place at a much higher temperature 
 180-230 Beilstein 1,395. 
 
 23 
 
Analyses: 
 
 (b) could not be analyzed. Nature of curve makes it prob- 
 able that composition given is correct. 
 
 (c) The two crystalline modifications have the same com- 
 position, 0.5846 gms. compound gave 0.2432 gm. acid equiva- 
 lent to 0.5285 to 0.5413 moles or 49.4 molecular r /< acid (Theory 
 1: 1 is 507r). 
 
 SYSTEM BARIUM FORMATE FORMIC ACID 
 
 There has been no work on solubilities in this system re- 
 ported in the literature. The barium salt was prepared from 
 the, carbonate and acid. It was recrystallized from water three 
 times, and dried at 140C. 
 
 The data are given below : 
 
 (a) Solid phase H 2 CO 2 . 
 
 % Base 0.00 0.91 1.74 2.30 3.73 4.67 5.12 6.95 
 
 Temp. 8.4 7.2 6.1 5.1 2.6 +0.5 0.3 4.9 
 
 (b) Solid phase Ba(HCO 2 ) 2 -H 2 CO 2 . 
 
 %Base 8.52 8.86 9.23 9.83 10.03 10.75 
 
 Temp. 9.5 15.5 19.0 24.9 26.5 31.8 
 
 Analysis: 
 
 (b) 0.4262 gm. cpd. contained 0.0733 gm. acid ; equiva- 
 lent to 0.1593 to 0.1552 moles acid and salt respectively. This 
 calculates to 50.66 mol. % acid. Theory, 1 : 1 is 50%. 
 
 SYSTEM CALCIUM FORMATE FORMIC ACID 
 
 No previous work in this system could be located in the 
 literature. 
 
 The calcium salt was prepared by a method analogous to 
 that used for barium formate. 
 
 The data are given below : 
 
 24 
 
(a) Solid phase H 2 CO 2 . 
 
 % Base 0.00 0.16 0.48 0.71 0.93 1.27 1.53 
 
 Temp. 8.4 8.1 7.7 7.4 7.2 6.9 6.6 
 
 (b) Solid phase Ca(HCO 2 ) 2 . 
 
 % Base 0.39 , 0.57 0.83 1.10 1.26 1.35 1.54 1.61 
 
 Temp. 128.6 100.0 79.0 61.0 49.7 45.5 35.0 30.0 
 
 Analysis proved (b) to be the neutral salt. 
 
 It is seen from the data that calcium formate in formic 
 acid exhibits retrograde solubility. This is not surprising, 
 since the same phenomenon occurs in aqueous solutions of 
 many calcium "salts 1 . 
 
 SYSTEM MAGNESIUM FORMATE FORMIC ACID 
 
 This system has not been previously investigated. The 
 salt was prepared by dissolving the oxide (Kahlbaum) in acid, 
 and dehydrating the crystallized salt (a dihydrate) at 110C. 
 The resulting salt was slightly basic, but not sufficiently so to 
 have any effect upon the results obtained 2 . 
 
 It was not possible to obtain a curve giving the solubility 
 of this salt. It was found that those mixtures which yielded 
 clear solutions would not crystallize to any solid phase other 
 than formic acid. Furthermore, on heating other mixtures, to 
 get a more concentrated solution, there occurred separation of 
 solid, which dissolved very slowly on cooling. Measurements 
 showed that the magnesium salt was soluble up to a concentra- 
 tion of 0.2 molecular per cent, at room temperature (about 
 25C). 
 
 SYSTEM LEAD FORMATE FORMIC ACID 
 
 No work on this system has been reported in the literature. 
 The salt was prepared by double decomposition from lead ni- 
 
 1 Washburn, Physical Chemistry, pg. 354. Lumsden, J. Chem. Soc. 
 
 (London), 81, 355 (1902). 
 
 2 Analysis showed a basicity of much less than 1%. Since the salt was 
 
 only very slightly soluble, the amount of base (or of water pro- 
 duced) is of a negligible magnitude. 
 
 25 
 
trate and sodium formate. The resulting precipitate was thor- 
 oughly washed and recrystallized twice from water. It was 
 dried at 140"C. All of the points, the data for which are given 
 below, were determined in bulbs, due to the relatively insoluble 
 nature of the salt. 
 
 Solid phase Pb(HCO 2 ) 2 . 
 
 % Base 0.21 0.30 0.42 0.51 
 
 Temp. 20.0 73.1 109.4 124.5 
 
 SYSTEM ZINC FORMATE FORMIC ACID 
 
 No previous work in this system has been reported. The 
 salt was prepared by the method used for the Ba and Ca salts. 
 The solubility was less than 0.1 mol. % at 140C. 
 
 SYSTEM COPPER FORMATE FORMIC ACID 
 
 No previous work has been reported. The salt was made 
 from a pure sample of basic carbonate and acid. The hydrate 
 was crystallized from water and dehydrated, at about 80C, giv- 
 ing the bright blue neutral salt 1 . The solubility was less than 
 0.1% at 135C. 
 
 SYSTEM NICKEL FORMATE FORMIC ACID 
 
 No work on this system has been reported in the literature. 
 The salt was prepared from the carbonate and c. p. acid. The 
 hydrate, which was crystallized from water, was dehydrated 
 at 140C. It was soluble to less than 0.1 molecular % at 140C. 
 
 SILVER FORMATE 
 
 Inasmuch as it was desired to show the variation in com- 
 pound formation below hydrogen in the electromotive series, 
 attempts were made to investigate the silver system. The salt 
 is described in the literature 2 as a white crystalline salt, de- 
 composed by boiling water to give Ag and CO 2 . The descrip- 
 
 1 Voss, Liebig's Ann a 1 en 266. 33 (1891). 
 
 2 Beilstein, I, 395. Richter, Organische Chemie, 1^ 266. 
 
 26 
 
tion is not strictly accurate. It was found impossible to keep 
 the salt in the presence of water for- the length of time neces- 
 sary to filter the solution by suction. The decomposition is not 
 due to light, as has also been suggested 1 , but undoubtedly is 
 caused by the aldehydic nature of the acid itself. 
 
 The salt was isolated by precipitating it in absolute methyl 
 alcohol. It 'is not, however, sufficiently stable to work with; 
 it decomposed in a dessicator over H 2 SO 4 , even though pro- 
 tected from the light. 
 
 ACETATE SYSTEMS 
 
 Acetic Acid : 100% acetic acid was prepared from glacial 
 acetic acid, by using the method of Gross 2 . From the freezing- 
 point of the acid, and DeVisser's 3 figures, the percentage of 
 water was calculated, and the amount of acetic anhydride re- 
 quired to react with that amount added to the acid. The mix- 
 ture was refluxed for about 30 hours and then distilled. 
 
 The acid used froze between 16.5 and 16. 6. 4 
 
 SYSTEM SODIUM ACETATE ACETIC ACID 
 
 This system has not been previously examined with re- 
 spect to solubility. The literature mentions the existence of the 
 acid salts NaAc-2HAc melting at 80 and the equimolecular 
 compound melting "above 140C". 5 The "cryohydrate" of this 
 system has been studied 6 , i. e., the location of the eutectic and 
 the composition of the solid phase on either side were deter- 
 mined. 
 
 The salt used was a pure Baker and Adamson hydrate, 
 which was recrystallized from water, and maintained at 140 
 for over a week. The resulting salt gave no evidence of even a 
 trace of water. 
 
 1 Richter, loc. cit. 
 
 2 Dissertation, Col. University, 1919, pg. 15. 
 3 DeVisser, Rec. Trav. Chim. Belg., 12, 101 (1893). 
 
 4 DeVisser, loc. cit., obtained 16.67 hydrogen, scale, equivalent to 16.6 
 
 mercury scale. 
 
 5 See table I. 
 
 6 Vasiliev, J. Russ. Phys. Chem. Soc., 41, 753-7 (1909). 
 
 27 
 
The data for the system are given below and in Fig. II B : 
 (a) Solid phase H 4 C 2 O 2 . 
 
 % Base 0.0 0.83 3.59 5.40 
 Temp. 16.5 16.1 14.3 13.1 
 
 
 (b) Solid phase NaH 3 C 2 O 2 -2H 4 C 2 O 2 . 
 
 
 
 % Base 7.11 8.92 12.17 15.27 16.58 
 Temp. 25.3 36.7 54.3 66.9 71.9 
 
 21.55 
 85.7 
 
 % Base 30.72 33.03 33.16 
 Temp. 9.61 96.25 96.3 
 
 
 (c) Solid phase NaH 3 C 2 2 -H 4 C 2 2 . 
 
 
 % Base 34.03 36.87 39.06 42.54 44.25 
 Temp. 112.0 132.3 145.2 157.0 160.6 
 
 46.28 
 162.3 
 
 (d) Solid phase NaH 3 C 2 O 2 . 
 
 
 % Base 48.76 49.49 
 Temp. 174.4 195.5 
 
 
 26.86 
 93.2 
 
 SYSTEM ZINC ACETATE ACETIC ACID 
 
 No previous work in this system has been located in the 
 literature. Zinc acetate (Kahlbaum) was dehydrated at 100"C. 
 The anhydrous salt was only very slightly soluble in acetic 
 acid being soluble to 0.1 mol. % at 130C. 
 
 SYSTEM FERRIC ACETATE ACETIC ACID 
 
 Neutral, anhydrous ferric acetate cannot be prepared. The 
 salt used was basic, probably as slightly so as it is possible to 
 obtain it. It was prepared 1 by treating a solution of ferric 
 acetate in about 90% acetic acid, with an excess of acetic anhy- 
 dride, and refluxing the mixture. The resulting crystals were 
 dried with ether. The value obtained in this experiment is 
 
 1 The salt was part of a sample made by Professor J. E. Zanetti, of Co- 
 lumbia, to whom the author wishes to express his thanks. 
 
 28 
 
the maximum solubility that ferric acetate could have, since 
 the neutral salt would probably be less soluble than the basic 
 salt used. 
 
 The salt was soluble to less than 0.07% at 140C. 
 
 SYSTEM NICKEL ACETATE ACETIC ACID 
 
 There has been no work on this system reported in the 
 literature. The salt was prepared by dissolving the carbonate 
 in c. p. acid, and crystallizing- the hydrated acetate from the 
 solution. This compound was dehydrated and treated with 
 acetic anhydride to prevent the formation of any basic salt. 
 
 The salt was soluble to 0.44% at 140C. 
 
 SYSTEM SILVER ACETATE ACETIC ACID 
 
 No previous work has been reported. Silver acetate was 
 prepared from silver nitrate and sodium acetate. The precipi- 
 tate was thoroughly washed, and recrystallized from water. It 
 was dried over 99% H 2 SO 4 in vacuum. The measurements 
 could not be carried very high (not above 115), as the acetate 
 underwent reduction at higher temperature. The silver salt is 
 not very soluble in acetic acid. The data are given below : 
 
 % Base 0.094 0.204 
 
 Temp.. 76.0 115,0 
 
 CONDUCTIVITY MEASUREMENTS 
 
 To complete the series of conductivities compiled by 
 Schlesinger and his collaborators 1 , measurements were made 
 of the conductivity of systems Ba, Pb, and Mg formates in 
 formic acid. The measurements were made in a cell of the 
 Freas type, with platinized electrodes. Inasmuch as the preci- 
 sion required for a qualitative comparison of conductivities was 
 not very high, it was not found necessary to balance out the 
 capacity of the cell by means of a condenser. The results are 
 accurate to better than 1%. 
 
 1 Loc. cit. 
 
 29 
 
Conductivities were determined in a thermostat main- 
 tained at 25.00 0.01. A Leeds and Northrup bridge of the 
 Kohlrausch type, with telephone receiver tuned to 1000 cycles 
 was used. The current was supplied by a constant speed high 
 frequency generator (1000 cycles/sec.). 
 
 No attempt was made to get formic acid of conductivity 
 as low as that obtained by Schlesinger and his co-workers 1 . 
 That used in this work had a specific conductivity of 7.3 
 7.5 X 10~ r> reciprocal ohms. 
 
 CONDUCTIVITIES 
 
 34 44 
 
 Base Cone. 2 S X 10 Sa X 10 
 
 Barium 0.2758 11.92 11.85 
 
 0.1869 8.551 8.478 
 
 0.0923 4.902 4.827 
 
 0.0474 2.649 2.575 
 
 Lead 0.1047 2.462 2.387 
 
 0.0477 1.370 1.295 
 
 Magnesium 0.0919 3.006 2.931 
 
 0.0479 1.758 1.683 
 
 DISCUSSION OF RESULTS 
 
 An examination of the experimental results will disclose 
 the extent of agreement between them and the corresponding 
 consequences of the hypothesis proposed in the beginning of 
 the paper. 
 
 (a) Compounds Isolated. 
 
 The curves in general, and the particular compounds iso- 
 lated may be first compared. The addition compounds crystal- 
 lized, generally, as needles, whereas the neutral salts give in 
 most cases crystals belonging to the rhombic system. Table 
 II gives the compounds actually isolated, with the freezing- 
 point relationships of each. 
 
 1 S. and Martin, J. A. C. S., 36, 1590 (1914). 
 
 2 Cone, in equivalents per liter. 
 
 3 S=Specific conductivity of solution in reciprocal ohms. 
 
 4 Sa=S Specific conductivity of solvent. 
 
 30 
 
Compound 
 KHC0 2 -3H 2 C0 2 1 
 KHCO^H.CCV 
 KHC0 2 -H 2 C0 2 
 NH 4 HCO 2 -3H 2 CCV 
 NH,HC0 2 -H 2 C0 2 2 
 
 NaHCO 2 -2H 2 C(V 
 
 NaHCO 2 -H 2 CO 2 
 
 Ba(HC0 2 ) 2 -H 2 C0 2 1 
 
 TABLE II. 
 
 Transition Temp. 
 18.0 
 4.0 
 Melts at 108.6 
 
 29.0 
 
 prisms 27.0 
 
 needles 14.8 
 
 35.6 
 
 70.5 
 
 undeterminable" 
 
 Composition solution 
 
 in equilibrium at 
 trans, temp. (moL %) 
 
 19.6 
 23.9 
 
 23.3 
 45.6 
 43.1 
 
 25.7 
 32.3 
 
 The order of increasing complexity, as well as of increa's- 
 ing stability of the complexes, is seen to be that required by the 
 hypothesis. Potassium forms the most complex compounds, 
 and of the eight different compounds isolated, the equimolecu- 
 lar potassium acid formate is the only one sufficiently stable to 
 give an actual melting point 4 . The form of the curve (Fig. 
 I A) in the neighborhood of the maximum indicates some dis- 
 sociation of the compound into its' components 5 . The more 
 complex components undergo transition before they melt. 
 
 Next in order of complexity and stability to the potassium 
 compounds, are those of ammonium formate (Fig. IB). A 
 comparison of the extent of compound formation 6 shows sod- 
 ium and ammonium formates to be solvated (in solution) to 
 practically the same extent. The increased complexity and 
 stability in the case of the ammonium compounds, is undoubt-' 
 edly due to the temperature factor 7 the lower temperature at 
 which the ammonium complexes exist, decreases their tendency 
 
 1 Compounds not previously mentioned in the literature. 
 
 2 Two modifications not previously noted. 
 
 3 See below. 
 
 4 Groschuff, loc. cit, did not obtain a melting-point, as noted. above. 
 
 5 The relation between stability of the compound and the sharpness of 
 
 the maximum is discussed by Kendall and Booge, loc. cit, pg. 1728; 
 also by Kremann, Monatshefte_25, 1215 .(1904). 
 
 6 See section (b), following. 
 
 7 See particularly, Gross, loc. cit., p. 29. 
 
 31 
 
to decompose. It is interesting that the equimolecular com- 
 pound exists in two crystalline forms, the stable form being 
 the only one previously mentioned 1 . 
 
 The sodium compounds are quite unstable, undergoing 
 transition into compounds of a lower order of complexity long 
 before their respective melting points are reached. 
 
 The same is true of the barium compound, the transition 
 point of which is given as indeterminable. The last few points 
 on the curve are probably metastable. On standing, crystals 
 separate which are probably the anacid salt. Not enough of 
 these could be obtained for an analysis, nor could they be made 
 to separate in fine enough form to enable one to determine a 
 melting-point, i. e., to locate the stable curve. Solutions more 
 concentrated than the last one could not be obtained at 140"C. 
 
 Lithium formate, though soluble in formic acid, does not 
 form any isolable complex with it. It is noteworthy that this 
 base, that of the least electropositive of the alkali metals, yields 
 no compound, whereas barium, the most electropositive of the 
 common alkaline earths, forms an equimolecular compound 
 with formic acid. Other properties also place these two metals 
 in this order 2 . 
 
 (b) Extent of compound formation from the relative slopes of the curves. 
 
 It has already been emphasized that addition compounds 
 may be formed in solution, and yet not be sufficiently stable to 
 allow of their being isolated 3 . By applying a more general 
 method of detecting compound formation, it was shown that* a 
 better estimate of relative degrees of solvation in solution was 
 obtainable. This method consists in the determination of the 
 extent of deviation of the curve representing the data collected, 
 from the ideal curve 4 , the equation of which is 
 
 1 Groschuff, loc. cit, who gives the transition temperature as 23.5, in- 
 stead of 27.0 obtained in this work. 
 
 2 Abegg, Hand. Anorg. Chem. 2-1, 117. Soddy, Chemistry of the Radio 
 Elements, p. 44, gives the order of the elements on the basis of 
 their physical properties as K, Na, Ba, Sr, Li, Ca. 
 
 3 Kendall, J. A. C. S, 36, 1731 (1914). Kendall and Booge, loc. cit., p. 
 
 1730, Kendall, Booge and Andrews, loc. cit., 2309. 
 
 4 Washburn, Phys. Chem. p. 174, Kendall and Booge^ibid. 
 
 32 
 
+10 
 
 -10 
 
 20 
 
 -30 
 
 40 
 
 10 
 Mol. % Base 
 
 15 
 
 20 
 
 Fig. III. 
 
 Freezing-point depressions. In order of increasing depression : 
 a Ideal b Lithium c Sodium d Ammonium c Potassium. 
 
 33 
 
. -.O g eX=^ 
 
 The factors affecting the precision with which an estimate 
 of the degree of hydration of electrolytes in aqueous solution 
 can be made, are fully discussed in the articles to which refer- 
 ences has been made 2 . It has been demonstrated that a quanti- 
 tative estimate of solvation is at present impossible. In a series 
 such as that being discussed, the relative extents of solvation 
 will be given by the respective deviations from the ideal curve. 
 
 The graphs of the ideal and observed curves are given in 
 Fig. III. As no substance was obtainable which would give an 
 ideal curve with formic acid, the ideal curve was calculated. 
 The known values of Q 3 and of T 4 were substituted in the 
 equation already given, and the value of T corresponding to 
 a given value of X determined. The data from which the ideal 
 curve was plotted, follow (solid phase is H 2 CO 2 throughout) : 
 
 Mol. fract. Base 0.00 0.02 0.05 0.11 0.14 0.18 0.22 0.08 
 Temp. 281.43 280.1 278.1 274.0 271.9 269.1 266.1 276.1 
 
 The curves indicate that the extent of divergence increases 
 in the order Li, Na, NH 4 , K, in agreement with the theory. The 
 bivalent bases of Ca and Ba lie below that of potassium in the 
 order named 5 . They are not comparable with the alkali metal 
 bases, because the nature of their ionization is unknown . It is 
 probable that three ions are^ formed to some extent. It is evi- 
 dent, however, that barium formate is more extensively sol- 
 vated than calcium formate. 
 
 1 X = Mol. fraction solvent ; Q, molar heat of fusion of the "solvent." 
 
 R= 1.988 cal. (gas constant). T Q is temperature (absolute) of 
 fusion of solvent and T that of the solution. 
 
 2 The factors considered by Kendall, Booge and Andrews, loc. cit., ob- 
 
 tain in the system under consideration, and the conclusions arrived 
 at in that paper are equally significant here. 
 
 3 Q 2421.2, Berthelot, C. R. 78, 716 (1874). 
 *T ft = 281.43 Peterssen, Ber. 13, 1191 (1880). 
 
 5 Ca and Ba are not given in Fig. Ill, at 1.5% the f. pt. is 6.6; Ba at 
 
 1.5% is at 6.3. 
 6 Schlesinger and Mullinix, loc. cit., pg. 75. Schlesinger and Bunting, 
 
 J. A. C. S. 4M945 (1919). 
 
 34 
 
The divergence which has been noted results essentially 
 from ionization and from solvation. lonization increases the 
 number of moles of solute and hence produces abnormal de- 
 pressions of the freezing-point. Solvation removes solvent, 
 thereby increasing the molecular fraction (1 X) of solute, 
 and consequently has a similar effect. It is not possible to 
 determine exactly what part of the total effect is produced by 
 each of these factors. In view of the fact that it has been 
 definitely established that ''ionization is preceded by combina- 
 tion between solvent and solute, and is indeed a consequence of 
 such combination" 1 , such an attempt appears to be superfluous. 
 It is noteworthy, however, that the alkali bases, which are 
 ionized to practically the same extent 2 , give curves which are 
 quite distinct showing that the second factor, solvation, is op- 
 erative. 
 
 (c) Solubility. 
 
 In addition to those bases which have been considered and 
 classified according to the extent of compound formation, there 
 remain those bases which were not sufficiently soluble to allow 
 a determination of relative degrees of solvation by either of the 
 methods described. For these it is necessary to use solubility 
 as the criterion, as will be evident from what follows. 
 
 In an ideal binary system, the equation given above 3 repre- 
 sents the equation of the solubility curve of either substance in 
 the other, according as the values substituted for Q and T be- 
 long to one component or the other. If, then, solubility meas- 
 urements in any solvent are made on a series of salts having 
 approximately the same values of Q and T , the solubility at 
 any given temperature will be the same in all cases where the 
 solutions resulting are ideal. 
 
 1 Kendall and Booge, loc. cit, p. 2324. 
 
 2 Schlesinger et al, loc. cit. 
 
 3 See, particularly, Washburn and Read, Proc. Nat. Acad. Sci. J[, 191 
 
 - (1915). C. A., 9, 1520 (1915). Washburn, Principles of Physical 
 Chemistry, pg. 172. 
 
 35 
 
If, however, there occurs combination between the solvent 
 and some of the substances, the curves for these will fall away 
 from the maximum more rapidly than the ideal 1 . ' The sub- 
 stances will be, at any given temperature, more soluble than 
 those which give ideal solutions. 
 
 In actual practice, a series of salts in which the values of 
 Q and T are the same throughout, is never encountered. This 
 fact removes the possibility of obtaining any quantitatively 
 comparable data. In the case of the series under consideration, 
 the values of T 2 are several hundred degrees above room tem- 
 perature. Because of the large value of T , the solubility will 
 of necessity be very small in all systems in which no compound 
 formation occurs. The order of solubility will accordingly 
 correspond qualitatively with the extent to which combination 
 between the two components of the system takes place. 
 
 The greater the extent of compound formation, the greater 
 the solubility ought to be (provided the comparison be made 
 of analogous compounds at the same temperature). Since it 
 has been postulated that compound formation should vary with 
 the position of the metal in the electromotive series, it follows 
 that a similar variation in solubility should occur. 
 
 Examination of the data (also Figures I and II) shows 
 that the solubility, at 25 say, does decrease to a minimum from 
 potassium through the alkaline earths to zinc and copper. Am- 
 monium formate is far more soluble than potassium formate, 
 but this is undoubtedly due to the low value of T (300 abso- 
 lute for the equimolecular compound). The only real excep- 
 tion is in the case of the lead salt, which is too soluble. While 
 no explanation can be offered at present, it is significant that in 
 sulfuric acid 3 , and in water, the corresponding lead salts are 
 also out of their proper place. 
 
 1 This is evident from Fig. III. 
 
 - The values of T Q are not determinable, except in the cases of Na, NH 4 
 
 and K. The melting-points of K and NH 4 are low (see above). 
 
 The temperatures of decomposition are generally above 300C. See 
 
 Berichte, 51. 399 (1918). 
 3 Unpublished work for Davidson, who discusses the subject of solubility 
 
 in greater detail. 
 
 36 
 
It has already been mentioned that it was not possible to 
 demonstrate with formates alone that the solubility of the bases 
 passed through a minimum and then, with increasing diverg- 
 ence between H and the metal (whose position is below hydro- 
 gen), increases again. The supplementary experiments with 
 the acetates are more satisfactory in this respect. 
 
 Sodium acetate (Fig. II B) is very soluble in acetic acid; 
 its solubility is of the same order of magnitude as that of sod- 
 ium formate in formic acid. Two compounds were isolated. 
 The compound NaH 3 C 2 O 2 -2H 4 C 2 O 2 undergoes transition at its 
 melting point, 96.3 0.1 (the literature gives SQC.) 1 . The 
 equimolecular compound undergoes transition at 163, before 
 reaching its melting point. The curve resembles very much 
 that for sodium sulfate in sulfuric acid 2 . 
 
 Zinc acetate is soluble to only a slight extent, 0.1% at 
 130C, and the results obtained with a slightly basic ferric ace- 
 tate indicate a still smaller solubility in the case of the neutral 
 ferric salt. Silver acetate is several times as soluble as the ace- 
 tates of these metals. The acetate of nickel is abnormally solu- 
 ble, 0.44% at 140, for which fact no immediate explanation is 
 available. 
 
 The experimental results taken collectively indicate that 
 compound formation and the related properties, decrease to a 
 minimum and then increase again, when the metal (R) in the 
 system HX-RX is varied from the upper to the lower end of 
 the electromotive series. 
 
 While the increase in compound formation below hydro- 
 gen in the acetate series is not as large as might be expected 
 (for example, it is very striking in the sulfate series), it is 
 doubly significant, because it brings up for consideration a 
 point which might otherwise be overlooked. On comparing the 
 data available in the different series RSO 4 in H 2 SO 4 , RHCO 2 
 in H 2 CO 2 , etc., it becomes evident that while the alkali bases 
 ire soluble to practically the same extent in each series, the rate 
 at which the solubility falls off as R is varied through the alka- 
 line eartns toward hydrogen, is different in the several series. 
 
 1 Lescoeurs, loc. cit. 
 
 2 Landon, loc. cit., pg. 23. 
 
 37 
 
Thus, the solubility of the sulfates falls off less rapidly than 
 that of the formates, the order being RSO 4 < RHCCX < 
 RH 3 C 2 O 2 <ROH 1 . There is an inverse variation in the extent 
 to which the solubility increases in the case of the bases of the 
 more noble metals. Here the sulfates show the most pro- 
 nounced increase in solubility, and the increase in the acetate 
 and hydroxide series is small. The complete significance of 
 these facts and their proper explanation may be evident after 
 several series of fairly strong acids have been examined, and 
 the influence of X completely determined. 
 
 (d) Discussion of Conductivity Measurements. 
 
 TABLE III. 
 
 Base 
 
 Cpds. Isolated 
 
 Cone. 2 
 
 A3 
 
 Cone. 2 
 
 A' 
 
 NH. 
 
 1:3; 1:1 
 
 *0.04646 
 
 67.93 
 
 *0.09293 
 
 65.74 
 
 K 
 
 1:3; 1:2; 1:1 
 
 '0.05480 
 
 65.74 
 
 *0.09464 
 
 63.37 
 
 Na 
 
 1:2; 1:1 
 
 *0.04381 
 
 62.19 
 
 fO.0923 
 
 60.01 
 
 Li 
 
 none 
 
 *0.9574 
 
 58.79 
 
 *0.0902 
 
 56.68 
 
 Ba 
 
 1:1 
 
 0.0474 
 
 54.32 
 
 0.0923 
 
 52.30 
 
 Sr 
 
 not examined 
 
 *0.0476 
 
 53.57 
 
 tO.0923 
 
 49.34 
 
 Ca 
 
 none 
 
 fO.0474 
 
 51.37 
 
 fO.0923 
 
 46.66 
 
 Mg 
 
 none 
 
 0.0479 
 
 35.14 
 
 0.0919 
 
 31.90 
 
 Pb 
 
 none 
 
 0.0477 
 
 27.15 
 
 0.1047 
 
 22.74 
 
 From this table it is evident that the variation of compound 
 formation with the' position of the metal in the electromotive 
 
 1 To test this statement, a few measurements on calcium acetate in 
 acetic 1 acid were made. At 30, calcium acetate is soluble to less 
 than 0.35% ; calcium formate is soluble to 1.6% ; calcium sulfate 
 (or rather, a 1:3 compound) to 5.4%, and calcium hydroxide (at 
 20) to 0.04%, each in the corresponding solvent. Calcium for- 
 mate, it has been noted, shows retrograde solubility, and therefore 
 the difference in solubility between it and calcium acetate would be 
 less at higher temperatures. 
 
 - Cone, in equivalents per liter. 
 
 3 A is equivalent conductivity in reciprocal ohms. 
 
 * Schlesinger and collaborators' results, loc. cit. 
 
 f Results marked thus f, are calculated by interpolation from the rate 
 of change of A with the concentration in the neighborhood of this 
 value of the concentration. It does not change the order of the 
 bases, although necessarily inexact. Also, the difference in concen- 
 tration in the different cases to be compared, is too small to affect 
 the order of the bases. 
 
 38 
 
series is paralleled by the change in conductivity. The agree- 
 ment between the freezing-point measurements and conduc- 
 tivity measurements is good. 
 
 It is again difficult to fix exactly the correct positions of 
 lithium and barium. Lithium has a higher equivalent conduc- 
 tance than barium, although the latter gives a stable compound 
 with formic acid while lithium does not. The light lithium ion 
 may be more mobile than the barium ion. 
 
 The exact position of ammonium formate among the bases 
 is not easily located. Although less extensively solvated than 
 potassium, it shows greater conductivity. This is probably 
 explained by the fact that the ammonium complexes are formed 
 are less stable 1 , and therefore undergo dissociation into ions to 
 a greater extent than do the analogous potassium complexes. 
 
 Lead, in the conductivity results also, takes the same ab- 
 normal position as before. Its solubility was slightly lower 
 than that of magnesium, and the conductivity results agree 
 with this fact. 
 
 The results indicate that the high conductivity of formates 
 in formic acid noted by Schlesinger and his collaborators, is 
 due, as predicted by the generalization, to extensive combina- 
 tion between solvent and solute. The extent of combination de- 
 creases in order from potassium through the alkali metals and 
 alkaline earths to lead, and is paralleled practically exactly by 
 the diminution of conductivity. 
 
 The agreement of the experimental results with the argu- 
 ment advanced at the beginning of the paper, affords consider- 
 able support to the general validity of the propositions there 
 stated. 
 
 SUMMARY: 
 
 The attempt has been made to extend the generalization 
 correlating compound formation and chemical diversity to a 
 series of non-aqueous systems of the type HX-RX. On the 
 basis of the generalization, the extent of compound formation, 
 
 1 Gross (dissertation, pg. 7), has shown that the extent of ionization de- 
 pends upon : (1) The extent of the compound formation AB + CD 
 ^ AB.CD and (b) The extent of dissociation of the complexes into 
 ions of opposite charge AB . CD ^ (AB . C) + + D-. 
 
 39 
 
and hence of ionization, should vary with the difference in char- 
 acter of the two components. 
 
 The use of the relative positions of the metals and hydro- 
 gen in the electromotive series as the criterion of diversity has 
 been proposed. From this it follows that as R is varied from 
 one extreme of the series to the other, the extent of compound 
 formation should diminish to a minimum (near hydrogen) and 
 then increase again. 
 
 This conclusion has been tested experimentally by the 
 determination of freezing-point curves of representative for- 
 mate-formic acid and acetate-acetic acid systems. In addition, 
 conductivity measurements have been employed to test the 
 validity of the argument advanced. 
 
 The results have agreed strikingly with the deductions 
 from the fundamental assumptions. Five new acid formates 
 were isolated. In the formate series, the extent of combina- 
 tion between the two components decreases to a minimum as 
 the position of R is varied from potassium through the series 
 toward hydrogen. The acetate series exhibits a similar varia- 
 tion, with a slight increase in the extent of combination in the 
 case of the silver system. The change in conductivity and in 
 solubility parallel the variation in extent of combination. This 
 experimental evidence indicates the probable validity of the as- 
 sumptions to which reference has been made. 
 
 40 
 
VITA. 
 
 Howard Adler was born in New York City on January 9, 
 1896, and attended the grade and high schools of that city. He 
 received the degree of B. S. from the College of the City of 
 New York in February, 1916. From 1916 to 1917 he taught in 
 the Department of Chemistry of the College of the City of New 
 York, at the same time attending the graduate school of Colum- 
 bia University. In June, 1917, he received from Columbia the 
 degree of M. A. From September, 1917, until February, 1919, 
 he was in the United States Army. In February, 1919, he re- 
 sumed graduate work at Columbia University. 
 
THIS 
 
 AN INITIAL FINE OF 25 CENTS 
 
 OVERDUE. 
 
 7" 300ct'56vv; 
 
 LD 21-95m-7,'37 
 
(iaylord Bros. 
 
 Makers 
 Syracuse, N. 
 PH. W. 21, 1908 
 
 UNIVERSITY OF CALIFORNIA LIBRARY