ii. 25 General and Physical Chemistry. The Arc Spectrum of Tantalum on the International Scale. HEDWIG JOSEWSKI (Zeitsch. wiss. Photochem. 1917 17 79-96) .-Accurate me'asurements have been made of the wave- lengths of lines in the arc spectrum of tantalum between A7000 and h2430. The observations which are compared with those obtained previously by Exner and Kaschek also afford information relative to t'he intensity and the sharpness of the lines. There is no evidence of the occurrence of pairs of lines with a constant difference of frequency as suggested by Paulson (compare A. 1915 ii 196). H. M. D. The Photographic Spectra of Meteorites. SIR WILLIAM CROOKES (PhiZ. Trans. 1917 [ A ] 217 411-430).-The spectra of thirty meteorites have been examined with the aid of a spectro- graph characterised by certain novel features which are described.The use of a quart'z slit has given very satisfactory results the difficulty attending the production of a true knifeedge being solved by making a very narrow bevel on the front of the quartz plate and thereby producing a jaw with an angle of 90°. The bevelled edge is quite opaque in consequence of refraction and t o prevent light passing t'hrough the flat part of the plate the plane surface of the quartz was coated with gold by cathode deposition. The most striking result derived from the spectral examination of the thirty aerolites is the similarity in composition and the small number of elements which are present. Making due allowance for differences in the photographic activity of the elements in their arc spectra it is found that only ten of the known elements are present. These elements are iron chromium magnesium nickel silicon sodium manganese potassium aluminium and cdciuin and of these the first four only are present in quantity.Excepting the results for three aerolites it is found that the relative proportions of the several elements are approximately the! same in all. This suggests that the aerolites have a common origin in the disruption of some stellar body for which the process of cosmical evolution is complete. The siderites would appear to have a different origin or may possibly have formed the solid nucleus or core from which the chromium and other elements have been separated leaving the magnetic elements iron and nickel as a residue in the familiar ferro- nickel meteorites.H. M. D. Hydrogen and Calomel Electrodes. GILBERT N. LEWIS THOMAS B. BRIGHTON and REUBEN L. SEBASTIAN (J. Amer. Chem. Soc. 1917 39 2245-2261. C0mpar.e Lewis and Randall A. 1914 ii 802).-The resultx of a number of investigations on (i) the poten- tial of the hydrogen electrode in solutions of hydrochloric acid and VOL. CXIV. ii. 3ii. 26 ABSTRACTS OF CHEMICAL PAPERS. potassium hydroxide of various concentrations (ii) the dissociation constant of water (iii) the potential of calomel and silver chloride electrodes in solutions of potassium chloride and hydrochloric acid (iv) methods of establishing definite and reproducible potentials at' the boundaries between solutions and (v) on improvements in the experimental technique of such deterniinations are recorded a t some length.An electrolytic hydrogen generator which supplies a continuous stream of pure dry hydrogen I s described. Electrodes of gold coated with a layer of iridium are recommended as the most suitable for measuring hydrogen ion potentials since they very rapidly acquire the correct potential and then remain constant. Pieces of apparatus are described in which the hydrogen calomel and silver chloride electrodes are most conveniently and accurately built up. A device for maintaining a constant and easily repro- ducible surface of contact betwefen two liquids is also described. All measurements were made a t 25O and the following results each the mean of many experiment8 are given in the paper 1. Hg,HgCl,KCl O*lMI\N.E.; e = - 0.0529 volt. 2. Hg,HgCl,HCl 0.12MflKCl O.lM,HgCl,Hg; e = 0.0278 volt. 3. Ag,AgCl,HC1 O-liMIIKCl 0*1M AgC1,Ag J e =0*0278 volt. 4. Ag,AgCl,HCl O.OlM(IMC1 O*OlM AgC1,Ag ; e =0*0272 volt. The values of the E.N.F. in 2 3 and 4 afford very good con- firmation of the validity of the formula of Lewis and Sargent (A. 1909 ii 369) for calculating the potential difference between liquid surf aces. 5. H,,HCl 0*1M HgC1,Hg; e=0*3989 volt. 6 . H,,HCl O*lXlIK@l 0*1M HgC1,Hg; e=0'4267 volt. The value 0.4267 found for the Combination 6 can be obtained by calculation from the experimental values of combinations 5 and 2. If this value is combined with that obtained from combination 8 the value of the combination 9 can be calculated. 8. Hg,HgCl,HCl O.OlMIIKC1 O*OlM 13gCl,Hg ; e =0*0272 volt.9. H,,HCl O*OlMIIKCl 0*01M HgC1,Hg; e = 0'5377 volt The difference between the E.M.F. of cells 5 and 7 that is 0'1116 volt multiplied by F' (23074) gives the free energy of dilution of hydrochloric acid which equals 2573 cals. From this value the degree of ionisation of 0.1M 0*03M and 0.01M potassium chloride is found t o be respectively a=0*780 0.865 and 0.930. Making use of the foregoing potential values the value of the E.M.F. of the normal calomel electrode is calculated in terms of the value of the normal hydrogen electrode. The following values are obtained 7. H,,HCl O'OlM HgC1,Hg; e=0*5105 volt. ( a ) Hg,HgCl,HCl O'lM{(KCl O*lM HgC1,Rg ; e = 0*0012 volt. ( C ) Hz,B*(Jf)]\AT.Bo; e=0*2828 volt. ( b ) H,,H'(M)IIHCl 0.01M H,; e = -0.0644 volt.GENERAL AND PHYSICAL CHEMISTRY. ii 27 I n all cases (a) ( b ) and (c) the contact potential is eliminated. Consequently if the potential of the normal hydrogen electrode is take11 as zero the potential of the normal calomel electrode is - 0.2828 volt.10. H,,KOH O.liM1IKCl 0*1M HgC1,Hg; e = 1.0833 volts. 11. H,,KOH 0-OlMIIKCl 0*01M HgC1,Hg; e = 1.0820 volts. The dissociation const'ant of water is calculated from the hypo- thetical combination H,,OH'(ilii)llH'(nl)H,; e = 0.8278 volt and the value Kw=1-012 x 10-14 obtained. It is shown that the values of the E.M.F. obtained when the gas pressure is changed do not vary more than 0~00001 volt from the values calculated from the thermodynamic equation. J. F. S. Free Energy of Hydrochloric Acid in Aqueous Solution.11. ARTHUR A. NOYES and JAMES H. ELLIS (J. Amer. Chem. Soc. 1917 39 2532-2544. Compare A. 1916 ii 369).-In continuatioil o i previous msasurements of the E.X.P. of cells of the type II,l HC1 I HgCl I Hg the authors have replaced the calomel electrode by a silver chloride electrode with the object of obtaining more accurate data for acid solutions of low concentration. Data are recorded for solutions varying in concentration from 0.3 to 0*001N a t 1 5 O 25O and 35O. The changes in the free energy and total energy attending the transfer of one gram-molecule of hydrogen chloride from solutions of varying concentration to a 0' IN-solution are calculated from the data and it is inferred that hydrochloric acid is by no means completely ionised in 0*003N-solution.On the assumption that the ionic activity and the ionic comeu- tration can be regarded as equal in the case of this dilute solution the authors have recalculated the activity coefficients for hydro- chloric acid in solutions varying in concentration from 0-003 t o 4.5N. AS before (loc. cit.) these activity coefficients are found to diverge from the conductivity ratio h / A by about 10% in the case of a O.1N-solution. The calculated activity coefficients diminish with increase in the concentration of the acid UP to 0 . 5 ~ but increase rapidly as the concentration of the acid is further increased. H. 11. D. Potential of the Bromine Electrode Free Energy of Dilution of Hydrogen Bromide Distribution of Bromine between Several Phases. GILBERT N. LEWIS and HYMAN STORCH ( J .Arner. Chem. SOC. 1917 39 2544-2554).-A platiiium iridium electrode immersed in a solution of potassium bromide or hydrobromic acid containing free bromine has been used in the determination of the potential of the bromine electrode by measurements of the E.M.F. of the cells obtained by combina- tion of this with the calomel or the hydrogen electrode. The acid cell affords the more trustworthy results and the value of the lnomiiie potential for a solution containing bromine and bromide ion in inolar concentration against the normal hydrogen electrode is found to be - 1.0872 volts. 3-2ii. 28 ABSTRACTS 03’ CHEMICAL PAPERS. By measuring the E.M.F. of the cell H,/HBrIAgBrIAg for HBr concentrations equal t o 0.01 0.03 and 0*1N it has been found that the ionic activity coefficients are very nearly equal t o those previously found for HC1 a t the same concentrations.The determination of the ratio of distribution of bromine between carbon tetrachloride and aqueous solutions of 0*001N- and 0.1N- hydrobromic acid has shown that the constantl K = [HBr,] /[HBr][Er,] has nearly the same value as the constant for a solution in which the hydrobromic acid is replaced by potlassium bromide. By passing a current of dry air through solutions of bromine in carbon tetrachloride it has been found that the vapour pressure of the bromine is proportional to the concentration of the solution when this is measured in terms of molar fractions. H. M. D. Ionisation and Polymerisation in Cadmium Iodide solutions. It. G. VAN NAME and W.C. BROWN (Amer. J. Sci. 1917 [iv] 44 453-468. Compare A. 1917 ii 455).-In the further investigation of the constitution of cadmium iodide solu- tions measurements have been made of the E.M.P. of cells in which iodine electrodes are in contact with iodine-cadmium iodide solutions and also (of the freezing p i n t s of solutions containing cadmium iodide and varying proportions of iodine. The freezing-point data show that the freezing point of a cadmium iodide solution is depressed to the extent of about 1.4O per mol. of added iodine. This lowering is very nearly the same for solutions in which the cadmium iodide concentration is varied considerably. The facts point to the existence of polymerised mole- cules in considerable quantity. From the E.M.P. data it) is possible to calculats the iodine ion concentration in iodine-cadmium iodide solutions and by extra- polation to zero iodine concentration to obtain the iodine ion con- centration for pure solutions of cadmium iodide.For the niore dilute cadmium iodide solutions examined (0.01 and 0.125 molar) the data are in agreement with the assumption that complex mole- cules are present but the behaviour of the stronger solutions does not appear to be compatible with this hypothesis. The assumption that complex molecules of the type (CdI,) are the only complex molecules formed is found to bs insufficient t o reconcile the observations which have been made according t o the distribution freezing-point and E.M.F. methods. H. M. D. A Comparison of the Activities of Two Typical Electrolytes. G.A. LISHART ( J . Amer. C’hem. SOC. 1917 39 2601-2605).-Measurements of the E.M.F. of cells of the type H2 I HCI 1 HgCl I Hg have been made for solutions containing from 0.01 t o 16.0 mols. of hydrogen chloride in 1000 grams of water. Prom the results the ionic activities are calculated and compared with the .corresponding values for potassium chloride. On the assumption that these thermodynamic quantities afford a measureCIENERAL AND PHYSICAL CHEMISTRY. ii. 29 of the degree of ionisation of the two electrolytes it is found that there is a considerable divergence between the degrees of ionisation even in dilute solution whilst in concentrated solutions the divergence is enormous. Specific Heats and Heats of Fusion of Triphenylmethane Anthraquinone and Anthracene.JOEL H. HILDEBRAND (MISS) ALICE D. DUSCHAK A. H. FOSTER and C. W. BEEBE ( J . Amer. Chem. Soc. 1917 39 2293-2297).-The specific heat and latent heat of fusion of triphenylmethane anthraquinone and arithracene have been determined in a calorimeter similar to that described by Lewis and Randall (A. 1911 ii 371). The specific heats were determined over several temperature ranges. The materials were contained in vessels of quartz glass or " pyrex " glass and consequently the specific heat of these substances had to be determined. The following values were obtained pyrex glass s = 0.174 + 0.00036t ; quartz glass mean value over the range 20-320° s = 0.2161 ; triphenylmethane solid s= 0.186 + 0*00277t liquid s = 0.479 ; anthraquinone solid s = 0.258 + Om0007t liquid s = 0.66 ; anthracene solid s = 0.280 + 0*0007t liquid s= 0.509.The following heats of fusion were also obtained triphenyImethane 17.8 cal.; anthraquinone 37.4 cal.; anthracene 38.7 cal. It is pointed out that the specific heat equation for triphenylmethane is not trustworthy for extrapolation to lower temperatures as the range of temperature from which i t was obtained was so small (20-60°) and it is also suggested that the temperature coefficient of the specific heat is much larger than would be expected. H. If. D. J. P. S. The Entropy of the Elements and the Third Law of Thermodynamics. GILBERT N. LEWIS and G. E. GIBSOX (c7. Amer. Chem. Soc. 191 7 39 2554-2581).-A theoretkal paper in which the authors have calculated the entropies of the elements and applied the results in testing the theorem of Nernst which may be regarded as equivalent to the statement that the entropy of every substance is zero a t the absolute zero of temperature.On the assumption that the entropy of a substance is known at one temperature the entropy a t any other temperature can be calcu- lated if the specific heat' is known for the interval of temperature concerned. It i= shown that' the entropy may be calculated by a qraphical method which does not necessibate any assumption in regard t o the exact form of the heat capacity equation. The calcu- lated atomic entropies show with respect to atomic weight or atomic number the 9ame kind of periodicity which characterises certain other properties of the elements. According to the equation AF-AH= - TLS in which AF is the increase in free energy A H the increase in total energy and AS the increase in entropy for any isothermal change i t is possible to calculate the free energy of formation of any compound from its elemmts i f tlhe entropies of the compound and of the elements ail({ii.30 ABSTRACTS OF CHEMICAL PAPERS. the heat of formation of the compound are known. Conversely the entropy change associated with the formation of a compound from its elements can be calculated from the equation if the changes in total and free energy are known The entropy differences thus calculated for a number of com- pound substances are found t o agree satisfactorily with those which are derived from the atomic and molecular entropies calculated according to the method referred t o above in which it is assumed that the entropies of the elements and compounds are zero at th0 absolute zero of temperature.This agreement is considered t o afford new support for the so- called third law of thermodynamics. H. M. D. Determination of Boiling Points in Capillary Tubes. FRIEDRICH EMICH (Mo?zatsh. 1917 38 219-223) .-An open glass tube 7-8 cm. long of external diameter 0*6-1*2 mm. and with B wall 0.1 mm. thick is drawn out at' one end to a fine capillary approximately 2 cm. long and of 0.05-0.1 rnm. diameter. The end of the capillary is immersed in the liquid to be examined a,nd when about half a cubic millimetre has entered the tip is sealed by contlact with a flame. I f this operation is successful the capillary will have a minute bubble a t the extreme end covered by a liquid plug nearly 1 millimetre in length The tube is attached to a thermometer and warmed in a bath in which the heating liquid is 4-5 cm.deep. As the temperature is raised the plug of liquid ascends the capillary and the In. p. is registered when the plug reaches the level of the surface of the heating liquid outside. The method is naturally restricted t o pure substances. D. F. T. An EEcient Apparatus for Fractional Distillation under Diminished Pressure. WILLIAM A. NOYES and GLENN S. SKINNER (t7. Amer. Chem. SOC. 1917 39 2718-2720).-A modified Claisen flask is used. The side-tube of the flask is bent upwards and fused on to a simple fractionating cohmn and into the side of the neck of the flask is fused a separating funnel.The flask can thus be used for large or small fractions by repidating the flow of liquid from the separat'ing funnel and successive fractions may be intro- duced without losing the vacuum. T I T . G. Studies in Catalysis. VIII. Thermochemical Data and the Quantum Theory. High Temperature Reactions. WILLIAM CUDMORE MCCULLAGH LEWIS (T. 1917 131 1086-1102. Compare ibid. 457).-According t o the radiation theory and the quantum hypothesis the heat of a reaction Q is given by the equa- tion Q=Nh(Zv,-Xv,) in which P is the Avogndro constant h the Planck constant Z v the sum of the critical frequencies of the react- ing substances and 2v2 the corresponding quantity for the resultant products. This relation has been previously deduced by Haber (Rer.Beut. ph?ysikrrZ. Ges. 1911 13 11171 who calculated tlheQENERAL AND PHYSICAL CHEMISTRY. ii. 31 critical frequencies of substances for which the requisite data were not available by means of the semi-empirical relation vV/vy= J.M/m in which vv is the characteristic ultra-violet frequency vl. the characteristic ultra-red frequency in terms of which the specific heat may be represented M the molecular weight of the substance and m the mass of an electron. The use of this equation involves considerable uncertainty in regard t o 171 aiid in Raber’s treatment of the problem in its application to the formation of a salt such as sodium chloride from its elements the quantity 21)~ is arbitrarily made equal to half the sum of the critical frequencies of the elements instead of the entire sum.By making the assumption that the ultra-violet quantuni breaks the bond between two adjacent atoms which are both thereby rendered chemically active i t follows that one quantum character- istic of sodium plus one quantum characteristic of chlorine will suffice to bring about the change represented by 2Na + 2@1= 2NaCl. Hence 2Nhvxac - Nh(v + vC,) should be equal to the heat of formation of two gram-molecules of sodium chloride. This relation is identical with that which follows from the introduction of the arbitrary assumption which is characteristic of Haber’s method of treatment. The application of the equation to the calculation of the heat of formation of sodium chloride may be briefly indicated. From the wave-length (A = 52 p) of the characteristic infra-red band v,.=0*0577 x 1014. By means of the square root relation (see above) v?) = 19.27 x 1014 from which the critical increment Nh:~N;IC = 182,290 ca!. Similarly the sum of the critical increments for 1 gram atom of sodium and 1 gram atom of chlorine is found t o be 85,000 cal. The lieat of formation of the salt is therefore 182,290-85,000= 97,290 cal. which agrees with the observed value 97,800 cal. Similar calculations have heen made in respect of other sub- stances fcr which the requisite data are available and the results when compared with the observed heats of reaction show a degrea of apreement which supports the validity of the equation connect- ing the heat of reaction aiid the critical freqnencies. Thermochemical Studies.DANTEL LAGERLOF (J. pr. Chem. 1917 [iil. 96 26-34).-A theoretical paper in extension of the earlier mathematical discussion (A. 1904 ii 382 605 ; 1905 ii 76 677) of the heat. of formation of carbon compounds the thermal effect of the intramolecular linkings being especially con- sidered from the author’s point of view. -__ H. M. D. D. F. T. Thermochemical Studies. The Constitution of Benzene and of some Condensed Aromatic Hydrocarbons con- sidered from the Thermochemical Point of View. DANIEL LAGERLOF (1. pr. Chew. 1917. rii] 96 35-49).-The relative stability of the cyclic hydrocarbons containing rings larger than cyclopentane as compared with the smaller rings such as cyclo- propane is attributed to the strain in reducing the aagle betweenii. 32 ABSTRACTS OF CHEMICAL PAPERS.the carbon valencies in the latter causing an endothermic effect whereas in the author's view the enlarging of the valency angle in the format'ion of hexatomic and bigger rings produces an exothermic effect if the ring is a plane one. Mathematical argu- inents are adduced in favour of this theory and the annexed formu12 are suggested for benzene naphthalene anthracene and phen- anthrene respectively. D. F. T. The Standard Unit in the Thermochemistry of Organic Compounds. W. SWIENTOSLAWSKI ( J . Amer. Chem. SOC. 1917 39 2595-2600).-The values obtained for the heats of combustion of naphthalene benzoic acid and sucrose in recent measurements are compared and discussed. I n terms of the 15O calorie the most probable values for the heat' of combustion of 1 gram of substance weighed in air are naphthalene 9612 Cal.; benzoic acid 6311 Cal.; sucrose 3945 Cal.To obtain satisfactory agreement in thermochemical data it is recommendeld that the heat capacity of calorimetric bombs should be determined by a standard method involving the use of a standard combustible substance. The question whether the heat of combus- tion is t o be expressed in terms of kilo-joules or calories should be determined by the International Congress. Until the standard substance has been decided on the calibration of calorimetric bombs should be based on one or other of the above values for the heats of combustion of naphthalene benzoic acid or sucrose. 33. M. D. Improved Form of Pyknometer. MARKS NAIDLE ( J . Amer. Cham. Soc. 1917 39 2357-2388).-A modification in the cap of tli3# side arm of a pyknometer to provide f o r any liquid which may be driven'out of the instrument by expansion during the weiqhing.J. F. S. Improved Victor Meyer Vapour Density Apparatus. D. A. MACINNES and R. G . KREILING (J. Arner. Chem. SOC. 1917 39 2360-2354).-1mpi-ovements to the Victor Meyer vapour density apparatus in connexion (i) with the means of introducing the substance (ii) with the vaporisation tube are described. It is pointed out thatl when the cork of the usual form of the apparatus is withdrawn t o allow of the admission of the substance under investigation there1 is a certain amount of cooling of the air inside the tube also there may be a certain amount! of spirting of the liquid on to the walls of the vaporisation tube.Both these effects will produce! errors. The authors suggest a means of introducing the substance a t the bottom of the bulb and a t the temperature ofGENERAL AND PHYSICAL CHEMISTRY. ii. 33 the vaporisation tube. A long glass tube reaching almost to the bottom of the vaporisation tube is fitted by means of a rubber stopper in the neck of the apparatus. A brass rod fitted with a hook a t its lower end passes down this tube and is made air-tight a t the upper end by means of a rubber tube. The substance under investigation is placed in a small bulb which has a long capillary neck (2-3 cm.) bent twice a t right angles. This is placed on the brass hook and the rod drawn up until the bulb neck just touches the enclosing glass tube. When the temperature of the vaporisa- tion vessel is constant the bulb is broken by drawing the brass rod sIightly further u p the tube.It is stated that the air in the ordinary vaporisation tube not being all a t the same temperature is the cause of niany errors. To obviate theHe the authors suggest a modified form of vaporisation vessel. This consists of a large test-tube 25 cm. long and 5 cni. diameter which carries a rubber stopper through which passes a capillary delivery tube and a straight tube 28 cm. 1on.q and 1.5 cm. diameter. The latter tube is placed centrally reaches almost t o the bottom of the outer vessel and carries the breaking apparatus described above. The whole apparatus is placed in a large boiling tube in the usual manner. It is to be pointed out that with thiq angaratus owing to the sudden rush of air when the tube of.materis1 is broken t.lie usual eudiometer and pneumatic trough are use- lees and must be replaced bv a gas burette.Trial experiments 2re described with numerical details €or bromine ethyl alcohol a n d diethpl ether. The resdts are in every way quite good. J . 5'. S. Convergence of the Liquid and Solid Volume Curves to Absolute Zero. GERVATSE LE BAS (Chem. Nms 1917 116 307-308).-1t is shomii t1mtm in general the solid and liquid ciirves c0nverg.e to absolute zero. This applies to types of substances where the liquid volume curve is stleeper than the solid curve where the volume of the resiiltant solid is greater than that of the liquid. w. P. s. The Relation between Temperature and Molecular Surface Energy for Liquids between -80" and 1650O. F.M. JAEGER (Zeitsch. anorg. C h ~ i n . 1917 101 1-214).-The author has determined the surface tension and molecular surface energy of about 200 organic iiquids between -80° and 250° and of about 50 inorganic substances in the molten condition between 300° and 1650O. The method employed was to determine the burst- ing pressure of bubbles of the liquid blown on the end of a capillary tube of known diameter just' immersed in the liquid. For high- temperature work the substance under examiliation was melted in a veFsel of platinum or platinum-rhodium heated in a resistance furnace the capillary tube being of the same material. The coni- plicated nppnratus used is described and illustrated in great detail.ii.34 ABSTRACTS OF CHEMICAL PAPERS. In the specially designed manometer normal octane was used in contact with mercury; it is strongly recommended as an ideal liquid for this purpose. For each of the substances examined a table is given in which is detailed for each temperature atl which observations were made (1) the surface tension x in ergs per square centimetre calculated from the equation x=r31/2 where T is the1 radius of the capillary and H is the bursting pressure in dynes; (2) the density d of the liquid; (3) the molecular surface energy p in ergs per squars centi- metre where p,=x(M/D)$; (4) the specific cohesion A2=2x/(g. d); (5) the quantity (A2M)/T where T is the absolute temperature of the melting point; (6) the temperature coefficient of the molecular surface energy dp/dt.Every substance examined was carefully purified and its density determined at' different temperatures special methods being developed for the high-temperature measurements. For many of the substances x-t and p-8 curves are given. It has been demonstra€ed by Eotvijs from van der Waals's law of corresponding states that for normal non-associated liquids dp/dt should be a constant=2*25 ergs per lo whilst for associated liquids the temperature coefficient should be smaller. Further it can be shown thermodynamically that if dpldt is a constant the specific beat of the surface layer must be the same as t h a t of the bulk of the liquid. The great number of p-t curves now examined illustrates well the constitutive character of molecular surface energy.The curves are rarely straight lines tho value of the temperature coefficient in the case of organic liquids generally fzlling but sometimes increas- ing with rising temperature. I n general in a series of related substances such as alcohols fatty acids or esters a t a given temperature the value of p increases with the molecular weight. The introduction of increasing quantities of halosen into hydro- carbons also increases the molecular surf ace energy. Isomeric sub- stances such as ethylene dicliloride and ethylidene chloride show wide differences. The value of dpld't is fairly constant for a series of related substances but marked exceptions sometimes occw ; thus formic acid has an exceptionally low value. I n the primary secondary and tertiary aliphatic amines the values of ,u and CJp/dt increase very markedly with increasing carbon content the lower members of the series haviiig ahnorr?n!ly low temperature coefficients.Of isomeric primary amines those with straiqht carbon chains have greater molecular surface enerqy than those with branched chains. Unsaturated substances such as allylamine Lave higher values of ,u than the corresponding saturated compounds. Formamide has an exceptionally low temperature coefficient 0.89 erg per lo similar to t h a t of water. This fact is probably to be associated with its high dissocizting power. The unsaturated character of aromatic compounds is accompanied by increased values of p. The halogenated compounds show in- creasing values of p with increasinq 'molecular weiqht.Position isomerides show marked differences ; for example of the three nitro-GENERAL AND PELYSICAL CHEMISTRY. ii. 36 phenols the para-compound has the greatest and the ortho-com- pound the least molecular surface energy. I n the nitroanisales the differences are much less marked probably because here there is no mobile hydrogen atom. The molecular surface energy of aniline and its homologues is much higher than that of any of the primary ali- phatic smines up to heptylamine. The introduction of halogen or nitro-groups into aniline increases the value of p as in the hydro- carbons. The surface energy relationships of many other aromatic compounds are discussed. Specially interesting are tlie p-t curves for such substances as p-azosyanisole p-azoxyphenetole and anisaldazine which form anisotropic liquids (liquid crystals).The curve consists of two distinct portions with a sharp minimum where the liquid passes from the anisotropic t o the normal form. The temperature coefficient of the anisotropic liquid is always greater than that of the normal liquid a fact which is contrary to Eotvgs’s conclusion that a lower temperature coefficient indicates a greater degree of molecul~r association. It is concluded that in substances of this class very complex and little understood changes are brought about by heat. The relationships among the haloids of phosphorus arsenic antimony and bismuth are in general similar to those among organic halogen compounds tlie values of x and p increasing with the molecular weight whilst the magnitude of d p l d t is more or less normal.Quite different however are the haloicis of the alkali metals lit,hium sodium potassium rubidium and czesium. For the same metal. the surface tension x of the molten salt! decreases with increasing atomic weight of the halogen from fluorine t o iodine and also decreases for the same halogen with increasing atomic weight of the metal. The values of p vary in an irregular manner whilst du./df is in all cases abnormally small. Other salts investi- gated include sulphates nitrates borates molybdates and tanestates of the alkali metals. It is concluded that a t high temperatures the law of correspond- ing states cznnot hold for mclten salts which are probably highly ionised and t h a t Eijtviis’s rule based on the va,lidity of this assump tion that a low value of dp/dt indicates a high degree of associa- tion in the liquid is therefore invalid.Also since d p / d t is bv no m2am constant f o r organic or inorganic liquids the1 specific hent of the surface larer must be different from that of the rest of the liquid. and the surface energy must be a t least in part of a kinetic nature. The author has investiqated t h e empirical rule discovered by Walden that for many non-acsccixted liquids the quot.ieEt obtained 13-7 dividinn the niolocular cohesion by the alxolute temperature of (n) the n?slt,inc point or ( 6 ) the boilinq point is a constant in the case of ( a ) 3.65 and of ( h ) 1.15 approx. The vaIuw of these two I f constants ” have been calculated pnd tabulated by the author for about 200 different substances which are divided into four groups according to the manner in which either value varies fromii.36 ABSTRACTS OF CHEMICAL PAPERS. the mean.. The rule is evidently only approximate and many variations occur which cannot be explained on the ground of mole- cular association. The greatest irregularities occur however among inorganic salts. E. H. R. Solubility and Internal Pressure. JOEL H. HILDEBRAND ( J . Amer. Clzem. SOC. 1917 39 2297-2331. Conipare A. 1916 ii 518).-Where Raoult's law is obeyed by a solution the solubility of the solid a t the absollite temperature T may be calculated by means of the expression log N = - LTT7,/4.58(T7 - T) where N is the solubility expressed in terms of molecular fractions L is the molecular heat' of fusion and T the melting point in absolute degrees.The author has calculated the solubility of anthracene anthraquinone p-dibromobenzene phenanthrene and tripheiiyl- methane a t 2 5 O by means of the formula and compared these values with the experimentally determined solubilities (see this vol. i 62) in some eight solvents. Iodine has been treated similarly. The divergence of the solubility from the calculated value is considered along with the interns1 pressure of the solvent and it is shown in the case of anthracene which has practically the calculated solu- bility in carhon disnlphide t h a t its solubility decreases with decreasing internal pressure for non-polar liquids. I n the case of alcohol a polar liquid the divergence from the calculated solubility is much greater a fact which agrees with the third rule previousIy stated (ZOC.c i t . ) . Formation of Additive Compounds in Aqueous Solutions. Stability of Hydrates and the Determination of Hydration in Solution. JAMES KENDALL JAXES ELIOT BOOGE arid JAJIES C ANDREWS ( J . A mw. C12~n2. ,COG. 1917 39 2303-2323).-The results of previous work on the formation of additive componnds are summarised and the general rules established in this work are recapitulated and extended to aqueous solution (A. 1914 i 858 1069; 1915 i 16 80; 1916 i 599 707). Since water can function both as a weak acid and a weak base the extent OF hydration in aqueous solutions should be found to increase with the increasine acidity or basicity of the solute. The known hydrates of acids and bases are reviewed in the paper and it is shown t h a t only the stronger acids and bases yield compounds with water which are capable of isolation.A number of freezing-point determinations with solutions of varying concentrations of the following acids acetic P-hydroxypropionic a-hydroxypropionic citric d-tartaric hydrofluoric phosphoric and hydrochloric have been carried out. The results in all cases conform exactly with the abov'e-mentioned prediction. A critical discussion is given of the factors which must be taken into consideration in the determination of hydration in solution by the freezing-point method. The question of the hydra- tion of salts is treated in a preliininary manner anti i t is shown that the hydration follows the order of the scheme Ce<Rb<T<<NHf<Na<T.i J.F. S. and NO < C1< Rr < I. J. F. s.GENERAL AND PHYSICAL CIiEMlSTRY. ii. 37 Mechanism of the Ionisation Process. J A 31 F:S K ENDALL aiid JAMES ELIOT BOOGE (,I. Rnzer. Chent. Soc. 1917 39 2323-2333).-The results of a number of papers by the authors (see preceding abstract) have indicated an intimate and general connexion between ionisation and the formation of compounds in solution. I n the present paper the authors advance the hypothesis that ionisation is preceded by combination between solvent and solute 2nd is indeed a consequence of such combination. This point of view combines the current ionic and hydrate' hypotheses re- f erring conductivity in solutions t o the dissociation of solutesolvent complexes into radicles of opposite charge.The actual mechanism of the ionisation process under this assumption with its relation t o phenomena such as unsaturation association and high dielectric constant is briefly discussed. It is shown that compound forma- tion between solvent and solute may be postulated in all conduct- ing solutions and that the distinction still commonly retained between the two components is arbitrary and misleading. The general evidence in favour of the hypothesis is given in a pre- liminary form attention being centred on a few fundavneiital points only. J. F. S. Soap Bubbles as Models of Crystal Structure. ill. J. MARSHALL ( J . Amer. Chent. SOC. 1917 39 2386-2387).-1t is shown that when small soap bubbles of uniform size are produced on the surface of a soap solution they form a symmetrical network which is in reality the simple face-centred lattice as found in crystals of pure metals.These bubble aggregates can readily be produced and projected on the screen and so serve to show causes of crystal structure and the method of building up of crystals. The best effects are produced by using a solution of sodium oleate to which glycerol has been added and blowing the bubbles by means of a glass tube which has been constricted to 1 nim. diameter by drawing and then further constricted t o a very fine tip by allow- ing the walls of the 1 mm. tube to fall together in a smoky flame. The jet should be at right angles t o the surface of the soap solutioii when the bubbles are blown. J. F. S. Formation of Crystals in Gels.HARRY N. HOLMIW ( J . Physical Chem. 1917 21 709-733).-The influence of silicic acid gels on the forination of crystals has been examined by experiments in which one of two reacting soluble substances was added t o a solution of silicic acid which was then allowed to set. The second substance dissolved in water to give a solution having a greater osmotic pressure than the jelly was then brought into contact with the upper surface of the jelly and slow diffusion allowed to take place. If an insoluble substance is produced by the reaction this forms within the jelly and the slow diffusion process leads to the formation of large well-developed crystals. Perfectly f orined tetra- hedral crystals of copper may for instance be obtained by the diffusion of hydroxylamine hydrochloride into a silicic acid gel con-E.38 ABSTRACTS OE' CHEMICAL PAPERS. taining copper sulphate. Other substances obtained in crystalline form by this method were silver dichromate gold lead iodide mercuric iodide basic mercuric chloride silver sulphate silver acetate and basic lead chromate. The capillarity associated with the gel structure is supposed t o be partly responsible for the observed crystal growth and a similar influence is brought into play when the fine-grained precipitation membrane begins to be formed. I n support of the theory that the capillary diffusion is the chief factor in the phenomenon i t has been observed that well-formed crystals may be obtained by allowing slow diffusion t o take place through flowers of sulphur barium sulphate or alundum.H. M. D. Properties of Mixed Liquids. 111. Law of Mixtures. I. J. LIVINGSTON R. MORGAN and M ~ R Y A. GRIGGS (J. Amer. Chein. Soc. 1917 39 2261-2275. Compare A. 1916 ii 224 296).- With the object of testing the validity or otherwise of the simple law of mixtures the surface tension of a number of homogeneous mixtures has been determined by the drop-weight method a t two temperatures in each case. The lower temperature was loo or 1 5 O and the higher temperature 40°. The folIowing mixtures in a series of compositions were measured ( a ) Binary mixtures benzene- toluene benzene-ethyl propionate benzene-chlorobeiizene benzene- methyl butyrate benzene-propyl acetate benzene-acetone toluene- ethyl propionate toluene-chlorobenzene toluene-methyl propionate toluene-ethyl formate acetone-chlorobenzene chlorobenzene-ethyl propionate ethyl lactatepropyl acetate chlorobenzene-methyl butyrate amyl p-phenylpropionate-methyl propionate and acetone- propyl acetate.( b ) Teriiary mixtures benzene-tolueneethyl propionate benzenetoluene-chlorobenzene benzenetoluene- methyl propionate and benzene-toluene-acetone. ( c ) Quaternary mixtures benzene-toluene-ethyl propionate-chlorobenzene benz- ene-t oluene-meth yl pr opionat e-et hyl lactate and benzene-t oluene- methyl butyrate-propyl acetate. ( d ) Quinary mixture benzene- toluene-methyl butyrate-propyl acetate-methyl propionate It is shown thatl ten of the above-mentioned mixtures follow rigidly the law of mixtures in the form = Z,P,+l,P,+ etc. (where the summation of the relative weights I Zb etc.is equal to unity. Of these mixtures six were of two constituents two of three con- stituents one of four constituents and one of five constituents. Where variations appear the observed value is invariably less than that calculated from the law. These deviations increase in magni- tude with increased temperature and are always at a maximum at both temperatures for that mixture which contains equal weights of the constituents. Although the deviations might be due t o the magnibude of the difference in the surface tension values of the constituents when pure the effect is probably negligible and merges into the more important' factor-the nature of the constituent. Ail example of the latter is chlorobenzene which renders every rn ixture in which it is present abnormal.The deviation of a complex mix-GENERAL AND PHYSICAL CHEMISTRY. ii. 39 ture is not a summation of the deviations of the pairs of liquids of which it could be made but is of the same order as these. It is further shown that chemical interaction for binary mixtures could not be the cause of the maximum deviation invariably found a t a composition of 50% by weight of the two constituenh whereas this behaviour is shown to be exactly what might be expected if the one liquid by its simple physical presence influenced the value of the property of the other and the conclusion is consequently drawn that the mixture law considered is a rigid law provided no chemical action takes place between the constituents and neither liquid influences the value of the property of the other.J. F. S. Properties of Mixed Liquids. IV. Law of Mixtures. 11. J. LIVINGSTON R. MORGAN and ANDREW J. SCARLETT jun. ( J . A naer. Clienz. SOC. 1917 39 2275-2293. Compare preceding abstract).-The surf ace tension of the following binary mixtures (i) water-acetone (ii) acetone-ethyl alcohol (iii) phenol-acetone (iv) phenol-ethyl alcohol (v) benzene-acetic acid (vi) benzene- ethyl alcohol (vii) benzene-methyl alcohol (viii) acetone-methyl alcohol (ix) ethyl alcohol-nikthyl alcohol and (x) benzene-phenol has been determined by the drop-weight method over a range of concentrations and temperatures. The curves representing the variation in surface tension with concentration are in general with- out maxima or minima but that of (v) shows a minimum whilst (ii) and.(viii) show maxima. The comparison of the experimental results with those calculated by nieans of the mixture law of Morgan and Griggs leads to the following observations which fall into three groups (a) the systems (i) (iii) (iv) (v) (vi) (vii) and (x) give values smaller than the calculated values; (6) systems (ii) and (viii) give values larger than the calculated values; (c) the remaining system (ix) gives a slight positive deviation a t OG no deviation a t 30G and a slight negative deviation at 45O. The position and magnitude of the maximum deviation from the mixture law found when the deviation is plotted against’ the con- centration of one constlituent divides the systems into two classes. I n one class the maximum deviation always very small is found €or a mixture containing 50% by weight of each constituent.The systems falling into this group are (ii) (viii) and (ix). The only explanation of this behaviour is that it is due to the physical effect of the one liquid on the other since an equal weight of the two constituents brings about the effect. Systems of the other class on the contrary exhibit a maximum deviation usually larpe and at some other concentration than 50% which corresponds always with some simple and even relation of the molecular weights of the constitueiits that is corresponds with a definite chemical formula. The cause of this according to the theory put forward by Denison (A. 1913 ii 30) is the actual production of a compound. The molecular compounds found to exist in the binary mixtures of liquids examined are COl!k2,10H,0 2PhOH,COMe PhOH,2EtOH C,H,,2CH,-C02H,ii.40' ABSTRACTS OF CHEMICAL PAPERS. 2C6H,,EtOH C6H6,MeOH and 4PhOH,3C&&. The existence of these compoands in solution is confirmed by density and vis- cosity measurements. The compound C,H6,2CH,*C0,H shown to exist in the system benzene-aoetic acid is particularly interest- ing when considered in the light of the results of other methods which lead to the conclusion that acetic acid is always polymerised into double molecules in benzene solution. These methods are such however its would fail to show a combination o€ the solvent with the polymerised solute even if it did exist; and hence the evidence found here is not only not inconsistent with other evidence but gives a wider point of view on the process which has been designated hitherto as a simple polymerisation System of Recording Rate of Chemical Reaction.JAMES W. MCBAIN (Chem. News 1917 116 315-316).-It is suggested that the usual expression for denoting the velocity constant of a reaction may be replaced by a number which has a direct physical significance. For instance it may be written k = 1 / t (remainder of thd expression) where k is the! present rate constant of the reaction and t is the time the reaction has proceeded or k t = (remainder of the expression). It is necessary to give the value of k and also the value of the unit of time (minutes or hours). The value of k is however always set a t unity and the unit of time is chosen accord- ingly; the equation then becomes t = (remainder of the expression). The chief advantage is that the proposed " unit of time" gives a direct idea of the rate of the reaction.J. F. S. w. P. s. Contact Catalysis. 111. WILDER D. BANCROFT (J. Physiccd Chem. 1917 21 734-775. Compare A. 1917 ii 566; this vol. ii 13).-A review of the literature relating to the action of poisons in contact catalytic reactions. The changes in over-voltage pro- duced by certain ions are supposed to be effects which are com- parable with retarded or inhibited contact catalytic reactions. n. M. n. Revision of Atomic Weights in 1916. E. MOLES ( J . Chinz. phys. 1917 15 433-469).-A review of $he work on the deter- mination of atomic weights published during 1916. Errors affecting the Determination of Atomic Weights.VI. Surface Actions as a Sourceof Errors in Weighing. PH. A. GUYE and E. MOLES (J. Chim. phys. 1917 15 360-404. Compare A. 1916 ii 385 386 432 445).-A further consideration of the errors involved in the accurate determination of equivalent weights in which the authors direct attention to the anomaly first pointed out by Hinrichs (compare A. 1893 ii 163 316; 1894 ii 276) that the value of the combining ratio is a function of the quantity of substance used in the determinations. It is considered that the available data afford clear evidence that such a relation does actually exist but the interpretation given by Hinrichs is considered to be unacceptable H. M. D.GfiNlCRAL ANI) 1'iIYSICAL CHEMISTRY. ii. 41 By reference t o data obtained in recent work 011 the deterniina- tion of atomic weights it is found that all series of measurements do not show the occurrence of such a connexion between the com- bining ratio and the quantity of substance operated on but that this is confined to series oi' determinations in which the quantit'ies of substance employed have been determined by weighing in air the reduction to a vacuum being effected by calculation.This suggests that the anomaly is due t o surface condensation of air water etc. and it is shown t h a t the curves which express the rela- tion between the experimental combining ratio and the weight of substance operated on can be satisfactorily accounted for on this hypothesis. The average relative deviation attributable t o this source of error is 1 in 20,000 but it.is sometimes as high as 1 in 10,000. H. M. D. Errors affecting the Determination of Atomic Weights. VII. Surface Actions as a Source of Errors in Weighing. PH. A. GUVE and E. MOLES ( J . Chini. ph~ys. 1917 15 405-432. Compare preceding abstract) .--Experiments have been made with silver in the form of a solid block and with finely powdered zinc oxide with the object of ascertaiiiiiig the magnitude of the errors which may be ascribed to the formation of a surface filin when these substances are weighed in the air. In the case of silver the error involved amounts t o 2 x 10-5 grain per gram of silver. This value is to be regarded as a minimum the actual error in practice depending on the humidity of the air in tlhe balance case on the nature of the surface of the metal and on other factors. The error attributable to surface condensation according t o the experi- inents with zinc oxide is of the order 0.2 t o 0.4 x 10-5 gram per gram of substance. It is shown that errors of this order of magnitude affect the value of the second decimal figure when the atoniic weight of the element under consideration is greater t,han 100. The errors in question should be eliniiiiated in atomic weight measurements by actually weighing the substances involved in a vacuuni. The possibility of making such weighings has been greatly increased as a result of recent improvements in the technique of various forms of micro- balance. H. M. D. Graphical Interpolation of Tabulated Data. HORACE G. DEMING ( J . Anacr. C k e m . SOC. 1917 39 2388-2392).-A method is described based on the principle of a triple parallel alignment chart' whereby the intersolated values required from data may be rapidly obtained by a graphic method. It is claimed for example that the adoption of this principle t o logarithm tables would reduce the amount' of space occupied by such tables to about 10% of that now necessary. .J F. S. VOL. cxiv. ii. 4