INORGANIC CHEMISTRY.1. ATOMIC WEIQHTS.THE separation from each other, by chemical means, of the scarcermembers of the rare-earth group of elements is attended by so manydifficulties that the finality of the values assigned to their atomicweights may well be questioned. For these elements, especially,the advantages of a control by physical methods are obvious, butsince the examination in 1925,l of lanthanum, praseodymium,cerium, and neodymium in the first mass spectrograph, no data havebeen available until this year. F. W. Aston has now succeeded inso improving the original method of accelerated anode rays, thathe has been able to obtain with this series of elements the requisiteintensity in the high-resolution mass spectrograph, and to determinethe isotopic constitution and atomic weights of the remaining tenelements.Considering the whole group of fourteen elements of atomicnumbers 57 to 60 and 62 to 71, the elements of odd atomic numbers,wiz., La (57), Pr (59), Eu (63), Tb (65), Ho (67), Tm (69), Lu (71), allprove to be simple with the exception of europium, which consistsof a nearly equal mixture of the isotopes 151 and 153.Of the evenelements, samarium (62) was found to be the most complex, and toconsist of seven isotopes. Neodymium, gadolinium, and ytterbiumcontain six isotopes each, and erbium and dysprosium four each.The mass-spectrum values for the atomic weights agree closelywith the international values for La, Ce, Pry Eu, Dy, Yb, and Lu,but are smaller than the chemical values in the other seven cases.These discrepancies are discussed in another part of this v01ume.~Since terbium and thulium are simple elements and the twin linesof europium are well defined and easy to measure, the chemicalvalues are almost certainly in error.I n the same paper, Aston calls attention to the case of osmium,for which he obtained the value of 190.31 in 1931, which was dis-tinctly lower than the then-accepted chemical value 190.9, but veryconsiderably lower than the new value of 191.5 adopted by theInternational Committee in their last report.* It will be interestingF.W. Aston, Phil. Mag., 1925, 49, 1191.PTOC. Roy. Soc., 1934, [A], 146, 46; A., 1150.J., 1934, 499.a P. 370.F. W. Aston, Proc. Roy. Soc., 1931, [A], 132, 487WHYTLAW-GRAY : ATOMIC WEIGHTS.95to see whether further work on the chemical side confirms the higherfigure, but up to the present, with the exception of tellurium,revision of the chemical atomic weights has invariably supportedthe results obtained with the mass spectrograph.The redetermination of the atomic weights of niobium andtantalum by 0. Honigschmid is another example of this, and hascleared up the discrepancy noticed in last year’s report.’ Tantalumbromide and niobium chloride were prepared in a high state ofpurity and, after vacuum sublimation, were weighed. They werethen brought into solution, and the silver equivalent determined bystandard procedure. The values obtained were Ta = 180.89, Nb =92.91, in close agreement with Ta = 180.89, Nb = 92.90 found byAstonY8 but very appreciably lower than the International values,181.84 and 93.3 respectively.A detailed account of this work onniobium has been published by 0. Honigschmid and K. Winters-berger.g Other investigations published recently which confirmAston’s values are those on the atomic weight of molybdenum by theanalysis and synthesis of its trioxide by R. Lauti6,lo which gave thevalue Mo = 96.01; and on the densities and atomic weights ofkrypton l1 and xenon by W. Heuse and J. Otto.12 The latter worksupports the value obtained previously for these two gases byH. E. Watson,13 and by R. Whytlaw-Gray, H. S. Patterson, andW. Cawood l4 by microbalance methods.Turning now to work along classical lines, there is not much toreport since last year, but mention should be made of an importantinvestigation by C.R. Johnson15 on the atomic mass of sodium,in which a very careful and critical study was made of the errorsinherent in the measurement of the ratios NaCl : Ag and NaCl : AgClby standard methods. Special attention was paid in this researchto accurate nephelometry and to the determination of the correctend-point both in titration and in the estimations of the washings.A potentiometric method was used in some cases to control theresults of the nephelometer. Among the many points of detailstudied may be mentioned the neutrality of the fused sodiumchloride, adsorption and solubility of silver chloride, the presenceof films of silver chloride or silver on the vessels used, and of invisiblecracks in the Pyrex flasks and beakers which might have retainedNaturwiss., 1934, 22, 463; A., 937.2.anorg. Chem., 1934,219, 161.Ann. Reports, 1933, 30, 1.* J., 1932, 2888; Nature, 1932,130, 130.10 Compt. rend., 1933,197, 1730; A., 127.11 Physikal. Z., 1934, 35, 67; A., 234.l2 Ibid., p. 628; A., 1053.14 Proc. Roy. SOC., 1931, [A], 134, 7; A,, 1932, 106.l5 J . Phyaicd Chem., 1933, 37, 923; A., 1934, 4; -also A., 1933, 773.l3 Nature, 1931, 127, 631 ; A,, 66696 INORGANIC CHEMISTRY.traces of liquids. The result of six closely concordant experimentsgave NaCl : Ag = 0.541817, and that of five gave the gravimetricratio NaCl : AgCl = 0.40779. From each of these, by taking Ag =1074380 and C1= 35.457, the same value Na = 22.994 was obtained :a value slightly lower than the international figure, Na = 22.997.G.P. Baxter and A. H. Hale16 have determined the ratio1205:Na2C0,. This is the first time since the work of T. W.Richards and C. R. Hoover,17 who measured the ratio Na2C03:NaBr: Ag, that the atomic weight of carbon has been linked upwith the silver standard. Richards and Hoover’s ratios wereunfavourable for the calculation of an accurate value for carbon,giving values ranging from 11.997 to 12.008. Baxter and Halesfound, as the mean of nine closely concordant experiments,I,O, : Na2C0, = 3.14950, but unless the atomic weights of iodineand sodium are known to a very high degree of accuracy, that ofcarbon remains uncertain. If, for inatance, we use the internationalvalues Na = 22.997 and I = 126.92, carbon becomes 12.005, butwith Na = 22.994 and the same value for iodine, C = 12-010.Thelatest value for iodine is 126.917, determined by Honigschmid andH. Striebe1,f” which agrees closely with the mass-spectrographvalue of 126-91, but really, at the present time, the accuracy ofexperiment is insufficient to discriminate between these smalldifferences.Evidence of a more direct character supports the new and highvalue for carbon advanced by M. Woodhead and R. Whytlaw-Gray lBin 1933. It will be remembered that these workers obtainedC = 12.011 by a comparison of the limiting densities of carbonmonoxide and oxygen with a microbalance, a value perceptiblygreater than C = 12.007 found for the same ratio by E.Moles andM. T. Salazar 2O using the standard methods of comparing gas densi-ties. The last two workers21 have now published a new series ofmeasurements of the densities of these two gases by an improvedmethod, which gives as the atomic weight of carbon 12.0065. Inthe meantime, F. W. Aston 22 has compared photometrically thesecond-order lines at positions 6 and 6.5 of C12 and C13 and hasfound the isotopic ratio to be C12 : C13 = 140 : 1, whence, allowancebeing made for the packing fraction and change of scale, the atomicweight of carbon = 12*0080. It may be mentioned thatl6 J . Amer. Chem. SOC., 1934, 56, 615; A., 469.l7 Xd., 1915, 87, 95; A., 1915, ii, 96.l8 Z. anorg. Chem., 1932,208,53; 2. physikal. CJwm., Bodenatein Festband,l9 J ., 1933, 846; A., 1933, 894.2o Anal. Fis. Quim., 1929, 27, 267; A., 1931, 407.* l Ibid., 1934 82, 054.1931, 282; A., 1931, 1208.az Nature, 1934, 184, 178; A., 937WHYTLAW-GRAY : ATOMIC WEIGHTS. 97F. A. Jenkins and L. S. Ornstein 23 found by a study of the bands ofthe Swan spectrum the ratio C12 : C13 = 106 : 1, which has beenconfirmed by the work of J. P. Tate, P. T. Smith, and A. L.Vaughan,% who found C13 present to the extent of 1%, which sup-ports the higher value C = 12.011 although it is difficult to assessthe probable error in these figures, There now seems, however,to be no reasonable doubt that the true chemical atomic weight ofcarbon lies close to 12.01 and that the international value of 12.00is too low.Just recently, important new evidence has appearedwhich supports the high value. H. S. Patterson and W. C a ~ o o d , ~ ~using an improved microbalance technique, have compared the limit-ing densities of carbon dioxide, ethylene, nitrous oxide, carbontetrafluoride, and oxygen and repeated part of their earlier work onmethyl fluoride. The final results are : N = 14.007, C = 12.012(from ethylene), C = 12.010 (from CO,), F = 18.995 (from CF,),and P = 18.995 (from CH,F).As a check on the method, the compressibilities of the gases weremeasured by means of an Andrews apparatus, as well as by themicrobalance, and a good agreement obtained. The two valuesfor fluorine agree very closely with F. W. Aston’s value 26 of 18.996,and negative definitely the possibility of the existence of a higherisotope.The value for nitrogen is rather lower than 14.008, theaccepted figure, and also than the recently published values ofE. Moles and J. Sancho 27 from ammonia, N = 14.008, and of Molesand Salazar 21 from nitrogen itself, N = 14.0083. A full accountof the work of Patterson and Cawood will appear shortly, but itmay be stated that it has been carried out with every modernrefinement and attention to detail.The discovery that hydrogen is a mixed element and that the iso-topes are easily separable has cast doubt on the accuracy of one ofthe fundamental atomic-weight ratios, viz., that of 0 : H. Thenumerous determinations made of the atomic weight of hydrogenwere summarised by E. Moles 28 in 1925, and the final mean calculatedfrom the results of five different workers was found to be 1.00777.Inall cases, the hydrogen was prepared by electrolysis of either acid oralkaline solutions, and usually was weighed on palladium. It isprobable then that this hydrogen was deficient in the heavier isotopeand that the mean atomic weight is too low. It seems unlikely thathydrogen of normal isotopic content was ever used in measuring23 Proc. K . Akad. Wetensch. Amsterdam, 1932, 35, 1212; A., 1933, 333.24 Physical Rev., 1933, [ii], 43, 1054.25 Nature, 1935, in the press.26 “ Maas Spectre and Iaotopos,” p. 150; PYOC. Roy. Soc., 1927, [A], 115,487.2 7 Anal. Pis. Quirn., 1934, 82, 931. 2. physikal. Chem., 1925, 115, 61.REP.-VOL. XXXI. 98 INORGANIC CHEMISTRY.this ratio, though the oxygen employed probably contained itsisotopes in the usual proportions.Now the ratio of the masses of 0l6 and H1 has been determinedwith great exactness with the mass spectrograph by F.W. Aston 29and by E. K. Bainbridge 30 and found to be 16.0000 : 1.00778.The mean mass number of chemical oxygen on the physical scalecan be calculated if the proportion of the heavier isotopes is known.This has been determined spectroscopically by a number of differ-ent observers, who obtain values varying from 0 l 8 : Of6 = 1 : 1250 31to 1 : 630,32 the proportion of 017 being negligible. Of these,the second ratio is generally regarded as the most probable, andgives a value of 16.0035 for the atomic weight of chemical oxygenon the physical scale.Hence pure H1 or protium has an atomicweight on the chemical scale of 1.00778 x 16.0000/16.0035 =1.00756. If, now, we know the proportions of the two hydrogenisotopes present normally in the element, we can calculate the atomicweight. Unfortunately, the data on this point are not very con-cordant. The earlier spectroscopic estimates of W. B l e a k n e ~ , ~ ~of H. Kallmann and W. LasareffF4 and of J. T. Tate and P. T.Smith 35 were too low and gave values for D : H of 1 : 30,000 to1 : 40,000. G. N. Lewis 36 and R. T. Macdonald?' from their ex-periments on the electrolytic preparation of heavy water, calculate1 : 6500. W. Bleakney and A. J. Gould 38 have recently redeter-mined the ratio spectroscopically, using hydrogen from rain water,which was decomposed completely by passage over heated iron;they found the value 1 : 5000.They also investigated the isotopicratio in hydrogen from freshly prepared electrolytic cells and foundit t o be about 1 : 25,000. A value distinctly lower than the lastthree has been found by (Mrs.) E. H. Ingold, C. K. Ingold, H.Whitaker, and R. Wh~tlaw-Gray,~~ who, by the action of metalson water and by electrolysis, have prepared specimens of watercontaining a low proportion of deuterium and have comparedtheir densities with that of standard water by using a delicate floatmethod. The value given in a preliminary notice is 1 : 9000.2B op. cit., p. 100. 30 Physical Rev., 1933, 43, 103 ; A., 1933, 203.31 H. D. Babcock, ibid., 1929, 34, 640; A., 1929, 971; 1930, 1232.32 R.Mecke and W. H. J. Childs, 2. Physik, 1931, 68, 362; A., 1931, 543;see also S . M. Naudd, Physical Rev., 1930, 36, 333; H. M. Kallmann and W.Lasamff, 2. Physik, 1933, 80, 237; A., 1933, 333; R. T. Birge and D. H.Menzel, Physical Rev., 1931, [ii], 37, 1669; A., 1933, 204; F. W. Aston,Nature, 1932,130, 21.33 Physical Rev., 1932, 41, 32.35 Physical Rev., 1933, 43, 672.37 J . Ckem. Physics, 1933,1, 341.38 Physicctl Rev., 1933, 44, 265; A., 1933, 994.34 Naturwiss., 1932, 20, 206.36 J . Amer. Chem. SOC., 1933, 55, 1297.3B Nature, 1934,134, 661WARDLAW : METALUC CARBONYL AND NITROSYL COMPOONDS. 99If, now, we calculate the atomic weight of hydrogen on thechemical scale, using the two extreme values for the isotopic ratio,viz., D : H = 1 : 5000 and 1 : 9000, we get respectively 1.00776 and1.00767, both of which are smaller than the mean value found bychemical methods, vix., 1.00777, which should itself be less than thetrue value.H.Muckenthaler,*o in an interesting paper, has fully discussedthis discrepancy, and contends that it is to be explained by the useof an erroneous value for the isotopic ratio in oxygen. He hasredetermined this ratio, using (Sir) J. J. Thompson's parabolamethod, and finds 0 l 8 : 0l6 = 1 : 1058, which lies nearer to theearly value of Babcock than to that of Mecke and Childs. It may benoted, however, that, even if we assume Babcock's value for theoxygen ratio, and that of Bleakney and Gould for the proportionof the hydrogen isotopes, the value for hydrogen is only raised to1.00787, which exceeds the mean chemical value by only 1 part in10,000.It is evident, then, that there is an unexpla'ined discrep-ancy between the chemical and the physical value for the atomicweight of hydrogen, and that further work on the chemical side isdesirable. R. W.-G.2. METALLIC CARBONYL AND NITROSYL COMPOUNDS.A particularly interesting chapter in inorganic chemistry dealswith the metallic carbonyl and nitrosyl compounds. Notableadvances have been made in our knowledge of these compoundsduring recent years through the numerous investigations of W.Hieber and W. Manchot and their collaborators. At the presenttime, amongst the simple carbonyls of the transition elements, fivemonometallic derivatives are known : Cr(CO),, Fe(CO),, Ni(CO),,Mo(CO),, W(CO),.In 1921, I. Langmuir 1 suggested that the volatilecarbonyls represent a type in which the metal atom takes on thenumber of electrons necessary to rcach the atomic number of thenext inert gas, nickel adding eight in Ni(CO),, iron ten in Fe(CO)5,and molybdenum twelve in Mo(CO),. This implies that eachcarbon monoxide molecule contributes two electrons to the centralmetallic atom. It will be noted that this principle applies equallywell to the more recently prepared carbonyls of chromium andtungsten.3 Moreover, it indicates why monometallic carbonyls40 Physikal. Z., 1934, 35, 851; A., 1283.1 Science, 1921, 54, 65; A., 1922, ii, 137.a A. Job and A. Cassal, Compt.rend., 1926,183,392; A., 1926, 1017; Bull.SOC. chirn., 1927, 41, 1041; M. M. Windsor and A. A. Blanchard, J. Amer.Chem. SOC., 1934, 56, 823; A., 589.3 A. Job and J. Rouvillois, Cowapt. Tend., 1928, 187, 664; A., 1928, 1201100 INORGANIC OHEMISTRY.are not formed by elements of odd atomic number such as cobalt(at. no, 27). It is well known that cobalt carbonyls show a greattendency to polymerise, so that the tetracarbonyl is really C O , ( C O ) ~ , ~whilst the tricarbonyl gives a molecular weight in Fe(CO), whichindicates the complexity CO,(CO),,.~Recent work has fully established that the carbon monoxidemolecule can be substituted by other molecules, with the productionof new substances of great theoretical interest. For example, onesuch group may be substituted by a molecule of ammonia, pyridine(py), or alcohol, and two groups by chelate groups such as ethylenedi-amine (en) (I), the diethyl ether of dithioglycol (th) (11), or o-phen-anthroline (phen) (111).In this way such substances are produced asCH2-NH241M H2F/ SEt ‘1M <*, H,C\ f Nh I CH,-NH2fSEt M(1.) (11.) (111.)Cr( CO),py,, Fe( GO),( NH,),, Ni( CO),phen, Fe2( CO),en,. It willbe seen that in all these cases the covalency of the metal is the sameas in the simple carbonyls from which they are derived. A secondseries arises by an alteration in the covalency of the metallic atom.The most remarkable examples of this type are iron carbonyl hydro-gen and cobalt carbonyl hydrogen and their derivatives. Thevolatile Fe(CO),H, 8 is an unstable yellow liquid (m.p. - 70’) formedby the action of alkalis on iron pentacarbonyl : Fe(CO), + 20H’ =Fe(CO),H, + (30,”. The extreme ease of oxidation of the CO-ordinately bound carbonyl group in alkaline medium is a distinctivefeature of the reaction. Methylene-blue is quantitatively reducedby Fe(CO),H,, and the active hydrogen may be determined in thisway, An outstanding property of this carbonyl hydride is theability fo form salts with organic bases such as pyridine and o-phenanthroline, whereby the stable compounds Fe( CO),H,, (C,H,N),and Fe(CO),H,,2C,,H,N2 are produced. The hydrogen in thelatter compound no longer reduces methylene-blue.In appropriate non-aqueous solvents, iron pentacarbonyl reacts4 L. Mond, H.Hirtz, and M. D. Cowap, J., 1910, 97, 798.W. Hieber, F. Muhlbauer, and E. A. Ehmann, Ber., 1932, 65, [B], 1090;A., 1932, 920.6 W. Hieber, Sitzungsber. Heidelberg. Akad. Wiss., 1929, 3, 4.‘7 W. Hieber and F. Sonnekalb, Ber., 1928, 61, 558; A., 1928, 510.W. Hieber and F. Leutert, Naturwiss., 1931, 19, 360; A,, 1931, 810; W.Hieber and H. Vetter, Ber., 1931, 64, [B], 2340; A., 1931, 1255; idem, 2.anorg. Chern., 1933, 212, 145; A., 1933, 686WARDLAW METALLIC CARBONYL AND NITROSYL COMPOUNDS. 101with halogens to give Fe(CO),X, where X = C1, Br, or I ; whilstwith mercuric sulphate in sulphuric acid solution, the pentacarb-onyl yields Fe(C0)4Hg.10 Other members of this series l1 whichare regarded as salts of Fe(C0)4H2 include Fe(C0),Na2, Fe(CO),Cd,[Fe(CO),H],,[Ni(NH,),], for dilute mineral acids liberate Fe(C0)4H2from them. Similar investigations with the other metallic carbonylsgave positive results only in the case of cobalt carbonyls.It hasbeen known for some time that the following reaction is realisable :2Co(CO),M + 2H' = CO" + Co(CO),H + 2CO + $H2 + M(M = alcohol or amine constituent.)Quite recently, however, E. A. Ehmann l2 has stated that if thereaction with CO(CO)~ is carried out in an aqueous solution ofalkali or baryta, a similar result is obtained to that with Fe(CO)5,but the mechanism is more complicated :3Co(CO), + 20H' = 2Co(CO),H + CO," + CO(CO)~ (polymerised).A part of the tetracarbonyl is decomposed simultaneously withliberation of carbon monoxide but without hydrogen evolution :3cO(Co), + 2H20 = 2Co(CO),H + Co(OH), + 4CO.The presence of Co(CO),H was proved by oxidation with hydrogenperoxide, as well as by titration with methylene-blue.Althoughthe isolation of such salts as [Co(CO),],[Ni(NH,),] and[Co(CO),],[Co(phen),] confirms the formula for the parent hydride,this has not yet been obtained. In chemical behaviour, the carb-onyl hydride of cobalt is distinguished from that of iron by itsmuch greater sensitivity ; e.q., spontaneous decomposition withformation of free carbonyl and free hydrogen occurs rapidly in acidsolution at the ordinary temperature. It is significant that, so far,no hydride of nickel carbonyl has been obtained, nor any saltderived from it. Examination of the following series shows thatthe E.A.N." of the metal is in each case that of the next inert gas :Fe ( CO ) 6, Fe ( CO ) ,H2 , Co ( CO ) 4H, Ni (CO ) 4.Lafstly, reference must be made to the interesting series of changesthat occur when the carbonyls react with nitric oxide.From thereaction of this gas and Fe(CO), dissolved in Fe(CO),, a red crystallinecompound Fe(CO),(NO), (m. p. 18.5") is obtained which decomposes9 W. Hieber and G. Bader, Ber., 1928, 61, [B], 1717; A,, 1928, 1202.10 H. HockandH. St;uhlmann, Ber., 1929,62, [B], 431,2690; A., 1929,412;11 F. Feigl and P. Krumholz, 2. anorg. Chem., 1933,215,242; A., 1934,159.12 See W. Hieber, 2. Ek&r&m., 1934,40, 168; A,, 611.* E.A.N. = effective atomic number, i.e., total number of electrons, sharedand unshared (see N.V. Sidgwick, " Electronic Theory of Valency," 1927,p. 163).1930, 47102 INORUmC CHEMISTRY.at 70°.13 Under appropriate conditions, this nitrosocarbonylreacts with iodine, pyridine, or o-phenanthroline to give Fe(NO),I,Fe2(NO)4(py)3, and Fe(N0);phen. l4 When iron pentacarbonyl isheated with nitric oxide under pressure, black crystals of irontetranitrosyl, Fe(N0)4, are obtained.14 Similar reactions are re-corded with cobalt carbonyls. Slowly a t room temperature,instantaneously at 40", nitric oxide reacts with Co(CO), to forma cherry-red liquid (b. p. 78.6") having the molecular formula,Co(CO),NO. With pyridine, this cobalt nitrosocarbonyl yieldsCo,(NO),CO,Bpy, and with o-phenanthroline, in benzene or methylalcohol, the complex substance CONO(CO),,C~,H,N,.~~ Otherresults of great interest have been omitted from this account,but sufficient has been recorded to indicate the substantial achieve-ments in this field of research.The constitution of the metallic carbonyl and nitrosyl compoundsis a fascinating but difficult problem.It cannot be said that a finaldecision has yet been reached, but an important paper by N. V.Sidgwick and R. W. Bailey l5 places the structures of these com-pounds on a much firmer basis. The experimental work alreadyreviewed shows that, in the carbonyls, a carbon monoxide groupoccupies one co-ordination position and thereby provides twoelectrons to form a link with the metallic atom. The carbonmonoxide molecule is obviously capable of co-ordinating througheither the carbon or the oxygen atom, but Sidgwick and Baileyconsider that it is more probable that the link is formed throughthe carbon, thus M+CO, since the 4-covalent condition is the normalstate of the carbon atom whereas it only rarely occurs with oxygen[e.g., in basic beryllium acetate, Be,0(CH3*CO-O),].They furtherstate that carbon monoxide is correctly formulated as :C;;zOz or%6. This formula, originally proposed by I. Langmuir,16 hasbeen shown by D. L. Hammick, R. G . A. New, N. V. Sidgwick, andL. E. Sutton l7 to be supported by the electrical dipole moment, theinteratomic distance, the heat of formation, the force constant, andthe parachor. N. V. Sidgwick has emphasised that the minutemoment (0-12) of carbon monoxide can be explained only bysupposing that the very unequal sharing of electrons between thecarbon and oxygen is offset by the transference of an electron froml8 W.Hieber 8nd J. S. Anderson, 2. anorg. Chm., 1932,208,238; A., 1932,1219.l4 Idem, ibid., 1933, 211, 132; A., 1933, 476.l6 Proc. Roy. SOC., 1934, [A], 144, 621; A., 833.l6 J . Amr. Chm. SOC., 1919, 41, 1543; A , , 1919, ii, 506.J., 1930, 1876; A., 1930, 1239; 880 also N. V. Sidgwick, Chem. Reviews,See N. V. Sidgwick, " The Covalent Link," 1933, p. 187.1931, 9, 77WAEDLAW : METALLIC CARBONYL AND NITROSYL COMPOUNDS. 103the oxygen to the carbon. The observation of L. E. Sutton andJ. B. Bentley,19 that the electrical dipole moment of nickel carbonylis zero, strongly supports the idea of a triple bond between thecarbon and oxygen, whilst the Raman spectrum of nickel carbonylconfirms the presence in it of a triple link of carbon to oxygen.Nevertheless, objections to this formulation have been raised recentlyby R.Samuel and his collaborators.21 They conclude that theevidence from molecular spectra and considerations of wave-mechanics favour the classical formula (3x0 ; their arguments,however, are based on an interpretation of molecular spectra whichdisagrees with those of W. Heitler and F. London, of L. Pauling,and of J. E. Lennard-Jones. Much more important arguments arebrought forward by L. Pauling,22 who concludes that carbonmonoxide is in resonance between the two forms E O and CEO,with the former predominating.The resonance would explain whythe length and heat of rupture of the link are those required by thetriple link; it would also justify the use of the triple link as thenormal structure. There is, therefore, a very strong case for theview that the triple-bonded structure predominates in carbonmonoxide and that the monometallic carbonyls are correctlywritten as M+-CS6 or M-C&. Sidgwick and Bailey havepointed out a curious regularity in the composition of the carbonyls,whether they contain one or more metallic atoms in the molecule.If we calculate the E.A.N. of the metallic atoms by adding 2 to theatomic number of the metal for each carbonyl group, then thedifference between the result and the atomic number of the nextinert gas is always one less than the number of metallic atoms inthe molecule. For a molecule M,(CO),, the equation G - :(xm + 29)= x - 1 is always true, where m is the atomic number of M and Gthat of the next inert gas.When x = 1 (monometallic), this, aswe have seen, holds in every case. It also holds with all the poly-metallic carbonyls known : Fe,(CO),, Fe,(CO),,, Co,(CO),, andCo,(CO),,. In these, the molecule must be held together by furtherco-ordination, obviously through the oxygen as donor, M-C=O-M,every such link involving the sharing of two more electrons. Ifwe may assume that, as in the simpIe carbonyls, this leads to eachmetallic atom in the polymetallic carbonyls having the E.A.N. ofthe inert gas, it follows from the equation that the number of suchID Nature, 1932, 130, 314; A., 1932, 1077; L.E. Sutton, R. G. A. New,and J. B. Bentley, J., 1933, 652; A., 1933, 765.2o J. S. Anderson, Nature, 1932, 180, 1002; A., 1933, 113.21 H. Leasheim and R. Samuel, Proc. Physicat Soc., 1934,46,523; A., 945;R. I?. Hunter andR. Samuel, J., 1934,1180; A., 1058; Nature, 1934,154,971.22 J . AM. Cbm. Soc., 1932,54,988; A., 1932, 561.+-104 INORaANIC CHEMISTRY.new links through the oxygen must be 1 when 2 = 2, 3 whenx = 3, and 6 when x = 4. This implies that the molecules inquestion are respectively linear, triangular, and tetrahedral, asshown in Figs. 1, 2, and 3. The crystal structures of Fe,(CO),FIU. 1 .-Fe,( CO ),. FIU. 2.-Fe,(CO),2.0c\and Fe,(CO)12 have been examined by R. and are compatiblewith the configurations proposed.The fact that these two substancesare diamagnetic, like Fe(CO), and Ni(CO),, is held to support theseassumptions.In their preparation and properties, the nitrosyl compounds23 2. Krbt., 1927,65, 85; 1931,77, 36; A., 1928, 108; 1931, 671WARDLAW : METALLIC CARBONYL AND NITROSYL COMPOUNDS. 105show a close analogy to the carbonyls. They are commonly formedby the same metals and may be prepared by the direct action ofnitric oxide, like the carbonyls by that of carbon monoxide. It hasbeen suggested that the nitrosyls are derivatives of hyponitrousacid OH*N:N*OH, but there is substantial evidence against such anidea. Nitrosyl compounds are formed from nitric oxide and notfrom hyponitrous acid (except under conditions where the latterforms nitric oxide).Again, hyponitrites in the presence of acidsyield nitrous oxide, whereas nitrosyls readily liberate nitric oxidewith acids, just as the carbonyls liberate carbon monoxide. More-over, those compounds containing one NO group to one metallicatom, such as the nitroprussides, would require a formula doublethe accepted one. There is no evidence in favour of this and muchagainst it. It can scarcely be doubted that each NO group isseparately attached to the metal. The great similarity betweenthe nitrosyls and the carbonyls suggests that their structures arevery similar and that there is a triple bond between the nitrogenand oxygen, giving M-NEO, corresponding to M-CEO.The structure of nitric oxide must, however, differ from that ofcarbon monoxide, for nitrogen has one more electron than carbon.Now it is known that the [NO]' ion can exist in polar compoundssuch as [N0]C104,24 [NO]S0,H,24 and [N0]BF4,25 and N.V.Sidgwick and R. W. Bailey assume that the NO group, like thecarbon monoxide molecule, is attached to the metal by a link oftwo shared electrons, but that, in addition, it transfers a furtherelectron and thereby counts as three towards the E.A.N. of themetal. The suggestion that NO contributes three electrons tothe E.A.N. of the metallic atom has been made already by otherobservers but the detailed mechanism of the linkage has not prev-iously been discussed. The structures proposed are [:N; ;;03+,f - - +- + N-0, M-N-0. Support for this theory is forthcoming in thesimple cases of the carbonyl and nitrosyl compounds of copper.The former 26 are all of the type Cu(CO)Br, where the copper iscuprous, but their nitrosyl 27 analogues are derived fromcupric copper,e.g., Cu(NO)Cl, and Cu(NO)SO,.The extra electron contributed bythe NO group to the copper satisfactorily explains this distinction.Although there is no ferric analogue of the complex compound24 A. Hantzsch and K. Berger, 2. anorg. Clbem., 1930, 190, 321 ; A., 1930,z 5 E. Wilke-Dorfurt and G. Balz, ibicZ., 1927, 159, 197; A., 1927, 120.26 0. H. Wagner, i b d . , 1931, 196, 364; -4., 1931, 581.27 W. Menchot, Annalen, 1910, 375, 308; A., 1910, ii, 956; Ber., 1914, 47,1601 ; A., 1914, ii, 567; W. Manchot and E. Linckh, ibid., 1926,59, 407; A.,1926, 462.1007.D 106 INORQANIC CHEMISTRY.Ms[Fe(CN)5C0], it is notable that ferric derivatives of this type,when they do occur, have the corresponding more stable ferrousanalogues.It is surprising, therefore, that if iron be consideredas ferric in the nitroprussides M,[Fe(CN),NO] no ferrous equivalentis available. However, the theory of the NO linkage advancedby Sidgwick and Bailey removes this anomaly, for on their inter-pretation the iron is ferrous in the nitroprussides. Their explana-tion of the colour reactions of the nitroprussides is particularlyinteresting. It is well known that reaction does not occur withhydrogen sulphide, but addition of alkali or alkaline sulphide to thesolution produces the deep purple colour of M,[Fe(CN),(NOS)] .28Again, hydroxyl ions convert nitroprussides into nitro-compounds,M,[Fe( CN),(N0,)].29 It is considered that the strongly positive+-+ N=O group attracts the negative OH’ or S” (or HS’) ions with theproduction of a nitro- or thionitro-group :0 .SN O N O+-+ Fe-N=O -+ Fe-Nf or Fe-N”The method of formulation adopted by Sidgwick and Bailey fornitrosy1 compounds leads to the maintenance of the inert-gas rulein the majority of cases, even in such complicated molecules as thoseof R~ussin’s ‘f red and black salts,” KFe,(NO),S, and &Fe,(NO),S,.Closely allied to the compounds just discussed are nickel deriva-tives of the type NiSR,NO and the related iron and cobalt compoundsMSR,2NO. Special importance is attached to these substances,for W.Manchot and his collaborators, who systematically studiedthem, consider that the metal atoms exhibit univalency. Beforeaccepting this conclusion, however, it is profitable to examine it indetail. It is obviously arrived a t by considering merely the so-calledprincipal valency of the atom in the sense of Werner’s theory, andneglecting the influence of the NO group. This naturally raises theimportant question as to what is meant by the absolute valencyof an atom in compounds of this type. A definition which willbe universally applicable is not easy to formulate, but on thewhole, the best seems to be that of H. G. Grimm and A. Sommerfeld,mwho consider that the absolute valency of an atom is numericallyequal to the number of electrons of the atom “engaged” inattaching the other atoms.Now, according to Sidgwick and28 J. F. Virgili, 2. anal. Chem., 1906, 45, 409; A., 1906, i, 637.an L. Cambi and L. Szego, Atti R. Accad. Lincei, 1927, 5, 737; A., 1927,30 H. (3. Grimm end A. Sommerfeld, 2. P h y d , 1926,38,36; A., 1926,660.917; (;rclzzettcc, 1928, 58, 71; A., 1928, 345.See also N. V. Sidgwick, ‘‘ The Electronic Theory of Valency,” p. 182WARDLAW: METALLIC CARBONYL AND NITROSYL COBI-POUNDS. 107Bailey’s theory, NiSR,NO should have the structure bd-Si-SR,so that the nickel atom has as many unshared electrons as theisolated nickel atom and hence the valency of the nickel is zero.This is one of the absurdities to which the system of Grimm andSommerfeld occasionally leads, but it does emphasise the difficultyof assigning to the metal atom in these compounds a numericalvalue of absolute valency.There is, of course, no doubt about theexistence of compounds of univalent nickel, as the isolation ofK,Ni(CN), has shown, but the series under consideration are onlycorrectly designated derivatives of univalent iron, cobalt , and nickelin the restricted sense of Werner’s principal valency. In theappended table, the nitrosyl compounds under discussion aresummarised. 31Type MSR,2NO. Type MSR,NO.FeSK, 2N0 FeS,03K,2N0 NiSE t,NOFe4S,K,7NO [F~(S,O~),IK,S~NO NiSPh,NOFeSEt,2NO CoSE t, 2N0 NiSPh,NO,pyFeSPh,2NO CoSPh,2NO “i(SaO,)aIKa,NO[Co(S,O,),IK,,2NOThe mechanism of the reduction to the univalent stage in thecase of the simple mercaptides is very interesting. The nickelmercaptide Ni(SEt),, for example, forms the unstable intermediateproduct Ni(SEt),,2NO, which loses nitrosyl mercaptide, NOmSEt,in accordance with the equations(1) Ni(SEt), + 2N0 = Ni(SEt),,BNO(2) Ni(SEt),,2NO = NiSEt,NO + NOSEtDepending on experimental conditions, the NOSEt undergoes moreor less rapid decomposition into disulphide and nitric oxide :(3) 2NOSEt = 2N0 + EtSmSEt.By similar reactions, ferrous and cobaltous mercaptides yield thecomplex compounds FeSEt ,2NO and CoSEt ,2N0.32 Thesederivatives are covalent, for they dissolve in organic solvents andhave low melting points (polar compounds would have appreciablyhigher melting points).Their intense colours are attributed tothe deformation of the nitric oxide molecule on the basis of Fajans’stheory.The complex thiosulphates are obtained when metallic salts,e.g., nickel chloride or cobaltous acetate, react with nitric oxide in31 See H.Schmid, Angew. Cbm., 1933, 46, 691; A., 14, for detailedreferences.34 W. Manchot and 5. Davidson, Ber., 1929, 62, 684; W. Manchot and H.Gall, Bw., 1927, 60, 2318; 1028, 61, 2393; A., 1928, 35; W. Mrtnchot and F.Kaem, &id., 1927, 60, 2175; A., 1927, 1157108 INORUANIC CHEMISTRY.presence of excess of sodium thiosulphate. The nickel salt,K,[Ni(S20,)2NO],2H,0, is fairly stable, but potassium cyanide 33converts it into the less stable complex cyanide, K,[Ni(CN),NO],which shows the characteristic reactions associated with univalencyof the nickel, vix., pronounced reducing properties and liberationof hydrogen from water.This complex cyanide may also be pre-pared by (a) the reaction between nitric oxide and K,[Ni(CN),] 33or (b) the replacement of carbon monoxide in the complex compoundK,[Ni(CN),CO] 34 by nitric 0xide.~5 It is possible to convert theionised compound K,[Ni(CN),NO] into the covalent Ni(CO),.This is done by addition of acid, which gives the very readily oxidisednickel monocyanide NiCN, which in turn absorbs carbon monoxideto form the unstable intermediate product (NiCN,C0),,34 whichthen decomposes into Ni(CN), and Ni(CO),. W. Manchot andH. Gall36 have suggested an ingenious process for obtaining nickelcarbonyl, based on the above considerations. The process consistsin having a carrier, barium sulphate, associated with a nickel saltwhich is first transformed into a hydrosulphide by hydrogen sulphide.By the action of carbon monoxide on this nickel hydrosulphide, anunstable intermediate compound of univalent nickel is formed whichproduces Ni(CO), in good yield :2Ni(SH), + 2xCO = SNiSH(CO), + H2S22NiSH(CO), = Ni(SH), + Ni(CO), + (2% - 4)CO3.MOLECULAR STRUCTURES.The publication of A. Stock’s Cornell lectures on the hydridesof boron and silicon again focuses attention on a remarkable seriesof investigations which has clarified the chemistry of this verydifficult subject. Included in the volume is a summary of thecontents of 69 papers by the author and his co-workers and 15 byother investigators.An important chapter deals with the structuresof the boron hydrides. These compounds were formerly preparedby the reaction of hydrochloric acid with the so-called “ magnesiumboride ” made from magnesium and boron trioxide. In 1930 B. D.Steele and J. E. Mills 2 found, in their experiments with aluminiumand cerium borides, that the yield of hydrides was better if phos-phoric wits used instead of hydrochloric acid; and A. Stock hasw. w.33 W. Manchot, Ber., 1926, 59, 2445; A., 1927, 33.34 W. Manchot and H. Gall, {bid., p. 1060; A., 1926, 698.35 Schuler, Dim., T. M. Munchen, 1928.3G Bev., 1929, 02, 678; A., 1929, 526; D.R.-P., 577,144.The George Fisher Baker Non-resident Lecturership in Chemistry atJ., 1930, 74; A., 1930, 437.Cornell University, Cornell University Press, 1933.o p .C i t . , p. 43WARDLAW : MOLECULAR STRUUTURES. 109since found that from '' magnesium boride," too, a distinctly betteryield (11 yo instead of 4-5%) is obtained with 8N-phosphoric acid.Silicon hydrides are generally present in the crude gas, but ifberyllium b ~ r i d e , ~ prepared from boron trioxide and silicon-freeberyllium, is used instead of " magnesium boride," this contamin-ation can be avoided. The boron hydrides, belong to two series :B5Hl1, and perhaps B6H12. This is no mere formal classification.The boranes, B,H,+4, are more stable and have relatively highmelting points ; the lower members form stable salts with ammonia.The hydroboranes, B,H12+6, on the other hand, dissociate much morerapidly and melt at considerably lower temperatures than theboranes; their ammonia compounds dissociate even at room tem-perature.The hydrides separated from the crude condensateinclude B,Hl0 (b. p. 1 8 O ) , B6H, and B6H10 (liquids), and BloHl,(m. p. ca. 100"). Other hydrides, B,H,, B,Hg (liquid), and perhapsB6H1,, are obtained by heating B,HIo. No hydride whose moleculecontains one or three atoms of boron is at present known. Themost important hydride B2H6 cannot be prepared directly from aboride and acid, for it is decomposed by water : B2H6 + 6H20 =2H,BO, + 6H2. Recently, however, A. B. Burg and H. I. Schlesin-ger 6 have discovered a new method of preparing it. They subjecta current of hydrogen mixed with gaseous boron trichloride, atreduced pressure, to a high-tension electric discharge.In additionto hydrogen chloride, boron, solid hydrides of boron, and someB,H6, much B,H,C1 is produced, which, on standing, dissociatesrapidly into B2H6 and BCl,. are ofopinion that better results are obtained by using the tribromideinstead of the trichloride. By the reaction of B2H6 and methylalcohol,7 the interesting substance dimethoxyborine (m. p. - 130-6')is produced, 4MeOH + B,H6 = 2BH(OMe), + 4H,, accompaniedby an unstable by-product, possibly a polymeride of BH,OMe.Water rapidly decomposes dimethoxyborine : BH( OMe), +3H20 = B(OH), + 2MeOH + H,.The structure of B2H, is a perpetual puzzle, and in spite of thegreat ingenuity displayed by numerous writers, it appears that nocompletely satisfactory solution has yet been reached.The mostimportant reagent for diagnosing the structure of the boron hydrides(l) B?ZH7&+4 ' B2H6, B,Hg, B6H10, BloH14; (2) B7&H??+6 ' B4H10,A. Stock and W. Siitterlino p . cit., p. 48.5 J . Amer. Chern. Soc., 1931,!53,4321; A., 1932,350; A. Stock, H. Martini,6 Ibid., p. 407 ; A., 497.7 A. B. Burg and H. I. Schlesinger, J. Amer. Chm. Soc., 1933, 55, 4009,and W. Sutterlin, Ber., 1934, 67, [B], 396 ; A., 497.4020; A,, 1933, 1257I10 INORGANIC CHEMISTRY.is ammonia. This forms a, series of salts such as B2H,,2NH3,8B ~ H ~ O , ~ N H ~ , ’ B5H9,4NH3,10 and BloH,4,6NH,,10 and their exist-ence is interpreted as indicating the presence of a proportion ofatoms of acidic hydrogen in the hydrides.Substitution productsmay be formed by electrolysis of the boron hydrides l1 in liquidammonia. When thus electrolysed, B2H6 has a conductivity whichis considered to be due to the diammine B2H,,2NH3 acting as the,salt [B,H,](NH,),. By a secondary reaction, the ions [B2H4]” and[NH,]’ yield hydrogen and B,H5,NH2, which st,ill can form a salt(NH,),[B,H,*NH,]. Electrolysis of this substance gives hydrogenand B2H4(NH2),. The unsaturated character of the boron hydridesis indicated by the observation that two atoms of sodium may beadded to B,H, 8 and B ~ H I , . ~ The above results are explained byassigning to B2H, and its derivatives the formulae : l2[$3---Byg] YH 2Na+ r NH, Hp=~<r2] 2NH,+I 1-Higher homologues of this series may be similarly formulated.Insupport of these formuh it is mentioned 13 that the ultra-violetabsmption spectrum of B2H6 resembles that of ethylene rather thanthat of ethane, and that the absorption of B,Hlo is similar to thatof the conjugated hydrocarbon butadiene, CK,:CH*CH:CH,. Whencertain ammonia addition products such as B2H6,2NH3 are heatedto about 200” for several hours, the exceptionally stable compoundB3N3H6 l4 is obtained. Its similarity to benzene in certain physicalproperties has suggested that its structure is probably representedby the inset formula : the double and single bonds can be inter-changed exactly as in Kekuk’s formula for +&~H-GH benzene.In 1856 J. Nessler proposed the use ofan alkaline solution of mercuric iodide andpotassium iodide as a reagent for the direct determination ofammonia, and many studies have since been made of the brown com-pound obtained in this reaction.Some recent observations of M. L.Nichols and C. 0. Willits l5 indicate that it has the composition* A. StockandE. Pohland, Ber., 1926,459, [B], 2210; A., 1926, 1317.A. Stock, E. Wiberg,and H. Martini, Ber., 1930, 63, 2927; A., 1931, 50.lo A. Stock and E. Pohland, Ber., 1929, 02, 90; A., 1929, 279.l1 Op. cit., Chap. 21.l2 E. Wiberg, 2. anorg. Chern., 1928, 173, 199; A., 1928, 036.l3 Op. cit., Chap. 26.l5 J. Amer. Chem. Soc., 1934,50, 769; A., 614.‘EH‘BH=b€’l4 Op. cit., Chap. 14WARDLAW : MOLECULAR S!t%UCTDRES. 111represented by the empirical formula, NH,*Hg,I,.It is very in-soluble and tends to separate in very minute particles which arenegatively charged and give colloidal solutions. Owing to theagglomeration of the particles, the yellow colour changes to redwhen ammonia solutions of higher concentration are utilised. Thismay be prevented, and the colour made permanent over awide range of ammonia concentrations, by adding a protectivecolloid, e.g., the addition of 1 C.C. of an 0.5% alkaline ash-freegelatin solution containing 1% of perhydrol to 50 C.C. of Nesslersolution.In connexion with the structure of hydrazoic acid and the azides,some observations by E. C. Franklin 16 are of interest. From hisexperimental results, he concludes that most of the reactions ofhydrazoic acid support the idea that it is an ammononitric acid,H - N = m .This linear structure, originally proposed byThiele, is nowadays more correctly written as H-N=lu-N.The action of hydrazoic acid on the metals bears a striking resem-blance to that of nitric acid. Cont'rary to the statements of previousinvestigators, which are reproduced in most text-books, Franklinfinds that no hydrogen is evolved when the acid is treated with zinc,iron, manganese, nickel, or copper. The products are metallicazides, nitpogen, ammonia, and small amounts of hydrazine. Withmagnesium, however, a small amount of hydrogen may be detected,recalling the fact that this metal yields some hydrogen with verydilute nitric acid. Ferrous azide is converted into ferric azide whenheated with excess of hydrazoic acid.Potassium azide may beproduced by heating a solution of potassium nitrate and potassamidein liquid ammonia : KONO, + SKNH, = KN=N=-N + 3KOH +NH,.Whilst the chemical evidence indicates a linear structure for theacid and its salts, the results from crystal structures l7 show clearlythat this applies also to the ion [NfiNfN]. Again, N. V.Sidgwick,lB in discussing the structure of organic azides, concludesthat an organic azide is a mixture of two open-chain forms (a)and (b) in resonance : (a) R-N=NtN, ( b ) R-NtN-N. Afascinating example of the linearity of the arrarigement of the nitro-gen atoms in the azide group was recently disclosed by an X-rayexamination of cyanuric triazide, C,N,(N,),, by (Miss) I. E. Knaggs.(Sir) W.H. Bragg19 has pointed out that the arrangement bearsl6 J . Amer. Chern. SOC., 1934,56, 668; A., 477.17 S. B. Hendricks and L. Prtuling, aid., 1925, 47, 2904; A., 1926,18 Trans. FaracEay SOC., 1934,30, 801.l9 Nature, 1934,134, 138; A., 948.113112 INORGANIC CHENISTRY.a resemblance to the arms of the Isle of Man, a row of three nitrogenatoms lying in the position of each leg from knee to ankle.N 7A number of substances are described in the literature as cadmouscompounds, e.g., a sub-halide Cd,Cl, which is supposed to be formedby fusing anhydrous cadmic chloride with metallic cadmium innitrogen. (Miss) W. R. A. Hollens and J. F. Spencer 2o find thatthis is really a mixture, for the observed mass susceptibility equalsthat calculated for the mixture Cd + 7CdC1,.They have alsoproved that the so-called cadmous hydroxide and oxide preparedfrom Cd4C1, are mixtures of cadmium and the corresponding cadmiccompound. The relationship between the colour and crystallhestructure of precipitated cadmium sulphide has been examinedby W. 0. Milligan.21 He infers from his X-ray measurements thatboth the cubic p-CdS and the hexagonal a-CdS may each be yellowor red depending on the conditions of precipitation, and attributesthese colour differences to variations in particle size and nature ofsurface. From the cadmium halides, the a-CdS is the main product,whilst from the sulphate and the nitrate (in hot, acid solution) the$-CdS is obtained.Some important observations on the so-called calcium sulphatehemihydrate, which is generally recognised as the active principleof plaster of Paris, have been recorded by W.A. CasparLZ2 Fromsolutions of calcium sulphate in hydrochloric, sulphuric, or nitricacid under proper conditions as to dilution of the solvent and tem-perature, crystals of “ hemihydrate ” 0.5-1 mm. thick and 3 4 mm.long can be obtained; these belong to the trigonal system, with adensity not far below that of anhydrite. In its air-dry condition,the crystal usually contains not more than 4.0--4*5% of water,corresponding rather to 3CaSO,,H,O than to 2CaS0,,H20. Themoisture content of the “hemihydrate” has been shown byprevious investigators, working upon less well-defined materials,to be held in the same way as that of zeolites.Experiments withthe trigonal crystals confirm this, for they may be made to giveao J., 1934, 1062; A., 978.22 Nature, 1934,133, 648 ; A,, 720.21 J. Physical Chem., 1934, 38, 797WARDLAW : MOLECULAR STRUCTURES. 113up water to within 0.1% or less of complete dehydration withoutloss of form or transparency. Exposure to moist air causes theoriginal degree of hydration to be gradually regained. “Deadburning ” converts these trigonal crystals into pseudomorphsconsisting of ordinary anhydrite. Caspari concludes that anhydriteis apparently dimorphous, there being an orthorhombic, com-pasatively inert modification, and a trigonal form, stable only upto ca. 200”, which can take up water zeolitically. It is the behaviourof the latter form in contact with water that causes plaster to set.The suggestion is made that there may be no essential differencebetween the “ soluble anhydrite ” and the ‘‘ hemihydrate ”mentioned in the literature of calcium sulphate.The diamond has been studied for a longer period than anyother natural stone, and its unique character has always beenassumed.It has been left to (Sir) R. Robertson, J. J. Fox, andA. E. Martin 23 to make the fascinating discovery that there are twotypes of diamond which show striking differences in a number ofphysical properties, while in other properties no differences whatevercan be observed. They find that diamonds showing a laminarstructure (I) differ in certain properties from ordinary diamonds(11); the latter have an infra-red absorption band at 8 p and areopaque t o ultra-violet light of less than 3000 8., whilst (I) have noband at 8 p and are transparent up to 2250 8.With (11) the con-ductivity induced by light is very small, high voltages having to beapplied before a current can be detected, whilst (I) give an appreciablecurrent without an applied voltage. Also, (I) are activated bylight of 2300 8., and afterwards give a current in the dark and alarge current when re-illuminated with light of more than 5000 8.These activated diamonds are deactivated by light of 2400-5000 8.X-Ray examination indicates that (I) have a mosaic structure;they are also more optically isotropic than (11), but the specificgravity, refractive index, dielectric constant, and Raman effectare the same for both types. It is considered that the differencebetween (I) and (11) is not due to impurities, but to different con-ditions resulting during their formation from the plastic state.A useful addition to the chemistry of zirconium has been madeby M.P i ~ o n , ~ ~ who has prepared three definite sulphides, ZrS,,Zr3S,, and Zr2S,, of which the last two are new. The method ofpreparation was to act on zirconium oxide at a high temperaturewith hydrogen sulphide. By heating first at 1100-1200” andthen raising the temperature to 1700”, a fused crystalline mass of23 Phil. Trans., 1934, [A], 232, 463; A., 583.24 Cmpt. rend., 1933, 196, 2003; 197, 151; A., 1933, 918; Bull. SOC. chim.,1933, [iv], 58, 1269; A,, 1934, 266114 INORGANIC CHEMISTRY.Zr3SG was obtained.On heating this at 900-1300" in hydrogensulphide, the black ZrS, was produced. Brown Zr,S3 was formedwhen Zr,S, was heated at 1400" for 2 hours in a cathode-ray vacuumor at 1700" for one hour in hydrogen. All the products were crystal-line. Evidence of the existence of Zr,S, was also obtained. Anexamination of the chemical properties of these substances indicatedthat the action of numerous reagents was less pronounced with thecompounds containing less sulphur.The main product of the action of gaseous fluorine on sulphur isthe hexafluoride, SF,, which is a highly stable gas. In addition,two other fluorides S2F2 and SF, are described in the literature,although N. V. Sidgwick 25 has directed attention to the possibilitythat SF, may not exist.In a recent communication by K. G.Denbigh and R. Whytlaw-Gray,26 the preparation and propertiesof a new fluoride, disulphur decafluoride, S,F,,, are described. Thisvery interesting substance was obtained by the fractionation ofa large quantity of the hexafluoride. Only small quantities are pro-duced in the reaction of fluorine on sulphur, but the yield is improvedby using plastic instead of rhombic sulphur. This new fluoride isstable, but less so than the hexafluoride. Its b. p. is 29", m. p. -92",and liquid density 2-08 g./c.c. The parachor is in fair agreementwith a sexacovalent structure which, moreover, appears mostprobable on chemical grounds. The two sulphur atoms are linkedtogether by a single bond, and each sulphur carries five fluorineatoms also linked by single bonds.Valuable information continues to accumulate about the chloridesof sulphur.Por a number of years, T. M. Lowry and his collaboratorshave been investigating the mechanism of the formation of thesecompounds. In 1927 27 they discovered that sulphur and chlorine,after being heated in sealed tubes at loo", gave a product with afreezing-point curve which showed, in addition to the familiarmaxima due to the mono- and tetra-chlorides, two well-definedbreaks which they attributed t o crystallisation of the dichloride anda new chloride S,Cl,. Later work demonstrated that the tetra-chloride could exist only in the solid state. The fact that thedielectric constant of the solid tetrachloride was much higher thanwould be expected for anything but a salt led Lowry and G.Jessop 28to suggest that it is a polar compound; they therefore assignedto the tetrachloride the structure [$C13fil, and to the new chlorideS,C1, the configuration c l e s ~ ~ ~ l or [S',Cl3]fi. The so-calledc1.s c125 Ann. Reports, 1933, 30, 126.2 7 T. M. Lowry, L. P. McHatton, and G. 0. Jones, J., 1927,746; A., 1927,505.28 J . , 1929, 1421; A., 1929, 978; J., 1930, 782; A., 1930, 666.a6 J., 1934, 1347WARDLAW : MOLECUL-4R STRUCTURES. 115sulphur monochloride, S,CI,, still presented an interesting structura1problem, and recently A. H. S ~ o n g , ~ ~ in Lowry's laboratory, re-viewed the relevant physical data available (e.g., Raman spectrum,parachor, and dielectric properties) and concluded that ordinarysulphur monochloride is probably a mixture of the two forms (1)"--S<Cl and (2) Cl*S*S*CI.He has also investigated the reactionbetween sulphur monochloride and chlorine, which is generallyrepresented by the simple equation S,C1, + C1, = 2SC1,. It is,however, recognised that the reaction is definitely more complicatedthan this, and A. H. W. Aten 30 had suggested that it is an auto-catalytic reaction, both the di- and the tetra-chloride being activein catalysing the process. A. H. Spong 31 now finds that no effectascribable to sulphur tetrachloride can be observed, but that thereaction velocity is markedly influenced by the concentration ofS,C14. He considers that the primary mechanism is probablyionic, the chlorine ion attacking the negatively charged sulphuratom in the modification of the monochloride with formula (1).The mechanism he proposes is :+ c1(a) S,C12 + Cl = SCI, + SCl(b) $&C13 f- sc1 = SCI,-s3CIzV.Zappi and V. Cortelezzi 32 examine the experimental data whichare held to favour polar structures for certain halogen derivativesof the nonmetals. The evidence for this possibility in the case ofphosphorus pentachloride was very slight ; according to A. Voigtand W. Biltz,= the compound is entirely non-conducting, andJ. H. Simons and G. Jessop support this c0nclusion.~4 Moreover,these investigators point out that a polar formula is definitelyruled out by their observation that, in carbon tetrachloride, phos-phorus pentachloride has zero or a very small dipole moment.The support for a polar structure is the experimental data of G.W. F.Holroyd, H. Chadwick, and J. E. H. MitchellY35 who found that thepentachloride had a small conductivity in nitrobenzene but none inbenzene and ethylene dibromide. Now Zappi and Cortelezziconclude from their experiments that such measurements in nitro-benzene are subject to large errors and are unreliable. They find,however, that solutions of phenyl dichloroiodide in carefully purified28 J., 1934, 485; A,, 605.31 J., 1934, 1283.33 2. anorg. Chem., 1924,133, 297; A., 1924, ii, 552.34 J . Amer. Chem. SOC., 1931, 53, 1263; A., 1931, 669.35 J., 1925, 127, 2492; A., 1926, 15.30 2. phy8ikal.Chem., 1905, 54, 55.33 Bull. SOC. chim., 1934, [v], 1, 509116 INORGANIC CHEMISTRY.nitrobenzene and in phosphorus oxychloride show a very feebleelectrical conductivity, whilst the cryoscopic molecular weight islow. They contend that this does not indicate that the structure is[PhICI'JCl but that the experimental results may be due to dissoci-ation in accordance with the equation : PhICl, 2 PhI + Cl,.They also examine critically similar experimental data given byother compounds such as iodine trichloride, and conclude that thefeeble conductivities which have been recorded may be morerationally explained as due to the dissociation of complexes formedwith the solvent. Cryoscopic dissociation they also regard as mole-cular and not ionic.The conclusion that phosphorus pentachlorideand iodine trichloride are covalent compounds will be generallyaccepted.The iodine in periodic acid tends to pass into the 6-covalent state,so that the ordinary periodic acid is H,[IO,]. The 4-covalent formHCIO,] has been only once described, but J. R. Partington and R. K.Bahl*6 now show that HIO, is a definite compound, although noevidence could be found of the existence of the anhydride I,O, or ofmesoperiodic acid H,IO,. Periodic acid, H510,, at 100" in a vacuumloses 2H,O and HIO, is formed. On heating at 80" in a vacuumtwo molecules of H510, lose 3H,O, H4120g being formed.In general, cupric copper does not readily form 6-co-ordinatedcompounds, and its covalency is normally 4. It was surprising,therefore, when in 1927 W.Wah13' stated that he had prepareda laevorotatory iodide [C~(en)~(K,0),]1, and thus established that6-covalent copper was of octahedral type. Now, C. H. Johnsonand S. A. Bryant 38 have reinvestigated this matter. All theirattempts at a resolution were unsuccessful, and moreover, theiranalyses show that there are no grounds for presuming that theion [Cu(en),(H,O),]" is present in the crystalline salt. Theyincline to the view that the constitution of the ion in the crystalis [Cu(en),]", where the covalency of the copper is 4 and not 6, andso optical isomerism cannot arise. Some earlier work of G. T.Morgan and his co-workers lends support to this structure.The valuable co-ordinating properties of the tridentate group2 : 2' : 2"-tripyridyl (trpy) have been applied with success to theproblem of the structure of 4-covalent compounds of platinum.G. T.Morgan and F. H. Burstal139 have succeeded in isolatinga red crystalline salt, [Pt(trpy)Cl]C1,2 or 3H,O, andl thereby pro-vided an elegant chemical proof that in this co-ordination compound313 J., 1934, 1088; A., 979.3 7 Acta Sci. Fennlzicae Comrn. Phys.-Math., 1927, 4, 1; A., 1928, 395; Ann.Reports, 1933, 30, 100.J . , 1934, 1783. Ibid., p. 1498WARDLAW : MOLECULAR STRUCTURES. 117the four valencies of the platinum atom must be planar, for a tetra-hedral structure is inadmissible. If a model of this molecule (I) isbuilt up on the assumption that the pyridine rings have the Kekul6structure, it will be found that chelation can take place withoutI C1undue strain only if in two of the pyridine rings the double bondsare fixed, i.e., not oscillating between the two possible Kekult5forms.F.G. Mann 40 has shown that pp’-diaminodiethylamine,(NH,*C,H,),NH, can act as a tridentate group with platinous salts,but as this triamine may occupy the three points of a triangularface, it is uncertain whether the salt [BrPt(NH,*C,H,),NH]Br isplanar or tetrahedral. Further interesting stereochemical questionsare raised in a paper by G. T. Morgan and F. H. Burstall 41 dealingwith 2 : 2’-dipyridylplatinum salts, and in a communication byJ. S. Anderson 42 on Zeise’s salt K[PtC1,,C,H4],H,0.In 1930, F. G. Angell, H. I>. K. Drew, and W. Wardlaw 43 broughtforward new experimental data about the two known forms of thethio-ether addition compound (Et,S),PtCI,, which indicated thatfurther investigation of this group of substances was necessary asthe chemical evidence available did not afford any confirmationof the planar structure.The results of recent chemical and X-rayexperiments by E. G. Cox, H. Saenger, and W. Wardlaw 44 with thedimethyl sulphide derivatives of platinous and palladous chlorides,[Pt(Me,S),CI,] and [Pd(Me,S),Cl,], prove that the two isomeridesof the former are planar cis-tram-compounds. The a-form is thetrans-compound, not the cis- as suspected by Werner, and it is nottetrahedral as suggested by others. The X-ray results with thep-isomeride are less definite, but it seems likely that the sulphuratoms are in cis-positions and that the compound is ionised in thesolid state.I n the case of the palladous compound, only one formwas obt’ained ; this is isomorphous with the a-platinous compoundand is therefore the plane trans-compound. The chemical re-actions of the substances differ very considerably, notably witahsilver oxide. The p-platinous compound reacts rapidly with this40 J., 1934, 466; A., 640.42 Ibid., p. 971 ; A., 994.44 J., 1934, 182; A., 397.41 Ibid., p. 965; A., 1113.43 J., 1930, 349; A., 1930, 559118 INORGSNIC CHBMISTRY.reagent, with production of silver chloride and a basic substancewhich forms an alkaline solution in water and yields the originalsubstance with acid. The a-form, on the other hand, reacts onlyslowly, with evolution of dimethyl sulphide and precipitation ofplatinum as hydroxide or oxide.The so-called third form ofPt(Me,S),CI, has been shown by L. Tschugaeff and W. Surbotin 45to be really the plato-salt, [Pt(Me2S!4][Pt.C14], a result confimedby the present investigators. The diethyl sulphide derivatives ofplatinous and palladous chlorides have also been submitted toa detailed chemical examination by H. D. K. Drew and G. H.Wyatt,46 who draw conclusions regarding the structures, sub-stantially in agreement with the results obtained for the dimethylsulphide derivatives.Palladium, like platinum, has been shown by X-ray47 methodsto give planar configurations in certain of its 4-covalent compounds.In no case, however, had the cis- and trans-isomerism demanded bytheory been established.Recently,48 from the reaction of glycineand potassium chloropalladite, two substances were isolated, vix. ,yellow prisms, Pd(NH,~CH,*CQ,)2,3H20, and glistening, light yellow,anhydrous plates. These have been shown by X-ray and chemicalexperiments to be different in structure and to have the cis-trans-planar configurationsA. A. Griinberg and V. M. Schulman 49 have also described two formsof Pd(NH,),CI, which they consider to be cis- and trans-isomerides.4. SOME RARER METALS.Considerable attention has been given in recent years to theso-called “rare elements,yy and results of great interest have beenobtained in what is undoubtedly a very profitable field of chemicalinvestigation.Nowadays, the description “ rare ” is not veryfitting, for most of them can be obtained in it pure form withoutdifficulty. Apparently, the rare earths of the yttrium group,especially those with odd atomic numbers, are still difficult to obtainpure. Workers on rhenium have been particularly active, but muchresearch remains to be done before the chemistry of this element issatisfactorily elucidated.w. w.45 Bw., 1910, 43, 1200; A., 1910, i, 354.4 7 Ann. Reports, 1933, SO, 108.40 J., 1934, 56; A., 284.F. W, Pinkard, E. Sharratt, W. Wardlaw, and E. G. Cox, J., 1934, 1012;A., 994.49 Compt. rend. Acad. Sci. U.R.S.S., 1933, 1, 218; A., 1934, 379WARDLAW : SOME RABER METALS. 119Although the account given in the following pages can in no senseclaim to be complete, it is hoped that it will provide an indicationof the kind of investigation that has been going on in the cases ofgermanium, gallium, indium, and rhenium.Germanium.-Discovered in 1886, germanium was consideredto be one of the rarest of elements until it was shown in 1916that the spelter residues from certain American zinc ores maycontain up to 0.25% of germanium dioxide. Pour years later, therewas found at Tsumeb, S.W. Africa, a sulphide ore, germanite,stated to have a germanium content of 5-6%, which has beenmade available in large quantities by the Otavi Minen undEisenbahn Gesellschaft.2 There is no doubt that the discoveryof these new sources of the element and the improved methods thathave been elaborated for its extraction have given an impetusto investigations of this interesting member of the fourth group.Prom the many contributions which have appeared in recent years,important relationships to carbon and silicon, on the one hand,and tin and lead, on the other, have been established.In an electrochemical investigation, J.I. Hall and A. E. Koenighave obtained coherent grey deposits of germanium, on copper, byelectrolysis of a solution of the dioxide in 3N-potassium hydroxideat 78-90', with a low current density. They find that germaniumwill displace silver from aqueous silver nitrate. An interestingobservation has been made with the dioxide, GO,.* Two crystallinemodifications have been identified, one isomorphous with quartzand the other with cassiterite (SnO,) and plattnerite (PbO,);a remarkable feature of these two forms is the difference in density(4.28 and 6.26).5 The literature relating to the hydrides and theirhalogen derivatives is very voluminous.C. A. Kraus and E. S.Carney have shown that, by treating magnesium germanide withammonium bromide in liquid ammonia, a yield of 60-70% ofmixed germanes can be obtained instead of the possible 22% whenhydrochloric acid is employed. Incidentally, it may be mentionedthat a mixture of silicon hydrides in good yield is obtained by drop-ping Mg,Si into a solution of ammonium bromide in liquid ammonia.'G. H. Buchanan, J. Ind. Eng. Chem., 1916,8, 585; A,, 1916, ii, 486.a See W. I. Patnode and R. W. Work, I d . Eng. Chm., 1931,23, 204; B.,1931, 495, for references to history of germanite and earlier methods ofextraction.Tran8.Amer. Electrochem. SOC., 1934, 65, 79; A., 735.4 V. M. Goldschmidt, 2. physikal. Chem., 1932, 17, 172; A., 1932, 681;A. W. Laubengayer and D. S. Morton, J . Amer. Chem. SOC., 1932,54,2303; A.,1932, 905.6 A. W. Laubengayer and D. S. Morton, Zoc. cit.Ibid., 1934, 56, 765; A., 615.W. C. Johnson and T. R. Hognew, ibid., p. 1262; A., 742120 INORGANIC CHEMISTRY.Particular reference, however, should be made to the isolation ofthe hydrides (GeH)z and (GeH2)r. The monohydride 8 is preparedby the action of cold water on sodium germanide, NaGe, as a darkbrown powder which yields germanium and hydrogen at 165".The dihydride results when calcium germanide, CaGe, is treatedwith acid.It is an amorphous yellow compound, quite stable whendry but explosively reactive with oxygen. Pyrolysis at 120-220"gives a mixture of GeH,, &,Ha, Ge3H8, and hydrogen with aresidue of germanium. The chemical behaviour of (GeH,),indicates an open-chain structure of high molecular weight, analogousto that of the polyoxymethylenes.A systematic study of certain derivatives of bivalent germaniumhas given much useful information. The sequence Ge, Sn, Pbindicates that the dichloride should be unstable, and so it is notsurprising that it was not isolated until 1929, when L. M. Dennisand H. L. Hunter 10 prepared it as a crystalline, colourless massby leading the tetrachloride (b. p. 86") over heated germanium andquickly cooling the vapour.Previously, in most text-books thiscompound was described as a liquid. Germanium sulphide, GeS,was first described by C. A. Winkler, and a study of its preparationby reduction of the disulphide GeS, with hydrogen was made byL. M. Dennis and S. M. Joseph; 11 by this method it is obtainedas a black crystalline solid. A red form is obtained when hydrogensulphide is passed into a hot solution of the dichloride, and the redprecipitate dried in nitrogen at 300". When heated for a few hoursin nitrogen at 450", the red reverts to the black form.12 It is interest-ing to notice that the monoxide, GeO, is a jet-black crystallinecompound which sublimes in nitrogen at 710".12Amongst the nitrogen compounds, two nitrides are known,Ge3N4 and Ge,N,.Germanic nitride l3 can be prepared by theaction of gaseous ammonia on germanium at high temperatures orby thermal decomposition of germanic imide, Ge(NH2),, which isa light white powder obtained by the ammonolysis of GeC1, inliquid ammonia. At 150" the imide loses ammonia and formsgermanam, &,N,H, for which the formula (I) has been proposed.148 L. M. Dennis and N. A. Skow, J . Amer. Chem. SOC., 1930,52,2369; A,,P. Royen and R. Schwarz, 2. anorg. Chern., 1933,211,412; A., 1933, 579;1930, 1007.ibid., 1933, 215, 295; A., 1934, 158.10 J . Amer. Chem. SOC., 1929,51, 1151; A., 1929, 662.l1 J . Physical Chem., 1927, 31, 1716; A., 1928, 33.la L. M. Dennis and R. E. Hub, J . Amer. Chern. SOC., 1930, 52, 3553; A.,lS W.C. Johnson, ibid., p. 516; R. Schwarz and P. W. Schenk, Ber., 1930,14' J. S. Thomas and W. Pugh, J., 1931, 60; A., 1931, 322.1930, 1387.63, [B], 296; A., 1930, 437WARDLAW : SOME RARER METALS. 121Germanous imide l5 may be prepared from the di-iodide and am-monia : GeI, + 3NH, = GeNH + 2NHJ. When heated to250-300" for several hours, it gives GR3N2,16 a finely divided brownpowder : 3GeNH = &3N2 + NH,.Germanium is now included in the list of elements which formheteropolyacids l7, l8 of the type H,[X(R,O,),], where R = Moor W, and X may be one of the following Group IV elements : Si,Ti, Zr, Th, Sn, Pb, Ge, and possibly Hf. The free acid,H8[Ge(Mo,0,),],aq., has a varying water content and crystallisesin yellow transparent octahedra readily efflorescing, and withm.p. ca. 65".18 The organo-metallic compounds of germaniumhave been widely investigated. Amongst the many interestingresults obtained, mention may be made of the fact that R. Schwarzand M. Lewisohn 19 have prepared an optically active phenylethyl-isopropylgermanium bromide, and have also described the aromaticgermanium compound (11).Ph Phph PhFinally, there is evidence of the existence of derivatives of per-germanic acid. The per-compounds of the Group IVu elements,titanium, zirconium, and thorium, with the general formula H,MO,where M = Ti, Zr, or Thy are well known. In Group IVb the ten-dency to per-acid formation is appreciably less, so that with leadsuch substances are unknown, and with tin, the perstannateNa,Sn,O,,SH,O is very unstable.,O Germanium, however, forms well-defined per-compounds, and Schwarz and Giese have announced thepreparation of K,&,07,4H,O, Na,Ge20,,4H20, and Na,Ge05,4H,0.These authors were unable to isolate a crystalline persilicate, butobtained an oil which may have contained some decomposedpersilicate.On the other hand, F. Krauss,21 by evaporatinga solution of sodium silicate treated with hydrogen peroxide,obtained a powder which he considers to be Na,Si03,H,0,2H20,.Evidently he does not regard this product as a true persilicate.Gallium and Indium.-As gallium is present in nearly all1 6 W. C. Johnson, G. H. Morey, and A. E. Kott, J . Amer. Chern. SOC., 1932,16 W. C. Johnson and G. H. Ridgely, ibid., 1934,56, 2395.l7 A. Brukl, Monatsh., 1930, 56, 179; A., 1930, 1538; R.Schwarz and H.54,4278; A., 1933, 38.Giese, Ber., 1930, 63, 2428; A., 1930, 1637.C. G. Growcup, J . Arner. Chem. SOC., 1930,52,5154; A , , 1931, 322.Ber., 1931, 64, 2352; A,, 1931, 1435.2. anorg. Chm., 1932, 204, 318; A., 1932, 350.2o R. Schwarz and H. Giese, Ber., 1930,68, 780; A., 1930, 720122 IXOR43ANICl CHEMISTRY.germanium-containing blendes and also occurs in germanite to theextent of ca. 0*6%, its extraction generally accompanies that ofgermanium. The 'University of Colorado has recently publisheda " Bibliography of Indium," 22 in which communications from thedate of the discovery of the metal in 1863 to 1933 are classified.Most of the work on indium appears to have been published inGermany and the United States, but the United Kingdom is repre-sented by the work of (Sir) H.C. H. Carpenter and S. TamuraBon twinned metallic crystals, some older work of Roberts-Austenand T. Carnelley, and an investigation of indium acetylacetone byH. D. K. Drew and G. T. Morgan.24Within the last few years, important additions to the fundamentalchemistry of gallium have been made. Its tri-bromide and tri-iodide were prepared for the fkst time in 1930,,5 and in the sameyear it was proved that, not only did sulphides of bi- and ter-valentgallium exist, but also the sulphide of the univalent element.26The yellow crystals of Ga,S, were obtained by passing nitrogen andsulphur vapour over the metal a t 1200" ; when reduced by hydrogena t 800°, it gave a glistening yellow sublimate of Gas, which onbeing heated in a high vacuum yielded Ga,S, and the volatile Ga,S.This new sulphide is greyish-black and readily oxidised.- A numberof selenides have been synthesised : Ga,Se, GaSe, Ga2Se3, In,Se,InSe, and In,Se,, and by the thermal analysis of the systems Ga-Teand In-Te, the existence of the following tellurides has beendemonstrated: GaTe, Ga2Tes, InTe, and In,Te,. The oxide ofunivalent gallium, Ga,O, has been obtained 28 but GaO is unknown.A remarkably stable nitride was isolated in 1932 by the interactionof ammonia and gallium at 900°.29 It is unattacked by concentratedhydrochloric, hydrofluoric, or nitric acid or by hot aqua regia.Even hot concentrated sodium hydroxide dissolves it but slowly.C. A.Kraus and F. E. Toonder30 have prepared organometalliccompounds of gallium which show the expected analogy to theaa '' Studiea," 1934, 21, No. 3.ad J., 1924, 1261; A., 1924, i, 941.26 W. C. Johnson and J. B. Parsons, J. Physical Ch., 1930,34, 1210; A . ,1930, 874.26 A. Brukl and G. Ortner, Naturwiss., 1930, 18, 393; A., 1930, 720;Monatsh., 1930, 56, 358; A., 1930, 1537; W. C. Johnson and B. Warren,Naturwiss., 1930, 18, 666; A,, 1930, 1138.Bull. Inat. Min. Met., 1928, No. 282; A., 1928, 603.27 W. Klemm and H. U. von Vogel, 2. anorg. Chem., 1934,219,46; A., 1081.28 A. Brukl and G. Ortner, 2;. anorg. Chem., 1931,203,23; A., 1932,238.2B W. C. Johnson, J. B. Parsons, and M. C. Crew, J . Physical Chem., 1932,Proc.Nat. Acad. Sci., 1933,19,292; A., 1933, 599; J . Amer. Chem. SOC.,36,2588; A., 1932, 1218.1933, 55, 3547; A., 1933, 1150WARDLAW : SOME RARER METALS. 123corresponding zinc derivatives. I n the third group of the PeriodicTable from boron to thallium, either the normal trimethyl or triethylcompounds or both have now been prepared. The series was com-pleted in 1934 by tho isolation of trimethylindi~rn,~~ a colourlesscrystalline solid which gives a molecular weight in benzene in accord-ance with the polymeride [In(CH,),],. I n the following table,based on that given by W. Klemm,32 the types of compound thatexist for the different valencies are summarised.Valenc y . Valency.Gallium. 1. 2. 3. Indium. 1. 2. 3.c1 ............ - + + c1 ............+ + +Br ......... - + + Br ......... + + + ............ I ............ + + + I - - . +0 (+) - + 0 (+I 3. s ............ (+I + + s ............ + + + ............ ............(+) = Prepared from the gaseous state by cooling but not stable; decom-posed on heating; + = prepared by synthesis and stable; - = not yetprepared and probably non-existent.The summary shows that for both elements uni-, bi-, and ter-valent compounds exist. The tendency to exist in the univalentstage is less for gallium than for indium, whilst with thallium it isgreatest. Bivalent compounds do not appear to be very stable.In the tervalent state, the halides are almost all colourless; ex-ceptions are GaI, and I d 3 , which are yellow. Klemm 32 has pointedout an interesting relation between colour and constitution in thecompounds of gallium and indium: those of the bivalent stageshow no appreciable colour deepening compared with the tervdentstage, whereas the unsaturated compounds of the univalentstage are quite dark.He concludes that this anomaly is explainedby the magnetic properties. The bivalent gallium and indiumcompounds are diamagnetic. Now as Ga" and In" ions containa free electron, they should show paramagnetism. Evidently theseions combine with spin-equalisation to the diamagnetic (Ga2)IV and(In,)m ions, just as two paramagnetic hydrogen atoms form a dia-magnetic molecule.Rhenium-From the many conflicting statements in the literature,it seems established that rhenium can form three oxides, Re20,,ReO,, and ReO,.The colourless heptoxide 33 is obtained by directoxidation of rhenium. I t s m. p. (in vacuum) is 301-5", and sublim-ation begins at 220". In contrast to manganese heptoxide it issl L. M. Dennis, R. W. Work, and E. G. Rochow, J . Amr. Chm. SOC., 1934,56, 1047.IP Angew. Chm., 1934, 47, 17.8s W. Biltz and G. A. Lehrer (with K. Meisel), Nach. Gea. Wise. Qcjttingen,1931, 191; Chm. Zentr., 1932, i, 1070; A., 1932, 708; 2. anorg. Chm., 1933,214,225; A., 1933, 1259; aid., 1932,207,113; A., 1932, 1008124 INORGANIC CHEMISTRY.very stable, and this is reflected inits large heat of formation, whichis approximately 296 kg.-cals. per g.-mol.M Under certain experi-mental conditions it is possible to obtain from the oxidation ofrhenium a white crystalline product which (Frau) I.and W. Nod-dack35 assumed was the peroxide ReO, or Re208. Later experimentsby H. Hagen and A. Sieverts36 have indicated that this is not aperoxide, but possibly only another form of Re207. Accordingto W. Biltz3, and his collaborators, a red trioxide is obtained bythe prolonged action of metallic rhenium on the heptoxide a t200-250", or better, from ReO, and Re,O, at 300". The crystalstructure of this trioxide has been examined by K. Meise1,3, whofinds it to be isomorphous with tungsten trioxide, WO,. Theblack dioxide Re0,38 is formed when Re,07 and Re are heated,first a t 300" and then a t 600-650". This dioxide decomposes at1000" in accordance with the equation : 7Re0, = 3Re + 2Re207.Certain inve~tigators3~ have claimed that the red oxide is a pentoxide,Re,05, but in view of the results obtained by K.Meisel 37 this appearsunlikely. A hydrated Re2O3*0 has been prepared by the hydrolysisof rhenium trichloride with aqueous sodium hydroxide. It isreadily oxidised and will liberate hydrogen from water. A blueoxide, obtained from the reduction of Re207, has been mentioned.It may be the analogue of the well-known molybdenum-blue.In last year's report attention was directed to the isolation ofcertain halides of rhenium. 0. Ruff and W. Kwasnik41 havecontinued their investigations on the rhenium fluorides, and haveisolated pure ReF, (m. p. 18.5"; b. p. 47.6"). This is readilyreduced t o ReF, (m. p. 124-5") at comparatively low temperaturesby a variety of reagents, e.g., hydrogen a t 200", carbon monoxidea t 300", sulphur dioxide at 400".They have also establishedthe existence of ReOF, (m. p. 39-7"), Re0,F2 (m. p. 156"),and the complex fluoride K,ReF,. By contrast, the literaturedealing with the action of chlorine on rhenium is very confusing.This confusion began in 1928 with W. Noddack's statement 42 that34 W. A. Roth snd G. Becker, 2. physikal. Chem., 1932, [A], 150, I ; A., 1932,35 Naturwiss., 1929,17,93; A., 1929, 411.36 2. anorg. Chem., 1932, 208, 367; A., 1933, 43.3 7 Ibid., 207, 121; A,, 1932, 903.38 W. Biltz, ibicl., 1933, 214, 225; A., 1933, 1259.39 H. V. A. Briscoe, P. L. Robinson, and A. J. Rudge, J., 1931, 3087; A.,1932,32; W. A. Roth and G. Becker, Ber., 1932,65, [B], 373; A., 1932,353.40 W.Geilmann and F. W. Wrigge, 2. anorg. Chern., 1933, 214, 239; A.,1933, 1259.469.-Ibid., 1934, 210, 65.42 2. ElektrocJlem., 1928, 34, 627; A., 1928, 1344WARDLAW : SOME RARER METALS. 125two volatile chlorides ReCI, and ReCl, are formed when rheniumis heated in chlorine. In 1931 H. V. A. Briscoe, P. L. Robinson,and F. M. Stoddart 43 were unable to confirm this, and from theirexperiments they concluded that the primary product of heatingthe metal in chlorine was the tetrachloride ReCl,. More recently,however, W. Geilmann, F. W. Wriggo, and W. Biltz 44 report thatno tetrachloride can be isolated from this reaction, but that thepentachloride is produced and can be purified by fractional sublim-ation in a vacuum. These investigators have also prepared a redtrichloride from the dark brownish-black pentachloride by heatingit in a current of nitrogen, The analytical data 45 for this trichlorideshow that it is oxygen-free and disprove the suggestion of W.Manchot and J.J. G . F.Druce 47 claims to have prepared trimethylrhenium from thetrichloride. Although a hexafluoride and a pentachloride of rheniumare now known, the highest bromide so far prepared is the tri-bromide, which H. Hagen and A. Sieverts48 have isolated as agreenish-black sublimate by heating rhenium at 500" in brominevapour. It is noteworthy that when the different compounds sofar isolated are arranged in the order of their highest valencies,the series becomes RezO,, Rep,, ReCl,, ReBr,.This is quitein accordance with the general rule that, for a given element, in theseries oxides, fluorides, chlorides, bromides, etc., the tendency toreach the highest possible valency decreases.The salts of per-rhenic acid offer some marked contrasts to thepermanganates. For example, the colourless Re,O, dissolves inwater to produce a colourless solution of per-rhenic acid, which formscolourless salts with the alkali and alkaline-earth metals. Inaddition, an investigation by E. Wilke-Dorfurt and T. Gunzert 49has revealed striking differences in solubility, crystal form, and con-tent of water of crystallisation between the per-rhenates and thecorresponding salts of permanganic, perchloric, and hydrofluoroboricacids. Per-rhenates of the type M'RO, are generally referred to asmeta-salts, for with excess of base it is possible to isolate yellowmesoper-rhenates, MiReO,.The barium salt Bas(ReO6),, whichhas been fully is decomposed by water into Ba(OH),that it might be Re,Cl,O.43 J., 1931, 2263; A., 1931, 1255.44 2. anorg. Chem., 1933,214,244; A., 1933, 1259.45 W. Biltz, W. Geilmann, and F. W. Wrigge, Annalen, 1934, 511, 301 ; A.,4 8 Ibicl., 500, 228; A., 616.4a 2. anorg. Chem., 1933, 215, 111; A., 1934, 44.49 Ibid., p. 369; A., 1934, 158.979.4 7 J., 1934, 1129; A., 995.(Frau) I. and W. Noddack, ibid., p. 129; A., 1934,44; B. Schasnow, ibid.,11. 185; A., 44126 INORUANIC CHEMISTRY.and Ba(ReO,),. A number of similar compounds containingrhenium of lower valency have been reported, but their identityis not yet fully established in every case.A brown rhenite,Na,ReO,, is known, and an unstable sand-yellow hyporhenate,possibly Na4Re20,, as well as an unstable green rhenate, BaRe04,have been recorded by (Frau) I. and W. N~ddack.~oThe stable sulphide of rhenium is the black disulphide, ReS2.51A black hydrated sulphide, Re2S,, is precipitated when hydrogensulphide or sodium thiosulphate reacts with potassium per-rhenate,but this is converted into the disulphide when heated in a currentof nitrogen. 52Some interesting results have been obtained by the electrolyticreduction of acid solutions of rhenium compounds. Reduction ofKReO, in 9N-hydrochloric acid, with either a bright or a platinisedplatinum cathode, gives a green solution containing quinquevalentrhenium.53 An olive-green solution of tervalent rhenium is obtainedby cathodic reduction of K,ReCl, in 2N-sulphuric acid.54 Theseresults are very similar to those obtained with molybdenum com-pounds, and further work on these electrolytically reduced solutionsshould yield valuable information about rhenium derivatives.Finally, it may be mentioned that rhenium forms a number ofco-ordination compounds displaying covalencies of four and six,e.g., PyH[ReBr,], K,[ReOC15],H,0, X2[ReC1,], X,[ReBr,], whereX = K, Rb, Cs, etc., and that various oxychlorides and oxybromidesare known.The chemistry of rhenium is much more complicatedthan this short sketch might lead one to suppose, but this state-ment can be fully confirmed by reference t o “Das Rhenium”(Leipzig, 1933) by I.and W. Noddack. w. w.5 . THE CORROSION OF METALS.Progress in research on the corrosion of metals has not beenreviewed in these Reports during recent years, although occasionalreference to results of particular interest has been made. Thepresent Report does not cover any specific period, therefore, butaims at reviewing the trend of research. Space does not permit ofthe discussion of all the aspects of this work; some of the morefundamental directions of research have been followed a t theexpense of leaving a certain amount of interesting work unmen-tioned. A feature of modern views on corrosion is the recognitionI. R. Juza and W. Biltz, 2. Elektrochem., 1931, 37, 498;’ A., 1931, 1128.52 W.Biltz and F. Weibke, 2. anorg. Chem., 1931, u)3, 3; A., 1932, 238.63 W. F. J a b and B. Jeiowska, W., 1933, 214, 337; A., 1933, 1254.W. Manchot and J. Dihing, Zoc. cit., ref. (46)HEDGES : THE CORROSION OF METALS. 127of the imporfant part played by films, especially those of a pro-tective nature. In order to emphasise this aspect, the results arepresented in a somewhat different order from that usually followed.Film Formation and Passivity.Progress in the study of passivity, whilst of primary importancefrom the theoretical viewpoint of corrosion, is one part of the fieldof corrosion in which considerable unanimity of outlook amonginvestigators all over the world has been reached. Although all arenot agreed as to the mechanism, yet it is now generally acceptedthat passivity is due to the presence of a protective film, generally,but not necessarily, of oxide.Air-formed PiZms.-The existence of air-formed oxide films, whichinterfere with the reactivity of the underlying metal, has beendefinitely established for copper,f aluminiumY2 irony3 and certainother metals.Reference has been made in a previous Report to theisolation of the air-formed protective film on iron. These films havesince been subjected to further study, particularly with regard tothe conditions of their breakdown and their structure. U. R. Evanshas shown that breakdown of the protective oxide film on iron,steel, zinc, or aluminium tends to occur where the specimen hasbeen bent or otherwise distorted.Corrosion occurs preferentiallya t the bend, especially on the convex side.U. R. Evans and J. Stockdale 6 have put forward a scheme repre-senting the structure of the surface film. This is pictured in fourzones : (u) an outer zone of oxide, (b) a zone containing oxide andmetal having a high resistance to attack, ( c ) a shattered zone,fairly free from oxide and having a low resistance to attack, and(d) the unchanged metal. They have also devised an improvedelectrolytic method of isolating the film, by dissolving away ( c ) .Flakes of iron oxide, thus isolated from iron heat-tinted to the first-order yellow, were found to have a thickness corresponding with2 x lo4 g. of ferric oxide per sq. cm., agreeing with F. H. Con-stable's ' determination of the thickness of the film on yellow-tintediron (0.46 x cm.).Passivation in Concentrated Nitric Acid.-Although it has notbeen possible to isolate an oxide film from iron which has beenimmersed in concentrated nitric acid, yet this classical example of1 W.H. J. Vernon, J., 1926, 2273; A., 1926, 1108 ; F. H. Constable,Nature, 1929, 123, 569; A., 1929, 503.a H. Sutton and J. W. W. Willstrop, J . Inst. Metals, 1927, 38, 259; U. R.Evans, J., 1927, 1039; A., 1927, 610.a Idem, ibid., p. 1021; A., 1927, 619.6 J., 1929, 92 ; A., 1929, 270.4 Ann. Reports, 1928, 25, 30.6 Ibid., p. 2661 ; A., 1930, 29.Proc. Roy. SOC., 1928, [A], 117, 376; A., 1928, 106128 INORGANIC CHEMISTRY.passivity has been brought into line with the oxide-film theory, forE.S. Hedges 8 obtained convincing evidence of the existence of afilm of ferric oxide. Shortly afterwards, C. Benedicks and P. Seder-holm 9 examined the effect of dilute alcoholic nitric acid solutionson carbon steels. Owing to the relatively slight dissociation ofnitric acid in alcohol, these solutions have something in common asan oxidising agent with concentrated aqueous solutions of nitricacid, and they were shown to render the steel passive and to producea film of ferric oxide, which was actually photographed. Thecritical concentration of nitric acid required to render iron and steelpassive has been reinvestigated by Y. Yamamoto.1° Hedges hasalso shown that other metals are rendered completely passive inconcentrated nitric acid a t - ll", and that copper l1 acquires anoxide film in concentrated nitric acid at ordinary temperatures,the existence of which is the cause of the practically completeinertness of this metal when it is kept in motion in this reagent.A.Kutzelnigg l2 has described the passivity of copper in a mixtureof nitric and sulphuric acids.Activating E8ect.s of 1ons.The activating influence of certainanions (especially chlorides) on passive metals has been studied, butit is not yet possible to state definitely whether the effect is due topenetration of the oxide film by anions of small size or whether theactivation is related to the well-known peptising effect of chlorides,thus loosening the protective film. The activating effect has beenstudied 13 by measuring the potential of the metal in the solutionagainst a standard reference electrode.S. C. Britton and U. R.Evans 1* have measured the penetrating powers of different anionsby determining the leakage current at an aluminium anode in asolution of potassium chromate, to which solut4ions containing theions investigated were added. The following decreasing sequenceof penetrating power was noted : chloride> bromide>iodide>fluoride>sulphate>nitrate>phosphate. These results have beenconfirmed by L. Tronstad and B. W. Bommen.15 In general, theactivating power decreases with increasing size of the anion, but it8 J., 1928, 969; Ann. Reports, 1928, 25, 30.9 2. physikal. Chem., 1928,138, 123; B., 1929, 21.10 Bull.Imst. Phys. Chan. Rm. Japan, 1934, 13, 375; A., 1934, 736.11 E. S. Hedges, J., 1930, 561; A , , 1930, 649.1 2 2. Elektrochem., 1933, 39, 67; A , , 1933, 365.13 A. L. McAulay and S. H. Bastow, J., 1929, 85; A., 1929, 270; U. R.Evans, ibicl., p. 92; A., 1929, 270; E. S. Hedges, ibirE., p. 1037; A., 1929,776; T. P. Hoar and U. R. Evans, J. Iron and Steel Inat., 1932, 26, 379;A., 1932, 989.14 J., 1930, 1773; A., 1930, 1258.l6 K . Norske Vidensk. Selsk., 1933, 45, 174HEDGES: THE CORROSION OF METALS. 129may be pointed out that a similar rule holds roughly for peptisation.A comparison of the activating effects of various anions on a passivemetal with the peptising effect of the same anions on the hydroxide.of the metal would be of much interest.The comparative feebleness of the fluoride ion in penetratling orloosening the protective oxide film on iron has been confirmed byA.W. Chapman.16 I n activating passive chromium by cathodicpolarisation, E. Muller and K. Schwabe l7 have shown that adefinite activation potential is required in different acid solutions,the negative potential increasing in the order : hydrochloric,hydrofluoric, hydrobromic, sulphuric, perc hloric, ort hophosphoric .The small ions are thus the most powerful. These authors haveput forward the view that chromium in the passive state is coveredby a network of chromic oxide molecules, anchored to the units ofthe chromium space lattice. As the cathodic polarisation isincreased, the small hydrogen ion is dragged through the oxidenetwork and the electrostatically-bound anion, if it is not too large,can follow it and dissolve the film.Activation is favoured atcorners and edges of the metal, because there the oxide network islooser. A short-circuited cell is then set up, and if the potentialreaches a higher negative value than the activation potential forthe particular acid, the whole metal will become active. Withchromium a t 20" this is realised only in hydrochloric acid.Anodic Films.-During the last five years or so a great amount ofwork on anodic passivity has been carried out by W. J. Muller 18l6 J., 1930, 1546 ; A., 1930, 1128.1 7 2. Elektrochem., 1931, 37, 185; A., 1931, 571; E. Muller, 2. physikd.Chem., 1932, 159, 68; A., 1932, 473.18 W.J. Muller, 2. Elektrochem., 1927, 33, 401; A., 1928, 135; Monatsh.,1927, 48, 61; A., 1927, 735; ibid., p. 559; A., 1927, 1145; 2. Elektrochem.,1928, 34, 571; A., 1928, 1319; ibid., p. 850; A., 1929, 270; ibid., 1929, 35,ibid., 1930,56, 191; A., 1930, 1527; 2. Elektrochem., 1930, 36, 550; A., 1930,Trans. Paraday SOC., 1931, 27, 737; A., 1932, 25; Korrosion U. Metallschutz,1932, 8, 253; A., 1933, 576; Trans. Paraday Soc., 1932, 28, 471; A., 1932,576; Z. Elektrochem., 1932, 38, 850; A., 1933, 30; Angew. Chem., 1933, 46,197; A., 1933, 468; Korrosion u. Metallschutz, 1934, 10, 1; A., 1178; Natur-wisa., 1934,22, 479; A., 968; 2. Elektrochem., 1934, 40, 119, 536, 578; A., 602,1072 ; W. J. Miiller with E. Noack, Monatsh., 1927,48,293 ; A., 1927,942 ; withK.Konopicky, ibid., p. 711 ; A., 1928,247; with 0. Lowy, ibid., 1928, 49, 47;A., 1928,713; with K. Konopicky, ibid., 1928, 50, 385; A., 1929,146; idem, 2.Ekktrochem., 1928,34, 840; A., 1929, 269; idem, ibid., p. 858; A., 1929, 146;idem, Z. physikal. Chern., 1929, [A], 141, 343; A., 1929, 770; with 0. Lowy,Monatsh., 1929, 51, 73; A., 1929, 402; with K. Konopicky, ibid., 1929, 52,289; A., 1929, 1241; with L. Holleck, ibid., 1929,52,409,425; A., 1930, 298;with K. Konopicky, ibid., p. 442, 463; A., 1930, 298; with W. Machu, ibid.,REP .-VOL. XXXI. E93, 656; A., 1929, 270, 1393; Momtsh., 1929, 52, 53, 221; A., 1929, 886;1257; ibid., p. 679; A., 1930, 1377; ibid., 1931, 37, 328; A., 1931, 915i30 1NORGANIC CHXMISTRY.and his collaborators in Vienna.This important wcrk has notpreviously been described in these Reports.An essential feature of the experimental conditions is the elimin-ation of disturbances a t the anode due to convection. By using ahorizontally disposed anode, sheltered from such disturbances, thereaction products remain where they are formed and their propertiescan be observed. Under these conditions, current density-potentialcurves for different anodes in various solutions can be made repro-ducible, and .the value determined at which the potential suddenlyrises,One of the most striking results of the work, which appears in theearlier papers, is the complete demonstration that the time duringwhich the current must flow before passivity sets in depends on thecurrent density; passivity can be brought about either by a smallcurrent acting for a long time or by a heavy current acting for ashort time. This relation suggests that the accumulation of somereaction product a t the surface of the anode is responsible forpassivity.For a typical case, such as the anodic treatment of ironin dilute sulphuric acid, the view developed is that some product(in this case ferrous sulphate) accumulates a t the surface of the anodeuntil the solution in the vicinity becomes supersaturated, whereupona solid crystallises on the anode surface. The deposit covers a largeportion of the surface, increasing the effective current density atthe uncovered portions to perhaps 100 or 1000 times the originalvalue. The film is not identified with the protective film causingpassivity, but, by screening a large area of the electrode, it maylead to conditions which cause true passivity.Muller calls the firststage " Bedeckungspassivitat " and the second stage " chemischePassivitiit .' 'To elucidat,e the relation between the two stages, the rate of fallof current with time has been studied, whilst the composition of thefilm responsible for " Bedeckungspassivittiit " has been confirmedin some cases by direct observations with the polarising microscope.In fairly concentrated acids, the films appear to be normal salts,but in more dilute solutions they may consist of basic salts or evenoxides or hydroxides.Assuming that the thickness of the deposited layer remainsp. 474; A., 1930, 298; idem, 2.physikal. Chem., 1931, Bodenstein Festband,p. 687; A., 1931, 1238; ibid., 1932, [ A ] , 161,147,411; A., 1932, 993,1208; withH. K. Cameron and W. Rlachu, Monatsh., 1932, 59, 73; A., 1932, 343; with w. fixachu, ibid., 1932, 60, 359; A., 1932, 1000; with E. Low, 2. Elektrochem.,1933, 39, 872; A., 1934, 33; with W. Machu, 2. physikal. Chem., 1933, [A],166,357; A., 1934,33; Monateh., 1933,63, 347; A., 1934,368; with E. Low,2. Elektrochem., 1934,40, 570; A., 1072HEDGES: THE CORROSION OF METALS. 131constant and that the film grows sideways only, the fall of thecurrent i with the time t is given by the expressionwhere i, is the initial current, i, the residual current, and C and Aare constants defined by the formula?SS2Kk(1 - u)wo C = "' and A = k(1 - u)i,s being the specific gravity of the film material, 6 the thickness,P the original area of the anode surface, k the electrochemical con-stant, u the anionic transport number, wo the resistance from thecathode to the boundary layer adjacent to the anode, and K theconductivity of the solution in the boundary layer.The sideways-growth relation has been confirmed experimentallyin sulphuric acid as the electrolyte for anodes of copper, and alsofor those of iron or nickel in the early stages.In the later stagesof anodic passivation, the last two metals follow a course whichcan be related to the growth of a film in thickness, without sidewaysextension. Then the currents i, and i, passing at times t, and t,are in accordance with the formulat , - t , = B(l/iI2 - 1/i22)where B is a .constant depending on the area uncovered.Thisrelation has been confirmed experimentally also for anodes of zinc,chronium, and lead.In the later papers of the series, Miiller definitely accepts theexistence of a protective film at passive anodes, but maintains that,when the metal can exist in more than one state of valency, theformation of the film is preceded by an electronic change in themetal. Thus, with an iron anode in dilute sulphuric acid," Bedeckungspassivitat " is due to the primarily-formed film offerrous sulphate; at unprotected spots, subject to a very highcurrent density, a change is brought about which induces the metalto go into solution as ferric ions; the product is easily hydrolysed,and forms a protective film of ferric oxide on the anode (" chemischePassivitat ").With aluminium, where no change of valencyoccurs, the whole process is regarded as " Bedeckungspassivitat ."Whether the change of valency observed with some metals is primaryor secondary, there is clearly agreement here that anodic passivity,like other forms of passivity, is characterised by the formation of aprotective film.Recent researches have examined quantitatively the degree ofprotection afforded by the film. Like previous investigators, wh132 INORGANIC CHEMISTRY.have discussed the problem only qualitatively, Muller regards thefilm as containing pores; the finer the pores, the more protectiveis the film. He has shown quantitatively how the potentialassumed by the metal is related to the number and size of thesepores. Qualitatively, the greater the number of pores and thelarger their size, the closer the potential approaches that of theunderlying metal ; on the other hand, when the pores are sufficientlysmall, corrosion no longer occurs.Thus, the activating influence ofchlorides is traced to the relative ease with which such small anionscan travel through the pores of the film, whilst larger anions, suchas sulphate and phosphate, can penetrate the pores only withdifficulty. A film may be efficiently protective, therefore, in solu-tions containing large anions, but not in solutions containing smallanions.Many of Miiller’s observations are confirmed by the contem-poraneous work of E.S. Hedges l9 on the formation of anodic filmson copper, silver, magnesium, zinc, cadmium, mercury, tin, lead,iron, cobalt, nickel, and aluminium. Tbe conditions of study weresuch that the film automatically appears and disappears periodically,and were thus peculiarly favourable for the study of film formation.I n many cases, two definite stages in passivation, corresponding ingeneral with those postulated by Muller, could be observed directlyin each recurrent period.The conditions for periodic anodic film formation or passivityhave been elucidated. The experiments show that periodicpassivity is simply a special case of periodic film formation, inwhich the film is highly protective, and that passivity is a generalproperty exhibited to various extents by all metals under suitableconditions, and to different degrees depending on the protectiveproperties of the film under the particular conditions prevailing.The work has led t o a general theory of anodic polarisation,which is confirmed by experiments on the influence of speed ofrotation of the anode on periodic anodic passivity, and receivessome support from the work of U.R. EvansY20 who has isolatedthe film responsible for the passivity of an iron anode in dilutesulphuric acid.E. Muller and K. Schwabe 21 have studied the formation of filmson anodes of zinc, cadmium, lead, and copper in saturated andunsaturated solutions of their salts. Complete passivity is neverproduced in these systems.1s J., 1926, 1533, 2580, 2678; 1927, 1077, 2710; 1928, 969; 1929, 102820 Nature, 1930, 126, 130; A., 1930, 1126.21 2.Elektyochena., 1932, 88, 407; A., 1932, 814.A., 1926, 807, 1213; 1927 85, 630; 1928, 23, 600; 1929, 775HEDGES : THE CORROSION OF METALS. 133The formation of films in the dissolution of nickel anodes in acidand salt solutions has been studied by I<. Georgi.22 The potentiala t low current densities is higher the greater the size of the anionin the series chloride, bromide, sulphate, chlorate. Three statesare recognised. The active state is favoured by low currentdensity, small anion, high hydrogen-ion concentration, and hightemperature ; the reverse conditions favour an impoverished diffusionlayer next to the anode and cause transition to a state of higherpotential.At the lowest current densities, nickelous ions enter thesolution at certain active centres, and hydrogen collects on thegreater portion and may be removed by depolarising influences.The second state is characterised by an invisible film of nickelichydroxide, and the third by a porous diaphragm of the samehydroxide. Georgi obtained similar results with ancdes of cobaltand ir0n.~3The passivity of copper anodes in sulphuric acid24 and of goldanodes in hydrochloric and sulphuric acids25 has been traced tothe formation of oxide or similar films. The mechanism of theanodic passivity of gold in chloride solutions has been furtherstudied by G. Armstrong and J. A. V. Butler.26 The nature of theanodic oxide film formed on aluminium in oxalic acid has beeninvestigated by S.Setoh and A. Mi~ata.~'Optical and Other Means of Investigation.-The air-formed filmhas been recognised by optical means by H. Freundlich, G. Pat-scheke, and H. Zocher,28 who prepared mirrors of pure iron by thethermal decomposition of iron pentacarbonyl in absence of air.When air was admitted, a fall in the reflecting power of the mirrorswas observed, and the chemical reactivity of the iron was reducedsimultaneously .L. Tronstad 29 has applied the optical method to the examinationof anodic passivity, and shown that the optical constants of nickeland iron rendered passive in sodium hydroxide solution, and ofnickel rendered passive in sulphuric acid solution, undergo a change22 2.Elektrochem., 1932, 38, 681, 714; d., 1932, 1000, 1093.23 Ibid., 1933, 39, 209, 745; A , , 1933, 468, 1016.24 M. Lignana, Nature, 1932, 130, 474; A., 1932, 1208.25 W. J. Shutt and A. Walton, Trans. Fayaday SOC., 1932, 28, 740; A.,1932, 1209; ibid., 1933, 29, 1209; A., 1933, 1242; ibid., 1934, 30, 914; A.,1178.26 Ibid.,p. 1173.27 Sci. Papers Inst. Phys. Chem. Res. Tokyo, 1932, 19, 189, 237; A., 1933,28 2. physikal. Chern., 1927, 128, 321; 130, 289; A., 1927, 1037, 1149.29 Nature, 1929, 124, 373; A., 1929, 1150; 2. physikal. Chem., 1929, 142,241 ; A., 1929, 1002 ; K . Norske Vidensk. Selsk., 1931, No. 1 ; Nature, 1931,127, 127; A,, 1931, 301; Trans. Paraday SOC., 1933, 29, 502; A., 1933, 469.29, 1254134 INORGANIC CHEMISTRY.resembling that observed when a clean iron surface is broughtfrom a vacuum into contact with air.This result points directlyto the formation of an oxide film during anodic passivation.L. Tronstad and C. W. Borgmann30 have shown by the opticalmethod that when iron, steel, or stainless steel is immersed inpotassium chromate solution or concentrated nitric acid, thenatural, air-formed films are strengthened or even replaced bydenser films, in accordance with the oxide theory of passivity.Approximate calculations give 100 8. for the thickness of the oxidefilm formed on steel in nitric acid, and 30-40 8. for that formed ina chromate-chloride solution. The thickness of the film on stainlesssteel in concentrated nitric acid was only 10 A. Similar experimentsindicate that the natural oxide film on aluminium 31 has a thicknessof 100 A.and undergoes only small changes in chromate solutions.The optical method also shows 32 that in dry ozone highly protectivefilms are obtained on silver, iron, and ordinary and stainless steels,whereas less protective films are acquired by copper and zinc. Inmoist ozone, no highly protective films are produced.On the whole, relatively litltmle help has been gained by the applic-ation of X-ray and electron-diffraction technique to the elucidationof the structure of the film. F. Mriiger and E. Nahring33 obtainedidentical X-ray diagrams for active and passive iron, nickel, andchromium in the finely-divided condition, and considered that anyoxide film present could not be thicker than lo-' cm.Actually,there is no reason why a really continuous protective film needexceed lo-' cm. in thickness ; moreover, in some cases the film maynot have a crystalline space-lattice.G. P. Thomson 34 has found no difference in the electron-diffractionpatterns of active and passive iron; but it should be pointed outthat neither aluminium covered with the usual air-formed film, norlead, freshly cut and heated in the air a t about loo", gave a diffractionpattern in his experiments. J. A. Darby~hire,~~ however, afterisolating the oxide films on heat-tinted nickel and copper by Evans'smethod, has demonstrated their crystalline nature by the electron-diffraction method. The spacings obtained indicate the formulieNiO and Cu,O for these oxides.The electron-diffraction patternof iron rust has been obtained by J. Cates.36 C . A. Murison373O Trans. Faraday SOC., 1934,30, 349; A., 486.31 L. Tronstad and T. Hoverstad, Trans. Paraday SOC., 1934, 30, 362; A.,32 Idern, ibid., p. 1114.34 Proc. Roy. SOC., 1930, [A], 128, 649; A., 1030, 1082.86 Trans. Faraday SOC., 1931, 27, 675.36 Ibid., 1933, 29, 817; A., 1933, 1022.37 Phil. Mag., 1931, [vii], 17, 96; A., 134.486.33 Ann. Physik, 1927, 84, 939HEDGES: THE CORZOSION OF METALS. 135has found that surface films on copper heated in air give diffracticnpatterns for cuprous oxide and a new form of cupric oxide. Dif-fraction patterns have been obtained from oxide films on zinc byG. I. Finch and A. G. Q ~ a r r e l l .~ ~Corrosion in Aqueous Xolutions.Neutral Media.-A vigorous campaign to determine the mechan-ism of the corrosion of metals in neutral aqueous solutions has beenmade during the last few years by Bengough and his collaborator^.^^The work has aimed at acquiring quantitative data for the extentand nature of corrosion under different conditions, and has beenconcerned especially with the corrosioii of zinc and mild steel inneutral salt solutions, such as potassium chloride, over a wide rangeof concentration. Special attention has been devoted to repro-ducibility of the results. The influence of various factors, such asconcentration, depth of immersion of the specimen, convection,access of oxygen, purity and treatment of the metal, and natureof its surface, has been studied, and it has been possible in manycases to ascertain which is the controlling factor in a given set ofcircumstances.Two types of corrosion have been recognised,corresponding with evolution of hydrogen and absorption of oxygen.Concurrently with the kinetic measurements, the course anddistribution of corrosion have been followed by microscopicalobservation. It is held that certain of the results cannot beaccounted for by the differential aeration theory. The initialbehaviour of zinc, when placed in a salt solution which does notform a passivating film, is to displace hydrogen a t numerous points.The possibility of the production of gaseous hydrogen being neglected,a polarising layer is formed a t the metal surface.The oxygenpresent is required for depolarisation; if the anions present form asoluble salt with the metal and are plentiful, corrosion proceeds a ta rate which is directly proportional to the oxygen supply.When zinc corrodes in solutions of potassium sulphate or chloridein presence of oxygen, the film of zinc oxide formed is in partscontinuous; impermeable to oxygen, and closely adherent to themetal, and in parts in the form of " domes," which are loose andpermeable to oxygen. Corrosion is most serious under the" domes." The authors point out that, if the protective type ofhydroxide film is formed over a part of the metal, corrosion can88 Proc. Physical Soc., 1934, 46, 148 ; p., 352.39 G. D. Bengough,,J. M. Stuart, and A. R. Lee, Proc.Roy. Soc., 1927, [ A ] ,116, 451; A,, 1928, 250; ibid., 1928, [ A ] , 121, 88; A., 1928, 1333; ibid., 1930,[A], 127, 42; A., 1930, 712; G. D. Bengough, A. R. Lee, and F. Wormwell,ibid., 1931, [A], 131, 494; A., 1931, 691; ibid., 1931, [A], 134, 308; A., 1932,27 ; G. D. Bengough and F. Wormwell, ibid., 1933, [A], 140,399; A., 1933,679136 INORUANIC CHEMISTRY.only occur elsewhere ; they maintain that the preferential corrosionunder the " domes " is not due to potential differences set up bydifference of oxygen concentration, but to the fact that ions canfreely enter solution there. Other portions of the metal surfaceare prevented from undergoing local dissolution by the protectivefilm, which, however, allows depolarisation of hydrogen displacedby metal entering the solution elsewhere.According to the " film-distribution " theory advanced, dissolvedoxygen is not really an inhibitor of corrosion, but a stimulator ; itmay act as a local inhibitor owing to the formation of secondaryproducts, which prevent access of oxygen to the underlying metaland the entrance of ions into solution, but corrosion is propor-tionally increased elsewhere.Thus, corrosion distribution isdetermined mainly by the distribution of protective films, whichcause the metal to be locally cathodic to bare, or less completelyprotected, metal. When the films are widespread, corrosion maybe sufficiently localised to be called " pitting." Among the factorsinfluencing the character, distribution, and continued adherence ofthe films are distribution of alkali, surface tension, presence ofspecially reactive areas, gravity, movement of liquid, alternate wetand dry conditions, the presence of foreign substances, and thenature of the solution.Simultaneously, Evans and his collaborators 40 have continuedresearches the results of which support the differential aerationtheory, some of the systems investigated being similar to thoseexamined by Bengough and his colleagues.It has been shownthat, under conditions favouring the complete tapping of theelectric currents flowing between the anodic and cathodic portionsof the corroding metal, the currents measured are equivalent to thecorrosion occurring, both when the anodes and cathodes are ofdifferent metals and when the anodic and cathodic areas of the samemetal are determined by differences in oxygen concentration.Theproblem of corrosion velocity is thus resolved into a study of theelectrochemical factors which determine the strength of the current.It is pointed out that the experiments of Bengough and his co-workers have been carried out under conditions which reduce thepossibility of differential %ration to the minimum.The objections of the Bengough school to the differential-aerationtheory are further answered by C . W. Borgmann and U. R. Evans 4 1in a paper describing work on the corrosion of zinc in chloride solu-40 U. R. Evans, L. C. Bannister, and S . C . Britton, Proc. Roy. Soc., 1931,[A], 131, 355; A., 1931, 691; U. R. Evans and T.P. Hoar, ibid., 1932, [ A ] ,137, 343; A,, 1932, 1003.4 1 Trans. Amr. Electrochem. SOC., 1934, 65, 249EEDGTES: THE COR,ROSION OF METALS. 137tions. There, it is shown that different results are obtained accord-ingly as the specimens are partly or wholly immersed in the solution.The apparent discrepancy between the results obtained with half-immersed sheet zinc and Bengough’s results with totally immersedzinc is traced to the avoidance of oxygen starvation in the partly-immersed specimens.U. R. Evans and R. B. M e a r ~ , ~ ~ replying to further criticisms byBengough, agree that the film-distribution theory and the differential-=ration theory have much in common, which is concealed by themode of presentation; in particular, the views agree in ascribingprotection largely to the cathodically-formed alkali.Differences ofopinion exist as to the effect of solid corrosion products on oxygentransport, and the possible inhibition of corrosion by excess ofoxygen. Further work in support of both the theories of differentialaeration and film distribution will be awaited with great interest;to many who are following the work it may appear that the twoschools of thought are by no means irreconcilable.In a study of the effect of non-metallic inclusions on the corrosionof mild steels, C. E. Homer 43 has shown that under mildly corrosiveconditions (e.g., in tap water, distilled water, etc.) sulphide andscale inclusions determine the initial points of attack. Only a smallproportion of such inclusions has any effect, however, and inclusionsof silicates or alumina are inactive. The action appears to beconnected closely with the breakdown of the protective oxide film.No evidence was obtained that either sulphide or scale inclusionsact as cathodes in the corrosion process.Under conditions favouringpitting, the initial breakdown may determine the sites of pitting.Similar conclusions have been reached by L. Tronstad and J.Sejer~ted,~4 who have investigated the effect of sulphur andphosphorus on the corrosion of iron.J. N. Friend and W. West45 have shown that the addition ofcopper up to 3.70% increases the resistance of nickel steel to alternatewet and dry sea action. F. Todt 46 has studied the influence ofoxidising and reducing agents on the corrosion of iron in bufferedsolutions.The corrosion of tin, tin-antimony , and tin-antimony-copper alloys in various tap-waters has been described byT. P.42 Proc. Roy. SOC., 1034, [A], 146, 153; A., 1181.43 Iron and Steel Inst., Cmnegie Schol. Memoirs, 1932, 21, 35; Special44 J . Iron and Steel Inst., 1933, 127, 425.4 5 Ibid., 1931,123, 501; B., 1931, 637.46 2. Elektrochem., 1934, 40, 536; A., 073.4 7 J . Inst. Metals, 1934, 55,135.Report No. 5, 1934, 225138 INORGANIC CHEMISTRY.The theory of the corrosion of metals has been discussed 48 andsome new views have been put forward. E. S. Hedges49 has out-lined a, theory involving primary reaction of the metal withwater molecules, the extent of the corrosion depending on the degreeto which the initially formed hydroxide film is dissolved or peptisedby the solution, and almost simultaneously A.L. McAulay andE. C. R. Spooner 50 have advanced the view that electrode potentialmust originate in interaction between the metal and water only.U. R. Evans and T. P. Hoar 5l have discussed the mechanism ofcorrosion in the light of both cathodic and anodic processes, andhave emphasised the important r61e of the relative solubility of thereaction products.The Dissohtion of Metals in Acids.-In the course of work on thedissolution of sodium amalgam in weak acid solutions, J. N.Bronsted and N. L. R. Kane 52 discuss the wider aspects and concludethat the dissolution of a pure metal in an acid is probably the resultof a chemical reaction between an electron of the metal and amolecule of the acid.G.Tammann and F. Neubert 53 have found that the rate of evolu-tion of hydrogen from dilute acids when acted on by zinc, iron, oraluminium can be expressed by v = a t + b t2, where v is thevolume liberated a t the time t . The constant a is characteristicof the metal and is unaffected by impurities, whilst the constant brepresents the accelerating influence of heterophase impuritiesforming local electrolytic cells.The velocity of the dissolution of zinc in acids has been thesubject of several papers. E. Miiller and J. Forster 54 have studiedparticularly the influence of the concentration of the acid and thenature of the anion. M. Centnerszwer and M. Straumanis 55 haveshown that electrolytic zinc dissolves much more slowly in dilutesulphuric acid than in hydrochloric acid of the same hydrogen-ionconcentration.The velocity of dissolution is related to the con-centration (C) of hydrochloric acid (up to 2N) by the linear equationdv/dt = k(C - Co), where Co is the threshold concentration of acidat which dissolution begins, and k is a constant. The value of thetemperature coefficient and the influence of stirring confirm that4a M. Stremanis, Korrosion u. Metallschutz, 1933, 9, 1, 29; 0. P. Watts,Trans. Arner. Electrochem. SOC., 1933, 64, 219; A., 1933, 1122.49 “ Protmtive Films on Metals,” London, 1932, p. 165.bo Proc. Roy. Soc., 1932, [ A ] , 138, 494; A., 1933, 28.51 Tram. Faraday SOC., 1934, 30, 424; A., 606.5a J .A m r . Chem. SOC., 1931, 53, 3624; A., 1931, 1373.sa 2. anorg. Chem., 1931, 201, 225; A., 1932, 128.b4 2. Elektrochem., 1932, 38, 901; A., 1933, 130.5 5 2. physikal. Chem., 1931, [ A ] , 167, 421; A., 370HEDGES : THE CORROSION OF METALS. 139the velocity of this reaction is controlled by the chemical process,not by the diffusion of the acid. The dissolution is preceded byan induction period, which can be eliminated by previously rubbingthe zinc with emery paper.56 The influence of various addenda,especially hydrophilic colloids, on the rate of dissolution of zincand iron in hydrochloric acid has been investigated by M. Schun-bert.57 0. Bauer and P. Zunker 58 have studied the influence ofsmall quantities of alloying elements on the rate of dissolution ofzinc in hydrochloric acid, and shown that the results are not neces-sarily parallel with those obtained in neutral salt solutions.L.Whitby 59 has found that, whilst the rate of dissolution of99-90 yo magnesium in 0-05N-hydrochloric acid is independent ofthe impurities in the metal, large variations are observed fordifferent samples in sodium chloride; these are traced to the effectof films formed at the cathodic parts of the surface, whilst in theacid solutions film formation is prevented. The rate of dissolutionof magnesium in dilute acids has also been studied by M. Elpatrickand J. H. Rushton,Go who direct attention to the part played bywater. R. Muller 61 concludes, from measurements of the velocitycoefficient of the dissolution of aluminium with hydrochloric acid,that the reaction occurs with HCl-H,O complexes.The Corrosion of Tinp7ate.-The use of tinplate as a food containerhas encouraged considerable activity during recent years in theinvestigation of the conditions of corrosion of tin, iron, tinplate,and the tin-iron couple in various aqueous solutions, notably thoseof weak acids.The existence of discontinuities in the tin coatingof tinplate might be expected to lead to serious local corrosion a tthe exposed steel. Experiments 62 have shown, however, thatunder certain conditions the potential of the tin-iron couple isreversed, tin becoming anodic to iron. Under these conditions theattack is not localised at small areas, and the tinplate gives moreuseful service.Different explanations of this phenomenon havebeen advanced, the reversal of potential being ascribed to filmformation on one or both of the metals, or to the high hydrogenoverpotential of tin. T. P. Hoar 63 states that tin is anodic to iron56 2. physikal. Chem., 1934, [ A ] , 167, 421 ; A., 370.67 Ibid., 1933, [A], 157, 19.59 Trans. Paraday Soc., 1933, 29, 415, 853 ; A., 1933, 233, 1017.60 J . Phys~cuZ Chem., 1934, 38, 269; A., 605.61 2. Elektrochem., 1934,40, 126; A., 605.62 C. L. Mantel1 and W. G. King, Trans. A ~ W . Electrochem. Soc., 1927, 51,40; B., 1927, 632; R. H. Lueck and H. T. Blair, ibid., 1928, 54, 257; B.,1928, 819; E. F. Kohman and N. H. Sanborn, Ind. Eng. Chem., 1928, 20, 76,1373; B., 1928, 159.68 2.Metallk., 1933, 25, 282.63 Trans. Furaduy Soc., 1934,30, 472; A., 735140 INORGANIC CHEMISTRY.in citric and oxalic acid solutions, owing to the removal of tin ionsas complexes, whilst in dilute sulphuric acid, where such complexesare not formed, tin remains cathodic to iron.An improved method of detecting discontinuities in the tincoating of tinplate, by observing the sites of rust formation whenthe carefully cleaned specimen is immersed for some hours in hotdistilled water within a certain p H range, has been described byD. J. Macnaughtan, S. G. Clarke, and J. C. P r y t h e r ~ h . ~ ~T. N. Morris and J. M. Bryan 65 have undertaken a systematicinvestigation of the corrosion of tinplate, studying in particularthe effect of a typical fruit acid (e.g., citric acid) in dilute solutionon (a) the steel base of tinplate, ( b ) tin, (c) the tin-iron couple,( d ) tinplate.The investigation has also included a study of theinfluence of pH, presence of oxygen, and the corrosion products onthe further corrosion of these materials; e.g., the influence of tinsalts on the corrosion of iron, and of iron salts on the corrosion oftin and iron in the presence and in the absence of air.The corrosion of tinplate in various dilute acids has been investi-gated by G. Gire,66 who has also emphasised the important r61e ofthe presence of oxygen.Atmospheric Corrosion.In the field of atmospheric corrosion the need of well-planned,comprehensive researches, extending over a period of many years,has been realised, and work is in progress in many parts of the world.Up to the present, insufficient time has elapsed to enable the finalresults to be anticipated. The earlier work of W.H. J. Vernonfor the Atmospheric Corrosion Research Committee of the BritishNon-Ferrous Metals Research Association has already been reportedupon.67A Joint Corrosion Committee of the Iron and Steel Institute andthe National Federation of Iron and Steel Manufacturers began in1928 a comprehensive series of field tests under well-defined con-ditions and over prolonged periods. A special feature of the workis that all the materials tested are of known origin and have usuallybeen manufactured in the presence of members of the Committee;full particulars have been recorded regarding their manufacture,casting, and rolling, and each specimen can be traced back to itsexact position in the ingot.The investigations cover the effects64 J. Iron and Steel Inst., 1932, 25, 159; B., 1932, 606.65 D.S.I.R., Food Investigation Bd. Special Report, No. 40, 1931 ; B., 1931,591 ; Report of the Director of Food Investigation, Section E, 1932 and 1933 ;Trans. Paraday SOC., 1933, 29, 395, 830; 1934,30, 1059.s 6 Rev. Trav. Ofice P&ches Maritimes, 1930, 3, 409; 1931, 4, 355; 1933, 6,305. 67 Ann. Reports, 1928, 25, 29HEDGES : THE C0R;ROSION OF METALS. 141of (1) surface condition, (2) the presence of rolling scale and of vari-ations in the type of scale, (3) the nature of the basis metal andmethod of preparation of the surface on the protection afforded bycoatings of paint, (4) copper content on the corrosion of mild steel,and also (5) comparisons of ingot iron and several types of wroughtiron and (6) tests on high-tensile steels.The tests are being carried out under widely different climaticconditions a t four main stations in Great Britain and ten subsidiarystations in Great Britain and abroad. Marine, moorland, highlyindustrial, and urban atmospheres are comprised, as well as testsin railway tunnels, etc., and under arctic and tropical conditions.Comparative tests under service conditions are being conducted indifferent types of atmosphere on railway sleepers of the same steelwith and without addition of copper. The specimens have beenweighed and will be removed periodically in batches in order toenable corrosion-time curves to be constructed.A Sub-Committee on Laboratory Corrosion Tests has also beenformed to develop laboratory tests to give a true index of servicebehaviour. An improved, automatic, spray test has been recom-mended. Further work has been done on corrosion testing byexamining the alteration of mechanical properties.When the final results of this comprehensive research programmeare available, it may confidently be expected that the knowledgeof certain aspects of atmospheric corrosion will be considerablyadvanced.Simultaneously, the Committees on Corrosion of Iron and Steeland on Corrosion of Non-Ferrous Metals and Alloys of the AmericanSociety for Testing Materials have continued and extended theirwork on atmospheric corrosion, which has been in progress formany years. Some of the more recent results have been set forthand discussed in a Symposium on the Outdoor Weathering of Metalsand Metallic Coatings.69 This discussion is of a practical natureand is intended to illustrate certain proper uses to which the testdata may be put by engineers.Csntinuing his work for the Corrosion of Metals Research Com-mittee of the Department of Scientific and Industrial Research,W. H. J. Vernon 70 has studied the corrosion of copper in certainsynthetic atmospheres in the laboratory, with special reference tothe influence of sulphur dioxide in air a t different relative humidities.He has described an air thermostat, suitable for use in work of thisIn the meantime two reports have been published.68Iron and Steel Inst., Special Report No. 1, 1931 ; Specid Report No. 6,1934; discussion, J . Iron and Steel Inst., 1934,129, 357.6s Amer. SOC. Testing Materials, 1934, Pre-print.70 Trans. Paraday Soc., 1931, 27, 255; B., 1931, 7631 42 INORGANIC CHEMISTRY.kind,71 and a method has been worked out 72 for the preparationof atmospheres of any desired relative humidity. The rale of themoisture in iron rust in determining the critical corrosion humidityis explained in colloid-chemical terms by W. S. Patterson andL. H e b b ~ . ~ ~ P. R. Kosting 74 has conducted accelerated weatheringtests on soldered and tinned-sheet copper in atmospheres rich insulphur dioxide and carbon dioxide.A study by W. H. J. Vernon and L. Whitby 75 of the greenpatina which forms on the surface of copper exposed to merenttypes of atmosphere, has shown- that, contrary to the former belief,the patina consists mainly of basic copper sulphate. In the courseof time its composition accords with that of the mineral brochantite,CuS04,3Cu( OH),. In seaside districts the patina may containcopper chloride, but the basic sulphate predominates when urbanand marine conditions coincide. The sulphate is obtained fromsulphurous and sulphuric acids brought into the atmosphere throughthe combustion of coal. There seems to be no doubt that thelongevity of copper exposed to the atmosphere is due to the pro-tective properties of the film of basic sulphate. Methods havebeen worked therefore, for the rapid production of an artificialpatina of this substance by anodic treatment of the copper in asuitable electrolyte. Other artificial methods have been describedby J. R. Freeman and P. H. Kirby.77L. Whitby 78 has shown that magnesium carbonate predominatesin the corrosion product of magnesium in indoor or outdooratmospheres. No indication of the formation of a protective filmwas observed. The rate of corrosion increases with the relativehumidity of the atmosphere.A reflectivity method for measuring the tarnishing of highly-polished metals has been described by L. Kenworthy and J. M.Waldram,79 and some results obtained by this method for tin andBritannia metal are given. E. S. H.E. S. HEDGES.R. WHYTLAW-GRAY.w. WARDLAW.7 1 Trans. Faraday Soc., 1931, 27, 241; A., 1931, 815.72 W. H. J. Vernon and L. Whitby, ibid., p. 1; A,, 1931, 816.73 Ibid., p. 278; B., 1931, 762.74 Bur. Stand. J . Res., 1932, 8, 365; B., 1932, 553.75 J . Inst. Metals, 1929, 42, 181; B., 1929, 855; ibid., 1930, 44, 389; B.,1930, 992; W. H. J. Vernon, ibid., 1933,52,93; B., 1933,633; J., 1934, 1853.76 Idem, J . Inst. Metals, 1932, 49, 153; B., 1932, 940.7 7 Metals and Alloys, 1932, 3, 190; B., 1932, 986; ibid., 1934, 5, 67.78 Trans. Farachy SOC., 1933, 29, 844; A., 1933, 1017.79 J . Inst. Metals, 1934, 55, 247