CRY STALLOCRAPHYTHE change from biennial to annual publication of the report onCrystallography occurs a,ppropriately at a time when attention hasbeen focused on the subject by Sir William Bragg’s presidentialaddress to the Royal Society summarising many of the recentadvances in crystal analysis. The past history of structure deter-mination has also been brought nearly up to date this year by theappearance of the long-awaited second volume and part of the thirdvolume of the “ Strukturbericht,” the first volume of which hasmeanwhile become a classic. The main lines of Vol. I have beenfollowed, and the results so far available include all inorganicstructures, but not alloys, organic compounds, and fibres.As it is impossible t o cover all phases of crystallography adequatelyin this survey, some subjects which appear to be more suitable forbiennial treatment have been left over for consideration in laterReports. This applies particularly to metals, which have beendiscussed at length in the two preceding Reports.1.THE TECHNIQUE OP STRUCTURE ANALYSIS.For all kinds of crystal structure work, more powerful X-ray tubesare highly desirable, and a nokble advance in the design of rotatinganode tubes has now been made by V. Linnitzki and V. G~rski.~They have combined the anode with a molecular pump so that therotation of the anticathode, which formerly introduced manydifficulties in construction and maintenance, is now turned topositive advantage. As far as the recording of the X-ray reflectionsis concerned, there are three main types of instrument that can beconsidered.G;. Kellstrom’s 4 new value for the viscosity of air maybe taken to show, inter aEia, the reliability of the X-ray method forthe measurement of e, and there is therefore the more justificationfor continued work on precision measurements of lattice spacings,1 Proc. Roy. Soc., 1936, A , 157, 697.2 Akademische Verlagsgesellschaft m.b.H., Leipzig, 1936 : Vol. I1 (1928-32) by C. Hermann, 0. Lohrmann, and H. Philip; Vol. I11 (1932-35) byC. Gottfried and F. Schossberger.8 Tech. Php. U.S.S.R., 1936, 3, 220.4 PhysicaE Rev., 1936, [S], 50, 190COX AND CROWFOOT: TECIENIQU’E OF STRUCTURE ANALYSIS. 197particularly by powder method^.^ M. 5. Buerger’s new Weissenberggoniorneter is likely to prove useful for the more general examin-ation of a crystalline species, and other cameras have also beendevisedy7 including some for work at high and low temperatures *and low pressure^.^ It does not appear to be generally recognised,however, that many of the advantages of a vacuum can be attainedmore simply by filling the X-ray camera with hydrogen.Finally,for the exact measurement of intensities a new automatic ionisationspectrometer has been designed by W. A. Wooster and A. J. P.Martin.lo The use of electrometer triodes (in conjunction with anionisation chamber) is well established ; a further advance whiehbas been proposed l1 is the replacement of the ionisation chamberor photographic plate by a Geiger-nlluller counter.The introduction of the Patterson method of synthesis was de-scribed in last year’s Report and it is already taking a definite placeits a first stage in the determination of many crystal structures.One example of its use is that of silver uranyl acetate.12 D.Harker 13has pointed out that improvements may be introduced by makingfull use of the symmetry of the crystal under investigation. In thegeneral case the Patterson series is of the formP(xyz) = ChC&JIF(hlcZ)J2 cos (hx + Icy + Z X ) . . . . (1)the maxima in the function P representing interatomic distances.To be manageable, however, this must be reduced t o a two-dimensional series&(xx) = C ~ C ~ ] l ? ( h O Z ) ~ ~ cos (hx + Z X ) . . . . . . . (3in which, since the P’s of one zone only axe used, the resolutionof the peaks is not very good.If, however, the b axis is a two-fold axis of symmetry, two equivalent atoms have co-ordinates(xyx) and (~yx’) and the vector between them has components (2xy0,2x)5 3. W. M. du Mond and V. L. Bollmann, ibid., pp. 383, 524; W. Kossel,Ann. Physik, 1936, [v], 26, 533; M. Straumanis and A. Ievins, 2. Physik,1936, 98, 461 ; E. R. Jette and F. Foote, J. Chem. Physics, 1935,3,605; M. U,Cohen, 2. Krist., 1936, 94, 288, 306.6 Ibid., p. 87.7 M. J. Buerger, Amer. Min., 1936, 21, 11; E. Sauter, 2. Krist., 1936, 93,* R. L. McFarlan, Rev. Sci. Instr., 1936, [ii], 7, 82.9 E. Franke, 2. physikal. Chem., 1936, B, 31, 454.10 Proc. Roy. SOC., 1936, A , 155, 150.11 W. Van der Grinten and H.Brasseur, Nature, 1936, 137, 657; D. P.Ie Galley, Rev. Sci. Instr., 1935, 6, 279 ; M. Pahl and A. Faessler, 2. Physik,1936,102, 562.93; 0. Kratky and G. Krebs, ibid., 95, 253.12 I. Fankuchen, 2. Krist., 1936,94,212.13 J. Chem. Phpics, 1936, 4, 381198 URY STALLOQRAPHY.and is represented by a maximum value of P in the plane y = 0.Evidently, in such a case, the x and the y co-ordinates of all theatoms can be determined by measurements of P for y = 0 only;(1) then becomesP(x0x) = ChCl COS (hx + ZZ) . (&p(hkz)l2) . . . . , (3)a two-dimensional Fourier series which can readily be computed.This series (3) has two advantages over (2) : it gives greater reaolu-tion because all the 3”s are used, and it shows only interatomicvectors perpendicular to the symmetry axis, so the risk of over-lapping is reduced.Similar improvements can be introduced wherethe crystal has other elements of symmetry, and in one of thesecases the method has been applied by D. Harker to determine thecrystal structure of proustite and pyrargyrite.These methods render it necessary to carry out Fourier synthesesin the very first stages of the analysis instead of once at the end;but by means of calculating devices, such as those of J. M. Robert-son l4 and A. L. Patterson l5 and the sine and cosine strips nowmade available by C. A. Bcevers and H. Lipson,16 a two-dimensionalFourier synthesis can be carried out in little more than a day.Robertson l7 has also described simplifications of the calculationsnecessary in the case of non-centrosymmetrical projections andsucccssf6lly used them on the s6ructure of rcsorcinol.A furthervaluable contribution to the literature of crystallography is (‘ Struc-ture Factor Tables ” by (Mrs.) ]EL Lonsdale.ls In these, thenecessary data for Fourier synthesis and for the calculation of thestructure factors have been presented for each of the 230 spacegroups in the form in which they can most readily be applied inactual practice. The most laborious part of structure analysis(after the experimental data have been obtained) is still the calcu-lahion of structure factors, both to derive an approximate structureby ‘( trial and error ” and to determine the phases of the coefficientsin a Pourier series ; the labour involved can be considerably lightenedby the adoption of W.L. Bragg’s proposal l9 to’use contoured graphsfrom which phase factors can be read off directly when the co-ordinates of the atoms are given. Since it is usual to investigateonly special planes of the type (hEO), the total number of graphsl4 Phil. Mug., 1936, 21, [vii], 176.15 Ibid., 1936, 22, 753.lG Proc. PhysicccE SOC., 1936, 48, 772; Nuture, 1036, 157, 825.18 “ Simplified Structure Factor and Electron Density Formula for tho 230I* Natu.re, 1936, 1318, 382; W. L. Bragg and H. Lipson, 2. Kri.Pt., 1937,95,Ibid., 138, 683.Space Groups of Mathematical Crystallography,” Bell and Sons, 1936.383COX AND CROWFOOT : TECHNIQUE OB STRUCTURE ANALYSIS. 199required is not very great, about twelve (for all values of h and kup to h + k = 8, say) for each of the seventeen plane groups beingsufficient;.Fig. 1 shows the graph for the plane (230) in any spacegroup of the tetragonal classes &2m, 4mm, 42, and 41mmm. Thecontour8 give the value of the functionS = cos 4-nx . cos 6x2~ +- cos 6nx . cos 47r~for all values of x and y between 0 and 1; the heavy lines arecontours with S = 0, and negative contours are dotted. ThusFIG. 1.0 X. 1.0Contours of S = cos 47rx. cos Giry + cos 6nx . COB 4ny.the quantity S (multiplied by a power of 2 according to the spacegroup) read off for the co-ordinates (x, 9) is the contribution of anatom in the general position (xyz) and of all the atoms in the cellrelated to it by the symmetry elements of the space group to the phasefactor of that particular plane.The contribution to the structurefactor is then simply ZnSfe. One very valuable feature of the methodis that it is possible to determine by inspection of the appropriategraph how the co-ordinates of an atom or atoms must be altered toobtain a desired change in the phase factor for any plane; thisshortens considerably the time necessary for trial and error analyses200 CRYSTALLOURAPHY.A complete set of graphs is being prepared and subjected to practicaltests in various laboratories.Several workers 2o have directed their attention to the determin-ation of the exact form of thef-curves for various atoms, vix., nickel,copper, zinc, cadmium, aluminium, silver, palladium, and sulphur.In the case of the hexagonal metals, cadmium and zinc, the atomicscattering factor has a low or high valuc according as the plane fromwhich reflection occurs is nearly parallel or perpendicular to thebasal.plane. This is interpreted by G . IV. Brindiey 21 as being dueto asymmetry of lattice vibrstions-a view which, though contestedby H. Hermann,22 receives support from the work of C . Zener.23Zener has investigahed the dependence of the Debye-Waller temper-ature coefficient e-M upon reflection plane orientation for the caseof metals of hexagonal symmetry, and finds tha6 the ratio of M forvibrations perpendicular to and parallel t o the c-axis is 1430 and1-73 for zinc and cadmium respectively. Other work on zinc,24while showing that the anomalous f-values are chiefly due to theanisotropy in the thermal vibrations, suggests that Hermann’s viewthat the atoms themselves are anisotropic may be partly correct.Other supposed cases of anisotropic thermal vibrations are men-tioned elsewhere in the Report.A critical test of atomic symmetrymight be possible through the use of polarised X-rays as developedby W. H. George.25Optical and magnetic methods have now definitely establishedthemselves as aids to structure determination, particularly in thefield of aromatic compounds, and in favourable cases it appearspossible t o fix the direction cosines of the plane of the molecules inthe lattice t o within lo. Somewhat uncritical use has, however,sometimes been made of magnetic data, and a detailed discussionby (Mrs.) K.Eonsdale and K. S. Krishnan 26 of the precise relation-ships existing between molecular susceptibilities and those of thecrystal as a whole is welcome. L. Pauling 27 has also discussed the4o J. Laval, Compt. rend., 1935, 200, 1605; E. Niihring, 2. Physilc, 1935, 93,197; C. M. Kotin and T. Losada, Anal. Pis. Quim., 1935, 33, 597; P. de laCierva and J. Palacios, ibid., p. 34; G. W. Brindley, Phil. Mag., 1936, [vii],21, 778: G. W. Brindley and F. W. Spiers, ibid., 20, 865.21 Proc. Leeds Phil. Soc., 1936, 3, 200; Phil. Mag., 1936, [vii], 21, 790;Nature, 1936, 138, 290.22 Ibid., p. 290.23 Physical Rev., 1938, [ii], 49, 122; C. Zener and 8. Bilinsky, ibid., 1936, 50,24 G. E. M. Jauncey and W. A. Bruco, ibid., pp.408, 413; R. D. Miller and2s Proc. Roy. Soc., 1936, A , 156, 96.26 Ibid., p. 597.27 J. Chem. Physics, 1936, 4, 673.489; see also idem, ibid., p. 101.E. S. Foster, ibid., p. 417COX AND CROWFOOT : TECHNIQUE OF STRUCTURE ANALYSIS. 201diamagnetism of aromatic molecules, and the relation betweenoptical anisotropy and structure has been treated by M. Ramanad-ham28 and K. S. S~ndararajan.~~ Although it is unlikely, in viewof the greater accessibility of other properties, that thermal con-ductivity will be used as an aid to structure determination, yet it isof great importance that the relationship of this property to crystalstructure should be understood; first step in this direction hasbeen made by W. A. Wooster,m who has collected the availabledata and attempted a correlation of thermal anisotropy withstructure.The possibilities and advantages of orienting moleculesor particles of markedly anisotropic form by streaming or similarmethods have long been realised ; an interesting example is affordedby the orientation of tobacco mosaic virus “molecules ” in quitedilute solutions by flow through a Lindemann glass capillary tube,31and a new procedure, vix., sedimentation from an aqueous solutionin an alternating electric field, has been applied successfully inobtaining highly oriented preparations of wool ~ells.~2 Suchmethods are capable of extension to many other imperfectlycrystalline substances.Two other applications of X-rays not concerned with structureanalysis may be mentioned.One is the determination of particlesize, which has been carried out particularly on graphite33 fromdifferent sources and also applied to the colloid chemical behaviourof vanadium pentoxide 34 and gold By measurements oncounted layers of fatty acid films on water, G. L. Clark and P. IN.Leppla 36 have been able to obtain a direct test of the Laue equationconnecting the broadening of X-ray lines with film thickness. Theagreement is very satisfactory down to distances corresponding toonly three or four fatty acid layers.Several papers have dealt with the systematic application ofmorphological crystallography to the identification of chemicalindi~iduals.~’ X-Ray methods provide both a simpler and a more28 Proc. Indian Acad. Sci., 1936, 3, A , 43.2D 2.Krist., 1936, 93, 238.31 J. D. Bernal and I. Fankuchen, Nature, 1936,138, 1051.33 H. J. Woods, Proc. Leeds Phil. Soc., 1935-6, 3, 132.33 U. Hofmann, D. Wilm, and E. Csalhn, 2. Elelctrochem., 1936, 42, 504;G. R. Levi and A. Baroni, 2. Krist., 1936, 93, 156; P. Corrier, Compt. rend.,1936, 202, 59; N. Ganguli, Phil. Mag., 1936, [vii], 21, 355; cf. G. I. Finch andH. Wilman, Nature, 1936, 137, 271; Proc. Roy. Xoc., 1936, A , 155, 345.30 Ibid., 95, 138.34 J. A. A. Ketelaar, Nature, 1936,137, 317.35 J. B. Haley, K. Soltner, and H. Terrey, Trans. Paraday Soc., 1936,32,1304.36 J . Amer. Chem. Soc., 1936,58, 2199.37 A. K. Boldirev and V. V. Dolivo-Dubrovolaki, 2. Krkt., 1936, 93, 321;C. Weygand, Angew. Chon., 1936, 49, 243; B. N. Delone, Ann.Sec. Anal,Phys. (;him., 1936, 8, 92 ; A. F. Kapustinski, ibid., p. 103202 CRYSTALLOGRAPHY.delicate means, and several attempts have been made to classifycertain groups of compounds by means o€ powder photographs.One such “ finger-print system ,” developed mainly for industrialincludes 4000 patterns from over 1800 inorganic sub-stances, and there are several examples of its use.39 Among organiccompounds, only the method of direct, ad hoc comparison of twocompounds suspected of being the same has been employed to anyextent, as in the identification of tetrahydroartimisia ketone,4O thebarium saltis of the pterins2l the hydrocarbon ‘f CZ1Hl6 ” from cholicncid,42 the phrenosiiiic acids,43 and 0thers.4~E. G. C.D. M. C.2. CRYSTAL CHEMISTRY.Metals.-Precision measurements of the lattice constants ofberylli~m,~~ cadmium, osmium, and ruthenium 47and of tantalum and vanadium 48 have been made.In addition t ohis usual annual survey 49 of lattice constants and other propertiesof the elements, M. C. Neuburger 50 has published “ Die Allotropieder chemischen Elemente uiid die Ergebnisse der Rontgenographie.”This monograph contains a critical discussion of the allotropy of allthe chemical elements and is notable for a list of over 1000 references.During the past year new results have accumulated regarding theallotropy of titanium, calcium, and boron. Titanium, like zir-conium, might be expected to occur in an a- and a p-form, hFxagonalclose-packed and cubic body-centred respectiveIy ; no convincingevidence for this had previously been obtained, however, the88 J.D. Hanawalt and H. W. Rinn, Ind. Eng. C‘hem. (Alzcal.), 1936, 8, 244.39 A. TV. Waldo, Amer. Min., 1935, 20, 575; J. N. Antipov-Karataev and40 L. Ruzicka, T. Reichstein, and R. Pulver, Helv. Ghim. Acta, 1936, 19,4 1 C. Schbpf and E. Becker, Annalen, 1936, 524, 49.4% W. E. Bachmenn, J. W. Cook, C. L. Hewitt, and J. Iball, J., 1936, 54.43 A. C. Chibnall, S. H. Piper, and E. F. Williams, Biochem. J., 1936, 30,100 ; (Miss) D. M. Crowfoot, J., 1936, 716.44 H. J. Backer, J. Strating, and A. J. Zuithoff, Rec. tmv. chim., 1936, 55,761 ; M. P. Wolarowitsch, G. 13. Rawihch, andK. F. Gussjew, KoZZoid-Z., 1936,76, 338.B. K. Brunowski, Kolloid-Z., 1936, 75, 325.646.45 A.Ievins and M. Straumanis, 2. physilcal. Chem., 1936, B, 33, 265.46 E. A. Owen and T. L. Richards, PhLiZ. Mag., 1936, [vii], 22, 304.47 E. A. Owen and E. W. Roberts, {bid., p. 290.48 M. C. Neuburger, %. Krist., 1936, 93, 312, 314.4D Ibid., p. 1. ,60 Sammlung chemischei- und chemische techniacher Vortriige, F. Erlke,S tuttgar t COX : CRYSTAL CHEMSTRY. 203irregularities in the resistance-temperaturo curve for $he metalbeing attributed to impurity taken up at high temperatures. J. H.de Boer, W. G. Burgers, and J. D. Fast 51 have now shown thatimpurities are indeed taken up, but that the effect of these is to maskthe transition which occurs (quite sharply in the absence of air) a tnearly the same temperature as for zirconium ($82’).X-Rayexamination confirms that @-titanium has an A2 structure witha = 3-32 A.Calcium is known t o occur in three forms, with transitions O! --+ pa t 300’ and p --+ y at 450°, the c(- and the y-structure being cubicand hexagonal close-packed respectively. The p-form was formerlyconsidered to be of lower symmetry, but tho measurements ofA. Schulze 52 indicate that it may possess an A2 lattice. Thediamond-like and the graphitic form of so-called crystalline boronformerly described have probably been aluminium boride or boro-carbide ; 8. von Naray-S~abo,~~ however, now reports an adamantinetetragonal boron with a = 12.55 and c = 10.18 A., and a graphiticor orthorhombic form with a = 17.64, b = 25.0, and c = 10.26 A.No outstanding advances have been made in the study of alloysor in the theory of metallic structures generally; a number ofbinary and ternary systems have been studied and further work hasbeen done on order-disorder transformations and related topics,but the results are deferred for more comprehensive discussion insubsequent reports.An excellent and very full account of thequantum thcory of metals, covering conduction, cohesion, magneticand optical properties, and crystal structure is given by N. F. Mottand H. Jones in “ The Theory of the Properties of Metals andAlloys,” 54 and W. Hume-Rothery 55 has written a most lucid reviewof the subject from a more descriptive point of view which shouldbe of the utmost value to chemists and metallurgists.0rides.-i. A. A. Ketelaar 56 has determined the structure ofvanadium pentoxide.The structure contains chains of oxygentetrahedra linked by shared corners, accounting for the formationof elongated micelles in solution.Structures of the spinel type have been the subject of severalinvestigations. A. E. van Arkel, E. J. W. T’erwey, and M. G. vanBruggen 57 have shown that various ferrites (AfO,Fe,O,) are able to5 1 ,€‘roc. K . Alzad. Wetensch. Amsterdam, 1936, 39, 515; ?V. G, Burgers andF. M. Jaeger, 2. Krist., 1936,94,299.53 Z. Metallk., 1936, 28, 55.53 Naturwiss., 1936, 24, 77.54 International Monographs on Physics, Clarendon Press, Oxford, 1936.55 “ The Structure of Metals and Alloys,” Institute of Metals, 1936,55 2. Krist., 1936, 95, 28.57 12ec. trav. chim., 1936, 55, 331, 340204 CRYSTALLOGRAPHY.dissolve excess Fe203 without change .of (spinel) structure.This isexplained by the stability of the anion framework, which is thesame for y-Fe,O, as for the ferrites, the former having on the average28 vacant cation positions per unit cell. I n a solid solution ofFe203 in ferrite the number of vacant positions diminishes, withconsequent stabilisation of the y-Fe203 structure. These workersshow that the maximum in the magnetisation curve correspondsto the maximum solubility; the dissolved Fe,03 takes up the ferro-magnetic y-lattice, but when the solubility limit is transcended, thesystem becomes a two-phase conglomerate containing non-magnetica-Fe203 so that the total magnetisation of the mixture is lowered.The coincidence of the solubility limit in certain cases with thecomposition 2M0,3Fe20, is probably accidental, and is apparentlynot due, as has been suggested,58 to the formation of a new type offerrite.Trimanganese tetroxide, Mn304, and ferric oxide, Fe203, form(above 800°) a continuous series of solid solutions from 100% to14% of the former with a gradual change of structure from thetetragonal (distorted spinel) structure of Mn,Q4 to ‘the true spinellattice; at lower temperatures solid solution still occurs but withthe structure of Rln203 (C-modification of the sesquioxides).Thesolubilities in ferric oxide of oxides with the sodium chloridestructure were also studied ; FeO-Fe,O, mixtures have only a verylimited range of homogeneity, but NiO and MgO appear to formsolutions over a wider range.Some fersites have rhombohedra1symmetry ; 59 they may possibly possess structures of the hEmatitetype. The dominating influence of the anion arrangement isillustrated by some observations of R. Mehl and E. L. McCandless 60on the orientation of oxide films on iron. Pe30, films formed bythe decomposition of ferrous oxide have identical orientation withthe parent crystal; the same relation holds when ferrous oxideis obtained on the surface of the magnetite crystals by reduction.suggest that, whereas Mn,O,and Co,O, have true spinel structures M1W$I04 (the former beingdistorted tetragonally on account of the lower symmetry of thehalf the tervalent cations occupying tetrahedral positions in thespinel structure, and the other half sharing the octahedral positionsa t random with the ferrous ions.Bond distances and X-ray5 8 R. S. Hilpert and R. Scheveinhagen, 2. physikal. Chem., 1935, B, 31, 1.59 W. Soller and A. J. Thompson, Physical Rev., 1935, [ii], 47, 644; A.Krause, A. Ernst, S. Gawrych, and W. Kocay, 8. anorg. Chem., 1936,228,353.60 Nature, 1936, 137, 702.61 Rec. trav. chim., 1936, 55, 531,E. J. W. Verwey and J. €I. de BoerMnIVion), yet Fe,04 is a solid solution J?eO,Fe,O, or FeIII(FeIIE”eII104COX : CRYSTAL CHEMISTRY. 205reflection intensities are considered to support these views, althoughit seems doubtful whether either can be relied upon to distinguishunequivocally between the very closely similar structures in question.With the proposed distribution of ions in magnetite, an electronpassing from Fe++ to Pe+++ is in the same energy state finally asinitially, and the probability OE a transition is therefore reasonablyhigh, in agreement with the high conductivity of Fe,O, as comparedwith Co,O, or Mn30,.It is of interest to enquire whether theseconsiderations can be applied quantitatively to other compounds,e.g., nickelous and ferrous oxides and ferrous, cobaltous, and cuproussulphides, in which there is normally a cation deficiency, and there-fore presumably a proportion of cations in higher valency states.Although experiments 62 suggest that there is a close relation be-tween cation deficiency and conductivity, the results for Pe304 andCo304 are very similar, and the conductivity of the latter is appar-ent,ly not so greatly inferior to that of magnetite as Verwey andde Boer have supposed; moreover, the necessity of assumingquadrivalent cobalt ions is a weakness in their hypothesis.Morequantitative measurements of conductivity (and also of magneticsusceptibility) in relation to composition, radius ratios, and otherfactors appear to be necessary before the structures of these oxidescan be regarded as fully elucidated.Oxides of tungsten [W,O,,, W,O,,(OH),, and W1,03,(OH)2] inwhich the metal appears to be present with a lower valemy thansix have been obtained by F. Ebert and 1%. Flasch 63; it is difficult,however, to exclude the possibility of the presence of hydrogen ionsin these compounds.Their cell dimensions show that they are veryclosely related to W 0,.It has been suggestedG4 that quartz assumes a new crystallineform below -183.6", but other workers 65 have not confirmed this.-Further work also appears to be necessary before the statementsthat the lattice of a-quartz is deformed to the extent of 2-30/, inagate and finely ground sand can be accepted. B. E. Warren and hisco-~orkers,67 continuing their studies of oxide glasses, have obtainedmuch more precise results by the application of a generalised Fourier62 C. Wagner, 2. tech. Physik, 1936, 16, 327; C. Wagner and E. Koch, 2.63 2. anorg. Chem., 1935, 226, 65.64 H. Osterberg, Physical Rew., 1936, [ii], 49, 552.65 H. Dobberstoin, Naturwiss., 1936, 24, 414; L.Balamuth, F. Rose, and65 N. A. Schischakow, Cornpt. rend. Acad. Sci. U.R.S.S., 1936,1,19.67 B. E. Warren and 0. Morningstar, Phy8icaE Rev., 1935, [C], 47, 808;B. E. Warren, H. Krutter, a i d 0. Morningstar, J . Amer. Cerarn. SOC., 1930,19,202.physikal. Ghem., 1936, B, 32, 439.S. L. Quimbg, Physical Rev., 1936, [ii], 49, 703206 CBYSTALLOGRAPHY.method; the existence of tetrahedral networks in silica withSi-0 = 1.60 A. is definitely confirmed, and vitreous boric oxide isshown to consist of a network in which each boron is surrounded bythree oxygens at 1.39 A. as in crystalline borates. In both casesthe oxygen atoms are shared between only two cations, thus con-ferring a considerable degree of flexibility upon the network, so thatthe irregular glass structure has a configuration almost as stable asthe crystalline arrangement.N. A. Schischakow 66 and N. Valenksfand E. Porai-Koschitz 68 have studied the transition from vitreoussilica to cristobalite.8iEicutes.Further work has been done on clay minerals; thestructures previously assigned to dickite and kaolinite have beenconfirmed by C. J. Ksanda and T. W. Barth 69 and by S. B. Hen-dricks 70 respectively, and the latter has also studied anausite.J. W. Gruner 71 has shown that glauconite has a mica-type ofstructure with a higher Si : A1 ratio ; some silicon may possiblyoccupy 6-co-ordinated cation positions. P. A. Bannister andM. H. Hey 72 have continued their studies of zeolites with measure-ments on scolecite, metascolecite, and ettringite ; the first is shownto be iso-structural with nairolite and its transition to meta-scolecite has been investigated.The cubic mineral pollucite(Cs2Si4Al,01,,H,0) appears to be related to the zeolites.73E. Podschus, V. Hofmann, and K. Leschewski 74 have made adetailed study of ultramarine-blue and several related substances.The highly symmetrical structure proposed for the ultramarinesby F. M. Jaeger,v5 although undoubtedly correct in essentials, wasnot entirely satisfactory from the point of view sf interatomicdistances or of X-ray reflection intensities. The structure nowdetermined is analogous to that of hauyne and nosean 76 and is lesssymmetrical and less rigid than that suggested by Jaeger, account-ing in a satisfactory manner for the variation in cell dimension whenthe sodium of ultramarine-blue is replaced by other alkalis.Leschewski and his co-workers consider it unnecessary to assumethat the sodium ions and sulphur atoms are mobile ; €or the sodium,they emphasisc that careful analyses never show more than eightions per unit cell, and these can be located definitely on cight three-Nature, 1936, 137, 273; cf.G. Peyronol, Z. Krist., 1936, 95, 274.Amer. Min., 1935, 20, 631.70 2. Krist., 1936, 95, 247.71 Amer. Mipz., 1935, 20, 699.72 Min. May., 1936, 24, 324; idern,ibid., p. 227.73 H. Strunz, 2. Krist., 1936,95, 1.74 2. anorg. Chem., 1936,228,305.76 F. Machatschki, Centr. Min., A, 1934, 5, 136.76 T m 8 . & b d t t Y SOC., 1929,=, 320COX : CRYSTAL CHEMISTRY. 207fold positions in the lattice.The sulphur is best accounted for (inagreement with the chemical evidence) by dividing it into (a)sulphur ions and ( b ) S, molecules, distributed statistically over thetwelve-fold positions; a sharp distinction between the two is notpossible. There is some evidence from these compounds that thethermal vibrations of the alkali ions may have the symmetry oftheir environment .7' The colour of ultramarine-blue is attributedto the S, groups.SuZphides.-That close crystallo-chemical relationships existbetween germanium and silicon is well known; 78 attention wasdrawn in last year's Report (p. 209) to the new type of fibrousstructure exhibited by SiS,, and the determination of the structureof GeS, by W.H. Zachariasen 79 is therefore of considerable interest.As in SiS,, each cation is surrounded by four sulphur atoms in aslightly distorted tetrahedron, the Ge-S distance being 2.19 A.(cf. 2.26 from sum of tetrahedral radii, and 2.47 in th& monosulphide),but the linking of tetrahedra is three-dimensional, giving rise to astructure which is much more like that of SiO, than t'hat of SiS,.The sulphur bond angle is 103".The high-temperature modifications of the cuprous and argentoussulphides, selenides, and tellurides have one of two structures,BOaccording as the catiorm/anion radius ratio is greater or less than 0.6.In the former category are a-Ag2S and a-Ag,Se, in which the silverions are distributed statistically in the interstices of a cubic body-centred arrangement of anions, whereas in the unit cells of a-Ag,Te,cr-Cii,Se, and a-Cu,S four anions and four metal ions form a zinc-blende structure, with the remaining cations distributed statistically.cc-Cuprous telluride is not cubic, and cuprous sulphide is deficientin copper, its formula being spproximately Cu,S,.Similarlycobaltous sulphide (nickel arsenide type) is said 81 to be stable onlywhen slightly deficient in cobalt; in both these cases it is note-worthy that the radius ratio is very near the limiting value for thcstructure concerned. Two other sulphides of cobalt, Co,S, andCoSSs, are reported; both are based on a cubic close-packing ofsulphur, and have similar cell dimensions. Photographs of Co,S,itre alrnos t identical with those of pentlandite,*2 suggesting theformula (Ni,Fe),S, for the latter, in better accord with the observeddensity.The transition from 7-NiS (millerite) to p-NiS (nickel7 7 See p. 200.78 E.g., W. Schiitz, Z. physikal. Cliertb., 1936, B, 31, 292.79 J . C h m . Physics, 1936, 4, 618.83 P. Rahlfs, 8. physikal. Chem., 1936, B, 31, 157.81 H. Hiilsmann, F. Weibke, and K. Meisel, 2. worg. Ckem., 1936,227, 113.8% M. Lindqvist, D. Lunclqvist, and A. Westgren, S v m b Kern. Tidskr., 1936,48, 156208 CRYSTALLOGRAPHY.arsenide type) has been studied by G. R. Levi and A. Baroiii,83and by W. Biltz et ~ 1 . 8 ~ ; the transition temperature is 396". W.Biltz and J. Laar 85 have confirmed the existence of Pd4S, Pd5S2,and PdS,, and a study of hauerite 86 (MnS,) has been made.M. J.Buerger 8' has discussed the arsenopyrite structure in detail, andshows how it may be regarded as a superstructure based on themarcasite type ; several new examples of this structure are recorded.D. Harker has applied his modified Patterson's method ofanalysis 88 to proustite, Ag3AsS3, and pyrargyrite, Ag,SbS,. Thetwo structures, which are almost identical, contain continuous(AgS), groups in the form of trigonal spirals, the bond angles forsulphur and silver being 83" and 165" respectively, while the lengthof the Ag-S bond is 2.40 A. Each arsenic (or antimony) is bonded tothree sulphur atoms in a flattened pyramid. W. V. Medlin 89 hasused B. E. Warren and E. Gingrich's method90 to find the radialdistribution of atoms in realgar, ASS, and orpiment, As,S,.Theresults, which are remarkably similar for the two compounds, areinteresting as showing the possibilities and limitations of resultsbased on visual estimation oE intensities in powder photographs ;for these two substances the first two or three peaks in the distribu-tion curve can be related to the appropriate atomic distances withoutdifficulty, but in other cases (e.g., CaHgBr,) interpretation is moredifficult.A study of berthierite, FeSb,S,, has been made.g1HaZides.-By means of a modified Debye-Scherrer methodM. Straumanis and A. Ievins 9, have determined the cell dimensionsof NaCl and rock-salt with, it is claimed, greater accuracy than anyyet attained. The value for rock-salt (5.6276, 0.00005 A.)differs by 0.00032 A. from that obtained by Siegbahn.Radium fluoride has the fluorite structure 93; the radius of theradium ion (C.N.= 6) is 1-62 A. On the other hand, thallousfluoride94 has a new type of (orthorhombic) structure which maybe regarded as a deformed sodium chloride lattice, and the iodide 95s3 2. Krist., 1936, 92, 210.84 2. anorg. Chem., 1936, 228, 275.86 W. Biltz and F. Wiechmann, ibid., p. 268.8 7 Z . Krist., 1936, 95, 114.a8 Seep. 197.Rg J . Amer. Chem. SOC., 1936,58, 1590.90 Physicd Rev., 1934, 46, 368.9 1 M. J. Buerger, Amer. Min., 1936, 21, 442.ga Z. Phyeik, 1936, 102, 353.aa G. E. R. Schulze, 2. physikal. Chern., 1936, B, 32, 430.g4 J. A. A. Ketelaar, 2. Krkt., 1935, 92, 30.g 5 L.Helmholz, ibid., 1936, 95, 129.8 5 Ibia., p. 257COX : CRYSTAL CHEMISTRY. 209is also orthorhombic with a semi-layer lattice in which each ionhas five near neighbours.G. Wagner and L. Lippert 96 have studied transformationsbetween the sodium chloride and the cmium chloride lattice;rubidium chloride when condensed from vapour on a surface a t-190O has the latter type, reverting to the normal former type atroom temperature. A thermodynamic study of the transformationof ammonium bromide at -39" has been made,97 and J. Weigle andH. Sainigs have shown how its tetragonal structure below thistemperature is obtained by slight deformation of the normal cubicstructure. The energy relations of super-lattices in mixed crystalshave been discussed by H.O'Daniel99 with special reference toalkali halides ; he concludes that both NsC1-AgCl and TlC1-CsC1should give rise to super-lattice structures, but the experimentalresults are not in agreement with the theory.Hydrous Oxides and Salts.-Although there is a tendency in somequarters to ignore the distinction between the true hydrogen bondand the less symmetrical hydroxyl bond, the significance of theselinkages is now generally appreciated, and examples of both inhydrated compounds continue to accumulate. A wave-mechanicaltreatment of the problem shows that both symmetrical andunsymmetrical states of the system 0-H-0 must exist.Manganite, Mn(OH)O, as anticipated, has a structure of thediaspore type, but the lattice now determined is monoclinic, witha larger cell than that formerly proposed.As in diaspore, the closeapproach of oxygen atoms (2.65 A.) indicates the existence ofhydroxyl bonds, which, however, link the structure together insheets, giving rise to (010) cleavage. The octahedron of oxygensaround each cation is so much distorted that the manganese hassquare co-ordination rather than octahedral. Unfortunately, it isdifficult to judge the accuracy of the parameters in this interestingstructure from the published qualitative comparison of intensities.Sb,O,, Sb,Ol,, and Sb,O, are all known to possess structureswhich are essentially that of senarmontite, and it is therefore notsurprising to find that Sb205,H,0 has the same space group andsimilar cell dimensions t o Sb203.3 It is noteworthy that the higher96 2.physikal. Chem., 1936, B, 31, 263; 33, 297.9 7 A. Smits, J. A. A. Ketelmr, and G. T. Miiller, ibid., 1936, A, 175, 359.as Helv. Physicas Acta;, 1936, 9, 516.8. Rh&!., 1935, 92, 221.1 R. H. Gillette and A. Sherman, J . Anter. Chew&. SOC., 1936, 58, 1135; cf.M, J. Buerger, 2. K&t,, 1936, 95, 163; cf. J. Garrido, Bu,Zl. SOC. fmw.M. L. Huggins, J . Physical Chem., 1936,40,723.Min., 1935, 58, 224.8 G. Natta and M. Baccaredda, Gazxettu, 1936, 66,3082 10 CRYSTAUOURAPHY .oxides of antimony as usually prepared contain a considerableamount of water, which probably plays a part in stabilising theoxygen-rich senarmontite structures by making possible the form-ation of hydroxyl bonds.The dihydrate of boron fluoride has almost identical cell dimen-sions with ammonium perchlorate, from which it is concluded 4 thatits structure is ( OH3)+ (BF,OH)- ; also, from the resemblance of thepowder diagrams of nitric acid monohydrate and phosphoric acid itis inferred 5 that the former may be orthonitric acid, H3N04.Substances of the type M(Hal),,3M(OH), (M = Co or Ni) hn17csimple layer lattices in which there is a statistical distribution ofhalogen and hydroxyl.6 Space groups of three hydrates of sodiumpyroborate, Na2B,D? , have been determined.V.Kassatochkin and V. Kotov 8report that “ potassium tetroxide,” KO,, has the same structure asstrontium and barium peroxides (CaC, type). This substance isthus definitely not K,O, but contains the singly charged 0; ion ; the0-0 separation is given as 1.28 A., a value which appears slightlyhigh by comparison with the distance 1-31 A.in the peroxides, since0; is presumably a resonance structure between O=O and -0-0-.A preliminary note 9 on the structure of NH4C1BrI suggests that itis closely similar to that of ammonium di-iodide ; the BrICl- ion islinear, with the iodine at its centre. L. K. Frevel lo has madequantitative studies of several azides, and finds that the N-Ndistances in the potassiuin, sodium, and ammonium salts are 1-145,1.150, and 1.166 A., respectively. These distances are in agreementwith the accepted formulation of the linear azide ion ; if it is assumedthat the empirical function of L. Pauling, L. 0. Brockway andJ.Y. Beach li can be applied to variations between N=N andNSN, 1-15 A. corresponds to about 30% of triple-bond function,and 1-165 A. to about 24%. It is noteworthy that although theradius of NH,+ is considerablygreater than that of K+, each ammoniumion in ammonium azide is surrounded tetrahedrally by four azidenitrogens at exactly the same distance (2.96 A.) as the shortest K-Ndistance in the potassium salt; this relative shortening of thecation-nitrogen distance indicates the formation of hydrogen, orComplex Ions.-Lznear ions.4 L. J. Klinkenberg and J. A. A. Ketelaar, Ree. trav. chim,, 1938, 54, 959.6 E. Zintl and W. Hauoke, 2. physikal. Chem., 1935,174,312.W. Feitknecht, Helv. C h h . Acta, 1936, 19, 467; W. Feitknecht and A.W. Mhder, Z.K ~ k t . , 1936, 92, 301.Collet, ibid., p. 831.a J. Chem. Phy8ic8, 1936, 4, 458.* R. C. L. Mooney, Phyaical Rev., 1935, [ii], 47, 807.lo J . Amer. Chem. Soc., 1936, 58, 779.11 Jbid., 1935, 57, 2705. See this vol., p. 45COX : CRYSTAL CHEMISTRY. 21 1rather “ imino,” bonds. Silver azide coutains linear azide ions withAX, Ions. The cell dimensions l3 of potassium hydrogen car-bonate suggest that its structure is related to that of the sodiumsalt, but the proposed orientation of the carbonate ions is verydifferent, and earlier magnetic measurements 1* show that the CO,groups arc considerably inclined to each other. Sodium carbonatemonohydrate has also been studied l5 by the Patterson method;the oxygen-water separation is giver1 as 2.69-2.72 A.The cell dimensions OE the anhydrous and the mono-hydrated snlphates of the magnesium series have been determinedfrom powder photographs by pr’.Rammel,l6 and the first case ofisomorphism between tellurates and sulphates (potassium salts) hasbeen rec0rded.l’ W. Schiitz 18 has demonstrated the isomorphismof the ions [Ge0,]4- and [GcF6I2- with [Si04]4- and [SiP6]2-respectively.The structure of silver phosphate, &,PO,, has been determinedin greater detail by L. Melrnh~lz,~~ who finds the P-0 distance tobe 1-61 A. This is greater than in dipotassium hydrogen phosphatea,nd is attributed to the formation of covalent bonds between oxygenand silver (Ago = 2.34 A.). In order to obtain satisfactory agree-ment between observed and calculated intensities, it was necessaryto assume the thermal vibrations of the silver atoms to have betra-gonal symmetry, and the ratio of the amplitudes along and per-pendicular to the tekragonal axis mas calculated (cf.the work ofBrindley and others, p. 200).Strong forces between anion and catioil are suggested also in thestructure of hydrated cadmium sulphate, 3CdS04,8H20,1Qa wherethe oxygen-oxygen linkages on the whole are weak.C. Finbalr and 0. Walssel 2o have studied the cubic high-tempera-ture forms of the alkali perchlorates and borduorides. By\ asmm-iiig rotation of the [C104]- and [BF,]- ions, they deduce a structurefor which both the interatomic distances (Cl-0 = 1-55 and B-33 =1.48 A,) and the agreement between observed and the calculatedintensities are more satisfactory than for the structures previouslyN-I4 = 1.18 rfr 0.04 A.12AX4 Ions.12 M.BassiArc, Bull. SOC. frarq. Jfh, 1935, 58, 333; Compt. r e d . , 1935,13 J. Dhar, Current Xci., 1936, 4, 867.14 A. Mookherjee, Physical Rev., 1934, [ii], 45, 844.1 5 J. P. Harper, 2. Krist., 1936, 95, 266.16 Qompt. rend., 1936, 202, 57, 2147.18 2. physikal. Chem., 1936, B, 31,292.10 J . Chem. Physics, 1936, 4, 316.1 9 5 H. Lipson, Proc. Roy. SOC., 1936, A , 156, 462.20 8. physisifcal. Chem., 1936, €3, 32, 130, 433.201, 735.l7 M. Patry, ibid., p. 16162 12 CRYSTALLOGRAPHY.proposed. They conclude that rotation of the anions also occursin the cubic hexafluorophosphates.The hydrates of calcium sulphate have been the subject of severalinvestigations during the past year.The structure of gypsum hasbeen worked out in detail by W. A. Wooster; 21 it is built up oflayers of sulphate alnd calcium ions parallel to the cleavage plane(OIO), linked together by layers of water molecules. Each watermolecule forms two hydroxyl bonds (length 2-70 A.) with oxygenatoms of sulphate groups in successive layers and one ionic linkwith a calcium atom (2.44 A.). A structure proposed for calciumsulphate hemihydrate 22 is based on cell dimensions and symmetrydifferent from those previously found,23 and is not in agreementwith the work of H. B. Weiser, W. 0. Milligan, and W. C. Eckholm,24who found hhat the water is not zeolitic, and that the dehydrationproduct has a lattice which, although closely similar to, is notidentical with that of the hemihydrate.It appears still to beuncertain whether the '' soluble anhydrite " first prepared byvan't Hoff by the action of nitric acid on gypsum is identical withthe dehydrated hemihydrate.[AX,] Ions. J. Beintema,25 from a study of the antimonates ofmagnesium, nickel, and barium, concludes that part of the waterwith which these compounds crystallise goes to form theoctahedral complex [Sb(OH),]-. A. I?. Wells26 has shown thatin Ag[Co(NH,),(N0.J4], contrary to the chemical evidence, thecomplex ion has a trans-configuration. The structure is a dis-torted cubic close-packing of the octahedral complexes as in trans-[PtCl,(NH,),] 27 with the cations in the interstices.As a result of the approximately spherical form of the large[RX,] ion, most structures involving it are based on cubic close-packing.Thus in many substances of the type A,M[RX,] (A =alkali metal ; n = 0, 1, or 2) the metals M and R occupy the positions(000) and (500) of sodium and chlorine in a rock-salt lattice, thealkali ions being at the centres ($a*) of small cubes formed by 4Mand 4R. Recent examples of this are afforded by complex nitrites(e.g., K,Pb[Ni(NO,),]) 2* and by Prussian-blue and related sub-21 2. Kriat., 1936, 94, 375.22 W. A. Caspari, Proc. Roy. SOC., 1936, A , 155, 41.23 P. Gallitelli and W. Riissem, Per. Min., 1933, 4, 1.z4 J . Amer. Ckem. SOC., 1936,58,1261; cf. P. Gaubert, Bull. Soc. f r a q . &!in.,p5 Proc. K . Akad. Wetemch. Amsterdam, 1936, 39,241, 652.ws 2.Krist., 1936, 95, 74.28 L. Cainbi and A. Ferrari, Gazzetta, 1935, 65, 1162; M. van Driel andH. J.1934, 57, 252.E. G. Cox and G. H. Preston, J., 1933, 1089.Verweel, 2. Krist., 1936,137, 677COX : CRYSTAL CHEMISTRY. 213stances.29 In the latter case the rock-salt structure of theFe[Fe(CN),] system appears to be particularly stable ; ferrousalkali ferrocyanides, A,FeI1[Pe1I( CN) J, readily lose half their alkali,and with very little change of the remainder of the lattice are oxidisedto Prussian-blue, AFe[Fe(CN)6], which again can be oxidised LoBerlin-blue, Pe1x1[Fe1x1(CN)6], without serious alteration in thestructure other than loss of alkali. Analogous compounds in whichruthenium or copper replaces part of the iron have been preparedand have similar structures.It should perhaps be pointed out thatthe X-ray evidence available does not distinguish between the€ormulz AFeI1CFe1I1(CN),] and AFe1IX[Fe1I( CN) ,] ; a further pointof interest is that, in the ferrous ferrocyanide and Berlin-blue, all theiron atoms appear to be equivalent. This might be taken toindicate that individual hexacovalent ferro- or ferri-cyanide groupsdo not exist in the crystal as such, and that a continuous three-dimensional network -Fe-CN-Pe- extends through the lattice,This view, while attractive in many respects, is not without itsdifficulties (e.g., the same crystallographic equivalence of atoms isobserved in the mercury atoms of K,Hg[Hg(No,),]), and the moredetailed analysis of these interesting compounds must be awaitedbefore any certain conclusions can be drawn.Co-ordination comjdexes.I n many cases the analysis of co-ordination compounds has been carried far enough to determinethe configuration of a central metal atom only; in others, completestructure determinations have been made. It is probable, however,that the study of such compounds will in future be more frequentlypursued to completeness, not only for stereochemical reasons, butalso in order to determine the structure of organic molecules, owingto the technical advantages resulting from the presence of asymmetrically placed metal atom.Much additional information on the distribution of the valenciesof quadricovalent metal atoms has been obtained and there isincreasing evidence in favour of the view that bond distribution andprincipal valency are closely connected, the configuration of thebonds in many cases changing from tetrahedral to planar as thestate of oxidation of the atom alters. The implications of thiswould seem to merit attention from a theoretical point of view.In addition to the complete structure determinations by J. M.Robertson on metal derivatives of phthalocyanins (see p.215),several reasonably detailed analyses have been carried out.D. Harker 30 has shown that cupric chloride dihydrate, CuC1,,2H20,29 J. F. Keggin and F. D. Miles, Nature, 1936, 187, 577,30 2, Krist., 1936, 93, 136,\/ \//', / 214 ORY STALL0GRAPH.Y .is a true quadricovalent compound in which, in agreement, withearlier work, the copper valencies are coplanar.From a, com-parison of interatomic distances, Harker was led to suggest thatthe CuC1,,2H20 complex occurs in K2CuCl4,2H,O, which shouldtherefore be written K2[CuC1,,2H,0]C1,. An analysis 3l of( NIX4),CnC14,2H,0 gives interatomic distances which confirm this,On the other hand, the water molecules in K[AuBr4],2€I,0 are notco-ordinated to the metalH. Brasseur and A. de Rassenfos~e,3~~ continllring their investiga-tions of complex cyanides and related compounds, have shownthat barium cadmichloride tetrahydrate, BaCdC1,,4H2O, and thecorresponding cadmibromide have cell dimensions very closelysimilar to those of barium platinocyanide tetrahydrate. Theirinference that the ion [CdCl,]” has nearly the same form anddimensions as [Pt(CN),]” must await confirmation by more detailedstudies, since it is quite possible that the anions in these compoundsare of the form [MX4(H20),]”.The higher hydrate of 12-phosphotungstic acid, H3BW,,0,,,29H,0,33is of special interest.The anions (PW,,04,)-3 (which are identicalwith those found by J. F. Keggin in the pentahydrah) are heldtogether almost entirely by water molecules ; anions and groupsof 29H,O lie on interpenetrating diamond lattices. A noteworthyfeature is the constancy of both the I-1,Q-0 and the H,O-H,Odistance, which scarcely vary by more than the experimental errorfrom the value 2.88 A. This is the more remarkable since a watermolecule may be attached by three, four, or seven such bonds toneighbouring oxygen atoms or water molecules.Other co-ordination compounds are discussed in the Report onInorganic Chemistry .34E.G. C.3. MOLECULAR CRYSTALS.Until this last year exact X-ray analyses of organic compoundshave always depended on some preliminary knowledge of thestructure of the molecules concerned. J. M. Robertson has ncwachieved in his work on the phthalocyanines the first absolutelydirect analysis of an organic molecule, and one which does notinvolve even the assumption of the presence of discrete atoms.31 A. Silberetsin, Compt. rend., 1936, 202, 1196.351 See this vol., p. 164.32a 8. Krist., 1936, 95, 474 ; Bull. SOC. roy. Sci. Li2ge, 1936, No. 5, 125 ;s8 A. J. Bradley and J. W. Illingworth, Proc.Roy. SOC., 1936, -4, 15’9, 113.a4 See this vol., p. 157.Nos. 8-10,199.1 J., 1936, 1195CROWFOOT : MOLECULAR CRYSTALS. 2115The success of this analysis depends upon the fulfilment of two con-ditions. I n the first place the phthalocyanines crystallise in a,molecular arrangement having cent'res of symmetry at which, bychemical meaizs, different metal atoms can be inserted withoutappreciably disturbing the crystal structure. The P values directlycalculated from the intensities of the X-ray reflections observed withthe metal-free compound and nickel phthalocyanine were conse-quently known to differ only by the contribution of the nickel atom,which geometrically had to be equal to the structure factor furFIG. 2.Projection along the b cczi8, showing one complete phthwlocyanine molaule. Theplane of the molecule is steeply indined to the plane of the projection, the Mdirection making an.angle 04 46" with the b axis, and the L direction 2.3".Each contour represents a density increment of one electron per A,a, the one-electron line being dotted. .nickel a t the angle of reflection. A plot of the sums and differencesof the observed P values for the two compounds against sin e/xshowed a congregation of certain valuea about the theoretical curveof the nickel scattering factor. These values fixed the phase con-stants of the reflected rays, and therefore a Fourier synthesis couldbe formed directly from the measured P values.The Fourier synthesis of the (h01) terms gave a projection of thestructure on (010) which shows clearly resolved a pattern of atomsjoined in six-membered and five-membered rings unitcd together(sec Fig.2). This pattern may be said to provide the first purel216 CRYSTALLOGRAPHY.physical demonstration of the truth of organic chemistry. It illus-trates also the second condition necessaq- for the success of thisdetermination of chemical structure-that the projection obtainableshculd be one in which little overlapping of atoms occurs, so that$an unambiguous view may be obtained of the pattern as a whole.For example, the projection of the phthalocyanine structure alongthe c axis shows no clem resolution and could alone give no direct,chemical information.The method of direct determination of the phase constants of theFourier terms by isomorphous replacement of metal atoms hasalready been used by J.M. Cork 2 and C. A. Beevers and H. Lipsonin their investigations of the alums. Owing to the peculiarlyf avourable chemical and crystal structure of the phthalocyaninesthe results are here much more striking, and it would seem worthwhile to explore further in the field of organic chemistry for othercompounds which may possess this particular combination ofcharacteristics. For most structures it will be necessary, of course,still to use the trial and error method--with the renewed confidencenow given for its results.The exact X-ray analyses discussed last year proved that themethod was capable of giving purely chemical information of twokinds : (1) the actual determination of the mutual orientation ofthe atoms in the molecule, and here Robertson's analysis providesus with a notable advance; (2) the diagnosis of the nature of thechemical bonds present from the measured interatomic distances.In this both theory and experiment are still far from complete, butthe function described by E.Bauling, L. 0. Brockway, and J. Y.Beach of the relation between bond length and single-bond-double-bond resonance does a t least provide us with a new empiricalstandard with which t o compare the experimental values.Simple Holecular Compounds.The crystal structures of most of the simplest quasi-sphericaldiatomic and triatomic molecules have already been measured, andthe remaining types are likely to prove more complicated.E.Pohland ti has found, for example, that solid hydrogen cyanide, nitricoxide, and sulphur dioxide are all doubly refracting. More compli-cated types of X-ray goniometer are needed for further work in thelow-temperature region,6 and with one of these W. H. Keesom and2 Phil. Mag., 1927, 4, 688.B Proc. Roy. SOC., 1936, A, 148, 664.4 J . Amer. Chem. SOC., 1935, 5'7, 2705.Angew. Clwm., 1936, 49, 482.W. H. Keasom and IS;. W. Taconis, Physica, 1035, 2, 463; R. L, MeFsrlan,Re,,. S a i . Instr., 1936, [ii], 7, 89CROWFOOT : MOLECULAR CRYSTALS. 2 17K. W. Taconis have studied y-oxygen and chlorine.* y-Oxygenis cubic in agreement with the work of L. Vegard,g and chlorinecrystallises at - 185' in the tetragonal system with eight (non-rotating) molecules in the unit cell.The C1-Cl distance calculatedIrom the rather limited X-ray intensities available is 1-99 in eachmolecule, and 2.79 A. between molecules. The lattice constant andX-ray intensities of solid hydrogen sizlphide do no& appear to changethrough a wide temperature range, though transitions, probably ofrotation and orientation of the molecules, are indicated by otherdata, e.g., dielectric constnnt.l* That some orientation of themolecules does exist at liquid-air temperatures has been suggestedto account for the Raman spectrum of the solid.11In a number of crystals of this class, where rotating forms com-monly occur, entropy calculations suggest that equilibrium isfrequently not established a6 low temperatures. Tetramethyl-methane has a zero-point entropy of 8 e.u.which may be due torandom orientation,12 and the residual entropy of ethane cansimilarly be correlated with a form of incomplete r0tati0n.l~ Thestructure of ice is perhaps the most important example, since thedifficulties of placing the hydrogen atoms in all varieties of ice sofar measured have led many authors to describe the structures asi0nic.14 The Bernal-Fowler model l5 requires definite orientation ofthe hydrogen atoms between the water molecules forming hydrogenbonds, each hydrogen atom remaining still most closely associatedwith one oxygen, and although it is possible to form such a structurein a regular way, yet there is no evidence of any superstructure inthe X-ray results.L. Pauling calculates that if the orientation israndom, the residual entropy would be 0.805 e.u., compared withthe observed value of 0.87 e.u.16The structure of the different polymorphic forms of ice should beof particular interest for a further test of these theories. Fromtime to time in the literature measurements have been reportedsuggesting that ordinary ice, ice-I, is rhombohedral, which mightbe due to a correctly oriented form of the Bernal-Fowler type.N. Seljakow l7 has, however, now shown that if distilled water isPhysica, 1936, 3, 141.* 2. Physik, 1935, 98, 1.10 E. Justi and H. Nitka, Physikal. Z., 1936, 37, 435.11 S. C. Sirkar and J. Gupta, Indian J. Phgsics, 1936, 10, 227.l2 J. G. Aston and G. H. Mesaerly, J.C h m . Physic8, 1936, 4, 391.14 CP. W. H. Barnes, Tyam. Roy. SOC. Canada, 1935, [iii], 29, 111, 53.15 J . Chern. Physics, 1933,1, 515.17 Compt. rend. Acad. Sci. U.R.S.S., 1936,1, 293; 2, 227.Ibid., p. 237.J. D. Kemp and K. S. Pitzer, ibid., p. 749.1% J . ~ m e r . mm. SO^., i935,57, 2680218 CRYSTALLOGRAPHY.frozen from the surface in the open air, two quite different modi-fications may be obtained. Ordinary ice-I, a-ice, appears when thetemperature of crystallisation is between 0" and - 8' and gives thcusual hexagonal Laue photograph. Below - 8O, i.e., with somedegree of supercooling, fhe is formed, which shows a quite differentrhombohedra1 Laue pattern and appears to be pseudo-cubic ( c I ~ = 1-33).The high-pressure forms of ice, ice-11 l8 and i ~ e - 1 1 1 , ~ ~ according toR.L. McFarlan, have orthorhombic structures of some complication.Both the molecular arrangements suggested show oxygen atomssurrounded by four nearest neighbours at 2.71 A., very little differentfrom the distance in ice-I, 2.74 A. The higher densities imposedby the higher pressure, 1.2 in ice-11, 1.105 in ice-111, are attained inboth by different distortions from the regular tetrahedral arrange-ment of nearest neighbours showd by ice-I, which permits thenext neighbour distance to decrease from 4-47A. in ice-I to about 3.4 A. in ice-I1 and ice-111.It does not seem necessary to correlate thisdistortion with a truly ionic structure, sincesuch alterations of the valency angle mayoccur under strain even with homopolarbonds such as those in diphenyl ether-I~o$' 9776" where the C-0-C angle is about 128°.20 I nmany hydrates the water molecules are sur-rounded by similarly distorted tetrahedra of* G m h oox5'P oxygen atoms, and the most significantfeature seems a t present to be, not the dis-tortion, but the even approximately tetrahedral nature of thcco-ordination.FIG 3.Dimensions of the OxalicAc;dMolecule.7'24 1 u123-j0Aliphatic Compounds.To the complete structure determinations in the aliphatic serieswe may now add that of oxalic acid dihydrate by J .M. Robertson.21The approximate structure found by W. H. Zachariasen showed thetwo oxygen atoms to be differently situated with respect to thewater molecules and suggested therefore that these were different incharacter. The results now obtained by the use of absoluteintensities and double Fourier syntheses fully conflrm and extendthis.The two C-0 distances are found to be different, 1.24 and1.30 A., and though neither corresponds to pure double-bond orsingle-bond distances, they may be correlated with links mainlyl8 J. Chem. Phy8leic8, 1930, 4, 60.le Ibid., p. 253.2o L. E. Sutton and G . C. Hampson, Tram. Paraday Soc., 1935, 31, 945.a1 J., 1936, 1817CROWFOOT : MOLECULAR URYSTALS. 218C=O and C-OH respectively. The C-C distance as measured isconsiderably shorter than that first proposed, vix., 1.43-1-45 A.,which corresponds to 30-25% double-bond quality. According tocalculations of J. E. Lennard-Jones,22 such a shortening is to beexpected in a singlo link between two doubly-bound carbon atoms,as, e.g., in a conjugated chain.Here, it seems that it could takeplace through the co-operation of the water molecules forminghydrogen and hydroxyl bond chains through the structure asxnggested by J. I). Bernal and (Niss) H. D. Megaw 23 (Type B).Exactly opposite effects appear to occur with the oxalate ion,particularly in ammonium oxalate monohyclrate .a* Here the dis-tance between the central carbon atoms is 1.58 A,, longer than thatexpected for a normal single link. The lengthening might be dueto repulsion between the negative parts of the CO, ions, and therecertainly seems no allowance for a partial double-bond character ofthe central link. Also, the two CO, groups are found to be no longercoplanar, but inclined at an angle of about 28' to one another, aresult which seems most reasonable if there is a pure single linka t the centre, about which free rotation should be possible.Inmoat other oxalates studied, e.g., those of potassium, sodium, andrubidium, the ion is planar within a probable experimental error of& lOO.25 The difference in ammonium oxalate may partly be dueto a new system set up by possible hydrogen-bond formationbetween the ammonium ions and oxygen atoms of tihe oxalate ion.The interatomic distances in ammonium and potassium oxalatesshow relations similar to those of animonium azidc compared withpotassium azide26 (see p. 210). Calcium oxalate dihydrate istetragonal, and it is interesting that this occurs with calcium citratein deposits a t the bottom of the WeddellInvestigations have begun on a number of the homologues ofoxalic acid.28 These crystallise commonly in two different poly-morphic modifications, cc and p, the cc being the stable form for acidsof more than nine carbon atoms, the p for those of less.The changeis probably associated with the differential effective interaction ofthe carboxyl groups : below C, the crystals are hard and shining;above, platy and waxy. Of several tartaric acid derivatives de-~cribed,2~ tartramide seems most promising for further exact2% Private communication to 5. M. Robertson.23 PTOC. Roy. SOC., 1938, A , 151, 384.24 S. B. Hendricks and M. E. Jefferson, J . Chem.Physics, 1936, 4, 102.25 S. B. Hendricks, Z . K r i s t . , 1935, 91, 48.38 L. K. Frevel, J. Amer. ('hem. SOC., 1936, 58, 770; 2. Krist., 1936,94, 197.27 F. A. Bannister, Discovery Reports, 1936, XIII, 60.28 P. de la, Tour, C'ompt. rend., 1936, 202, 1935.2s J. Wyart and Y. Hi-Heng, ibid., 203, 96220 CRYSTALLOGRAPHY.information. A number of measurements are recorded on long-chain acids and salts,3O and also the ac-monoglycerides,31 which haveinteresting liquid-crystal properties. The choleic acids are some-what of a mysbery, since the same X-ray photographs are obtainedwith considerable variation in fatty acid present .32The measurements on oxalic acid and oxalates provide us with aseries of G O distances which can be described as due to varyingsinglc-bond-double-bond mixtures.That in metaldehyde probablyrepresents the pure single link and is given as 1.43 & 0.02 Met-aldehyde crystallises in the tetragonal system, each molecule having afourfold axis of symmetry. The molecule consists of a puckeredeight-membered ring of alternate oxygen and carbon atoms, similarFIG. 4.Metaldehyde.FIG. 6.Resorcilzol.@Carbon 0 Oxygent o the six-membered ring of paraldeh~de,~~ the methyl groups herelying npproxinmtely in the plane of the ring (Fig. 4). A Fourierprojection along the four-fold axis has been obtained, but theparameters at right angles to this are not so accurately fixed.The whole series of C-0 distances hitherto found in organiccompounds, including that in resorcinol (p.221), are shown in thetable.InteratomicMain bond type. Compound. distance, A.C-0-C Metaldehyde 1.43*C-OH Resorcinol 1.36*GOH Oxalic ac;d dihydrate 1.30GO- Ammonium oxalate monohydrate 1.25 *c= 0 Urea 35 1-25 *c= 0 Oxalic acid dihydrate 1.24 c=o Benzoquinone 36 1.14 * Probably most accurate values.--30 P. A. Thiessen and J. Stad, 2. phyalsikd. Chem., 1936, 176, 397; with81 T. Malkin and N. R. el Shurbagy, J., 1936, 1628.W. Wittstadt, Alzgew. Chem., 1936, 49, 641 ; R. Brill, ibid., p. 643CROWFOOT : MOLECULAR URYSTALS. 221Aromatic Compounds.The exact structure of the molecule of resorcinol 37 is shown inFig. 5 . The most interesting feature of this crystal structure is thespiral method of packing, which is able to bring the hydroxyl groupswithin that distance of each other, W ~ Z .2.66-2-74 A., associatedwith hydroxyl-bond formation without placing the carbon atomsnearer than the customary 3-5 A. Both theory and experimentnow show that in the type of hydrogen- and hydroxyl-bond form-ation common in organic compounds there is not usually completedegeneracy, hydroxyl and keto-groups still preserving individualcharacteristics. This is true in oxalic acid (above) and also,apparently, in quinhydrone, where more exact work shows that thesymmetry first found is a pseudo-symmetry, and that the benzo-quinone and the quinol molecules can actually be distinguished inthe crystal cell.38 Their mutual arrangement is necessarily almostexactly the same as that first put forward, and the lines of attractionbetween them are still those of the C:O .. . HO bonds running throughthe whole structure. In iaatin, the question of a difference of thiskind is bound up with the old problem of a distinction between thelactim and the lactam st.ructure. E. G. Cox, T. H. Goodwin, and(Miss) A. I. Wagstaff 39 find an orientation of the molecules in thecrystal unit which indicates a hydrogen bond between the nitrogenatom and the carbonyl group of neighbouring molecules; butwhether this is to be written mainly as of the type NH . . . 0:C(lactam) or =N . . . HOC (Ilactim), only a complete analysis candecide. The present work implies that to a certain degree thestructure is intermediate between the two.I n the 'phthalocyanine molecule there appears to be the firstexample of an internal hydrogen bond.The exact molecularstructure found (see p. 215) is illustrated in Fig. 6. The directPourier projection first obtained showed that the ring system issomewhat inclined to the plane of projection, but the regularity ofthe pattern proves that the molecule is essentially planar. Theangle of inclination adopted agrees with the observed intensities.32 Y . Go and 0. Kratky, 2. K~ist., 1936, 92, 310.33 L. Fading and D. C. Carpenter, J . Amer. Clhem. SOC., 1936, 58, 1274.34 D. C. Carpenter and L. 0. Brockway, ibid., p. 1270; P. G. Ackermann35 R. W. G. Wyckoff and It. B. Corey, 2. Krist., 1934, 89, 462.36 J. M. Robertson, Proc. Roy. Soc., 1936, A, 150, 106.37 Idem, ibid., 1936, A , 157, 79; cf.(Mrs.) K. Lonsdale, Nature, 1936,a8 J. Palacios and 0. R. Foz, Anal. Fis. Qui,m., 1935, 33, 627; J. Bijvoet,39 Proc. Roy. SOC., 1936, A, 167, 399.and J. E. Mayer, J . Chem. Physics, 1936, 4, 377.137, 826.private communication222 ORYSTALLOGRAPHY.Chemically thc structure is that assigned by R. P. Lin~tead,~O butthe interatomic distances indicate the additional regularity of acomplete resonance system. There are no systematic differences,e.g., in the C-C distances of the benzene ring that might indicateeither o-quinonoid forms or fixation of one of the Kelculh modes bythe fusion of the five-membered sing; but the molecule as a wholeshows a slight departure from tetragonal symmetry, the centralnitrogen atoms being drawn more closely together dong two out ofthc four possible lines.The distance between them, vix., 2-85 A,FIG. 6.Dinwasions of the PhthaZocyanine Molecule.suggests that this is due to hydrogen-bond formation and thehypothesis is being tested by further examination of the metallicderivatives ?1It is comparatively rarely that the full symmetry of a molecule isshown in that of the crystal if the symmetry is at all high. In thephthalocyanines this is due to the molecules being packed with theplanes of neighbouring ring systems slanting at a considerable angleto one another; and this behaviour appears to be common amongthe many aromatic compounds o€ which preliminary measuremcnt,s40 See Ann. Reports, 1935, 32, 360.41 J. M. Robertson, J . , 1936, 1736UROWPOOT : MOLECULAR ORYSTALS.223arc recorded this year 42 and also in reduced ring systems such asthose of the sex One exccption is hexabromomethyl-benzene which crystallises in a very simple rhombohedra1 structurewith one molecule in the unit cell.44 This molecule must crystallo-graphically possess a thrce-fold axis of symmetry, and the nearidentity of the intensities of equatorial and layer-line reflectionsshows that all the atoms fall nearly into a single plane. A planarcharacter is shown also in the crystal structure of hesaethylbenzene,45which is, however, triclinic like hexamethylbenzene. Molecularcentres of symmetry can more readily appear than axes, e.g., indiben~anthracene,~G r~brene,~7 and q~aterphenyl.~~ In the optic-ally active substitutcd diphenyls 48 the crystal centre of symmetrydisappears, as would be expected.The exact analyses described show several examples of slightdistortions of the external valency angles of the benzene carbonatoms, towards substituent atoms.7!hat in resorcinol is very smalland may not be real, and that in the phthalocyanines is forced bythe fusion of the five-membered ring. Both these distortions arein the plane of the ring, but in two other structures examined,p - toluidine 49 and flu or en^,^^ preliminary calculations of intensitiessuggest that deviations also occur a t right angles to this plane. Influorene, however, where such a distortion would seem mostplausible, the dipole moment of the solution indicates a planarstructure,51 and it is evident that further work is necessary toestablish with certainty the correct configuration in the solid.42 J.Iball, Z. Krist., 1936, 92, 293; 93, 47; 94, 7; 95, 282; J. Dhar andA. C. Guha, ibid., 1935, 91, 123; M. Milone, ibid., 1936, 93, 113; M. Prasadand J. Shanker, J. Indian Chem. Soc., 1936, 13, 123; with M. P. Lakhani,ibid., p. 519; R. Hultgren, J. Chem. Physics, 1936, 4, 84; E. Hertel andH. W. Bergk, 2. physilcal. Chem., 1936, B, 33, 319; L. Rivoir and R. Salvia,Anal. Pis. Quim., 1935, 33, 314; B. K. Blount and (Miss) D. Crowfoot, J.,1936, 414; J. 35. Robertson, M. Prasad, and (Miss) I. Woodward, Proc. Roy.SOC., 1936, A, 154, 187.43 J. D. Bornal and (Miss) D. Crowfoot, 2. K h t . , 1936, 93, 464.44 H. S . Backer, Rec.trav. chim., 1935, 54, 745; J. Beintema, P. TorpsLre,45 H. K. Pal and A. C. Guha, 2. Krist., 1036, 92, 392; N. Ganguli, ibid.,46 J. Iball, Nature, 1936, 137, 361.47 M7. H. Taylor, 2. Krist., 1936, 93, 151; cf. E. Bergmann and E. Her-linger, J. Chem. Physics, 1936, 4, 532.48 L. W. Pickett, J. AmeT. Chem. Soc., 1936, 58, 2299.49 J. Wyart, Bull. SOC. f m q . Min., 1936, 58, 281.50 J. Iball, I;. Krist., 1936, 94, 397; J . W. Cook and J. Iball, Chem. and61 E. D. Hughes, (Mrs.) C. G. Le FBvre, and R. J. W. Le FBvre, ibid.,and W. J. van Weeden, ibid., p. 962.93, 42.Ind., 1936, 467.pp. 646, 581224 CRYSTALLOGRAPHY.Fibre Structures.The most important contribution of the first application of X-raycrystallography to the problem of fibre structures was the demon-stration that the patterns obtained corresponded to units withinthe molecule many times smaller than the estimated molecular size.52Most subsequent work has concentrated on the examination of thescsmall repeating units, and with the introduction of such methods ofanalysis of fibre diagrams as that proposed by E.Sauter thisyear,53 it is to be hoped that still more accurate information will bemade available along this line. With synthetic polymers, however,such as the polyoxyrnethylenes, two types of X-ray interferencecan be obtained, one due to the fibre repeat unit and the other tosmall angle reflections corresponding to the molecular length.54Most natural polymers have very much larger molecules than these,and long spacings due to regularity in chain length would be muchmore difficult to observe; but the constant reports of long spacingsin the litcrature-particularly on proteins 55-suggest that a morecomprehensive search should be made in the region of small anglescattering.This has been now initiated by R. W. G. Wyckoff andR. 13. Gorey, using for their first attempt apparatus capable ofshowing spacings up to 150 A.55a They find that, on the basis ofthe scattering visible a t small angles, the compounds examined fallinto three classes, of which rubber, cellulose, and the proteins maybe taken as typical.For stretched rubber, Wyckoff finds no scattering a t all a t smallangles, which is not surprising since, even if chains of a single lengthwere present, it would be unlikely that the orientation producedby stretching would be sufficient to demonstrate them.However,in p-rubber, in which a rather different collfigura'tion of the chainappears, a spacing of 115 A., 24 times 6he simple identity period,has been observed.56 This may, of course, represent no more thana purely crystallographic superstructure. The exact configurationof the residues in ordinary, " crystalline," a-rubber has beenreinvestigated by K. H. Meyer and W. LotmarY5' who find amonoclinic cell in which the chains themselves have a twofold52 Cf. J. R. Katz, Tram. Paraday Soc., 1936, 32, 77.53 8. KriRt., 1936, 93, 93.54 H. Staudinger, H. Johner, R. Signer, G. Mie and J. Hengstenberg, 2.55 See Ann. Reports, 1935, 32, 241 ; W.T. Astbury, Nature, 1936, 13'4, 803.55a J . Biol. Chem., 1936, 114, 407.56 G. W. Pankow, Helu. Chim. Acts, 1936,19, 221.67 Sitzungsber. A M . Wiser. Wien, 1936, IIb, 145, 721; Arch. Sci. &YS.pkyerikal. Chem., 1927, 128, 425.nd., 1936, [v], 18, Suppl., 61CROWFOOT : MOLECULAR CRYSTALS. 225screw axis of symmetry, the crystal being essentially a racemate ofright- and left-screwed molecules. The density found is still ratherlower than that calculated for the measured unit, but it seemsprobable that this is due to holes between the rubber crystallites.The same discrepancy appears with the “ inorganic ” rubber,phosphonitrile chloride.58 In this, too, the crystal structuresuggests long chains formed on a screw-axis pattern, the spacegroup here being orthorhombic.Both rubber and phosphonitrilechloride give very similar X-ray scattering curves in the amorphousstate. That of rubber has now been submitted to a Fourier analysis,and the radial density found is in good agreement with the con-figuration of the rubber chain suggested for a - r ~ b b e r . ~ ~ Thestrongest peak at 5 A. corresponds to distances between neigh-bouring chains and moves out to 6-15 A. in phosphonitrile chloride.In thin rubber films the a-rubber structure again appears, but maybe rather differently oriented on stretchingCellulose, according to Wyckoff, shows diffuse scattering at smallangles but no definite line pattern. This is contrary to manyprevious reports of long spacings, but some at least of these havebeen proved by W.A. Sisson, G. L. Clark, and E. A. Parker 61 tobe actually not diffraction lines but absorption edges, and resultsmight differ with cellulose from different sources. 0. L. Clark andA. F. Smith 62 have also investigated chitin and various deriv-atives such as its nitrate and chitosan, with which, again, longspacings seem to appear. For the repeat structure of chitin, theirresults agree with those of K. H. Meyer and G. W. Pank0w,6~ theunit cell being similar in dimensions to that of cellulose, and largerthan that put forward by A. N. J. Heyn; 64 but the latter’s workdoes show the orientation of the chitin chains in the sporangiophoreof the fungus phycomyces. It is interesting that plant and animalchitin appear identical,65 whereas among the starches from thesetwo sources there are marked differences.Glycogen has been madeto give only a single amorphous ring, and the patterns from severalvegetable starches show a number of Debye lines and differ from5 B K. H. Meyer, W. Lotmar, and G. W. Pankow, Helv. Chim. Acta, 1936,19, 930; K. H. Meyer and G. W. Pankow, Arch. Sci. phys. nat., 1935, [v],17, 139-59 G. L. S h a r d and B. E. Warren, J . Amer. Chem. Soc., 1936, 58, 507.60 K. I. Krilov, Physikal. 2. Sowietunhn, 1935, 8, 136.61 J . Amer. Chem. Soc., 1936, 58, 1635.62 J . Physical Chem., 1936, 40, 863.63 Nelv. C h h Acta, 1935, 18, 589.64 Proc. K . Akad. Wetensch. Amsterdam, 1936, 39, 132.66 K. EL Meyer and W. Lotmar, Arch. Sci. phy8. not., 1935, [v], 17, 287;with G.van Iterson, Rec. trav. c h h . , 1936, 55, 61.REP.-VOL. XXxm. 226 CRY STALLOGIRAPHY.plant to plant.66 The actual laying down of these fibres in plants isa fawinating problem.67 In Valonia, protoplasm streams can beobserved running nearly at right angles to the primary cellulosefibre axis,68 and these probably determine the orientation of thesecondary axis as found by X-ray studies. Perhaps the time is notfar off when X-ray cinematographs will be taken of this process.They have already been successfully made by M. Mathieu for thenitration of cellulose,69 and could obviously be applied widely,especially in the many other cellulose reactions awaiting elu~idation.~~In the nitration of cellulose with gaseous nitrogen pentoxide, thewhole process can be observed in the course of one hour from thefirst disappearance of regularity in the direction of the cellulosefibre atxis to the final appearance of the new fibre period, 25.1 A., oftrinitrocellulose.In the third group, the proteins, more and mare evidence appearsof a surprising degree of natural orientation and complexity ofstructure.71 The only natural protein fibre to show an almostamorphous state appears to be " byssus " or silk of oysters, whichmay be a mixture of different ~haiiis.7~ In most cases the a-keratinoriented state appears even in the actual arrangement of the proteinin cell walls,73 e.g., of wool cells, 74 and the structures present maybe much more complicated than this, as the patterns obtained fromtendon, collagen, quill, and feather keratin In all these,spacings over 100 A.long have been observed, and the real fibrerepeat of tendon is probably at least 330 A. Truly crystallineDebye-Scherrer patterns are given by both the Bence Jones protein 75and chymotrypsinogeii 55 suspended in water, but the most interest-ing results of all have been obtained on the tobacco mosaio virus.The preparation of biologically active crystals was first effected by66 F. May and L. Graf, 2. Biol., 1936, 97, 167.87 W. Wergin, Angew. Chem., 1936, 49, 843; W. A. Sisson, J . PhysicalChern., 1936, 40, 343; W. K. Farr, Paper Trade J., 1935,101, T.A.P.P. 1.Sect. 183; F. Worschitz, Afagyar Chem. Pol., 1934, 40, 60; T. Pujiwara andY. Imanaka, J. Sci. Hiroshina Univ., 1936, A, 6, 237.68 G. van Iterson, Nature, 1936, 138, 365. 68 Ibid., p. 824.70 M. Mathieu and (Mdlle.) T. Petitpas, Corn@. rend., 1936, 203, 46; A. J.Barry, F. C. Peterson, and A. J. King, J . Amer. Chem. SOC., 1936, 58, 333;K. Hess and C. Trogus, 2. Elektrochem., 1936, 42, 696, 705, 710; J. B.Calkin, J. Phy8iCal Chem., 1936, 40, 27; G. N a t h and M. Baccaredda, AttiR . Accad. Lincei, 1936, [vi], 23, 444; M . Iaihara, J . SOC. Chem. Id. Japan,1936, 39, 62, 65, 68, 70; A. Frey-Wyssling, Hdv. Chim. Ada, 1936, 19, 901.71 Cf. W. T. Astbury, Nature, 1936, 157, 803.72 G. Centola, Gazxetta, 1936, 66, 71.73 A. Giroud and a. Champetier, Bull. SOC. C h h . biol., 1936, 18, 666.7 4 H. J. Woods, Proc. L e d Phil. Soc., 1036-36, 5, 132.75 A. Magnus-Levy, K. H. Meyer, and W. Lotmar, Nature, 1936, 187, 616CROWFOOT : MOLECULAR CRYSTALS. 227W. M. Stanley,76 and the Debye-Scherrer pattern obta,ined from asuspension of these by R. W. G. Wyckoff 77 and 1%. B. Corey issimilar to, though somewhat more complicated than, those of othercrystalline proteins. This is true both of preparations obtained bythe usual methods of protein fractionation and of “ crystalline ”pellets isolated directly by centrifuging clear juice pressed fromvirus-infected plants.78 The most remarkable effects are shownby the highly purified protein solutions prepared by F. C. Bawdenand N. W. Pi15e.~~ Therme separate on standing into two layers;the lower layer, which may be waher-clear, is liquid-crystalline,whereas tho upper layer shows to a high degree the phenomenon ofanisotropy of flow. On drying, at f i r s t a “ wet gel ” is formed witha much higher birefringence than the liquid but this graduallyshrinks by about 50% to form a “ d r y ” gel. J. D. Bernal andI. Fankuchen 79 have obtained X-ray patterns both from the wetand the dry gel, and from the liquid down to 13% concentrakionoriented by flow. All Hhow approximately the same large anglescattering, which oan be correlated with that given by the crystdls--a protein pattern of some complexity with a repeat unit of3 x 22-2 A. This must be due to the internal structure of theprotein molecule. There is also an inner pattern corresponding toa hexagonal paclsing of long rods; and the unit of the long spacinghere varies from 131.8 A. in the dried gel to 398 A. in the liquid,and corresponds to distances between the protein moleculc~,. Therelations between the various patterns permit Bernal and Fanlrucbcnto deduce Che presence of rods 100 A, across roughly triangular insection. Both the sharpness of the X-ray reflections and the shapeof the lenticols formed in the liquids suggest a length for thGse P&of a t least 1000 A., which would agree with Svedberg’s molecularweight of about 17,000,000 ; but there is no X-ray evidence atpresent against the rods being quite indefinite in length. Crystalo-graphically, the gel and the liquid structures may be awigned f oC. Herrnann’s liquid-crystal class R I1) BD,81 there being regularitywithin the long rods and in theiz arrangements normal to theirlength without accurate co-ordination between the $wo; but, theobservation of such regularities is indeed an achievement in amolecule of such a high molecular weight and with at least a, suspicionof life. D. M. C.E. G. Cox.D. M. CROWFOOT.76 Science, 1935, 81, 644; J. Biol. Chem., 1936, 115, 673.7 7 Ibid., 116, 61.78 Science, 1936, 84, 513.80 J . Arner. Uhern. SOC., 1936, 58, 1863.2;a€ure, €936, 3.38, 1051,81 2. Krist., 1931, 78, 186