ATOMIC MODELS PART 3.-SOME STEREOCHEMICAL PROBLEMS IN DYEING BY CONMAR ROBINSON Courtaulds Limited Research Laboratory Maidenhead Berks Received 22nd July 1953 Models of dye molecules and of cellulose and polypeptide chains have been built with the type of atomic model previously described by Hartley and Robinson. Their use in investigating stereochemical problems related to dyeing has been illustrated. The anisotropic flexibility of the cellulose chain has been demonstrated. The shape of a number of dye molecules in relation to the shape of the cellulose chain has been discussed and also the part played by intra- and intermolecular hydrogen bonds in both dyes and cellulose. The models suggest that contrary to the usual assumptions hydrogen bonding to cellulose can take place irrespective of the position of those groups in the dye molecule which are capable of forming such bonds.When an amide group links two aromatic rings together it appears unlikely that the group can be in the cis configuration when attached to cellulose as this configuration involves great steric hindrance. The possible part played by steric hindrance in the dyeing of a synthetic polypeptide is discussed. Although atomic models have frequently been used to investigate stereochemical problems in dyeing their usefulness has been limited by shortcomings in the designs of the models available. Recently Hartley and Robinson 1 have described a development of the Stuart type of model. In these the atoms are linked to- gether by a combination of link and collar which allows the valency bond to be distorted a few degrees from the normal when strained but which provides a restoring force which insures that the valency angle shall have its normal value when the model is not under strain.This linking mechanism not only allows molecules to be built in which the valency angles are known to deviate from the normal values but also allows an estimate of the distortion involved when a model of a configuration involving strain is built. Another advantage of the distortable link is that it allows the full value of the van der Waals radius to be used in designing the atoms while in the Stuart models 293 where the valency angles are fixed it is necessary to reduce this radius by an arbitrary percentage (e.g.15 %) for otherwise no structures could be built in which there was even a small deviation from the normal bond angles or van der Waals radii. The use of atomic models for quantitative work is analogous to that of a calculating machine. However mechanically perfect the models may be any result obtained (e.g. in determinink the relative positions of two atoms imposed by a certain configuration) can only have a degree of accuracy corresponding to that of the data used in their design. For this reason it is a great advantage to be able to use models for which the nearest approach cor- responds to the minimal energy conditions pertaining in crystals since then any distortion of the valency bonds which is found in a configuration will be a measure of the degree of strain involved and the probability of its occurrence can be judged accordingly.In this paper some results will be given which were obtained by building models of dye and fibre molecules and considering the steric factors which are involved when they are bound together in the dyeing process. Examples were chosen to illustrate the possibilities of using such models and to show that results are sometimes obtained which are contrary to what has been assumed as the result 125 126 carbon. benzene benzene ring (Ca) carbonyl graphite naphthalene ring (Clo) tetrahedral hydrogen , bonded nitrogcn amide ,# STEREOCHEMICAL PROBLEMS I N DYEING of neglecting to use models or using inadequate models. In some cases the facts observed could have been obtained from fairly simple scale drawings or calcula- tions but in other cases the number of degrees of freedom involved in manipulating the large molecules is so high as to make calculation of the required results tedious or even impracticable.Correctness of the results obtained will depend on the suitability of thc data used in the design of the atomic models employed in building the molecules which are under consideration. MODELS usED.-Table 1 summarizes the values for the radii and bond angles used in designing the models of all the atomic species which were here employed. Some of these have already been described in part 1 and 2,19 4 while others were designed since. The design of the later models will be understood from the principles given in part.1 and the following notes. TABLE 1 .-DESCRIPTION OF ATOMIC MODELS USED IN THE INVESTIGATION atomic species covalent radii van der colour bond arrangement b $ ~ ~ ~ ~ ~ Waals radii black , ,I , I 9 aluminium S blue , white s. , yellow 120.0 1 single 2 partial double {e:: 6 single 2 single 1 double 3 partial double - 124 124 112 120.0 - 8 singlc 4 single 1 single 1 single 120.0 2 single 1 partial double 1 single 1 doublc 1 double 2 single 1 partial double 2 single (4 single sockets) double bond 0.695 (partial) 0.665 0.715 (partial) azo oxygen carbonyl , ethcr , negatively charged sulphur di-univalent , tctrahedral , Benzene ring (C~).-This is a block of hexagonal cross-section based on the 1-38 Apartial double bond length for benzene and having a press-stud socket mounted on each of its six sides.The design is such that when univalent atoms are attached to these sockets their positions will be those demanded by the co- valent radius of single-bonded carbon (0-77A). The height of the block normal to the benzene ring corresponds to 3.6A at its centre but thc edges are chamfered so that the height at the edges is only 3.308,. Benzene carbon.-This is a prism of equilateral-triangle cross-section two sockets are mounted on one face and one on each of the others. The model is designed in such a way that if six of them are linked together by alternate double and single links the resulting model will correspond in outline to the c6 block described above.(Making the section of the model an equilateral triangle has not involved any inaccuracy in the single and partial-double bond lengths but the lines passing through the links forming the ring do not pass through the “ nuclei ” which are 0-058 nearer the centre of the ring.) The alternate single and double links not only give the required stability to the ring but they also allow the various canonical forms to be represented and the building of quinonoid structures. “ Graphite” carbon.-This is a prism the section of which is an equilateral triangle designed to give a covalent radius of 0-71 5 A while the height corresponds to a van der Waals radius of 1.70A-thc values found in graphite.5 The socket 105.0 124.0 - - 100.0 109.5 109.5 - - - 1 *o 0.30 - 1.0 and 0.30 0.5 1-50 1 a70 - 1 *oo (compromise value) (A) single bond - 0.77 centre 0.77 0.77 1.65 1.80 0.77 0.77 0.70 0.53 (partial) 0.60 0.55 0.70 0.66 - 1-50 1 -40 1 -40 - (partial) 0.50 - 1 *04 1 -40 1 *85 1.85 CONMAR ROBINSON 127 arrangement is the same as in benzene carbon.In building fused-ring systems this model is used to represent any carbon shared by two or three six-membered rings while the benzene carbon model is used for all unshared carbons. Con- sequently the three possible combinations of the two models give three bond lengths of 1.38 1.415 and 1.43 A.The use of the two atoms is well illustrated in the model of Coronene shown in fig. 5d. In this hydrocarbon six six-membered rings are symmetrically arranged about a central ring. Robertson and White 6 found that the bond lengths in the central rings and those radiating from it are 1.43A while those round the pcriphery are either 1.41 or 1.38 8 according to whether or not a shared carbon atom is involved. The model therefore gives all the bond lengths correct to + 0.005 A. In models of less symmetrical hydrocarbons such as anthracene or pyrene the bond lengths are7 still all within 0.02 In ovalene the discrepancy between the model and the observed bond lengths is higher but here the model is in better agreement with the calculated values. In this investigation these two atom models have been used for building the condensed ring systems (other than naphthalene) in dyes.Naphthalene block (Clo).-The observed bond lengths in naphthalene 8 are not in accordance with theoretical expectations,g the bond joining the two shared carbon atoms having the same value as the two bonds parallel to it. A naphthalene block (Clo) was therefore designed based on the X-ray data of Abrahams Robertson and White.8 The angles of the block were made equal to 120" which was within one degree of the observed vzlues. The plan of the resulting block was a hexagon with two opposite sides longer than the other four. A socket is mounted on each of the four short sides while two are mounted on each of the longer sides. The model has six sides instead of eight as a result of the cutting back of the van der Waals radius to an extent determined by the length of the links.An accurate design demanded a V-shaped cut in the centre of each long side which being less than 0.2A deep could be eliminated for practical purposes. Negatively charged oxygen.-This resembles the double-bonded carbonyl oxygen model but is provided with only one socket while the surface i s cut back so as to give a covalent radius of 0-50 A instead of 0.66 A. When attached to the carbonyl carbon model it gives the correct value of the C-0 bond in a carboxyl ion. Tetrahedra( sulphur.-This was designed as a compromise atom to allow the building of sulphates sulphones and sulphonates. It is a regular tetrahedron with one socket on each face each giving a covalent radius of 1.00h;.Only data for sulphones were available. Combined with the previous model and tetra- hedral carbon it gives C - S = 1.77A and S - 0 = 1-50 compared to the values 1.80 A and 1.43 A for diinethyl sulphone. In this investigation it was used combined with three negatively charged oxygens to represent a sulphonic group. Strained rings.-With the link and collar system described in part 1 the maximum deviation of any bond angle from the normal is only 6". In order to extend the usefulness of the model the design of the brass link was modified so as to allow a deviation of 15" from the normal if the usual collar which provides the restoring force was omitted. By using these links with a sponge rubber collar which gave reasonable stability to the ring though no longer providing a restoring force it was possible to represent even four-membered rings when using the tetra- hedral carbon atom models.The use of this device in building the %- I / C-N/ \C- ring is shown in fig. 5a. The model of such a ring is necessarily imperfect since the valency bond directions no longer pass through the atom centres and the distribution of the deviation from the normal bond angles in a given ring cannot be expected to give the exact values. The device however will often provide a satisfactory first approximation and illustrates the adaptability of the linking STEREOCHEMICAL PROBLEMS IN DYEING 128 mechanism. Where a more accurate representation of a ring is required and the data are available the bond angles could be fixed at the correct value by wedges while bond lengths could be modified by using links of several lengths differing CH3 I say by 0.02A.Choice of models.-The planar amide nitrogen model already described in part 2 4 was used to build the -CONH- groups. This is a prism of triangular cross-section with a single socket mounted on each face. One face is cut back to give a covalent radius of 0.50 A so that when it is joined to a single-bonded carbon atom it gives a partial double-bond distance of 1-30 A. This model was also used to build a planar aromatic amino group instead of the pyramidal amino nitrogen described in part 1. The cut-back hydrogen atom with a cylindrical surface of 0.5A radius described in part 2 was used for all hydrogen bonds.The formation of the bond was judged to be possible if this surface could be brought into contact with the van der Waals radius of the oxygen or nitrogen atom considered. If table 1 is consulted no difficulty should be found in seeing how the atomic models were used as the choice was in each case determined by the valencies shown in the ordinary chemical formula. The standard link and collar were used throughout except when building the five-membered ring shown in fig. 5a. CELLuLosE.-The cellulose chain contains only pure single bonds and one would therefore expect to find close agreement between the dimensions of the model and those found for the actual molecule. The repeat distance in the model was 10.2 to 10.3A.The model draws attention to several facts which would not at once be obvious without it. Cellulose is frequently referred to as a comparatively rigid polymer chain. The model shows that a greater degree of flexibility is to be expected normal to the plane of the ring than in the plane of the ring. Movement in the plane of the ring is restricted by the steric hindrance of the hydroxyl groups. In spite of this a ring could be formed without strain containing 20 glucose rings and 3514 in diameter. However as fig. 2a shows intramolecular hydrogen bonds can be formed by bonding each primary alcohol group to the nearest secondary alcohol group in the next ring. The nature of the environment will decide whether or not such bonds are formed as their formation will have to compete with a tendency to bond to solvent or other cellulose molecules.Under conditions where these intramolecular bonds are formed they would still further restrict the flexing of the chain in the plane of the rings since very little bending is possible without breaking them. NormaZ to the plane of the ring the model was found to be much more flexible and it could be arranged in a circle some 15 A in diameter and containing 9 or 10 glucose rings (fig. 3). Flexing in this direction is not re- stricted by the presence of intra-chain hydrogen bonds which do not tend to separate when the molecule is bent in this way. This anisotropy of flexibility would be an important factor in determining the arrangements assumed by cellulose chains when subjected to orienting forces.Hermans 10 has suggested that the rings in cellulose might be able to undergo a boat-chair transformation which would contribute to the flexibility of the chains. The transformation cannot be carried out in the model without breaking the ring. This suggests that even if it is possible it would require considerable energy to overcome the steric hindrance. SUBSTANTIVE DYES.-Benzopurpurine 4B Q T N = N \ - L - \ ‘f CH3 >- /-< N=N/\/\ SOjNa SO3Na w NH2 CONMAR ROBINSON 129 may be chosen to illustrate some of the steric factors arising in direct cotton dyeing since this dye has been the subject of so many researches. Fig. 4dshows a modcl of the dye. In this thc benzene and naphthalene rings are joined by links giving single bonds.Since these bonds actually have partial double bond character the length of the model is not strictly accurate. The data for trans- azobenzene 11 shows that in the model the C-N-N-C distance is 0.16 A too long while in diphenyl 12 the bond joining the ring is 1-48 A instead of the single bond length of 1.5481. Conscquently the model of the dye would seem to be 0.38A too long which should be taken into account in considering how the dye fits on to a cellulose chain. The dye is shown in the photograph in its coplanar configuration the ability to assume such a planar configuration being as is well known a general property of direct cotton dyes. There is a small amount of steric hindrance from the hydrogens attached to the diphenyl group in the 2 2’ positions which in the model has caused the C-H bonds to be slightly distorted to an extent that can be accommodated by the flexible links.This is probably a good reprcsentation of reality since electron diffraction analysis has shown the separation of the 2 2’ hydrogen in the coplanar diphenyl molecule to be only 1.8481 instead of 2.0A as would be expected if there was no distortion showing that energy is required to force the molecule into coplanarity.12 The well-known effect of substituents such as methyl groups in the 2 2‘ positions destroying the substantivity by making it impossible for the molecule to take up a planar configuration is illustrated by the model of 2 2’ dimethyldiphenyl in fig. 4a in which it will be seen the methyl groups force the benzene rings considerably out of the coplanar position.The diazo groups may be arrangcd in one of two trans configurations (in the cis configuration the rings are not coplanar) as shown in fig. 4b and 4c. In one of these positions a hydrogen bond can be formed it while in the other position the N-H . . . N distance is too great for this to between the diazo group and the amino group which is in the ortho position to be possible. The resulting chelate ring would contribute to the coplanarity of the dye molecule and is shown formed in fig. 46 and 4d. -NH2 or -OH groups that could bond in this way are frequently in this position in azo dyes. The benzene and naphthalene rings and the diazo and amino groups would be all parts of one resonating system with the 7~ electrons in one plane so that both diazo and amino groups would be planar and coplanar with aromatic rings.It follows that the height of the azo and amino nitrogen atoms normal to the plane of the valency bonds should be the same as that of the benzene ring and not less as in the models. The dye will then be coplanar and of uniform thickness except for the bulky -SO3- groups which in the model has a greater height than the benzene ring. If the SO3- group is rotated so that two of the oxygens are parallel to the plane of the naphthalene rings the whole molecule can lie undistorted on a plane surface. THE ATTACHMENT OF THE AZO DYE TO THE CELLULOSE.-Fig. 2b shows model of Benzopurpurine 4B placed on that of the cellulose chain.The remarkable flatness of the cellulose molecule which allows the dye atoms to make contact with cellulose atoms at many points is brought out in the photograph. Attempts have been made to correlate the length of the cellulose repeat unit or multiples of it with the distance between groups which are capable of bonding to cellulose. It is however unlikely that there is any significance in such a relationship since as the model shows if a group (on a dye molecule) capable of forming a hydrogen bond is made to travel parallel to the long axis of the cellulose chain it is never in a position in which it cannot bond with one or other of the OH groups of the cellulose. This can be shown for groups that are either donors or acceptors of protons.Iiefcrence may here be made to attempts to estimate the number of hydrogen bonds formed between the dye molecule and the cellulose from the heat of dyeing. Willis Warwicker Standing and Urquhart conclude that thc heat of dycing of E STEREOCHEMICAL PROBLEMS I N DYEING 130 - 14 kcal/mole which they found for Chrysophenine G and cellulose corresponds to the heat of formation of two hydrogen bonds for each molecule. It should however bc remembered that the heat of dyeing so calcuIated must correspond to the diference between the heat of formation of the bonds which are formed and of any bonds such as cellulose-water or dye-water bonds which must be broken before the dyeing can take place. It would seem therefore that Willis Warwicker Standing and Urquhart's results show that there are forces in addition to those arising from the two hydrogen bonds which can be formed with the diazo groups.Such additional forces seem to be connected with the r electron system which in dircct cotton dyes runs through the aromatic rings and the groups connecting them along the whole length of the molecule. This is borne out by the sub- stantivity of certain vegetable dycs which are characterized by a chain of con- jugated double bonds (bixin etc.14) and of the leuco compounds of anthroquinone vat dyes whcre hydrogen bonding through polar groups can make little or no contribution to the substantivity. Whatever the exact nature of the binding forces which seem to be associated with the resonating system it is of importance to notice that with typical direct dyes the models show that not only the group capable of hydrogen bonding but the entire surface of the dye molecules can comc in contact with the surface of thc cellulose chain.An intcresting exampic of a condensed ring system in a vat dye is given by Calcdon Jade Green 16.17 (dimcthoxydibenzanthrone). A model of the leuco derivative is shown in fig. 5c. The condensed ring system was built with the bcnzeiie and graphitc carbon atoms using graphite carbon for all atoms shared by two or three rings as already described for the coronene model. A drawing of the planar molecule 15 shows considerable overlapping of the van der Wads radii of the methoxy groups (only one methoxy group has been attached to the model as shown in fig.5c). The resonance energy tending to keep the molecule coplanar incrcases with the number of rings in the condensed ring system and therefore would be considerable in such a molecule and thcre would be a marked tendency for the coplanar configuration to be maintained by distortion of bond angles and the van der Waals radii of the inethoxy groups. However if the second methoxy group is attached and the model distorted out of the plane is laid along the cellulose chain it is found that all the carbon atoms can be arranged on an uiidulating surface in which the carbon atoms are little removed from a common plane and which can be followed by the cellulose chain without straining it. Actually since the model allows no distortion of the van der Waals radii and distortion of bond angles would probably be greatcr than the collar and link arrangement allows (6" from the normal) the dye molecule would probably be considcrably more planar than the modcl suggests.Fig. 5b shows a model of Chlorantine Fast Green BLL. This dye is cited by Venkataraman 15 as an example of one which is far from linear. In the model the dye is posed in a configuration which is considerably more linear than that shown by Venkataraman and it will be seen that four out of six of the aromatic nuclei lie on one line with a naphthalene and a benzene nucleus on either side of it (the latter is the only part of the molecule which does not lie in the common plane). There are four intramolecular hydrogen bonds which contribute to the coplanarity of the greater part of this configuration.is known to contribute to sub- AZOIC DYEs.-The amide group -CONH- stantivity in a dye molecule containing it and it is somctimes depicted in the cis configuration with both the -NH- and -CN- groups hydrogen bonded to cellulose.16 Vickerstaff accepts such an arrangement 17 for the method of attachment of azoic dyes to cellulose when referring to the sign of the dichroism observed by Morton 18 in ccllulose dyed with certain dyes showing it when writing the formula in the ordinary conventional manner (fig. 1). Fig. 4e shows a model of the corresponding configuration in which the steric hindrancc was found to be so great that it was impossible to attach the OH group FIG. 2a.-Cellulose straight chain. 2h.-Benzopurpurine 4B on cellulose chain FIG.3.-Cellulose curved in plane normal to rings. [To face p. 130 FIG. 4a.-2 :2”-dimethyl diphenyl. 4h.-trans-diazo group bonded to ortho-amino group. 4c.-trans-diazo group in position where bond is impossible. 4d.-Benzopurpurine 4B. 4e.-cis-configuration of amide group in azoic dye. 4f-trans-configuration of amide group in azoic dye. FIG. 5a.-Solophenyl Yellow FFL. Sb.-Chlordntine Fast Green BLL. 5c.-Caledon Jade Green. 5d.-Coronene. 5e.-Duranol Red 2B. FIG. 6a.-Copolymer of L-leucine and ~-fi-phenylalanine folded as a 3.7 residue Pauling helix. Benzopurpurine 4B shown for comparison in size. 66.-Copolymer of L-leucine and L-/3-phenylalanine in extended /3 configuration. CONMAR ROBINSON -!! P H- lose (Vickerstaff).o=ci3-N=Nu to the naphthalene ring and two of the hydrogen atoms on the benzene ring. The high degree of distortion that would be required by this configuration seems un- likely and suggests that the attachment may not be of this nature (similar steric hindrance will arise when forming the cis configuration in any molecule where any two aromatic rings are attached to an amide group). Fig. 4f shows the corresponding trans con- figuration. Here the NH group can hydrogen bond to the OH group which can also form a bond with the diazo group. It would then apparently be pos- sible for these three groups each to form a bond with an OH group of the cellulose chain. In this case the benzene and naphthalene rings would lie flat on the cellulose chain and parallel to its long axis an arrangement which would also be in agree- ment with the nature of the dichroism since the rest of the conjugated chain would be at right-angles to the cellulose chain as Morton postulated.Fig. 5e shows a model of Duranol Red 2B re- ferred to by Bamford Boulton Hanby and Ward.20 Hydrogen bonds are shown formed between the NH2 and OM groups and the quinonoid oxygens. 131 s o1 FIG. 1.-Azoic dye on cellu- THE DYEING OF SYNTHETIC POLYPEPTIDES.-when dealing with the polypeptide chain in either proteins or synthetic polypeptides we have to consider the dye molecule in relation to a polymer chain which may be folded in more than one way and in which the attached side chain may assume different configurations.The steric problems therefore become much more complicated and for this reason the models are all the more valuable since without them it may be extremely difficult to decide whether or not there is a configuration which will fulfil certain conditions. The use of the models in studying the polypeptide chain has already been described.4319 We shall here consider the dyeing of the 1 1 Copolymer of DL-P-phenylalanine and m-leucine which has been studied by Bamford Boulton Hanby and Ward20 It is of interest to see from models whether steric hindrance of the side chain could prevent a dye molecule from coming into contact with the backbone of the polypeptide chain. A model of the 1 1 copolymer was built in which the leucine and phenylalanine residues were arranged alternately along the chain (this was of course an arbitrary choice for the amino-acid arrangement).The model was built as an L instead of a DL configuration. In fig. 6b the polymer is shown in the p configuration. It will be seen that it would be possible for a dye molecule to lie parallel to the backbone and in contact with it (the model of Benzopurpurine 4B shown above it allows the spatial arrangement to be visualized). This of course does not necessarily mean that a dye molecule will reach the backbone since the penetration of the dye may be prevented by intermolecular forces. In fig. 6a the same copolymer is shown folded as 3.7 residue Pauling helix 21 a fold which may form the basis of the a fold in some polypeptides.The alternate leucine and phenylalanine side chains now appear as two helices as on a two-start screw. Even if a dye is chosen whose greatest width is no more than the diameter of a benzene ring the space between these side-chain helices would not allow it to lie flat on the backbone. This is so even if the side chains are pushed out of their symmetrical positions as much as the model allows. On the other hand the end of a dye molecule could be inserted between the side-chains. It will be noticed that the position of the side chains could not prevent water molecules reaching the backbone in either the a or the configurations of the polypeptides. For any method of attachment other than to end groups the model of the Pauling helix would not lead one to expect the close similarity in the dyeing properties of the a and ,8 fold found by Bamford Boulton Hanby and Ward.It is however 132 by no means certain that the Pauling helix is the fold involved in the dyeing of the cc helix. It should be remembered that the dyeing is probably largely confined to the isotropic regions where the polypeptide chains are not necessarily in the same configuration as in the crystal and further that its configuration may well be modified by contact with the dye. In the 27b fold,4 for instance the backbone would be exposed to the dye in much the same way as in the /? configuration. Thanks are due to Mr. J. E. Goodwin for the photography and his assistance in assembling the models. I am grateful to Mr. E. J. Ambrose for introducing the use of the more flexible link in strained rings and to Dr.G. S . Park for sugges- tions leading to the design of the negative oxygen and tetrahedral sulphur models. ADSORPTION OF DYES BY CRYSTALS 1 Hartley and Robinson Trans. Faruday SOC. 1952 48 847. 2 Stuart 2. physik. Chem. B 1934 27 350. 3 Briegleb Chem. Forschung. 1950 1 642. 4 Robinson and Ambrose Trans. Faraduy SOC. 1952 48 854. 5 Pauling The Nature of the Chemical Bond (Cornell University Press New York 1945). 6 Robertson and White J. Chem. SOC. 1945 607. 12 Karle and Brockway J. Amer. Chem. SOC. 1944 66 1974. 7Robertson Proc. Roy. SOC. A 1951 207 101. 8 Abrahams Robertson and White Acra Cryst. 1949,2,233. 9 Daudel and Daudel J. Chem. Physics 1948 16 630. 10 Hermans Physics and Chemistry of Cellulose Fibres (Elsevier Publishing Co.Inc. 11 Lange Robertson and Woodward Proc. Roy. SOC. A 1939,171 398. 1949). 13 Willis Warwicker Standing and Urquhart Trans. Furaduy SOC. 1945 41 506. 14 Shirm J. prakt. Chem. 1935 144 69. 15 Venkataraman The Chemistry of Synthetic Dyes vol. 2 (Academic Press Inc. New Y ork) . 16 Krzikalla and Eistert J. prakt. Chem. 1935 143 50. 17 Vickerstaff The Physical Chemistry of Dyeing (Oliver and Boyd London 1950). 18 Morton J. SOC. Dyers Col. 1946 62 272. 19 Robinson Nature 1953 172,27. 20 Bamford Boulton Hanby and Ward this Discussion 1953. 21 Pauling Corey and Branson Proc. Nat. Acad. Sci. 1951 37 205. ATOMIC MODELS PART 3.-SOME STEREOCHEMICAL PROBLEMS IN DYEING BY CONMAR ROBINSON Courtaulds Limited Research Laboratory Maidenhead Berks Received 22nd July 1953 Models of dye molecules and of cellulose and polypeptide chains have been built with the type of atomic model previously described by Hartley and Robinson.Their use in investigating stereochemical problems related to dyeing has been illustrated. The anisotropic flexibility of the cellulose chain has been demonstrated. The shape of a number of dye molecules in relation to the shape of the cellulose chain has been discussed and also the part played by intra- and intermolecular hydrogen bonds in both dyes and cellulose. The models suggest that contrary to the usual assumptions hydrogen bonding to cellulose can take place irrespective of the position of those groups in the dye molecule which are capable of forming such bonds.When an amide group links two aromatic rings together it appears unlikely that the group can be in the cis configuration when attached to cellulose as this configuration involves great steric hindrance. The possible part played by steric hindrance in the dyeing of a synthetic polypeptide is discussed. Although atomic models have frequently been used to investigate stereochemical problems in dyeing their usefulness has been limited by shortcomings in the designs of the models available. Recently Hartley and Robinson 1 have described a development of the Stuart type of model. In these the atoms are linked to-gether by a combination of link and collar which allows the valency bond to be distorted a few degrees from the normal when strained but which provides a restoring force which insures that the valency angle shall have its normal value when the model is not under strain.This linking mechanism not only allows molecules to be built in which the valency angles are known to deviate from the normal values but also allows an estimate of the distortion involved when a model of a configuration involving strain is built. Another advantage of the distortable link is that it allows the full value of the van der Waals radius to be used in designing the atoms while in the Stuart models 293 where the valency angles are fixed it is necessary to reduce this radius by an arbitrary percentage (e.g. 15 %) for otherwise no structures could be built in which there was even a small deviation from the normal bond angles or van der Waals radii.The use of atomic models for quantitative work is analogous to that of a calculating machine. However mechanically perfect the models may be any result obtained (e.g. in determinink the relative positions of two atoms imposed by a certain configuration) can only have a degree of accuracy corresponding to that of the data used in their design. For this reason it is a great advantage to be able to use models for which the nearest approach cor-responds to the minimal energy conditions pertaining in crystals since then any distortion of the valency bonds which is found in a configuration will be a measure of the degree of strain involved and the probability of its occurrence can be judged accordingly. In this paper some results will be given which were obtained by building models of dye and fibre molecules and considering the steric factors which are involved when they are bound together in the dyeing process.Examples were chosen to illustrate the possibilities of using such models and to show that results are sometimes obtained which are contrary to what has been assumed as the result 12 126 STEREOCHEMICAL PROBLEMS I N DYEING of neglecting to use models or using inadequate models. In some cases the facts observed could have been obtained from fairly simple scale drawings or calcula-tions but in other cases the number of degrees of freedom involved in manipulating the large molecules is so high as to make calculation of the required results tedious or even impracticable. Correctness of the results obtained will depend on the suitability of thc data used in the design of the atomic models employed in building the molecules which are under consideration.MODELS usED.-Table 1 summarizes the values for the radii and bond angles used in designing the models of all the atomic species which were here employed. Some of these have already been described in part 1 and 2,19 4 while others were designed since. The design of the later models will be understood from the principles given in part. 1 and the following notes. TABLE 1 .-DESCRIPTION OF ATOMIC MODELS USED IN THE INVESTIGATION atomic species carbon. benzene benzene ring (Ca) carbonyl graphite naphthalene ring (Clo) tetrahedral hydrogen , bonded nitrogcn amide ,# azo oxygen carbonyl , ethcr , negatively charged sulphur di-univalent , tctrahedral van der colour bond arrangement b $ ~ ~ ~ ~ ~ Waals radii (A) black 1 single 120.0 -2 partial double centre , 6 single - {e::: 1.65 ,I 2 single 124 124 112 1.80 1 double , 3 partial double 120.0 1 a70 I 8 singlc 9 4 single aluminium 1 single S 1 single blue 2 single, 1 partial double , 1 single 1 doublc white 1 double s.2 single , 1 partial double - -109.5 - - 1 *o - 1.0 and 0.5 120.0 1-50 124.0 1-50 - 1 -40 105.0 1 -40 - 1 -40 yellow 2 single 100.0 1 *85 , (4 single sockets) 109.5 1.85 covalent radii single bond 0.77 0.77 0.77 0.77 0.77 0.30 0.30 0.70 0.70 0.66 --1 *04 1 *oo double bond 0.695 (partial) 0.665 0.715 (partial) 0.53 (partial) 0.60 0.55 0.50 (partial) (compromise value) --Benzene ring (C~).-This is a block of hexagonal cross-section based on the 1-38 Apartial double bond length for benzene and having a press-stud socket mounted on each of its six sides.The design is such that when univalent atoms are attached to these sockets their positions will be those demanded by the co-valent radius of single-bonded carbon (0-77A). The height of the block normal to the benzene ring corresponds to 3.6A at its centre but thc edges are chamfered so that the height at the edges is only 3.308,. Benzene carbon.-This is a prism of equilateral-triangle cross-section two sockets are mounted on one face and one on each of the others. The model is designed in such a way that if six of them are linked together by alternate double and single links the resulting model will correspond in outline to the c6 block described above.(Making the section of the model an equilateral triangle has not involved any inaccuracy in the single and partial-double bond lengths but the lines passing through the links forming the ring do not pass through the “ nuclei ” which are 0-058 nearer the centre of the ring.) The alternate single and double links not only give the required stability to the ring but they also allow the various canonical forms to be represented and the building of quinonoid structures. “ Graphite” carbon.-This is a prism the section of which is an equilateral triangle designed to give a covalent radius of 0-71 5 A while the height corresponds to a van der Waals radius of 1.70A-thc values found in graphite.5 The socke CONMAR ROBINSON 127 arrangement is the same as in benzene carbon.In building fused-ring systems, this model is used to represent any carbon shared by two or three six-membered rings while the benzene carbon model is used for all unshared carbons. Con-sequently the three possible combinations of the two models give three bond lengths of 1.38 1.415 and 1.43 A. The use of the two atoms is well illustrated in the model of Coronene shown in fig. 5d. In this hydrocarbon six six-membered rings are symmetrically arranged about a central ring. Robertson and White 6 found that the bond lengths in the central rings and those radiating from it are 1.43A while those round the pcriphery are either 1.41 or 1.38 8 according to whether or not a shared carbon atom is involved.The model therefore gives all the bond lengths correct to + 0.005 A. In models of less symmetrical hydrocarbons, such as anthracene or pyrene the bond lengths are7 still all within 0.02 In ovalene the discrepancy between the model and the observed bond lengths is higher but here the model is in better agreement with the calculated values. In this investigation these two atom models have been used for building the condensed ring systems (other than naphthalene) in dyes. Naphthalene block (Clo).-The observed bond lengths in naphthalene 8 are not in accordance with theoretical expectations,g the bond joining the two shared carbon atoms having the same value as the two bonds parallel to it.A naphthalene block (Clo) was therefore designed based on the X-ray data of Abrahams Robertson and White.8 The angles of the block were made equal to 120" which was within one degree of the observed vzlues. The plan of the resulting block was a hexagon with two opposite sides longer than the other four. A socket is mounted on each of the four short sides while two are mounted on each of the longer sides. The model has six sides instead of eight as a result of the cutting back of the van der Waals radius to an extent determined by the length of the links. An accurate design demanded a V-shaped cut in the centre of each long side which being less than 0.2A deep could be eliminated for practical purposes. Negatively charged oxygen.-This resembles the double-bonded carbonyl oxygen model but is provided with only one socket while the surface i s cut back so as to give a covalent radius of 0-50 A instead of 0.66 A.When attached to the carbonyl carbon model it gives the correct value of the C-0 bond in a carboxyl ion. Tetrahedra( sulphur.-This was designed as a compromise atom to allow the building of sulphates sulphones and sulphonates. It is a regular tetrahedron, with one socket on each face each giving a covalent radius of 1.00h;. Only data for sulphones were available. Combined with the previous model and tetra-hedral carbon it gives C - S = 1.77A and S - 0 = 1-50 compared to the values 1.80 A and 1.43 A for diinethyl sulphone. In this investigation it was used combined with three negatively charged oxygens to represent a sulphonic group.Strained rings.-With the link and collar system described in part 1 the maximum deviation of any bond angle from the normal is only 6". In order to extend the usefulness of the model the design of the brass link was modified so as to allow a deviation of 15" from the normal if the usual collar which provides the restoring force was omitted. By using these links with a sponge rubber collar, which gave reasonable stability to the ring though no longer providing a restoring force it was possible to represent even four-membered rings when using the tetra-hedral carbon atom models. The use of this device in building the ring is shown in fig. 5a. The model of such a ring is necessarily imperfect since the valency bond directions no longer pass through the atom centres and the distribution of the deviation from the normal bond angles in a given ring cannot be expected to give the exact values.The device however will often provide a satisfactory first approximation and illustrates the adaptability of the linking %- \C-I / C-N 128 STEREOCHEMICAL PROBLEMS IN DYEING mechanism. Where a more accurate representation of a ring is required and the data are available the bond angles could be fixed at the correct value by wedges, while bond lengths could be modified by using links of several lengths differing, say by 0.02A. Choice of models.-The planar amide nitrogen model already described in part 2 4 was used to build the -CONH- groups. This is a prism of triangular cross-section with a single socket mounted on each face.One face is cut back to give a covalent radius of 0.50 A so that when it is joined to a single-bonded carbon atom it gives a partial double-bond distance of 1-30 A. This model was also used to build a planar aromatic amino group instead of the pyramidal amino nitrogen described in part 1. The cut-back hydrogen atom with a cylindrical surface of 0.5A radius described in part 2 was used for all hydrogen bonds. The formation of the bond was judged to be possible if this surface could be brought into contact with the van der Waals radius of the oxygen or nitrogen atom considered. If table 1 is consulted no difficulty should be found in seeing how the atomic models were used as the choice was in each case determined by the valencies shown in the ordinary chemical formula.The standard link and collar were used throughout except when building the five-membered ring shown in fig. 5a. CELLuLosE.-The cellulose chain contains only pure single bonds and one would therefore expect to find close agreement between the dimensions of the model and those found for the actual molecule. The repeat distance in the model was 10.2 to 10.3A. The model draws attention to several facts which would not at once be obvious without it. Cellulose is frequently referred to as a comparatively rigid polymer chain. The model shows that a greater degree of flexibility is to be expected normal to the plane of the ring than in the plane of the ring. Movement in the plane of the ring is restricted by the steric hindrance of the hydroxyl groups.In spite of this a ring could be formed without strain containing 20 glucose rings and 3514 in diameter. However as fig. 2a shows intramolecular hydrogen bonds can be formed by bonding each primary alcohol group to the nearest secondary alcohol group in the next ring. The nature of the environment will decide whether or not such bonds are formed as their formation will have to compete with a tendency to bond to solvent or other cellulose molecules. Under conditions where these intramolecular bonds are formed they would still further restrict the flexing of the chain in the plane of the rings since very little bending is possible without breaking them. NormaZ to the plane of the ring the model was found to be much more flexible and it could be arranged in a circle some 15 A in diameter and containing 9 or 10 glucose rings (fig.3). Flexing in this direction is not re-stricted by the presence of intra-chain hydrogen bonds which do not tend to separate when the molecule is bent in this way. This anisotropy of flexibility would be an important factor in determining the arrangements assumed by cellulose chains when subjected to orienting forces. Hermans 10 has suggested that the rings in cellulose might be able to undergo a boat-chair transformation which would contribute to the flexibility of the chains. The transformation cannot be carried out in the model without breaking the ring. This suggests that even if it is possible it would require considerable energy to overcome the steric hindrance. SUBSTANTIVE DYES.-Benzopurpurine 4B NH2 I CH3 CH3 >- /-< N=N/\/\ w Q T N = N \ - L - \ ‘f SO3Na SOjN CONMAR ROBINSON 129 may be chosen to illustrate some of the steric factors arising in direct cotton dyeing since this dye has been the subject of so many researches.Fig. 4dshows a modcl of the dye. In this thc benzene and naphthalene rings are joined by links giving single bonds. Since these bonds actually have partial double bond character the length of the model is not strictly accurate. The data for trans-azobenzene 11 shows that in the model the C-N-N-C distance is 0.16 A too long while in diphenyl 12 the bond joining the ring is 1-48 A instead of the single bond length of 1.5481. Conscquently the model of the dye would seem to be 0.38A too long which should be taken into account in considering how the dye fits on to a cellulose chain.The dye is shown in the photograph in its coplanar configuration the ability to assume such a planar configuration being as is well known a general property of direct cotton dyes. There is a small amount of steric hindrance from the hydrogens attached to the diphenyl group in the 2 2’ positions which in the model has caused the C-H bonds to be slightly distorted to an extent that can be accommodated by the flexible links. This is probably a good reprcsentation of reality since electron diffraction analysis has shown the separation of the 2 2’ hydrogen in the coplanar diphenyl molecule to be only 1.8481 instead of 2.0A as would be expected if there was no distortion showing that energy is required to force the molecule into coplanarity.12 The well-known effect of substituents such as methyl groups in the 2 2‘ positions destroying the substantivity by making it impossible for the molecule to take up a planar configuration is illustrated by the model of 2 2’ dimethyldiphenyl in fig.4a, in which it will be seen the methyl groups force the benzene rings considerably out of the coplanar position. The diazo groups may be arrangcd in one of two trans configurations (in the cis configuration the rings are not coplanar) as shown in fig. 4b and 4c. In one of these positions a hydrogen bond can be formed between the diazo group and the amino group which is in the ortho position to it while in the other position the N-H . . . N distance is too great for this to be possible.The resulting chelate ring would contribute to the coplanarity of the dye molecule and is shown formed in fig. 46 and 4d. -NH2 or -OH groups that could bond in this way are frequently in this position in azo dyes. The benzene and naphthalene rings and the diazo and amino groups would be all parts of one resonating system with the 7~ electrons in one plane so that both diazo and amino groups would be planar and coplanar with aromatic rings. It follows that the height of the azo and amino nitrogen atoms normal to the plane of the valency bonds should be the same as that of the benzene ring and not less as in the models. The dye will then be coplanar and of uniform thickness except for the bulky -SO3- groups which in the model has a greater height than the benzene ring.If the SO3- group is rotated so that two of the oxygens are parallel to the plane of the naphthalene rings the whole molecule can lie undistorted on a plane surface. of Benzopurpurine 4B placed on that of the cellulose chain. The remarkable flatness of the cellulose molecule which allows the dye atoms to make contact with cellulose atoms at many points is brought out in the photograph. Attempts have been made to correlate the length of the cellulose repeat unit or multiples of it with the distance between groups which are capable of bonding to cellulose. It is however unlikely that there is any significance in such a relationship since, as the model shows if a group (on a dye molecule) capable of forming a hydrogen bond is made to travel parallel to the long axis of the cellulose chain it is never in a position in which it cannot bond with one or other of the OH groups of the cellulose.This can be shown for groups that are either donors or acceptors of protons. Iiefcrence may here be made to attempts to estimate the number of hydrogen bonds formed between the dye molecule and the cellulose from the heat of dyeing. Willis Warwicker Standing and Urquhart conclude that thc heat of dycing of THE ATTACHMENT OF THE AZO DYE TO THE CELLULOSE.-Fig. 2b shows model 130 STEREOCHEMICAL PROBLEMS I N DYEING - 14 kcal/mole which they found for Chrysophenine G and cellulose corresponds to the heat of formation of two hydrogen bonds for each molecule. It should, however bc remembered that the heat of dyeing so calcuIated must correspond to the diference between the heat of formation of the bonds which are formed and of any bonds such as cellulose-water or dye-water bonds which must be broken before the dyeing can take place.It would seem therefore that Willis Warwicker, Standing and Urquhart's results show that there are forces in addition to those arising from the two hydrogen bonds which can be formed with the diazo groups. Such additional forces seem to be connected with the r electron system which in dircct cotton dyes runs through the aromatic rings and the groups connecting them along the whole length of the molecule. This is borne out by the sub-stantivity of certain vegetable dycs which are characterized by a chain of con-jugated double bonds (bixin etc.14) and of the leuco compounds of anthroquinone vat dyes whcre hydrogen bonding through polar groups can make little or no contribution to the substantivity.Whatever the exact nature of the binding forces which seem to be associated with the resonating system it is of importance to notice that with typical direct dyes the models show that not only the group capable of hydrogen bonding but the entire surface of the dye molecules can comc in contact with the surface of thc cellulose chain. An intcresting exampic of a condensed ring system in a vat dye is given by Calcdon Jade Green 16.17 (dimcthoxydibenzanthrone). A model of the leuco derivative is shown in fig. 5c. The condensed ring system was built with the bcnzeiie and graphitc carbon atoms using graphite carbon for all atoms shared by two or three rings as already described for the coronene model.A drawing of the planar molecule 15 shows considerable overlapping of the van der Wads radii of the methoxy groups (only one methoxy group has been attached to the model as shown in fig. 5c). The resonance energy tending to keep the molecule coplanar incrcases with the number of rings in the condensed ring system and therefore would be considerable in such a molecule and thcre would be a marked tendency for the coplanar configuration to be maintained by distortion of bond angles and the van der Waals radii of the inethoxy groups. However if the second methoxy group is attached and the model distorted out of the plane is laid along the cellulose chain it is found that all the carbon atoms can be arranged on an uiidulating surface in which the carbon atoms are little removed from a common plane and which can be followed by the cellulose chain without straining it.Actually since the model allows no distortion of the van der Waals radii and distortion of bond angles would probably be greatcr than the collar and link arrangement allows (6" from the normal) the dye molecule would probably be considcrably more planar than the modcl suggests. Fig. 5b shows a model of Chlorantine Fast Green BLL. This dye is cited by Venkataraman 15 as an example of one which is far from linear. In the model the dye is posed in a configuration which is considerably more linear than that shown by Venkataraman and it will be seen that four out of six of the aromatic nuclei lie on one line with a naphthalene and a benzene nucleus on either side of it (the latter is the only part of the molecule which does not lie in the common plane).There are four intramolecular hydrogen bonds which contribute to the coplanarity of the greater part of this configuration. AZOIC DYEs.-The amide group -CONH- is known to contribute to sub-stantivity in a dye molecule containing it and it is somctimes depicted in the cis configuration with both the -NH- and -CN- groups hydrogen bonded to cellulose.16 Vickerstaff accepts such an arrangement 17 for the method of attachment of azoic dyes to cellulose when referring to the sign of the dichroism observed by Morton 18 in ccllulose dyed with certain dyes showing it when writing the formula in the ordinary conventional manner (fig.1). Fig. 4e shows a model of the corresponding configuration in which the steric hindrancc was found to be so great that it was impossible to attach the OH grou FIG. 2a.-Cellulose straight chain. 2h.-Benzopurpurine 4B on cellulose chain FIG. 3.-Cellulose curved in plane normal to rings. [To face p. 13 FIG. 4a.-2 :2”-dimethyl diphenyl. 4h.-trans-diazo group bonded to ortho-amino group. 4c.-trans-diazo group in position where bond is impossible. 4d.-Benzopurpurine 4B. 4e.-cis-configuration of amide group in azoic dye. 4f-trans-configuration of amide group in azoic dye FIG. 5a.-Solophenyl Yellow FFL. Sb.-Chlordntine Fast Green BLL. 5c.-Caledon Jade Green. 5d.-Coronene. 5e.-Duranol Red 2B FIG. 6a.-Copolymer of L-leucine and ~-fi-phenylalanine folded as a 3.7 residue Pauling helix.Benzopurpurine 4B shown for comparison in size. 66.-Copolymer of L-leucine and L-/3-phenylalanine in extended /3 configuration CONMAR ROBINSON 131 to the naphthalene ring and two of the hydrogen atoms on the benzene ring. The high degree of distortion that would be required by this configuration seems un-likely and suggests that the attachment may not be of this nature (similar steric hindrance will arise when forming the cis configuration in any molecule where any two aromatic rings are attached to an amide group). Fig. 4f shows the corresponding trans con-figuration. Here the NH group can hydrogen bond to the OH group which can also form a bond with the diazo group. It would then apparently be pos-sible for these three groups each to form a bond with an OH group of the cellulose chain.In this case the benzene and naphthalene rings would lie flat on the cellulose chain and parallel to its long axis an arrangement which would also be in agree-ment with the nature of the dichroism since the rest of the conjugated chain would be at right-angles to the cellulose chain as Morton postulated. Fig. 5e shows a model of Duranol Red 2B re-ferred to by Bamford Boulton Hanby and Ward.20 Hydrogen bonds are shown formed between the NH2 and OM groups and the quinonoid oxygens. THE DYEING OF SYNTHETIC POLYPEPTIDES.-when dealing with the polypeptide chain in either proteins or synthetic polypeptides we have to consider the dye molecule in relation to a polymer chain which may be folded in more than one way and in which the attached side chain may assume different configurations.The steric problems therefore become much more complicated and for this reason the models are all the more valuable since without them it may be extremely difficult to decide whether or not there is a configuration which will fulfil certain conditions. The use of the models in studying the polypeptide chain has already been described.4319 We shall here consider the dyeing of the 1 1 Copolymer of DL-P-phenylalanine and m-leucine which has been studied by Bamford Boulton Hanby and Ward20 It is of interest to see from models whether steric hindrance of the side chain could prevent a dye molecule from coming into contact with the backbone of the polypeptide chain.A model of the 1 1 copolymer was built in which the leucine and phenylalanine residues were arranged alternately along the chain (this was, of course an arbitrary choice for the amino-acid arrangement). The model was built as an L instead of a DL configuration. In fig. 6b the polymer is shown in the p configuration. It will be seen that it would be possible for a dye molecule to lie parallel to the backbone and in contact with it (the model of Benzopurpurine 4B shown above it allows the spatial arrangement to be visualized). This of course does not necessarily mean that a dye molecule will reach the backbone since the penetration of the dye may be prevented by intermolecular forces. In fig. 6a the same copolymer is shown folded as 3.7 residue Pauling helix 21 a fold which may form the basis of the a fold in some polypeptides.The alternate leucine and phenylalanine side chains now appear as two helices as on a two-start screw. Even if a dye is chosen whose greatest width is no more than the diameter of a benzene ring the space between these side-chain helices would not allow it to lie flat on the backbone. This is so even if the side chains are pushed out of their symmetrical positions as much as the model allows. On the other hand the end of a dye molecule could be inserted between the side-chains. It will be noticed that the position of the side chains could not prevent water molecules reaching the backbone in either the a or the For any method of attachment other than to end groups the model of the Pauling helix would not lead one to expect the close similarity in the dyeing properties of the a and ,8 fold found by Bamford Boulton Hanby and Ward.It is however, P H-o=ci3-N=Nu lose (Vickerstaff). FIG. 1.-Azoic dye on cellu- s o1 -!! configurations of the polypeptides 132 ADSORPTION OF DYES BY CRYSTALS by no means certain that the Pauling helix is the fold involved in the dyeing of the cc helix. It should be remembered that the dyeing is probably largely confined to the isotropic regions where the polypeptide chains are not necessarily in the same configuration as in the crystal and further that its configuration may well be modified by contact with the dye. In the 27b fold,4 for instance the backbone would be exposed to the dye in much the same way as in the /? configuration. Thanks are due to Mr. J. E. Goodwin for the photography and his assistance in assembling the models. I am grateful to Mr. E. J. Ambrose for introducing the use of the more flexible link in strained rings and to Dr. G. S . Park for sugges-tions leading to the design of the negative oxygen and tetrahedral sulphur models. 1 Hartley and Robinson Trans. Faruday SOC. 1952 48 847. 2 Stuart 2. physik. Chem. B 1934 27 350. 3 Briegleb Chem. Forschung. 1950 1 642. 4 Robinson and Ambrose Trans. Faraduy SOC. 1952 48 854. 5 Pauling The Nature of the Chemical Bond (Cornell University Press New York, 6 Robertson and White J. Chem. SOC. 1945 607. 7Robertson Proc. Roy. SOC. A 1951 207 101. 8 Abrahams Robertson and White Acra Cryst. 1949,2,233. 9 Daudel and Daudel J. Chem. Physics 1948 16 630. 1945). 10 Hermans Physics and Chemistry of Cellulose Fibres (Elsevier Publishing Co. Inc., 11 Lange Robertson and Woodward Proc. Roy. SOC. A 1939,171 398. 12 Karle and Brockway J. Amer. Chem. SOC. 1944 66 1974. 13 Willis Warwicker Standing and Urquhart Trans. Furaduy SOC. 1945 41 506. 14 Shirm J. prakt. Chem. 1935 144 69. 15 Venkataraman The Chemistry of Synthetic Dyes vol. 2 (Academic Press Inc. New 16 Krzikalla and Eistert J. prakt. Chem. 1935 143 50. 17 Vickerstaff The Physical Chemistry of Dyeing (Oliver and Boyd London 1950). 18 Morton J. SOC. Dyers Col. 1946 62 272. 19 Robinson Nature 1953 172,27. 20 Bamford Boulton Hanby and Ward this Discussion 1953. 21 Pauling Corey and Branson Proc. Nat. Acad. Sci. 1951 37 205. 1949). Y ork)