24 M. H. F. WILKINS 



relation to the dyad. In each nucleotide pair the base-pair will be arranged 

 lop-sided. There will be two arrangements: with base-pair extending as 

 in Fig. 8(rt) from bottom left to top right, or from top left to bottom right. 

 If every base-pair is arranged in the same way, a regular helical model 

 will be built. If this model is turned upside down the lopsided arrange- 

 ment of the base-pairs will appear to have changed — the helical system 

 of base-pairs will have rotated relative to the phosphate groups. If the model 

 is viewed as a simple helix, the rotation is equivalent to a displacement in 

 the helix axis direction. In a fibre of DNA there will on the average be an 

 equal number of molecules up one way or the other. If one molecule 

 passes through the repeat unit in the structure, as in the A structure [i8], 

 each repeat unit will at random contain base-pairs in one position or the 

 other. This random arrangement will give rise to continuous diffraction 

 along the layer lines on the diffraction pattern. The A type diffraction 

 patterns of DNA give no trace of this continuous diffraction (Fig. 9). The 

 possibility that the molecule contains an appreciable degree of asymmetry, 

 as is required by the Donohue scheme, is therefore excluded. 



There may be a possible way out of this difficulty. If the model 

 were not built in a regular helical fashion but if successive base-pairs in 

 a molecule were placed at random lopsided one way or the other, the 

 base-pairs would on the average be symmetrical. The shape of the resultant 

 average base-pair, consisting of two lopsided base-pairs lying criss-cross 

 (Fig. 8(6)), would be extended considerably in the plane of the base-pairs. 

 Such a model is unattractive because of its considerable irregularity 

 (probably it would not be possible to form the phosphate-ester chains) 

 and because the base-pairs would be stacked on one another only to a 

 small extent and as a result their hydrophobic surfaces would be largely 

 exposed. 



The Hoogsteen base-pairs have the desirable feature, like those of 

 Watson and Crick, of symmetrically-placed glycosidic links. The distance 

 between the glycosidic links is smaller than in the Watson-Crick scheme 

 by about 2 A. (This difference could be larger — we have used a minimum 

 value estimated by Dr. M. Spencer.) The other main difference in the 

 base-pairs is that the six-membered ring in purines is placed, in the 

 Hoogsteen scheme, to one side of the pair of hydrogen bonds (at A' in Fig. 

 10), whereas in the Watson-Crick scheme this ring is at V in line with 

 the bonds. 



A molecular model resembling DNA has been built consisting of 

 adenine-thymine pairs of Hoogsteen type [19]. The diffraction pattern of 

 the model resembles that of DNA but the agreement between calculated 

 and observed intensities is not so good as that obtained with Watson- 

 Crick pairing. However, the best way of comparing the agreement between 

 calculated and observed intensities for different models is to use the Fourier 



