22 M. H. F. WILKINS 



Although the two pairs give almost the same separation of the glycosidic 

 bonds, the direction of the bonds cannot be made exactly the same. Some 

 of the hydrogen bonds deviate appreciably from linearity. However, the 

 stereochemical shortcomings are not so great that, considered from this 

 viewpoint alone, the scheme would appear unlikely to apply to DNA. 

 Other somewhat unfavourable features of the Donohue scheme have been 

 discussed [lo] e.g. considerations of the directions of atomic sequence in 

 the two phosphate-ester chains in the molecule. The main difTerence 

 between the Donohue and Watson Crick schemes lies in the symmetrical 

 relationship of the glycosidic links. In the Donohue scheme these bonds 

 are not symmetrical with respect to a dyad axis perpendicular to the helix 

 axis : as a result a double helix molecule with a structure of this kind will 

 appear different when turned upside down, i.e. rotated i8o about a line 

 at right angles to the helix axis. In other words a Donohue-type molecule 

 may be specified with respect to a direction along the helix axis while 

 Watson Crick molecules are symmetrical and are the same in both 

 directions. 



The second scheme (Fig. 7) involves a hydrogen bond arrangement 

 found between adenine and thymine [14, 15] and derives from a suggestion 

 made by Professor Linus Pauling. We will refer to the scheme as the 

 Hoogsteen scheme. The base-pairing is symmetrical as in the Watson- 

 Crick scheme. The scheme requires cytosine to have a not very probable 

 tautomeric form in which a hydrogen atom is moved from an amino group 

 to a ring nitrogen atom. 



Possible ways of constructing DNA models incorporating base- 

 pairs alternative to those of Watson and Crick 



In considering molecular models of DNA involving base-pairing 

 schemes other than that of Watson and Crick, it is necessary that the 

 position of the base, sugar, and phosphate parts of the molecule be placed 

 in approximately the same position as in the Watson Crick type model 

 we have described [3, 4]. I think we can safely assume that no other 

 arrangement would be compatible with the X-ray data. The idea that our 

 model is at least in this sense unique is confirmed by comparison of the 

 B and C configurations of DNA. We have shown that a small change in the 

 configuration (mainly a displacement of the nucleotide position in the 

 helical structure) causes the B diffraction pattern to change into the C 

 pattern [16, 17]. 



Let us attempt to build, with Donohue pairs, a DNA model as required 

 with the phosphate group positions related by the dyad axis referred to 

 above. The glycosidic links will not be related by the dyad axis. Hence 

 the base pair and the sugar groups will both be placed asymmetrically in 



