MOLECULAR CONFIGURATION OF NUCLEIC ACIDS 27 



molecular models with alternative base-pairs or calculating diffraction 

 from the models. If the base-pairs in the DXA model require modification, 

 the nature of the modification should appear when the F'ourier synthesis 

 is examined. 



Consider first the Donohue scheme. There is no indication in the 

 syntheses in Figs. 3 and 4 that the average base-pair should be made 

 larger, in the plane of the base-pair, than the average Watson Crick pair. 

 Therefore it is unlikely that the Donohue scheme exists in DNA. We may 

 note, however, that a double-helix structure with glycosidic links arranged 

 as in the Donohue scheme has been established for the two-chain complex 

 of polyriboadenylic acid ([2] and private communication). The structure 

 is clearly distinguishable from that of DNA, and the sequences of atoms 

 in the two phosphate-ester chains run in the same direction and not in the 

 opposite direction as experimental evidence (other than that obtained from 

 X-ray diffraction) indicates is the case for DNA [i]. 



Examination of the Fourier syntheses also shows that the Hoogsteen 

 scheme almost certainly cannot exist in DNA. First, it seems most un- 

 likely that, as required in the Hoogsteen scheme, the distance between 

 glycosidic links could be 2 A less than in our model : if any alteration is 

 required it is that the distance should be increased slightly. Second, there 

 is no indication that the position of the purines in the base-pairs should be 

 altered to that in the Hoogsteen pairs. This is shown most clearly in the 

 Fourier difference synthesis (Fig. 4), Such a synthesis corresponds roughly 

 to the difference between the real structure and the model used in the 

 synthesis. If the real structure contained Hoogsteen pairs, the difference 

 synthesis would have positive sign at AA Fig. 4, w^here the six-membered 

 purine ring occurs in the Hoogsteen scheme, and negative at BB, where the 

 ring occurs in the average Watson-Crick pair. The observed signs are 

 opposite to those expected if the Hoogsteen scheme were correct. Hence, 

 once more, the existence of the Hoogsteen scheme appears unlikely. We 

 hope to confirm this conclusion by calculating a Fourier difference 

 synthesis for the Langridge-Rich model containing Hoogsteen adenine- 

 thymine pairs. If our approach is correct the synthesis will indicate that 

 the Hoogsteen base-pair in the model should be replaced by the Watson- 

 Crick pair. 



It might be thought that because the X-ray data cannot resolve 

 spacings less than 3 A and as a result show individual atoms, the data 

 could not be used to distinguish between two tvpes of base-pairing in 

 which the positions of the atoms differed in the main bv less than 3 A. 

 This, however, is not so : the data can be used to distinguish between a 

 structure that is nearly correct and one in which the atoms are displaced 

 by I A or even less. In the case of well-defined groups, such as the phos- 

 phate group, the position may be determined to within 0-5 A. We are, 



