MOLECULAR CONFIGURATION OF NUCLEIC ACIDS 1 5 



power by recording diffraction at angles corresponding to spacings as small 

 as I • I A. As mentioned later, we have not yet analyzed these new data 

 but expect that they will improve considerably the results described here. 

 In our earlier work we had in the main analyzed the diffraction data in 

 two dimensions and by trial had adjusted a molecular model until it was in 

 agreement with the diffraction data in two dimensions [3, 4]. We have 

 now used the three-dimensional data to check the accuracy of the model 

 and to find the extent to which it is a unique solution. The Fourier 

 synthesis method is convenient for this purpose. The syntheses which I 

 will describe have been obtained by Dr. D. A. Marvin (aided by a computer 

 programme written by Dr. O. S. Mills). 



Principles of the Fourier synthesis method of structure analysis 



The Fourier synthesis method [5] may in simple terms be described 

 as follows. The diffracted X-rays have three characteristics : 



1. Direction of diffraction. 



2. Amplitude (given by the measured intensities). 



3. Phase. 



The direction of diffraction corresponds to the spacing of the electron 

 density in the structure. A Fourier synthesis is produced if the various 

 spacings, with correct amplitudes, are added together or subtracted. The 

 result is that one obtains directly a picture of the structure. The phases of 

 the diffracted beams, roughlv speaking, tell one which amplitudes are to 

 be added and which subtracted. Without knowledge of the phases the 

 structure cannot be derived. The main difficulty with the X-ray diffraction 

 method is that the phases are not given directly by the X-ray photograph. 

 In favourable cases, phases can be derived by measuring intensities with 

 and without a heaw atom placed in the molecule. This method has been 

 used in the remarkably successful structure analysis of myoglobin [6]. 



Another approach is as follows. If one has already a roughly correct 

 idea of the structure or of an appreciable part of it, one can obtain a com- 

 plete and exact structure, provided that the X-ray intensity data are com- 

 plete and exact. The procedure is to calculate the phases from the rough 

 structure, in this case our molecular model, and perform a Fourier synthesis 

 using the experimentally determined amplitudes and the calculated phases. 

 The picture given by this synthesis has the following characteristics : 



1. If the model is correct the picture is like the model (except that the 

 limited resolving power mav not enable all the detail of the model 

 to be seen.) 



2. If the model requires modification the picture is intermediate 

 between the model and the correct structure. As a result one can 

 see how to adjust the model to make it more nearly correct. 



