134 J. C. KENDREW [9 



Hg(NH3)2+ and finally the triple complex containing Au, PCMBS and 

 Hg(NH3)2+ (see for example Fig. 6). Theoretically two separate replace- 

 ments would suffice, but in practice it is a great advantage to use three or 

 four, since experimental errors are considerable, and for any particular re- 

 flexion the contributions of one or more of the heavy atoms may happen 

 to be so small as to preclude their being used for phase determination. 



Fig. 6. Difference-Fourier projection of double heavy-atom derivative of Type A myo- 

 globin along y, containing one mole each of PCMBS and Hg(NH3)2 + per mole of protein. 



In addition, it is a matter of some crystallographic interest that with certain of 

 the complexes a failure of Friedel's Law is observed, in other words, it is found 

 that with these crystals the intensities of two so-called 'equivalent reflexions', 

 020 and 020 for example, are different. It can be shown that this effect will be pro- 

 nounced, if the wavelength of the X-rays used is in the neighbourhood of that at 

 which one or more of the atoms of the structure have an X-ray absorption band. 

 Pepinsky and Okaya^ have shown how it can be used to resolve the ambiguity of 

 phase determination with a single isomorphous replacement; in protein crystals 

 the anomalous effect is not large enough for this to be done, but it can be a valu- 

 able confirmation of the phase determined by other methods. This anomalous 

 dispersion effect has also been used by Blow in his work on the x projection of 

 haemoglobin. 



In our phase determinations with myoglobin we have so far confined 

 our attention to reflexions of spacing greater than 6 Â., the intention being 

 to compute in the first instance a three-dimensional synthesis with this re- 

 solution. The number of reflexions involved is rather more than 400, of 

 which about one-quarter are real, with signs already determined in the 

 two-dimensional work. We are at the present time calculating the phases 

 of the remaining 300 reflexions. This task is not yet completed, but as a 

 first instalment of the results, a Fourier projection along the z axis of the 

 crystal is shown in Fig. 7 ; this is computed from 50 terms of which 40 have 

 general phase. This projection is, of course, subject to the same confusions 

 due to overlap as is the y projection illustrated in Fig. 2; it has been repro- 

 duced mainly to demonstrate the practicabihty of general phase determina- 

 tion in a protein crystal. An analogous projection of haemoglobin has 

 recently been calculated by Blow. 



