GENETIC CONTROL 37 



be arranged. There are many different languages operating on this principle, 

 and several types of comma-less codes have been examined (Freudenthal, 

 1958; Golomb et al., 1958). Such systems could code any sequence and 

 would impose no constraints or restrictions upon the sequence of amino 

 acids. But strong restrictions would exist in the arrangement of the 

 nucleotides in DNA. 



It should not be forgotten that this is pure speculation. If taken too 

 seriously these entertaining schemes might obscure the situation rather 

 than to clarify it: the more clever they look, the greater is the danger that 

 they might blind us to reality. On the other hand schemes of this sort will 

 help to discuss and analyse the experimental data which will soon be 

 obtained. 



Data relevant to the coding problem already emerge from the compara- 

 tive study of the fine structure of proteins. Thus Brenner (1957) showed 

 that among the sequences of amino acids known to exist in proteins, there 

 are more possibilities than an all overlapping code could allow (Gamow 

 et al., 1956). Morowitz (1959), studying the amino acids which are found 

 in the protein of E. coli next to a given amino acid, concluded that there are 

 no constraints as to the nature of the amino acid which is contiguous to 

 glutamic or aspartic acid or which precedes arginine or lysine. But leucine 

 or isoleucine are found next to arginine and lysine less frequently than 

 would be predicted from equally probable distribution (Morowitz and 

 Barra, 1959). These limited constraints may or may not be due to the 

 coding system. 



Comparison of abnormal haemoglobins in man shows that in mutations 

 which result in the production of Hb S or Hb C instead of Hb A, one 

 glutamic acid is replaced by valine or lysine respectively, and the contigu- 

 ous amino acids are not affected. In the mutation of Hb A to Hb G, the 

 amino acid which is replaced is the one next to the glutamic acid residue 

 which was replaced in the mutation of Hb A to Hb S or Hb C. These facts 

 are not compatible with an overlapping code; they show that the actual 

 system allows for a large degree of independence of contiguous amino 

 acids. 



In haemoglobin again, single mutations can cause the replacement of 

 glutamic acid by either valine, lysine or glycine (see Fig. 11). This suggests 

 that the ciphers for these four amino acids differ only slightly from one 

 another, perhaps by one nucleotide in each case. The number of different 

 substitutions of one given amino acid, e.g. glutamic acid, by another as a 

 result of single mutations might also be a test for certain coding systems ; 

 for instance, in the 'code without commas' the number of possible substitu- 

 tions which still 'make sense' is very limited (Levinthal, 1959). The dis- 

 covery of Tsugita and Fraenkel-Conrat (1960) that a mutation brought 

 about by deamination in virus RNA can cause the substitution of leucine 



