VI. RNA AND CODING PROBLEMS 285 



would be a three-letter one where 64 different sets are possible. This 

 scheme was first proposed by Gamow (1954), who, on the basis of DNA 

 structure, considered the code to be overlapping; that is, the letters of 

 one triplet can be used sequentially in adjacent triplets. This coding 

 mechanism would predict that a change in one base would result in 

 changes in three neighboring amino acids in a protein. However, such 

 observations that nitrous acid mutants generally are reflected in the 

 change of only one amino acid at a time (Tsugita, 1961 ; Tsugita and 

 Fraenkel-Conrat, 1960), and that the variety of abnormal human hemo- 

 globins differ by only single amino acid changes, (Watson and Kendrew, 

 1961) seem to rule out the overlapping code. Helinski and Yanofsky 

 (1962) and Henning and Yanofsky (1962) recently found that an 

 ultraviolet-produced mutant of E. coli K12 contains the A protein of 

 tryptophan synthetase differing from normal A protein by only a single 

 amino acid substitution, glutamic acid for glycine. This same glycine 

 is replaced by arginine in the A protein of a closely linked mutant. 



Since 64 triplets can determine only 20 amino acids, it was felt that, 

 with the exclusion of an overlapping mechanism, there must be a 

 scheme to select the correct triplets along a continuous sequence. Crick 

 et al. (1957) suggested that, although all possible sequences of amino 

 acids can be coded, only those nucleotide triplets that sit side by side 

 make sense, and the overlapping triplets so formed must be nonsense. In 

 this way, they could demonstrate that the maximum number that can 

 be coded is 20. Since then. Crick et al. ( 1961), on the basis of ingeniously 

 conceived genetic experimentation with the B cistron of the rll region 

 of bacteriophage T4, have proposed a non-overlapping triplet code where 

 selection of the correct sequence of bases depends on reading at a fixed 

 starting point. In this way, by starting at such a point and reading 

 toward one direction, the insertion or deletion of a base would result in 

 incorrect further reading of the code. However, an additional insertion 

 or deletion (double mutant) will result in a correction of the code- 

 reading in all sequences following the second mistake. Thus, recovery 

 of most of the original information (in their experiments recoveiy to 

 pseudo-wild type) can be obtained. It should be noted that by use of 

 the proper triple mutants they obtained firm evidence for a probable 

 coding ratio of three. Insofar as the region over which the suppressors 

 they investigated covered about a quarter of the B cistron (the protein 

 produced by this cistron may contain about 200 amino acids), the 

 authors feel that the code is probably degenerate; otherwise, nonsense 

 reading, which should have been encountered at a much closer distance, 

 has been compensated for by the flexibility afforded by a degenerate 

 code. It will be of interest to see whether as predicted (Crick et al., 



