In the Beginning was Information

If four letters (quartets) are represented in binary code (n = 2), then(4 letters per word)x(1 bit per letter) = 4 bits per word, which isless than the required 4.32 bits per word. This limit is indicated bythe hatched boundary in Figure 19. The six fields adjacent to thisline, numbered 1 to 6, are the best candidates. All other fields lyingfurther to the right, could also be considered, but they would requiretoo much material for storage. So we only have to considerthe six numbered cases.It is in principle possible to use quintets of binary codes, resultingin an average of 5 bits per word, but the replication process requiresan even number of symbols. We can thus exclude ternarycode (n = 3) and quinary code (n = 5). The next candidate is binarycode (No 2), but it needs too much storage material in relation toNo 4 (a quaternary code using triplets), 5 symbols versus 3 impliesa surplus of 67 %. At this stage we have only two remaining candidatesout of the large number of possibilities, namely No 4 and No6. And our choice falls on No 4, which is a combination of tripletsfrom a quaternary code having four different letters. Although No 4has the disadvantage of requiring 50 % more material than No 6, ithas advantages which more than compensate for this disadvantage,namely:– With six different symbols the recognition and translation requirementsbecome disproportionately much more complexthan with four letters, and thus requires much more material forthese purposes.– In the case of No 4 the information content of a word i W is 6 bitsper word, as against 5.17 bits per word for No 6. The resultingredundancy is thus greater, and this ensures greater accuracy forthe transfer of information.Conclusion: The coding system used for living beings is optimalfrom an engineering standpoint. This fact strengthens the argumentthat it was a case of purposeful design rather than fortuitouschance.95