Benzer and Champe studied the properties of deletions that cover the boundary between rIIA and rIIB. As expected, most of them cannot provide either function during infection of a non-permissive strain. One deletion however was highly unusual and was still able to provide rIIB function even though it lacks 10% of the rIIB sites. This deletion was instrumental in confirming the general nature of the genetic code proposed by Crick et al. In the discussion, the authors evoke the notion of bi-functional enzymes such as tryptophane synthase. In bacteria, the two catalytic activities are performed by individual proteins encoded by adjacent enzymes. In eucaryotes, both reactions are performed by a single protein: the product of the first reaction does not diffuse out but is “funneled” into the second active site. This is just one exception to the one gene one enzyme model discussed in the first session.

Crick et al. start their paper by presenting the evidence that the code must be non-overlapping. One evidence, provided by Brenner, is the founding work of what will become bioinformatics. Starting with a single rIIB frameshift mutation, now known to involve the addition of a single base pair that displaces the reading frame of the mRNA, they isolate many intragenic suppressor mutations that restore rIIB function. The original mutation is given a + sign, and its suppressors a - sign. They then isolate suppressors of these suppressors that are themselves + mutants. All tested combinations of two + and two – mutations lack rIIB function. Most combinations of a + and a – mutations have rIIB function; the other combinations are proposed to generate a stop codon between the two frame shifts. Finally, a number of triple mutants were shown to have rIIB function. They use the unusual deletion that fuses the two rII genes to demonstrate that frameshift mutation located in the rIIA portion of the fused gene abolish its rIIB function. The simplest interpretation is that + mutants have one more (or one less) base pair and that – mutants have one less (or one more) base pair. Although the general nature of the code is 3n base pairs per amino acid, they belive that the code is a three rather than a six letters code.