We had this problem of genetic suppression. And this is a… another interesting story because it is really our beginnings with genetic engineering; it's what I used to call genetic steam engineering, because we had to do things by brute force. And one of the things we wondered was a consequence of our… of suppression is that when we put back an amino acid into the amber codon what we've changed is a normal transfer RNA. And we… and that would mean of course that there had to be either many… many… two… at least two different genes, because we would take it away from its normal reading and allow it to read this and make a misprint, so to speak. And that we had different suppressors, which put in different amino acids, and therefore these suppressors were therefore related, their anti-codons would be related to the codons… would be… the anti-codons of these amino acids would have the same relation to the… to reading the suppressed form as the codons of the chain termination had to the codons for these amino acids. Thus, if you changed… in a protein you changed CAG glutamine to TAG or UAG amber mutant – same chain termination – then by definition you could change the amino acids, the transfer RNA which reads CAG – which presumably has a G where there's a C, a… a G to recognise the C – you could change that G into an A so it could recognise the U in this. It was a very simple model that... of tit for tat, so to speak, with what could be… would have to be proved genetically. And, of course, one way of proving it was of course to work out the sequence of transfer RNA in the modern in this… these… in these beasts. Fred Sanger had been developing RNA sequencing techniques and in fact the structure of a… of an RNA had been determined by Holley, and of course that… that RNA had been purified from yeast by an enormous amount of work using countercurrent distribution, it had been sequenced by a very elaborate, very… well, I shouldn't say elaborate… a very simple means, but it required a lot of material. And Fred was working out micro methods for doing this, had succeeded in doing this, and so our very simple idea which is: you just do it for proteins. If you have a mutant RNA, you can just determine its sequence and... and just work it out. A question: how do you find this RNA? You have to purify it. And it didn't seem reasonable to us; it just seemed very difficult to us. However, I was very much imbued with what I'd done with the phage protein because I had avoided purification simply by just using something that was a huge amount of the synthesis and that could be… so to speak, dominate everything else. We didn't have to purify it, because as long as everything else is spread over hundreds of species, if yours is a half or even a third you only see yours as the intense thing, because everything else is background. So we felt we could do these things with rather impure samples. And furthermore, we could… with phage infection you could just put the label in at the time this was growing and so you could get that very much labelled.