We had been doing experiments on C. elegans [Caenorhabditis elegans], and we'd already known... we'd already worked out how much DNA there was, John Sulston had done this by cot curves. We found quite a small value, which of course has proved to be very interesting. We found very little repetitive DNA, which made it different from higher organisms, but there still was a lot. There was still at least twenty times the amount in a bacterium. And the situation was that we would still have to explain quite a lot for C. elegans which didn't have all of this... these dispersed repeats. The dispersed repeats were supposed to, in the theory of regulation, be the different controlling elements in combination. Now I think the... looking back on it, and of course at the time the thing that I did not like about this and said so, is that it assumes what is called the...  what... that there's only a two-valued logic in developmental systems. Namely, you need to have it, you don't need to have it, namely on or off. And if you have to specify that only the on things are relevant, then you have a great problem of turning everything off in every other case. So suppose you have 100 cases, for example, 100 different cell types, you have to make a list for each case. Following genes will be on, following genes will be off. Those lists will be disjointed, we say. So you'll have to have different ways of regulating combinations. That takes a lot to do, and in fact you can see this, that the best way to do this for the extreme cases: let's take one case that one gene is on in all cells... but one, right? So what you do is you let it be on anyway, and you turn it off in the case that it doesn't; you don't have 99 ways of saying you've got to be on. The other... and of course if a gene is only on in one cell it is best to turn it off everywhere and make an exception of this. And of course the payoff there gets into problems when you get about halfway. What are you going to do if it's in 30? You know, where does it pay you to do this? Furthermore, it makes no allowance for the fact that there may be don't care conditions. That is, if you think that a muscle protein turned on in a neuron has no effect, there'll be no reason to worry about turning it on or off, so you can just leave it. And in fact, I mean, there was a remarkable experiment done which illustrates this, which is that someone grafted the eye of a tadpole to just above its tail, and of course when the tail shrank down, then the eye got positioned above the... the cloaca and it grew back. What was remarkable is it grew back in one column of the spinal cord and one column only. So you see, in that case we didn't have to specify for the retina, don't grow in any of these columns, because it never encountered them in the normal course of development. I mean, nature can't worry about cases where the eye suddenly occurs above the anus. I mean, it's not a natural situation. Therefore it became clear that there could be many don't care situations, and therefore the control didn't have to have the versatility. So it was not a strong argument, and I never liked it. And we had many arguments about this.