That brings me to the fact that, by that time, having had this privilege of colleagues of such remarkable talent, I decided that I didn't want to stay in antibodies and in immunology. I never thought of myself as an immunologist anyhow and I never even thought of myself as a physical chemist, although that was a closer draw, but at that point I started to think in a way that has pervaded everything I've done ever since, plus or minus. I started to think of recognition, the problem of recognition, and that turned my attention to a field which is in some sense overlapping with immunology, at least in the early stages, and that is the whole field of embryogenesis and morphogenesis: how do you inherit your great granddad's nose; how do you make structure out of cells and cell systems and tissues etc., etc.?

Well, the problem isn't so remote because some great scientists had specified how you might do it. So I have to go back here to history and sort of say, well, one scientist, Haeckel, had this ontogenetic law – the law that ontogeny recapitulates phylogeny. If you accept Darwin – and he was a great fan of Darwin's – then he said, ‘When you go through and look at embryos, they go through the early stages of evolution until they come to... so you get the frog stage and the chicken stage, I suppose, and whatever.’ It turns out that of course that's not true, and there's a great book by Stephen Jay Gould called Ontology and Phylogeny which describes why it's not true in a very beautiful way. But then there was the fellow, von Baer, who was actually before Haeckel, who said, ‘No, each one does it individually’, and that came out closer to be the truth.

But then embryology in its later history took a funny turn because it became mechanized and the scientists His and Roux who opened a whole journal, concerned with what they called Entwicklungsmechanik – the mechanics of structure; how do you put the whole structure together? You will remember that during embryogenesis, what happens if you have a flat plate or a round bowl of cells, something called gastrulation sets in and you make a tube, and it has a head end and a tail end, at least for our kind of creature, and they concerned themselves with that. But then they moved away from the genetics, and so the true problem on morpho... morphogenesis – not fully solved by any means today – is how do you in fact inherit the structures that this morphogenetic mechanics gives rise to? How do you reconcile the one dimensional structure DNA and its message with the three and four dimensional structure folding up an embryo so that, in some way, the... for instance, if I had twins and I had identical noses, at least from a plastic surgeon's point of view, and I had the ability to color the cells in each twin's nose, or number them, how come the noses are alike but the cell color distributions are different, and yet you have it?

Well, as I said before, some rather great scientists – for example, Sperry for the nervous system – came up with notions, the so-called chemoaffinity hypothesis, which looked a little bit like antibody recognition... not quite. But the idea was that right down almost to the level of each single nerve cell and neuron you'd have a specific marker that would say, ‘Plug this guy in, that guy in, that guy in.’ Well, that of course fascinated me. At the same time I was fascinated by the fact that people who had been working on this business of putting together an embryo really had some rather vague, not vague but general... too general, non-biological theories – namely that the way you made an embryo was by non-specific forces, so-called Van der Waals interactions, and hydrogen bonds, and it was never attributed to a molecule. I remember, and this connects me up with my latter-day career, going to the Neurosciences Research program... I'll come to that later because I'm now a chairman of it... going and hearing Dick Sidman from Harvard talk about a mutant in a mouse... staggerer: a mutant in a cerebellum in which they could have point mutations and it would cause big differences in the structure of that brain and finally the behavior of the mouse. And I remember on the airplane back thinking, well, look, if it's a molecule then there has to be molecules that do it because you can't get point mutations under Van der Waal's force. So let's look for the molecule. And that's what I started doing. And I remember the same kind of reaction from some of my colleagues, which is... this is a waste of time. And for a year and a half it did seem that way.