Everybody knew there was this book which he was famous for – Theory of Harmonic Integrals [sic] – but we didn't study it, and even when I started working with him he gave… he started me off on characteristic classes sort of things, work that he was doing with Allendoerfer and Chern – Chern classes and things like that. But pretty quickly I suppose I had to pick up the harmonic integrals theory at least to know the results, know the outlines. But there was never a course on harmonic integrals, it would have been much too difficult, and it wasn't regarded as, sort of, a suitable book even for Part III students. But I must have been exposed to it in my first year of research. I pretty soon had to learn, know all about harmonic forms, and when sheaf cohomology came along which it did in my first year, I had to know. So I'm not quite sure how I picked it up, but it was known. Mind you the results are easy and the proofs are difficult, so you can learn the formalism. I did actually read through it some stage in part, but I didn't learn it directly from him and there were no courses.

So you pick these things up. My memory is a bit vague now as to exactly how, but… and later on, after a while, it was… clearly it was central to the subject and you have to relate to sheaf cohomology and so on, and so by that time you had to know more about it. And later on, I learnt a bit more of the analysis. Early stages I didn't know the analysis, but later on when you got more involved I brushed up on the analysis and learnt about it. The early stages, you just know the applications, the algebraic geometry side and things like that.