He kept a little notebook with useful formulae, which were all exact - perfect. Every factor of two correct, every sign correct, every dimension correct, everything was right. He never put anything into his little notebook that he carried around with him without making sure it was perfectly correct. And that made it possible for him to do a lot of practical problems just on being asked because he had worked them out before and some of the critical formulae were in his little book. Most things he had thought about before in some form or other, most questions in physics were questions, were new forms of questions that he had already answered. And so if you asked him something he would begin at the top left of the blackboard and write line after line and then later on, perhaps at the bottom right or else before that, he would get the answer and put a box around it. And the answer was always correct; he never made any mistakes. A greater contrast with Viki Weisskopf could not be imagined. Everything Viki wrote was modular powers of two, powers of pi, powers of i - the numerator, denominator could well be interchanged. I remember when in a nuclear theory class he tried to derive the resonance formula, the Breit-Wigner ? The Breit-Wigner formula? The Breit-Wigner formula, which in atomic theory was called the Weisskopf-Wigner formula. Breit and Wigner sort of generalised it to nuclear theory, but it was exactly the same formula as the Weisskopf-Wigner formula in atomic theory. Well, he got it all wrong, just completely wrong. The numerator was in the denominator, the denominator was in the numerator, the sign was wrong, the values there were factors of two and pi floating around. So he said well, he said, this time I haven't prepared, I have to admit, I haven't prepared this lecture, so - but next time I'll come in prepared and I'll derive it correctly. Well he came in next time and again he tried to derive it and failed again. Now of course every student in that class learnt how to derive the formula. So it was much more effective than Julian Schwinger's smooth presentation, which really didn't leave you learning anything because he glossed over all the difficulties and presented only a very smooth picture of what was happening. You know, he - really didn't leave you with much of an idea of how to do it yourself, Whereas Viki's mistakes were, very educational. Now Fermi got everything right, but he wasn't a formalism person at all. He just calculated things usually with arithmetic, and got the answer by some trick that was very, very simple. However, it was a trick, and it was based on the fact that he had solved the problem several times before in different guises and put the answers in his little notebook. So that it was a little hard to learn from him also because you would - if you were to do the problem yourself you would have to invent that trick, which was not necessarily so easy. He didn't do it in - he didn't do a problem in a general way so that you would immediately recognise that that was the same way that you would do the problem. He has his own little twist, which you would have had to invent if you were to do it yourself. Anyway, it was all fine. But if you asked him a question to which he didn't know the answer, then things got much more difficult. He was not so happy about that and the discussion became difficult, and very interesting but difficult. And of course I asked him fairly often things that he didn't - questions to which he didn't really know the answer and we had some very interesting discussions as a result.