This has been a week of farewells from Queen Mary, after 28 years; certainly for me the end of an era.

#### A leaving party

First, on Monday, we had a leaving party, for four of us: apart from me there was Rosemary Bailey, Shaun Bullett and Reza Tavakol. Four people representing the four areas of the School before it was dismembered by the new Vice-Principal. Technically Reza is now in the Physics department, but since he spent almost all of his long and distinguished career in Maths, it was fitting that he should join the other three of us. We four felt very comfortable sharing a party.

I did have the chance to say a few words. Here are some extracts, with some extra comments which I didn’t make on the night (but hope my colleagues picked up).

I told two stories. The first concerned me, Ann Cook, Ian Percival, and Carl Murray: you can find it here, so I won’t repeat it. What it shows is the complete lack of walls and boundaries between different areas of the School at that time. I was amazed, coming to Queen Mary from Oxford, where something of that kind would never have happened.

The second story concerns Thomas Prellberg (a mathematical physicist) and Dudley Stark (a probabilist). I had been teaching for a semester in Prague, and had occupied myself (among other things) with working out the asymptotic behaviour of certain counting numbers related to wreath products of permutation groups. I succeeded, using a method based on group theory and the combinatorics of Stirling numbers, but was not terribly confident of my solution. So, when I got back to QM, of course I talked about it in the Combinatorics Study Group. Thomas and Dudley both got interested. Thomas tackled it with hard-core steepest descent; Dudley found an ingenious way to choose one of the counted objects uniformly at random, and deduced the asymptotics from this. Fortunately we all arrived at the same answer! We wrote a paper which simply consisted of three different proofs of the formula. I am quite fond of that paper, but it was published in a conference proceedings (dedicated to Tony Guttman, the eminent statistical mechanist), so it would not be eligible for submission to a modern research assessment!

I think there is another moral that should be drawn from these two stories. These two pieces of research really did cross the artificial boundaries between different areas of mathematical sciences (and beyond, in Carl Murray’s case), and they happened quite spontaneously. A top-down structure dividing a department into research groups answerable to a line manager is, in my view, very unlikely to produce results like these.

#### Last undergraduate lecture

On Wednesday I gave my last Mathematical Structures lecture to the firwt-year undergraduates. It was quite an emotional occasion for me. After telling them a few things that they had heard in the course but mightn’t have picked up specifically, I presented the prizes to two students for good project work, and then asked for questions. Since there were none, we stopped a bit early.

The students were, in the modern fashion (not quite mine, I am afraid, but I went along with good grace), very anxious to have their pictures taken with me, so we spent quite some time doing that in various combinations. I suspect that my picture now occurs in several pages on Facebook, but I certainly don’t have time to go looking.

#### The “booksale”

As advertised here, I gave away many of my books on Friday afternoon. Most customers were students, many of whom were keen to get a book that would help them with Mathematical Structures (to which my standard reply was, there is no book that is as good as the course notes!), but several colleagues came, and some went away very happy with some books they had really wanted.

In the middle of all this pandaemonium, I had an office hour, so found myself asking students’ questions while other students searched through the bookshelves.

I had put about 60% of my books out to be taken; about two thirds of these went. As a sop to vanity, I have kept one copy of each of my books, including the Russian translation of *Graph Theory, Coding Theory and Block Designs* and the Farsi translation of *Introduction to Algebra*. (Pure vanity, since I am afraid I don’t read either language.)

#### Last Combinatorics Study Group

Well, not the last CSG (it will continue under new management; now Caesar has fallen, a triumvirate of Leonard Soicher, Robert Johnson and David Ellis will run it). Whenever I am in London on a Friday afternoon, I will turn up; but I will sit inconspicuously in the back row.

Anyway, I talked (I had hoped that Robert Schumacher might do some of the work, but he was ill). I spoke about the serendipidous discovery we made last week, that the number of acyclic orientations of a complete bipartite graph is given by a poly-Bernoulli number. I will discuss this in more detail sometime.

The CSG itself deserves some historical notes, which I will try to get around to writing before too long.

#### What remains?

I haven’t finished clearing out my office yet; this has to be done by the end of the month, and so I may find myself spending the holiday break throwing out masses of paper, reprints, and old journals that nobody wants.

By the start of next month, when I am in Queen Mary, you will probably find me sitting in the common room talking to passers-by. The old men’s room on the fifth floor is somehow lacking in appeal …

Reading through what I wrote, I haven’t made it anywhere near enough how wonderful the students have been to me. They have brought me the most amazing bunch of flowers, a Pizza Hut voucher, a loaf of homemade bread (delicious!), a cake, and more chocolates than I can shake a stick at. The chocolates will certainly make me a nicer person: I can’t eat them all (that would be self-inflicted death by chocolate) so I will have to share them with my loved ones.

Thank you all, very much. I don’t know how to say it better than that.

From the many comments, it seems clear that I have done what every teacher dreams of, and given my students something of more than passing value.