Yesterday evening my train from Ascot to Reading was delayed for over an hour by a signal failure. Fortunately I had my commuter's survival kit with me: thick pullover, woolly hat, pencil and paper. So I dressed up warm, sat down and set to work thinking about a problem that has been worrying me for a while: "What does the root class of an object-oriented system actually represent?". I already have what I think is a satisfactory way of understanding the non-root classes as specifications of sets of objects, but with root classes this breaks down: there seems to be no set of objects that root classes correspond to. I find theories which have exceptional cases unsatisfactory and, when using them, I often get distracted trying to generalise them to remove the exceptions. Anyhow, I didn't solve the root class problem that evening, but I did work through several possibilities far enough to eliminate them, and by the time my train came, though quite chilled, I was feeling pleased with myself. These days it is quite rare that I get the chance to think about a single problem for such a length of time.
In the first volume of his 'Narrow Roads of Gene Land', evolutionary biologist William Hamilton writes that he worked out some of his early theories while waiting for trains at Waterloo. He was then working at Silwood Park near Ascot. I wonder if, in the evenings, he waited for his return train at Ascot or at Sunningdale? While I am waiting on platform 2, I like to imagine his ghost sitting on one of the benches over on platform 1 scribbling equations.