The Math-Fun list is a family list. We have several teenage subscribers. Referring to the Prom Theorem, there are several things that could spoil it: a) Averages must match, but medians don't need to. Nothing says that the distributions of the two groups are the same. b) There could be recall bias, especially at higher numbers. c) The Prom Theorem assumes a definite beginning and end to the dance. If it's a permanent on-going affair, snapshots don't have to add up. d) Interview subject selection bias. Some people have left the dance early and moved out of state. They don't get interviewed. e) Ambiguity about "What's a dance?". f) Assumption of a bipartite graph, with edges. g) Possible confusion between number of dances, and number of different dance partners. I'm sure that Prof. Gale knows all of these, but perhaps Ms. Kolata has overlooked some of them. Rich