The proposition that the higher-level sciences are in principle reducible to physics is defended less often than it is assumed, and the assumption does a good deal of work. Its plausibility rests on the thought that everything is made of the same constituents, which is not in dispute and does not establish the conclusion; that a system is composed of particles does not entail that its behaviour is explicable in the vocabulary of particles, any more than knowing the alphabet explains a sentence. Reductions that have actually been carried out — thermodynamics to statistical mechanics being the favourite example — turn out on inspection to be partial, approximate, and to require bridge assumptions that are not themselves derived from the lower level. The promissory note has been outstanding for a long time and shows no particular sign of being redeemed.