The topic to be considered is the recovery of a detailed structure of known general form from a multitude of minute fragments of information, each one of which is embedded in irrelevant material and may be corrupted by error. The simplest specific example is the problem of 'seriation on the basis of incidence of types', several automated solutions for which are now available and work well. These yield provisional solutions which have to be carefully cleaned up by the archaeologist, but they have the advantage (over earlier 'hand' methods) that the degree to which subjective judgements are employed can be carefully controlled. A more formidable problem is that of reconstructing the geographical layout of a mediaeval manor from information derived from contemporary cartularies and later deeds. This, too, can be solved 'by hand', but recently it has become clear that automated solutions are also possible, as in the seriation problem, and again offer certain advantages.