In 1966 Keith Thomas wrote an article in the Times Literary Supplement which laid down the axiom, ‘All historical propositions relating to the behaviour of large groups, for example… religious activity, are susceptible of treatment in this [statistical] way, and indeed permit of no other.’ Social historians have increasingly adopted this approach ever since. It has obviously great merits: I am sometimes afraid, indeed, that the principle is in some ways not being carried far enough, and that only the letter, and not the law, is being implemented. For instance, in my own field, individual village studies, often of high quality, are produced. But their authors frequently omit to indicate the nature of the agricultural and marketing region in which their community lies, and therefore fail to establish either its probable social structure or its ‘normality’ or ‘abnormality’ within its area. Quantified studies of great complexity and considerable value are therefore produced; but this value is diminished by the total ignorance of the reader, and sometimes it seems of the author, of the ‘typicality’ of the example.