Connoisseurship is the identification of resemblances between works of art so close as to indicate identical authorship, the finding of likenesses between artist and artist, so great that they point to the existence of a personal connection, which connection the relative ages of the artists and other considerations must determine. Comparison is therefore the apparently essential function of the connoisseur, and essential it would be, provided the connoisseur always knew just what factors to consider while making comparisons. Few connoisseurs, however, get beyond the elementary considerations of type, composition, and superficial detail. But as all of these are easily copied, such resemblances by themselves are not only insufficient to prove identity of authorship, but even of school. Neri de Bicci, for instance, precisely copied Fra Filippo more than once, and his copies, far from being mistaken for possible works of the Frate, do not so much as suggest his school, but write themselves down immediately as mere copies, because their technique and touch are so different. The evidence the connoisseur is therefore required to consider deals chiefly with questions of technique and touch, and he can deal with them only when he is a real connoisseur, a person who to peculiar endowments adds a special training. He must, to start with, have a fine eye, a great power of visualisation, a good memory, and he must take natural pleasure in the merely specific qualities of a work of art. So endowed, many years of study of one school enables him to distinguish with almost unfailing accuracy the precise relations existing -- to speak only of painting -- between picture and picture and painter and painter. No mere erudition, no luck in discovering documents, no skill in deciphering them, no subtlety in stringing far-fetched hypotheses, can fit the person who lacks endowment and training for dealing with the only evidence of primary consequence in an unauthenticated work of art: the evidence of technique and touch. It is because the mere scholar, palaeographist, or logician never gives due consideration to this evidence that he is always running away from the point in search of data that are either unnecessary, or incapable of yielding the required information.