This easy method of classification has much power. For example, the four possible kinds of simple spirals (see the 1st, 2nd, and the 5th and 6th diagrams in the lowest row of Plate 11, Fig. 17) are wholly determined by the letters _oj_, _jo_, _ij_, _ji_ respectively. The two forms of duplex spirals are similarly determined by _oi_ and _io_ (see 4th and 5th diagrams in the upper row of Fig. 17), the two slopes of loops by _oo_ and _ii_ (3rd and 4th in the lower row). It also shows very distinctly the sources whence the streams of ridges proceed that feed the pattern, which itself affords another basis for classification. The resource against uncertainty in respect to ambiguous or difficult patterns is to compile a dictionary of them, with the heads under which it is advisable that they should severally be classed. It would load these pages too heavily to give such a dictionary here. Moreover, it ought to be revised by many experienced eyes, and the time is hardly ripe for this; when it is, it would be no difficult task, out of the large number of prints of separate fingers which for instance I possess (some 15,000), to make an adequate selection, to enlarge them photographically, and finally to print the results in pairs, the one untouched, the other outlined and classified. It may be asked why ridges are followed and not furrows, the furrow being the real boundary between two systems. The reply is, that the ridges are the easiest to trace; and, as the error through following the ridges cannot exceed one-half of a ridge-interval, I have been content to disregard it. I began by tracing furrows, but preferred the ridges after trial. _Measurements._--It has been already shown that when both plots are present (Plate 4, Fig. 8, ~4~), they form the termini of a base line, from which any part of the pattern may be triangulated, as surveyors would say. Also, that when only one plot exists (~3~), and the pattern has an axis (which it necessarily has in all ordinary _ii_ and _oo_ cases), a perpendicular can be let fall upon that axis, whose intersection with it will serve as a second point of reference. But our methods must not be too refined. The centres of the plots are not determinable with real exactness, and repeated prints from so soft a substance as flesh are often somewhat dissimilar, the one being more or less broadened out than the other, owing to unequal pressure. It is therefore well to use such other more convenient points of reference as the particular pattern may present. In loops, the intersection of the axis with the summit of the innermost bend, whether it be a staple or the envelope to a rod (Fig. 14, second and third rows of diagrams), is a well-defined position. In spirals, the centre of the pattern is fairly well defined; also a perpendicular erected from the middle of the base to the outline above and below (Fig. 8, ~4~) is precise and convenient.