At the bottom of Fig. 14, ~49-54~, is given a series of rings, spirals, and plaits, in which nearly all the clearly distinguishable varieties are included, no regard being paid to the direction of the twist or to the number of turns. ~49~ is a set of concentric circles, ~50~ of ellipses: they are rarely so in a strict sense throughout the pattern, usually breaking away into a more or less spiriform arrangement as in ~51~. A curious optical effect is connected with the circular forms, which becomes almost annoying when many specimens are examined in succession. They seem to be cones standing bodily out from the paper. This singular appearance becomes still more marked when they are viewed with only one eye; no stereoscopic guidance then correcting the illusion of their being contour lines. Another curious effect is seen in ~53~, which has the appearance of a plait or overlap; two systems of ridges that roll together, end bluntly, the end of the one system running right into a hollow curve of the other, and there stopping short; it seems, at the first glance, to run beneath it, as if it were a plait. This mode of ending forms a singular contrast to that shown in ~51~ and ~52~, where the ridges twist themselves into a point. ~54~ is a deep spiral, sometimes having a large core filled with upright and nearly parallel lines; occasionally they are bulbous, and resemble the commoner "monkey" type, see ~35~. When the direction of twist is described, the language must be unambiguous: the following are the rules I adopt. The course of the ridge is always followed _towards_ the _centre_ of the pattern, and not away from it. Again, the direction of its course when so followed is specified at the place where it attains its _highest_ point, or that nearest to the finger-tip; its course at that point must needs be horizontal, and therefore directed either towards the inner or the outer side. The amount of twist has a strong tendency to coincide with either one, two, three, four, or more half-turns, and not to stop short in intermediate positions. Here are indications of some unknown fundamental law, analogous apparently to that which causes Loops to be by far the commonest pattern. * * * * * The classification into Arches, Loops, and Whorls is based on the degree of curvature of the ridges, and enables almost any pattern to be sorted under one or other of those three heads. There are a few ambiguous patterns, and others which are nondescript, but the former are uncommon and the latter rare; as these exceptions give little real inconvenience, the classification works easily and well. Arches are formed when the ridges run from one side to the other of the bulb of the digit without making any backward turn or twist. Loops, when there is a single backward turn, but no twist. Whorls, when there is a turn through at least one complete circle; they are also considered to include all duplex spirals. [Illustration: PLATE 9.