Personal Identification and Description 213 patterns. I transcribe two-thirds of it for the benefit of those who can no longer obtain Galton's original work*.
"The chief peculiarities of individual Arches, Loops and Whorls having now been described, it becomes easy to discuss the frontiers of the primary classes and the debatable country between them.
"A to L [i.e. Arch to Loop]. The frontier between A and L ceases to be distinct at the point where A is just short of developing into a nascent loop. In the Figures 169 to 172 that point is just, but only just passed, so all those figures should count as loops with an a suffixed. The debatable ground lies between these and unmistakable arches, and in that debatable ground, A is held to predominate over L under any one of the following conditions:
" 1. When the loop is formed by no more than one complete bend or staple, which may, however, be perfectly distinct, and may also enclose a rod (Fig. 21).
"2. When it consists of two or even three imperfect bends (Figs. 19, 20), especially if they converge and unite.
"3. Offsets at acute angles (Fig. 10) from the same ridge or from the same furrow do not rank as heads to loops.
" 4. When two symmetrically disposed loops are enclosed in the same curved ridge (Figs. 173, 174) they are counted as an imperfect form of tented arch, being noted as A with the suffix t or tur.
"Generally speaking A is held to predominate whenever the pattern has no continuous contour, even though there may be a fairly distinct delta (Fig. 20), but it would be proper to unite the suffix 1 to this." (pp. 108-9.)
Clearly since Arches form a relatively small group, it would be to the
advantage of the indexer, if frontier cases were allotted as far as possible to Arches.
"A to W [i.e. Arch to Whorl]. Between A and W a very small, or else an imperfect circle, or dot sometimes appears between two ridges of a pattern which is an arch in all other respects (Figs. 15, 17 and perhaps 18, which is ambiguous, and might be called a loop). If the diameter of the whorl does not exceed the width of one of the adjacent ridge intervals, the pattern does not lose the right to be called an A, but should for distinction's sake have a y suffixed to it. W is certainly reached when the little circle contains a central dot as in Fig. 175 which I should' call Wky.
"L to W [i.e. Loop to Whorl]. Between L and W a large class of transitional cases have been sufficiently discussed in speaking of complete and incomplete circuits t. See Figs. 180-183.
"The specimens Figs. 176 to 179 show the relationships between whorls to which the suffix sb is applied (Fig. 178), with loops. In Fig. 176 we see a loop that throws off a curious crest from the upper part of its outline, and which is here and elsewhere a striking appearance; but in Fig. 177 the same peculiarity is much less distinct, while the number of cases that exist between extreme distinctness and extreme indistinctness is so great that crests are not allowed to have a suffix. Their conspicuousness in individual cases certainly depends to a considerable degree on the printing, whether more or less ink and pressure are used. When, however, the ridges cease to be given off from the outside of the contour of the loop, and recurve upon themselves as in Fig. 178, forming a blunted end to that part of the pattern, the result is a well-defined whorl. Another intermediate form between a loop and a whorl is produced in another way, and is recorded by vy as already explained." (pp. 109-10.) [See our p. 206.]
Lastly Galton refers to the case in which a real whorl may be mistaken for a loop because enough of the finger ridges have not been imprinted by rolling. This is especially a danger with "dabbed" prints. See our Plate XXVI.
Chapter VII (pp. 111-115) is entitled Suggested Improvements. Here, as I have said, Galton gives up his special finger arrangement in favour of the
* I have retained Galton's figure numbers, and the figures to which he refers will be found on our Plates XXIII-XXX.
t See our p. 204 footnote.