We now come to the great difficulty in all classifications; that of transitional cases. What is to be done with those prints which cannot be certainly classed as Arches, Loops, or Whorls, but which lie between some two of them? These occur about once in every forty digits, or once in every four pairs of hands. The roughest way is to put a mark by the side of the entry to indicate doubt, a better one is to make a mark that shall express the nature of the peculiarity; thus a particular eyed pattern (Plate 10, Fig. 16, _n_) may be transitional between a loop and a whorl; under whichever of the two it is entered, the mark might be an _e_ to show that anyhow it is an eye. Then, when it is required to discover whether an index contains a duplicate of a given specimen in which a transitional pattern occurs, the two headings between which the doubt lies have to be searched, and the marked entries will limit the search. Many alternative ways of marking may be successfully used, but I am not yet prepared to propose one as being distinctly the best. When there are two of these marks in the same set, it seldom happens that more than two references have to be made, as it is usual for the ambiguity to be of the same kind in both of the doubtful fingers. If the ambiguities were quite independent, then two marks would require four references, and three marks would require nine. There are a few nondescript prints that would fall under a separate heading, such as Z. Similarly, as regards lost or injured fingers. I have tried various methods of sub-classification, and find no difficulty in any of them, but general rules seem inadvisable; it being best to treat each large group on its own merits. One method that I have adopted and described in the _Proc. Royal Soc._, is to sketch in a cursive and symbolic form the patterns of the several fingers in the order in which they appear in the print, confining myself to a limited number of symbols, such as might be used for printer's types. They sufficed fairly for some thousands of the finger marks upon which they were tried, but doubtless they could be improved. A little violence has of course to be used now and then, in fitting some unusual patterns to some one or other of these few symbols. But we are familiar with such processes in ordinary spelling, making the same letter do duty for different sounds, as _a_ in the words _as_, _ale_, _ask_, and _all_. The plan of using symbols has many secondary merits. It facilitates a leisurely revision of first determinations, it affords a pictorial record of the final judgment that is directly comparable with the print itself, and it almost wholly checks blunders between inner and outer slopes. A beginner in finger reading will educate his judgment by habitually using them at first. [Illustration: PLATE 2. FIG. 3. Form of card used for impressions of the ten digits. 11-1/2 x 5 inches. FIG. 4. Roller and its bearings, of a pocket printing apparatus.]