Finger Prints

by Francis Galton

A third method of classification is determined by the origin of the ridges which supply the interspace, whether it be from the thumb side or the little-finger side; in other words, from the Inner or the Outer side.

Lastly, a translation from the Latin is given of the famous Thesis or Commentatio of Purkenje, delivered at the University of Breslau in 1823, together with his illustrations. It is a very rare pamphlet, and has the great merit of having first drawn attention to the patterns and attempted to classify them.

In the sixth chapter we reach the question of Persistence: whether or no the patterns are so durable as to afford a sure basis for identification. The answer was different from what had been expected. So far as the proportions of the patterns go, they are not absolutely fixed, even in the adult, inasmuch as they change with the shape of the finger. If the finger is plumped out or emaciated, or variously deformed by usage, gout, or age, the proportions of the pattern will vary also. Two prints of the same finger, one taken before and the other after an interval of many years, cannot be expected to be as closely alike as two prints similarly made from the same woodcut. They are far from satisfying the shrewd test of the stereoscope, which shows if there has been an alteration even of a letter in two otherwise duplicate pages of print. The measurements vary at different periods, even in the adult, just as much if not more than his height, span, and the lengths of his several limbs. On the other hand, the numerous bifurcations, origins, islands, and enclosures in the ridges that compose the pattern, are proved to be almost beyond change. A comparison is made between the pattern on a finger, and one on a piece of lace; the latter may be stretched or shrunk as a whole, but the threads of which it is made retain their respective peculiarities. The evidence on which these conclusions are founded is considerable, and almost wholly derived from the collections made by Sir W. Herschel, who most kindly placed them at my disposal. They refer to one or more fingers, and in a few instances to the whole hand, of fifteen different persons. The intervals before and after which the prints were taken, amount in some cases to thirty years. Some of them reach from babyhood to boyhood, some from childhood to youth, some from youth to advanced middle age, one from middle life to incipient old age. These four stages nearly include the whole of the ordinary life of man. I have compared altogether some 700 points of reference in these couplets of impressions, and only found a single instance of discordance, in which a ridge that was cleft in a child became united in later years. Photographic enlargements are given in illustration, which include between them a total of 157 pairs of points of reference, all bearing distinctive numerals to facilitate comparison and to prove their unchangeableness. Reference is made to another illustrated publication of mine, which raises the total number of points compared to 389, all of which were successful, with the single exception above mentioned. The fact of an almost complete persistence in the peculiarities of the ridges from birth to death, may now be considered as determined. They existed before birth, and they persist after death, until effaced by decomposition.

In the seventh chapter an attempt is made to appraise the evidential value of finger prints by the common laws of Probability, paying great heed not to treat variations that are really correlated, as if they were independent. An artifice is used by which the number of portions is determined, into which a print may be divided, in each of which the purely local conditions introduce so much uncertainty, that a guess derived from a knowledge of the outside conditions is as likely as not to be wrong. A square of six ridge-intervals in the side was shown by three different sets of experiments to be larger than required; one of four ridge-intervals in the side was too small, but one of five ridge-intervals appeared to be closely correct. A six-ridge interval square was, however, at first adopted, in order to gain assurance that the error should be on the safe side. As an ordinary finger print contains about twenty-four of these squares, the uncertainty in respect to the entire contents of the pattern due to this cause alone, is expressed by a fraction of which the numerator is 1, and the denominator is 2 multiplied into itself twenty-four times, which amounts to a number so large that it requires eight figures to express it.

A further attempt was made to roughly appraise the neglected uncertainties relating to the outside conditions, but large as they are, they seem much inferior in their joint effect to the magnitude of that just discussed.

Next it was found possible, by the use of another artifice, to obtain some idea of the evidential value of identity when two prints agree in all but one, two, three, or any other number of particulars. This was done by using the five ridge-interval squares, of which thirty-five may be considered to go into a single finger print, being about the same as the number of the bifurcations, origins, and other points of comparison. The accidental similarity in their numbers enables us to treat them roughly as equivalent. On this basis the well-known method of binomial calculation is easily applied, with the general result that, notwithstanding a failure of evidence in a few points, as to the identity of two sets of prints, each, say, of three fingers, amply enough evidence would be supplied by the remainder to prevent any doubt that the two sets of prints were made by the same person. When a close correspondence exists in respect to all the ten digits, the thoroughness of the differentiation of each man from all the rest of the human species is multiplied to an extent far beyond the capacity of human imagination. There can be no doubt that the evidential value of identity afforded by prints of two or three of the fingers, is so great as to render it superfluous to seek confirmation from other sources.

The eighth chapter deals with the frequency with which the several kinds of patterns appear on the different digits of the same person, severally and in connection. The subject is a curious one, and the inquiry establishes unexpected relationships and distinctions between different fingers and between the two hands, to whose origin there is at present no clue. The relationships are themselves connected in the following way;—calling any two digits on one of the hands by the letters A and B respectively, and the digit on the other hand, that corresponds to B, by the symbol B1, then the kinship between A and B1 is identical, in a statistical sense, with the kinship between A and B.

The chief novelty in this chapter is an attempt to classify nearness of relationship upon a centesimal scale, in which the number of correspondences due to mere chance counts as 0°, and complete identity as 100°. It seems reasonable to adopt the scale with only slight reservation, when the average numbers of the Arches, Loops, and Whorls are respectively the same in the two kinds of digit which are compared together; but when they differ greatly, there are no means free from objection, of determining the 100° division of the scale; so the results, if noted at all, are subject to grave doubt.

Applying this scale, it appears that digits on opposite hands, which bear the same name, are more nearly related together than digits bearing different names, in about the proportion of three to two. It seems also, that of all the digits, none are so nearly related as the middle finger to the two adjacent ones.

In the ninth chapter, various methods of indexing are discussed and proposed, by which a set of finger prints may be so described by a few letters, that it can be easily searched for and found in any large collection, just as the name of a person is found in a directory. The procedure adopted, is to apply the Arch-Loop-Whorl classification to all ten digits, describing each digit in the order in which it is taken, by the letter a, l, or w, as the case may be, and arranging the results in alphabetical sequence. The downward direction of the slopes of loops on the fore-fingers is also taken into account, whether it be towards the Inner or the Outer side, thus replacing L on the fore-finger by either i or o.