THE SYLLABIC CLASSIFICATION OF FINGER-PRINTS
Having secured some technical knowledge of how to print, and how to
read old finger-prints correctly and with confidence when they turn
up again in experience, we are faced now with the problem of how to
classify and arrange them for secure preservation and prompt and easy
reference, whatever may be our object.
In natural history, in biological facts generally, it is not always
easy to define the objects of study strictly, so as to classify them in
a practical way. Dealing, however, with printed finger-patterns which
are no longer living and changing things, we can hope to secure some
of the advantages of a mechanical method. Verworn, in his _General
Physiology_ (p. 71) says, very justly: “The fixing of sharp limits and
definitions must contain, finally, a more or less arbitrary element,
[that], indeed, all limits and definitions are only psychological helps
towards knowledge.” Bearing this principle in mind, then, what is the
end or object we aim at in a system of finger-print classification?
The objects of identifying a person with some one who has had a
name and left a history are of various kinds, as criminal, civil,
military, naval, medical, legal, scientific, and insurance purposes.
Now, in regard to the use of finger-prints for so many ends in
view, a difficulty presents itself. It occurred to me at the outset
of my studies, that if the system were to prove trustworthy and
useful, even in a minor degree, immense numbers of people in civil
life, in army, navy, and mercantile services, or under criminal
conviction, would require to have their prints correctly classified,
indexed, and arranged for easy reference. How could it be possible
in so vast a collection or series of collections to find the one
single record wanted? To ransack--unaided by a scientific method
of classification--the register of an army containing some 500,000
soldiers would involve the search of a much larger number of cards or
sheets than 500,000, according to the duration of regular service,
and other possible conditions. To do this would obviously be quite
as hopeless and futile a task as groping for a lost needle in a huge
hay-field. The problem was to find a system which would facilitate
the search in a _high degree_. Any mere slight assistance would still
leave the essential problem unsolved. Now, we might have found in
finger-prints mere variety without persistence, or mere persistence
without initial variety, and in either case the study could yield
little practical result. Again, mere diversity, however persistent,
without some elements of underlying resemblance, would not have yielded
a basis for such a methodic arrangement as was obviously required.
Much aid came to me from the first, as I have already hinted, from
five years’ daily laborious experience in sorting and comparing
analogous but artificial patterns in the now obsolete Paisley shawl
trade, but in the case now in view colour did not come in as an aid
to arrangement. This problem, moreover, was not one of those the poet
derides as of mere “gold or clay,” but as I saw, it concerned itself
with human lives, and was a task, indeed, that might awaken in the
dullest mind a keen sense of moral responsibility in proposing its
general use as a new and quite trustworthy method of criminal and other
modes of identification. The expert in charge might suddenly be called
upon after a little expansion of the system to prove the identity
of some evil-doer out of many thousands of possible persons, or to
subject a suspected person, on the evidence of a few smudgy streaks
of ink or blood, to life-long servitude, or to the irremediable doom
of a shameful death. In my own case, at this early stage, the mere
possibility of a single serious false identification by a method as yet
untried became really terrible to contemplate. After closer study, a
clear path began to open through the tangled jungle.
Some familiarity with the equipment of a Far-Eastern printing-press
had been afforded me while editing _The Chrysanthemum_, a monthly
magazine published in Tokyo, and devoted to the discussion of Japanese
topics of literary, scientific, or antiquarian interest. There were
some hundreds of thousands of different forms of type, all classified
in so convenient a way that any compositor, by running about a little
more actively than would be quite compatible with the grave dignity
of an English printing establishment, could soon find the character
in whatever form of fount he desired. The idea suggested itself
then, that analogous qualities as a basis for classification of the
finger-patterns might be revealed by a closer study of Chinese. I do
not know Chinese--some years’ close study has convinced me of that.
However, each Chinese ideograph, for dictionary purposes, is supposed
to be built up around an element called by western lexicographers its
_key_ or _radical_, and of these there are two hundred and twelve.
You look for the radical in an unknown character, and then look for
that radical in its serial place in the two hundred and twelve. It
is a question then, as in finger-prints, of counting strokes, and if
the strokes are alike in number in any two instances, of looking then
as to how they are arranged. Two characters with the same number of
pen-strokes under the same radical or key, may bear quite a different
aspect.
A Chinese character is defined and limited, but a finger-print pattern
often, or usually, trails off into indeterminate lineations of little
value for classification purposes. Hence we seek in the latter to
isolate for study the central part of the pattern, where the intricacy
of the ramifications usually rises to a maximum. The space covered
by the lineations that matter is not usually greater than, often
not so wide as, the space occupied by the head of the Sovereign on
an English postage stamp. Into this brief compass is compressed a
world of significance. A courteous and intelligent young detective in
Scotland Yard asked me (in 1886 or 1887, when I was advocating the
adoption of finger-print identification), did I really propose to
rest identification on features contained within so small a space? I
answered him, in pointing to a railway map of London, to consider a
net-work of junctions which I indicated, if he would not be justified
in saying if that fragment, torn away from its context, were presented
to him, that it was a portion of a map of London? After a little
scrutiny, he admitted that was so. I had no difficulty in showing him
then, that the condensed ramifications of a single finger-print within
the very limited area proposed by me were much greater than that of the
significant portion of the London map I had just pointed out to him.
In tracking a criminal by a single impression made by a finger, the
lineations in so small a space would require to have been clearly
imprinted, and to have what many finger-print patterns have not, some
notable or significant characteristics about it. Then, when enlarged by
photography into a picture of some thirty inches, the measurements from
fixed points in the pattern should correspond with those of the person
in custody, on suspicion, and the curves should be shown to concur
in all their sinuosities. But, in comparing two official imprints of
the ten fingers properly and clearly impressed, there should be no
difficulty, the points of comparison being overwhelming.
In a possible collection of half-a-million or a million complete sets
of finger-prints, can the one before me, of one Thomas Atkins, John
Doe, or Richard Roe--under whatever alias--be promptly found if it is
there, or, if not there, can its absence be conclusively determined?
We have seen, I think, that if two such patterns are confronted,
common-sense, and the use of fine measurements, will soon determine
whether they be of the same original, or different. The problem, then,
is to get this swift and sure confrontation effected.
This problem engaged my attention from the first, or at least not many
months after I first began to attend to finger-patterns, and in 1880,
when I proposed the printing and recording of the ten fingers of old
criminals, I had thought out the same method now outlined in this
chapter. It would be impossible to compress all the details necessary
to work out the matter officially, without producing a work as large,
and perhaps as expensive, as a Chinese dictionary, of which the
probabilities are that one or perhaps two copies might be sold.
I laid this matter in outline before Inspector Tunbridge, in his
official capacity, in 1888, and again before the War Office Committee,
at which an Under-Secretary of the Home Office was present, taking
diligent notes. The system now in official use--an improvement made by
Sir Edward R. Henry upon Sir F. Galton’s very premature attempt (after
a few years’ study in old age) seems to work practically, and therefore
I have no criticism to offer, further than to suggest, that if in our
system of mercantile book-keeping we had retained the use of Roman
numerals, fortunes might continue to be made or lost. I cannot think,
however, that our merchants would now give up the Arabic notation for
the more complex and clumsy one of ancient Italy. Nor is nature likely
to resume her interest in the kangaroo and its future.
Science seeks simplicity, and the _Syllabic system_, now familiar
to every one who uses a telegraphic code, is what I proposed for
finger-print registers. In this I simply followed the method of
transliterating Japanese and Chinese words into syllables of the Roman
alphabet, a condition originally imposed by the old Japanese language
itself, in which consonants do not occur singly, but are followed by
vowels. Purkinje’s first analysis of the finger-print patterns was
not known to me, nor, I believe, to anyone in Europe or America, when
I first wrote, although I often in those years suggested that he
had probably written something on the subject. My first article in
_Nature_, as sent up, contained a kind of analysis of patterns, with
many types, named as whorls, ovals, deltas, loops, junctions, and the
like. Some are referred to in the text, but the editor expressed his
regret that he had not been able to insert the figures, and their
lack made the references in my article obscure. We shall deal with
a few of these elementary or typical figures presently. But, let us
now come to the main aspect of the syllabic system, in contrast with
that devised by Sir F. Galton, who looked upon it as merely ancillary
to the anthropometric system of Mons. A. Bertillon, of the French
police. Galton was supremely anxious to have his natural facts, his
finger-print records, arranged precisely in similar parcels, so
that one would not be excessively rich in records compared with its
neighbour. Now, what does it matter to the keeper of records, or even
to the tax-payer, whether one class of patterns is big or little? The
whole absurd complexity arising now, and increasing from year to year,
grows out of this essential misapprehension from the first of the vital
problems of finger-print classification. Advancing a stage for the
moment, let us suppose that a rich register exists, arranged on the
syllabic system. A type-writer, not necessarily a very intellectual
creature, or a boy-clerk, is in the room, and has the call to find
_A-bra-ca-da-bra_. I use here for convenience only five syllables,
representing one hand. The sheets or cards (sheets have been found best
by experience) are not separated in bundles except as to a convenient
size. It does not take long to look along the shelves till _A-bra-_
etc., is reached, and then the cabalistic word itself. It may prove
that there are some ten sheets on the register under this syllabic
title. These are transmitted, all in a few moments, to the expert
keeper of the records. At a glance an expert eye like his perceives
that, perhaps, seven out of the ten can have no possible relation to
the case now being enquired into. Of the three, one is perhaps now
in prison and cannot be the suspect. Of the two remaining forms, the
details of the first two fingers compared may diverge completely in
many ways, as determined by counting lines, measuring curves, and so
on. I am sure this would be no fancy description, from the many tests I
have applied. The whole strain of the recognition lies on the expert,
as the strain of the primary classification of records had lain upon
him at the time they were being made. Of course, more than one expert
might be needed.
It will be noticed, perhaps, that the syllable _bra_ occurs twice on
the same hand register. It by no means follows that the finger-print
represented by the second _bra_ is very like that of the first one. In
the same way, none of the patterns indicated by _bra_ in the cards of
similar syllabic index may much resemble the others, even broadly. The
pattern simply is of a certain typical form with which _bra_ is to be
linked for registration purposes. The same word, so to speak, might be
divided in a different syllabic way, thus:--
_Ab-ra-cad-ab-ra_;
_Ab-rac-ad-ab-ra_; and so on.
Hence the necessity of separating the syllables by hyphens.
The divergence of cards will be greater, of course, in the case of a
two-hand register, and even in one which comprehended, say, one million
of complete sets there would be very few repetitions of the same
arrangement of syllables.
One great advantage of the syllabic form is the help given to the
memory in transferring the eye from one sheet to others which may be
wanted. In the system now in use the symbols do not rivet themselves in
the same way, and have a monotony that becomes very tiring.
A general view of the precise intention aimed at in the particular
register must determine the extensiveness of the form the register is
to compass. Are the numbers likely to be large? Must the registers
extend over long periods? Are infants to be kept in view over adult
life, if that is reached? Many enquiries of this kind may have to be
met before the exact form of the cards or sheets is determined. For
such civil and social purposes as life insurances, signatures of deeds,
benefit of friendly societies, and the like, a comparatively simple
form of register and limited number of finger imprints might be all
that would be required for an effective service. The number of cards
would not be very great, and the probabilities of personation would
likely be restricted to a few local residents whose finger-prints would
not often be found even to approach coincidence in a slight degree.
To serve such needs, an elementary form of classification would go
a long way to overtake ordinary requirements, and would be easy of
reference. Few of the difficulties involved in graver conditions of
legal identification need be raised as an objection to the general use
in banking and ordinary business of this new mode of identification.
In forming a system, even with a very wide range, the whole amount of
possible complexity in finger-patterns need rarely be called upon, and
could not conceivably be exhausted. I speak confidently on this point.
The central part of the pattern used is generally very limited, and its
area may be widened whenever an enlargement of the primary requirements
may demand more complexity in the factors of identification. The
ramifications will usually provide variety enough to satisfy the most
avaricious register.
Some of the main conditions on which the problem of alphabetic
arrangement of the index depends may now be set forth, before we
proceed to consider how those conventional syllables are to be formed
which indicate patterns.
1.--Distinction is not made between capital and lower-case letters.
Simple letters are too soon exhausted in a register of any considerable
size. It is obvious that syllables give a much greater variety. As far
as possible, commensurate with the dimensions of the register, the
syllables should be kept few, simple, compact, and pronounceable. The
vowels have the Italian sound. No syllables should contain more than
four letters at the utmost.
2.--When a doubt arises as to the proper syllabic reading of a
finger-pattern, the earlier letters of the Roman alphabet have the
precedence, thus _b_ before _d_, _l_ before _x_.
3.--Where the core of a pattern seems to contain two or more clusters
of significant lineations, choose for the index syllable that on the
right side of the pattern, or, if that is difficult to determine,
next that which is highest in position. In such a case, reference to
orientation or position refers to the usual or official pattern. In
dealing with a smudge of unknown origin, the various possibilities may
be tried, assuming relative order of position, as above.
4.--When spaces or figures, such as ovals or circles, are described
as “_large_,” that means wider than the space occupied by two average
lineations in that finger-print.
5.--When a finger-pattern has been permanently defaced or obliterated
by injury or disease, the missing mark may be denoted by an asterisk
(*). If the finger itself is missing, by deformity or mutilation, the
asterisk may be encircled with an =O=. A special compartment of the
register might be kept for the reception of all such cases.
6.--Badly-printed or obscure patterns should be held in reserve under
a special register classified according to probabilities, aided by
cross indexing, and receiving special attention from the higher
experts. Official patterns badly printed should at once be repeated, if
possible, before confusion arises.
7.--Registers for naval or military, and banking, insurance, and
general purposes, should be kept strictly free from any police
supervision or control.
The syllables in my system, viewed as lexicographic elements, consist
of the ordinary Roman vowels and consonants, the vowels being
pronounced, as already said, as in the Italian language. I hold in
reserve for additional official purposes a few additional characters,
such as the Greek letter _delta_ Δ. Those, however, need not be dealt
with in the brief space now available, and would only be required,
I believe, in pretty extensive registers. The functions of the
conventionally fixed vowels may be better understood after we have
sampled a few of the consonants.
As suggested to me by Sir Isaac Pitman’s system of phonography, learned
in student years, I arranged the consonants in co-related pairs,
thus: _p, b_; _t, d_; _s, z_; _h, f_; _l, r_; _k, g_; _v, w_; _ch_
(considered as a consonantal character), _j_; _m, n_.
I have already pointed out, in dealing with problematic smudges, the
need of understanding patterns apart from their actual orientation,
which, in an unknown person’s case, may have to be assumed, an attitude
which may be determined by official bias. This I have entered more
fully upon in the _Guide to Finger-Print Identification_.
Holding this principle in view, then, let us now take some of the
simpler elements of patterns in their very simplest forms, and first
consider those grouped under the paired consonants.
=Ch= and =J=.
Each of these characters is taken to represent a hook with a short
leg. =Ch= is considered as one consonant, and as =C= is not otherwise
wanted, it might have been used alone but for its pronunciation being
indefinite. If in the usual form of official imprint the hook, with
its curve below, has its short leg facing to the left, thus, =J=, it
is duly represented by the Roman letter of that shape. Observe that
if you invert this character, or the type which represents it, thus
={inverted J}=, it will still point the observer to the =J= part of
the index, on getting the curve set right.
If the short leg of the figure points to the right it comes under =Ch=.
If that happens to confront one in its inverted position it cannot
be mistaken for a =J= figure, but must be looked for under =Ch=. In
all cases the degree or direction of slope in the figures, with a
few peculiar exceptions, is of no concern whatever, simplicity and
directness of appeal being aimed at from the first.
=B= and =P=.
These consonants are used to denote a bow. =B= is the form of a simple
bow with one lineation, or if two or more lineations blend into one,
they are found on the _left_ side when the convexity of the curve is
upwards. =P= is such a bow, but strengthened, as it were, by one or
more blended lineations on the _right_ side, with the same position of
the curve. A single line bow is never represented by =P=. If a bow with
a plurality of blended lineations is inverted the reading is not at all
affected.
=T= and =D=
represent pear-shaped, or battledore-like figures. =T= denotes such a
figure free from attachment to environing lineations, while =D= stands
for a similar figure fixed by its stem. Reverse the position of the
figure or turn it upside down and its index quality is not affected.
=K= and =G=
represent spindle-like forms, like the above but with _two_ (opposite)
stems instead of one. When the figure is moored by one stem it is
denoted by =K=; when fixed at both ends or free at both, by =G=.
Position does not affect these figures.
=V= and =W=.
These letters stand for whorls or spirals, a kind of figure that often
presents much difficulty in finger-print classification. =W= is a
whorl in which, tracing its course from the centre outwards, the pen
goes round as a clock-hand turns, or as one looking towards the south
perceives the sun to cross the sky. =V=, on the other hand, is one
which, traced in the same way (from within outwards), the pen goes like
the clock-hand backwards, or _widdershins_. Alteration in position
makes no practical difference whatever in the reading of those figures
into their proper syllables for an index.
=O= and =Q=
Although =O= is a vowel and will be met with again under that class, it
is paired in a kind of way with =Q=.
=O= denotes a _small_ circle or oval, or opaque, round, or ovoid dot,
_contained in the core of a pattern_.
=Q= denotes a _large_ circle or oval, containing, usually within
itself, other pattern elements of small dimensions.
A circle or ovoid is called _large_ when it occupies a space wider than
two average lineations of that finger-pattern in which it occurs. If
any doubt exists, by the principle previously mentioned, the figure is
referred to =O= as prior to =Q= in alphabetical sequence.
=M= and =N=
denote figures somewhat resembling mountain peaks, =M= signifying
an outline like that of a typical volcanic peak, while =N=, though
similar, ends in a rod-like form, as of a flag-staff on a mountain top.
Invert either of those typical forms and they can be read as before.
A curved cliff-like form, like a wave with a curling crest, may be
indicated by the Spanish ñ.
=L= and =R=
denote loops in which curvatures are apt to occur. =L= is a loop, the
axis of which is straight, while =R= is one the axis of which is curved
or crooked.
Note that if the legs of a loop widen out beyond the parallels, it
is no longer a loop, but a bow or a mountain. They may narrow again
and yet remain loops till at last they coalesce, when the figure is
transformed into a spindle or a battledore (=T, D=; _or_ =K, G=). If
the bend is more than that of a right angle, it comes under a new
definition, and has some qualities of the whorl or spiral, but is more
complex. This need not be entered upon here.
=S= and =Z=
I have used these two consonants to indicate certain patterns of a
sinuous, undulating, or zig-zag type, the sinuous or purely undulating
figures coming under =S=, but under =Z= if there is at least one
distinct angularity in the pattern.
=X=
This letter, long familiar to the student of algebra as the symbol
of the undetermined, I have reserved for the inclusion of various
nondescript and anomalous patterns. Those might become fairly numerous
in an extensive register, and in such case there would, no doubt, be
found a good basis for fresh sub-classification.
=F= and =H=
These two aspirates are made to do useful service, not unlike that of
vowels, but not of sufficient interest to be noted in a work like this.
We have thus, with the use of consonants alone, built up a kind of
osseous or skeletal system, and we have now but to add the vowels to
make those dry bones speak. Let us now consider this element in the
syllabic method.
=A=
This vowel indicates that the interior of a given loop, whorl, circle,
or containing pattern of any kind, is empty or vacant. Dealing here
only with the simpler conditions in which combinations of vowels and
consonants are found, such a figure will be indexed as _Ra, La, Ta,
Da_, as the dominant consonant may require. Such combinations as _ar,
al, at, ad_, etc., may occur, but this would lead us into too many
intricate ramifications for a work like the present.
If a pattern is very simple--consisting, for example, of almost
parallel lines--it may be denoted by the letter =A= alone. There are
such patterns, and they seem to be somewhat commoner among certain of
the negro tribes. I have mentioned in a previous chapter such a pattern
on the toe of a lady, and they are typical almost in some monkeys.
=E=
When we find in the interior of some loop, bow, or other pattern, a
group of not less than three short detached lines, or dots, this is to
be indicated by the use of =E= with the ruling consonant, as _te, re,
me_, and so on.
=I=
stands for a simple detached line, or not more than two parallel lines,
in the heart of an encircling pattern.
=O=
stands for a little oval or circle, or for a round or oval-shaped dot
in a core. If the circle, oval, etc. is _large_, extending over a width
occupied by two lineations, then it is treated as a consonantal form.
[See also =Q=.]
=U=
indicates a fork with two or more prongs within a core, forking towards
the bend of bow, loop, mountain, etc. A single prong or spur standing
out like a twig is to be distinguished from a fork.
=Y=
is for a similar fork as described above, but turning its two or more
prongs away from the concavity of its enclosing loop, bow, etc.
Besides the direct combination of simple vowels and consonants, which
arrangement by itself gives great variety to the index registers, an
immense number of syllables are formed by combinations of two or more
consonants, while some few of the vowels are treated as long or short
where the pattern needs further discrimination; as, for example:--
_bra_, _spo_, _art_, _prīd_, _prĭd_, _nut_, _nūt_.
By this method the most extensive register is gripped _and needs no
other index than its own essential structure_. If the sheets or cards
are kept in their proper sequence, and it would require to be the duty
of some one--not necessarily an expert--to see that the alphabetic
syllables were kept in serial order, there should be no difficulty in
finding the document sought for, if it is there at all.
In translating fresh finger-prints into syllabic form, one has to
catch the ideal design, so to speak, in the pattern. The consonantal
skeleton, in one of its duplicate forms, is then examined for its
containing vowel, and the syllable is complete. The work can be done
with amazing rapidity after one is familiar with the patterns, which
soon appeal direct to the eye as the type does in a printed book.
Let us now look at a few examples tabulated to show how the system
works in detail.
[Illustration: VOWELS AND CONSONANTS IN SYLLABIC CLASSIFICATION
with typical specimens of figure elements.]