8127/frame/fm

particular test, the description of DNA fingerprint, drug fingerprint, or ret-
inal fingerprint is invoked. Interestingly, there is no agreed-upon standard
for individualization by fingerprint comparison (Cole, 1998; 1999). Although
a number of mathematical models have been proposed, starting with Galton,
none has proved fully adequate to describe the process, significance, and
limitations of fingerprint comparison (Stoney and Thornton, 1986a). One
of the models, proposed by Balthazard in 1911, may have provided the
historical basis for what was to become widely accepted rules regarding
fingerprint individuality* (Stoney and Thornton, 1986a), and in 1918
Edmond Locard wrote that if 12 points (Galton’s details) were the same
between two fingerprints, it would suffice as a positive identification. This
may be where the often quoted 12 points originated (Moore, 1999). Other
theories have also been proffered (Kingston and Kirk, 1965). One of those is
detailed in Sidebar 2.
To this day, no comprehensive statistical study has ever been undertaken
to determine the frequency of occurrence of different ridge characteristics
and their relative locations (Saferstein, 1998; Cole, 1998; 1999). In 1973,
frustrated by the imposition of artificial thresholds, the North American
fingerprint community, under the auspices of the International Association
for Identification (IAI), adopted a resolution rejecting an arbitrary number
of corresponding points as the basis to accept or reject an individualization.
The International Association for Identification assembled in its 58th annual
conference in Jackson, Wyoming, this first day of August, 1973, based upon
a three-year study by its Standardization Committee, Hereby states that no
valid basis exists at this time for requiring that a pre-determined minimum
of friction ridge characteristics must be present in two impressions in order
to establish positive identification. The foregoing reference to friction ridge
characteristics applies equally to fingerprints, palmprints, toeprints and
soleprints of the human body.
* Using a world population of 15 billion human fingers, Balthazard determined that 17
corresponding minutiae would be needed to conclude a unique association between a print
and a finger. (Under his model, 17 corresponding minutiae would be found with a frequency
of only about 1 in 17 billion.) Balthazard considered a lesser number of corresponding
minutiae (for example 11 or 12) to be sufficient for an unequivocal identification if one
could be certain that the fingerprint donor was restricted to a particular geographical area
(Stoney and Thornton, 1986a).
8127/frame/ch02 Page 46 Friday, July 21, 2000 11:51 AM

---

The Evolution of Forensic Science
47
In 1995, the International Fingerprint community endorsed this standard in
the 
Ne’urim Declaration
(Israel National Police, 1995; Cole, 1999).
In other words, the decision to “make” a print has been left entirely up
to the experience and intuition of the examiner.*
In the first edition of 
Crime Investigation
(1953), Kirk, in an attempt to
explain the probabilistic nature of fingerprint evidence, gives an example of
how individual fingerprint traits may be multiplied to produce a composite
frequency.
Assume that, on the average, one person in every twenty is found to have
a whorl on the left thumb. Assume further that one in every ten is found
to have a whorl on the right index finger. Then the probability that an
individual picked at random will have both whorls is 1/20 times 1/10, or
1/200. To generalize, the probability of any combination of characteristics
being found in a given person is the product of the probabilities of each of
the individual characteristics. With the ordinary methods of classifying
fingerprints, the probability of any but the one person in question having
a particular fingerprint classification is so minute as to be negligible. For
this reason only, fingerprints may be accepted as definite identification.
The problem, which Kirk sidesteps along with everyone else, is the lack of
scientific basis for the actual frequencies of the individual traits; Kirk gives
no foundation for his 1/20 or 1/10 assumption** and fails to even address
the issue of independence. Perhaps it is telling that, although his premise
remains the same in the second edition, edited by John Thornton, the specific
example has disappeared.
In 1986, Stoney and Thornton (1986b) published an article in which
they reviewed fingerprint individuality models proposed throughout history
and defined a list of features that they resolved should be incorporated in a
comprehensive and utilitarian model. They conclude with the caution that
the value of any fingerprint for identification is inversely proportional to the
chance of false association and that this chance depends on the number of
comparisons that are attempted. These ideas have yet to be developed and
put to practical use.
One consequence of the growing emphasis on individualization has been
the tendency to overinterpret evidence that does not necessarily hold indi-
vidualizing potential. The history of microscopic hair comparison exemplifies
* It should be noted that the IAI resolution is not binding and many agencies continue to
adhere to a minimum number of corresponding points to conclude identity. For instance,
Great Britain continues to hold tenaciously to a standard of 16 points. 
** Some basis for these estimates may be found in historical fingerprint models such as
that of Gupta (Stoney and Thornton, 1986a), but Kirk provides no reference.
8127/frame/ch02 Page 47 Friday, July 21, 2000 11:51 AM

---

48
Principles and Practice of Criminalistics