Science
and fingerprinting
Some
people describe fingerprinting as a science. There are many areas of human
activity where practitioners study the subject and discuss theory but the term
"science" is usually reserved for those subjects where theories and
opinions are expected to be formally tested, with empirical evidence as the
ultimate arbiter of truth.
What
struck me after watching snippets of the Scottish Parliament SCRO inquiry on TV
was the absence of evidence-based reasoning.
For example, fingerprint experts from the Netherlands and USA said that
latent print Y7 could not have been deposited by Shirley McKie because they
could point to an area of difference between Y7 and McKie's inked print. On the
other hand Mr Peter Swann and the SCRO said that the discrepancy can be
explained by pressure distortion while the latent print was being
deposited. I don't know if these
differing opinions are due to differences in training or tradition or something
else but it could make the difference between someone being declared innocent
or being sent to prison.
I
would expect a science to resolve this by referring to evidence. I don't know if it is feasible to find the
limits of discrepancy between donor and latent that can be attributed to
pressure distortion by a formal study of latents deposited by known
donors. The handling of discrepancies
in day to day fingerprint work would then be governed by the results of these
studies rather than being left to the judgement of each fingerprint department.
Scientific knowledge is never defined by opinions alone, no matter how eminent
and experienced the people offering the opinions are. Medicine discovered this 400 years ago. In a profession where
public safety is involved any opinion that is not firmly based on evidence
should be looked on as potentially dangerous.
I
would like to see some sort of empirical (not just theoretical) evidence that
individualisation occurs in the overwhelming majority of fingerprint
cases. I would say that the 'Town
Fingerprint Project" shows that you don't need to have true
individualisation with an error rate of zero for fingerprinting to be
reasonably safe in normal situations. In the absence of such proof I would say
that the precautionary principle should apply and it should be accepted that
fingerprinting lives in the world of probability. Statistical modelling could then be used to discover if there are
any other especially unsafe practices (I have found one).