by This e-mail address is being protected from spam bots, you need JavaScript enabled to view it , PhD, WEMT-B

As we better understand the sciences of scent and of search, our practices for training and deploying SAR dogs may need to be refined. For example, the methods used by bloodhound handlers in the hunt for the 2001 anthrax mailer generated controversy. Just because dogs have the ability to identify individual humans doesn't mean that they'll do it reliably on command for us, as some lineup identification experiments have shown.

Lawrence Myers, a dog scent researcher at Auburn University in Alabama, is concerned at the way training practices and standards vary from state to state, and even county to county.

 "There are dogs who are trained to match human scents," he says. "Can they do that? Yes, but how reliably can they do it -- that's the question."

Myers believes that questions of handling scent evidence and how dogs are trained and tested deserve a national consensus, if not actual national performance standards.

"To me, there's never been a problem with using dogs for probable cause" -- in other words, to justify obtaining a search warrant -- Myers says. But states like California will accept dog identifications as direct evidence in court, despite scant scientific evidence backing up its reliability. Myers doesn't necessarily believe this evidence is unreliable: just that it hasn't been proven.

Such proof would require improved testing that will take time to develop, he argues. "Controlled conditions" -- setting up test problems for different dogs in conditions that are as identical as possible, and in which both dog and handler being tested can't know or be inadvertently tipped off as to where the target scent has been hidden, are a beginning.

But another step is necessary, Myers says: results from a proposed test must be compared with dog teams' performance in real search tasks to make sure that the test is testing the right things. Alone, real-world performance can't be reliable -- we can never know how many times a dog fails to detect a target that's never found, for example. Alone, a test might not be examining exactly the right skills. But when the two are cross-checked, a workable system may emerge.

"A ... reliable measure in the controlled setting should be predictive of performance in the real world, but that would have to be determined statistically, using large sample sizes," Myers says.  And of course, the points he makes don't just apply to scent discrimination -- they are valid for just about every testing and deployment practice for SAR dogs and scent dogs in general.

*An earlier version of this article appeared as a sidebar to "Who Goes There?" in Advanced Rescue Technology,  Vol. 7 No. 3 pp. 45-50. Copyright 2004 by Summer Communications, Inc. Reproduced with the permission of the publisher.