ROANOKE TIMES

                         Roanoke Times
                 Copyright (c) 1995, Landmark Communications, Inc.

DATE: TUESDAY, January 14, 1992                   TAG: 9201140081
SECTION: EXTRA                    PAGE: 1   EDITION: METRO  
SOURCE: JOE KENNEDY STAFF WRITER
DATELINE:                                 LENGTH: Long


FACING FACT

JOE Gibbs is not a suspect. As far as we know, the Washington Redskins' coachhas committed no crime, other than what some might describe as criminal mischief toward Lions.

But pre- and post-game interviews and commercials for a battery manufacturer have made him instantly recognizable in these parts. He is, therefore, the ideal subject for a composite sketch put together by the Virginia State Police, just to see how it's done. And how difficult it is to do.

Gibbs is a white male, age 51, of medium height and broad build. He has brown hair, aviator-style eyeglasses and a high voice. Identifying marks include a burgundy, baseball-style cap with a gold R written in script across its peak.

Our composite artist is George Watts, a state police special agent headquartered in Salem. His tool is an Identi-Kit, a book of disembodied chins, ears, noses, lips, eyes, hairstyles and other features that, when viewed each among its kind on page after page, begin to look absolutely creepy.

The kit comes from Smith & Wesson, manufacturer of the famed handguns that make some people instantly eligible to have their composites drawn - for real.

Watts doesn't know whose face I'm trying to put together. He begins by asking me a few basic questions about our subject. I describe him as I did above. Watts writes it all down and then consults a card from his kit. This gives him the numbered foils he needs to create the general description I've provided.

"The first time I put it together, it's not supposed to look like anything like the person you're describing," Watts says.

Even in the final analysis, he says, "the composite does not look like a portrait of the person you're trying to describe." Rather, it's a compilation of general characteristics from which investigators can work.

Watts lays down the chin, creates a face with metal-framed glasses and leaves the hair off.

"Bill Cochran?" he guesses, naming the Roanoke Times & World-News outdoor editor. Watts is an accomplished bowhunter. They are acquainted.

"Nope," I say.

He fiddles a little more, allowing me to choose the hair. Then he turns the sketch toward me.

Immediately I notice that the subject's hair is too light. Watts changes it. I'm not sure it's parted that way, either. It was tough to tell, with that hat. And his neck is too wide - like a pad-popping lineman's, not a deskbound coach's. Watts thins it out.

We look at the sketch. Watts has no clue. I'm not happy. I look through the mouth and hair samples, but can find none better. "Give up?" I ask. Watts nods.

"Joe Gibbs."

The agent studies our work and concludes, "That's a realistic likeness of Joe Gibbs. Really."

He explains: If I were in a bar with all the Redskins and Detroit Lions, and a crime were committed, and I provided this description, it would enable investigators to eliminate 95 percent of the people associated with each team.

Police could concentrate on the three or four who remain. Presumably, by then, Gibbs would be wearing contacts and a mustache.

I have to confess. I didn't fall into the trap of trying to be too fine with my description, because I'd already done that on a previous visit to headquarters. Then, I interviewed Special Agent Randy Dyer, another of the four staffers who work with the Identi-Kits. Dyer had me try to describe myself. The result, after a much more arduous effort, resembled Tom Brokaw with a hangover.

The kit isn't perfect, Dyer said: "It's only as good as your witness." The artist, after all, has never seen the suspect and has no idea what he or she looks like.

Even with my poor description of myself, Dyer said, "Out of a crowd of 100, we could pick you out, or your neighbor could tell from a picture in the paper."

Person-to-person identifications are only one tool in the investigator's arsenal, both men said. Physical evidence and other clues are often more reliable.

Bad descriptions are the norm, says David Simon, author of "Homicide," a book about homicide detectives in Baltimore.

"In a crowded world people just don't have the facility to commit a new face to memory," he writes. "Many veteran detectives don't bother to include preliminary descriptions in their reports for that reason: A description of a 6-foot-2, 220-pound suspect will hurt you in court when the guy turns out to be 5-7 and 150."

Simon says law enforcement studies have shown that interracial identifications - blacks of whites, whites of blacks - tend to be weakest. Both races have trouble distinguishing members of the other.

But all three men have seen instances in which the composites have proved startlingly similar to the alleged perpetrators. Watts says one was the Roanoke police composite of Carolyn Snyder, the woman suspected of being the mother of Baby Isaiah, the infant found in a Roanoke trash bin in December 1990. Snyder's attorney eventually convinced a judge she was unable to bear children.

Dyer recalls a case from Henry County. A woman was raped on a Saturday night. On Sunday morning, she described her attacker to him for a composite. On Monday morning, the sketch was passed out at the sheriff's department lineup, and someone said, "We locked him up last night in a stolen car."

They checked the jail: same guy.

In Baltimore, police often have an artist do the sketch, rather than use Identi-Kits, Simon says. Nationally, computerized sketching is the latest trend.

Composite sketches always seem eerie when you see them in the paper, and you always wonder how accurate they are. The answer is: sometimes very accurate, sometimes not very accurate and sometimes accurate enough.

But no matter what the method composition might be, only a good witness can make it work. Otherwise, it's garbage in, garbage out.



by Archana Subramaniam by CNB