tmp45A9 Page 11

  points, and as I went through fingerprint cards every day, I could stop and compare any print I found that contained those five points. However, these five points were rare enough that in actual practice I have never seen another fingerprint, out of the hundreds of thousands of individual prints I have looked at, containing these points.

  Which brings us to another common question.

  Points of Similarity

  How many points are required for certainty? Some countries have ruled on this; some of them require twenty-four points, which I consider an idiotic rule. To the best of my knowledge, no jurisdiction in the United States has set any standard. Very few technicians would be satisfied with fewer than eight points unless the points were extremely unusual; I will add that I once saw two prints that had seven points of similarity and that were not made by the same hand. A detective from another police department brought me the prints to look at, and it was only after he left that I realized I should have photographed the prints. I've wished ever since that I had, because that many points of similarity on nonidentical prints is extremely rare. In fact, as odd as this may seem, it is rare that there are any points of similarity on nonidentical prints, no matter how identical their classifications may be.

  Determining a Thumbprint

  Examine your thumbs carefully. You'll notice that at the top of your right thumb, the ridges tend to flow up and in toward the body. On the left thumb, the ridges at the top tend to flow up and in toward the body as well. This means that when the print happens to include the upper area, it is always possible to tell that it is a thumb, and which hand it belongs on.

  The Most-Used Fingers

  In burglaries and robberies, the fingers that most often show up are the index and middle fingers of the right hand, with the ring finger and the thumb running a close second and third. Remember, these are fingers 2 and 3 first, 1 and 4 second, according to the fingerprint numbering system. In forgeries, we get fingerprints far less often than we get palm prints — and you'll remember the discussion of palm prints in chapter four.

  Among other things, all of this means that any time you have a thumbprint, or any time you have two fingers together, you can

  always tell which hand was used. If you have just one finger and the print is of a loop, you can be almost certain which hand it is, because radial loops—loops sloping toward the thumb rather than toward the little finger—are rare, and they occur almost solely on the index finger. Therefore, you can make an educated guess as to whether the person who left the print was right-handed or left-handed. With whorls and only one finger, you are on less sure ground; with tented arches, you can rarely make more than a guess; and with plain arches, you'd really have to have at least the smudge of another finger to be sure which hand the print belongs on.

  Can Fingerprints Be Forged?

  When I first began fingerprint work, we were all assured that fingerprints could not be forged well enough to fool a good identification technician. That may still be true. But what most of us never considered was that a good, but dishonest, identification technician could forge a print and get away with it, as long as no other good identification technician studied his work. And we went on believing that until a case broke that made front pages across the country and was written up in Reader's Digest. The answer now is, yes, fingerprints can be forged. And at least one man went to prison on the basis of forged fingerprints.

  Here is how the technician did it: First, he photocopied a known fingerprint of the suspect. Next, while the photocopy was still hot, he put fingerprint tape down on it and lifted the print. (This, of course, was perfectly easy to do; remember that powdered photocopier toner is used for fingerprint powder in some situations.) Then he put the tape down on the object he wanted to "find" the print on. And then he photographed the print, with the explanation—perfectly reasonable—that the tape had been put on the print to protect it.

  The forgery finally was detected in the following manner: The suspected forgery was compared with all known prints of the onetime suspect, now victim. Although identification points are the same on all prints from the same finger, no two prints themselves should ever be identical, because of slight or great differences in position, pressure, and so forth. Therefore, when an inked print identical in every way to the purported latent was discovered, a strong presumption of forgery was created.

  When the story broke in the media—when Doc was out of town and Butch not yet in ident—I knew every person in the police department was going to ask me questions about it. In order to forestall having to answer a hundred-and-eighty-odd different people, as soon as I read the article, I pulled my own fingerprint card out of the applicant file, photocopied it, lifted a print from it, taped it down on a lift card, typed an explanation under it, and put it on the bulletin board. It took me about five minutes, counting typing. That's how easy it was—and if the dishonest technician had destroyed the card he used to create the forgery, he might have gotten away with it.

  This means that more than ever, the honor of the technician is on the line when s/he testifies in court.

  And this certainly gives you something to play with in fiction.

  New to be lair—Things are possible in small departments, small jurisdictions, that may not be possible in larger ones. When I commented that I had made a nonsuspect ident from a palm print, a task the FBI says is impossible, I will hasten to add that for the FBI it would certainly be impossible. It took me several weeks, and I was working with a universe of fewer than 1,000 palm prints.

  In Albany it was practicable —though not very practical —to search a single latent print manually through the entire files as well as through the specialized files, although it might take upwards of a year. In Salt Lake City, with closer to 90,000 fingerprint cards, it would be on the outside edge of practicable, and might take ten years. In New York it would be totally impossible, and the mind boggles at the thought of trying to manually search a single print through the massive FBI files, which at least in theory include every person who's ever been arrested, served in the armed forces, or applied for security clearance in the entire United States. That's why older novels in which a single print is rushed to the FBI without a suspect, and the FBI comes right back and tells who made the print, are totally unrealistic.

  Using AFIS

  The situation now is quite different. Using its AFIS system, the FBI can make up to one hundred single-print searches through its twenty-three-million-card criminal file in a single night and have the

  results the next morning. Generally, however, the entire file is not searched; rather, cards fitting within logical parameters are searched. For example, if a bank is robbed in Utah by a perp described as a white male in his mid-forties, the search might begin with a look at all known white male burglars and robbers between thirty-five and fifty-five in Utah, Idaho, Nevada and Colorado. If necessary, the parameters would be repeatedly enlarged until the whole file had been searched.

  In the past, even an ambitious and very hard working police department in a very large city found making nonsuspect idents difficult if not impossible, simply because of the quantity of cards that would have to be searched. A department in a town with many transients, no matter what the town's size, would have a terribly difficult time. Only a moderate-sized town with a stable population stood much chance of making the kind of record Doc and I created in Albany, Georgia, and other departments made in such places as Galveston, Texas (despite its large transient population) and Shreveport, Louisiana.

  AFIS has changed all that. AFIS, first thought of when computers began to make a large impact on law enforcement in the sixties, was intensively studied in the seventies and first implemented in the late eighties. Although the acronym stands for Automated Fingerprint Identification System, let me say at the outset that the computer itself does not make the identification. The best the computer can do is toss out one or more—up to ten—possibles. The identification itself must be made by the human eye.

/>   Because I left identification work before any AFIS systems were in use, I am indebted for this information to published Department of Justice documents, to Lynn Bergen of the Salt Lake City Police Department crime laboratory, and to the FBI latent fingerprint section.

  To start with, there are several different AFIS systems and AFIS networks, and like any other dissimilar computer systems, these cannot "talk" to each other. Furthermore, no state or local police departments are able to tie into the FBI AFIS system, so the dream of being able to search every fingerprint file in the country in minutes or days is still no more than a dream. However, the FBI hopes that by 1995 all police departments will be able to tie into what they call a "latent cognizant data base," which will contain at least a large part, if not all, of the FBI criminal data base.

  But from Salt Lake City, it is possible to search more than four million prints inside the state of Utah, and to access the fingerprint files of Alaska, Oregon, Washington, Idaho, Wyoming, Nevada and (with special permission only) California. Call your state crime laboratory to find out what states can be accessed from your jurisdiction, or the nearest real jurisdiction to your imaginary jurisdiction.

  And here's how it's done.

  Figure 5-7 shows a single fingerprint. The first arrow points to the core of the print. The second arrow points to the axis of the print, which is the midline axis of the finger that made the print. In order to be put on an AFIS system, a latent print must be in a black-on-white format. If the print was processed with black powder and mounted on a white card, it is already in the correct format. If it is in any other format, the following steps must be taken:

  First, the technician photographs the latent. The photograph is then enlarged five times, and the technician carefully traces the latent in black ink on white tracing paper. The tracing is then photographed and reduced back down to a one-to-one size, and the result is taken to the computer.

  The unknown latent, or the reproduction of the tracing, is placed in the computer, and the technician manually sets cross hair references, similar to the cross hairs on an aiming device, on the core and the axis. The computer then rotates the print to approximately 15 degrees left and right of the axis, plotting where minutiae—those points of identification we've already talked about— fall. The larger the number of points, the larger the numerical rating the computer assigns the latent, and the larger the numerical rating, the higher are the chances of an identification — although a few AFIS hits have been made with as few as eight points.

  In addition, the computer may be given specific parameters, such as whether this print is from a male, a female, or sex unknown; what race it is or race unknown; a specific or approximate birthdate or birthdate unknown; and so forth.

  Distinguishing Sex and Race From Fingerprints

  It's impossible. Although there are very slight statistical differences in race, the differences are nowhere near enough to determine what race a person is. There is no discernable difference whatever in sex; although size is sometimes an indicator, a large woman will leave larger prints than a small man. A witness might help, but remember what I said about eyewitness testimony ... and thereby hangs a tale.

  A woman was attacked and raped in her bedroom. Although all her lights were out, there was good light coming in from a street light just outside her bedroom window. The MO fit a known rapist in her neighborhood, but he was black, and she said the man who attacked her was a white man with a bushy beard.

  Detective Larry Grey picked up an extremely good latent from her window glass, a thumb print outside and a smudged index fingerprint inside, so although the index fingerprint was unreadable, we were sure the prints had been made after the glass was removed from the window. I searched the print through known white sex criminals and known white burglar files. Finding no match, I filed the print.

  Several months later, when things were quiet, I was re-searching old latents. When I came to that one, I felt a little lazy. It happened that because the number of sex criminals in Albany was fairly small, we had black and white filed together, and I knew I had a right thumbprint. The right thumb is always in the same place on every fingerprint card, but because of changes in the design of fingerprint cards, race can be noted in any of several places. I decided it was easier just to check the print than to check the race.

  I made a very quick ident. She was attacked by the man Larry and I had thought of from the first - and she was so panicked that she could not remember, by the time Larry talked with her, what race her assailant had been. Her white man with a bushy beard was in fact a black man with a bushy beard.

  I do not like eyewitness testimony.

  The computer then begins to search for similar prints in its data base. It produces up to ten candidates and then displays the candidate prints beside the latent on a dual screen, so that the identification technician can make the preliminary comparisons. If the prints seem close enough, the technician then, still with the computer, goes into the data base and pulls out an image of the card itself, to make a definitive comparison. Eventually, if the identification seems posi-

  tive, the technician will compare the fingerprint card itself with the latent.

  It is important to remember that no identifications are made by, or even from, the computer itself. A computer is not capable of making fingerprint identifications; rather, the computer reads what are called events—that is, breaks in ongoing patterns.

  And what have the results been?

  Most fingerprint technicians—including me—on first hearing of the plans for AFIS, back in the seventies, felt extremely dubious. We felt, correctly, that it would be impossible for a computer to take into account and allow for all the distortions, changes and so forth that result from position and pressure, that the human eye adjusts for so easily.

  Every technician who has ever worked with AFIS has been wild about it. It does not, as we assumed correctly from the start, replace the trained eye of the skilled fingerprint examiner. What it does do is eliminate about 95 percent of the scut work, the time-consuming work, and allow the examiner to accomplish far more in less time. A latent can be searched through 50,000 cards in less than an hour; the technician then has to actually compare only about ten individual prints. The total expenditure of time, not counting getting the latent and doing the charting to take it into court, might be no more than two hours. Searching manually, squeezing the work in among all the other work an identification bureau is constantly doing, it might take five years to search a single latent through 50,000 cards.

  Fingerprint cards, involving all ten fingers, also may be searched through AFIS. According to the Department of Justice, in a pilot study on AFIS, which is listed in the bibliography, "Search time for a ten-print search (rolled print to rolled print comparison) in a file of under 500,000 is a matter of minutes" (5). This is extremely important, because criminals like to give false names. In the past, although a fingerprint card on every arrestee was routinely sent to the FBI, it might take several weeks for the card to be classified and manually searched, and in that time a badly wanted person, picked up for a minor offense, might be back on the street and gone. And it's frustrating to find out that cop-killer you picked up for public intox, and released after he served his three days, has vanished again. That no longer happens in a jurisdiction that uses AFIS.

  The Department of Justice pilot study says that in the first two years of development, the California Identification System "entered 420,000 fingerprint cards into the file that is used for latent searching. The file includes persons born in 1950 or later who have been convicted of a felony, a group that is estimated to represent only 34 percent of the AFIS data base but 47 percent of daily AFIS activity____In its first year of operation, the San Francisco Police Department's AFIS system conducted 5,514 latent print searches and

  made 1,001 identifications____[They] cleared 816 of those cases,

  including 52 homicides, compared to 58 cases cleared the previous year on the basis of latent pr
int identifications" (13). That set of statistics alone says all that needs to be said on the value of AFIS.

  Bear in mind that all of these are nonsuspect identifications, what the Department of Justice refers to as "cold makes." Generally one hit on a nonsuspect search will clear from one to twenty or more crimes, because criminal MO's are fairly recognizable, and because a search warrant, issued on the basis of such a hit, is likely to find the proceeds of several crimes.

  You may, legitimately, wonder whether the attention of ident people ever wavers. Do matches slip by? How often?

  I can't answer that, of course. I have never known a match to slip by, but if it slipped by me, of course I wouldn't know. Probably matches do slip by. That's not ideal, of course. But I do know that no legitimate identification person ever, under any circumstances, goes to court and claims a match when there is not one.

  That's what's most important, when you're talking about someone's liberty, perhaps someone's life.

  What does this mean in fiction? It means what you want to make it mean. You can let your criminal leave his print and be caught by AFIS. You can let your criminal be too smart to leave his prints, or you can let his prints be so pressure-distorted that AFIS can't read them. You decide—but the technology is there, and as a citizen, you certainly have the right to demand that your jurisdiction use it, no matter what you do in fiction.

  TABLE 5_

  How You Can Use Experts

  A theory says that provided you do your library research first you should be able to locate any known fact (as long as it is neither a government secret nor a proprietorial trade secret), in a maximum of five telephone calls.

  When I heard that, I couldn't believe it. So I tested it—and it worked! I then began passing the theory on to my students, and they also tested it. It worked for them, too, and it will work for you.