By Thomas J. Baker, Consultant, Arlington, Virginia
“Is that him?” Standing on a street corner is a group of young males, and only one of them is the man sought by law enforcement officers for an interview or arrest. The question is, which one?
This is a scene that happens frequently in cities and towns around the world. The challenge faced by law enforcement officers is whether to extract the individual from his group, likely antagonizing the others in the process, and how to be sure the individual extracted is the actual suspect.
Similar scenarios have ended in a riot, in individuals being injured, and in officer deaths. Sometimes, a commander decides the mission is too difficult to complete, and nothing is done.
iometrics is the study of different characteristics in the human body and their measurements. It comes from the Greek words bio (life) and metry (to measure). As a system for identifying individuals, it is a relatively new word and a new concept.1
There are currently more than six billion people on Earth, and there were another six billion here before the current population over the last 100,000 years. With very few exceptions, humans are all basically the same. Yet each is also characteristically unique, and each can be distinguished in many ways from any other person alive today or who ever lived in the last 100,000 years.
Human differences are the result of what scientists call “chaotic morphogenesis,” a term used to describe the process of random variation in growth and development by which each human becomes physically individualized. This is even true for twins who share DNA for the same physical traits, as some of those traits will be manifested in different ways. That is why, for instance, everyone’s fingerprints are unique.2
These differences go to the heart, literally, of who each human is. Three years ago, I had a heart attack and subsequent bypass surgery. During a follow-up visit, my cardiologist used a model of the human heart to explain what had happened. Pointing to one spot on the model, he said, “The blockage happened here.” Then he added, “But your heart is not like that; your artery forks there.” The doctor said that some people have two branches at that specific point, and some have three. Pointing to his own face and then mine, he summed this up: “We’re as different on the inside as we are on the outside.”
This is chaotic morphogenesis, and this is what biometrics are all about.
Artery patterns on a heart may be hidden inside, but some unique patterns found in other body parts—those on the outside, such as the iris and retina of human eyes, the ridges and valleys of human fingerprints, and even the contours of human ears—are different from person to person and can serve as built-in identification tags.
Verification versus Identification
Almost all use of biometrics can be classified into one of two operational areas: verification and identification.
The uses most often depicted in the media or seen in commercial contexts relate to verification: an individual registers a biometric, usually a fingerprint, in a database ahead of time, and this biometric then functions like a key or a password that permits the individual to use a bank machine or authorizes the individual to log on to a computer. This is a one-to-one comparison to determine, for example, whether the person claiming to be Mr. Smith really is Mr. Smith.
Most law enforcement applications relate to identification. The biometrics used in this context are by no means limited to fingerprints. For identification, a biometric is obtained, either directly from a person or found at a crime scene, and it is compared to a database of thousands or even millions of other biometrics in search of a match. This is a one-to-many comparison in which an unknown person’s biometric data must be checked, for example, not only against that of Mr. Smith, but also against that of Mr. Brown, Mr. Jones, Mr. Doe, and many others.
The distinction between verification and identification has some important policy considerations for law enforcement.
The standard biometric for law enforcement has long been, and still is, fingerprints. Indications exist that fingerprints were used as identification tools by people 6,000 years ago in Babylon. In the fourteenth century, the Chinese used footprints, palm prints, and fingerprints to sign important documents such as treaties and contracts.3
The first organized approach to fingerprints for identification purposes was made in 1858 by Sir William Herschel, 2nd Baronet, an English civil servant working in India. He started with palm prints, and then he went to fingerprints.4 (Today, the value of palm prints is once again being recognized.)
Sir Francis Galton in the 1890s devised a fingerprint system, making use of what we now call “minutiae”: the major features of a fingerprint. In the late nineteenth century, Galton’s contemporaries called these characteristics Galton’s Details. He gave us the concept of the loops, arches, and whorls in fingerprints. This approach was first adopted in Argentina, and the Galton system is now used in much of the Spanish-speaking world.5
Sir Edward R. Henry, a British police official in India in the 1890s, built on Galton’s work and created a system of classification based on fingerprint patterns.6 We now know this as the Henry System. Henry eventually returned to the United Kingdom and became an assistant commissioner of the Metropolitan Police in London. In 1901, he started using the Henry System to classify fingerprints collected by his police agency. The Henry System came to be used in almost all the English-speaking world.
In the United States, the first systematic collection of fingerprints by a federal authority began at the federal prison in Leavenworth, Kansas, in 1904. Nearby, the Saint Louis, Missouri, Police Department began collecting fingerprints at about the same time, making it one of the first police agencies to do so. Upon hearing about the agency’s efforts, other police agencies started to send their fingerprints to the Saint Louis Police Department. Eventually, the IACP assumed responsibility for the Saint Louis database. The IACP later asked the Federal Bureau of Investigation (FBI) to take on the management of these growing fingerprint files. That database, as along with the fingerprint files from the federal prison system, became the foundation of the FBI fingerprint files in 1924.7
Decades later, the use of computers made possible the establishment of the Automated Fingerprint Identification System (AFIS), which has sometimes been referred to as the Automated Fingerprint Imaging System or Automated Fingerprint Information System. In 1999, the FBI’s Integrated AFIS (IAFIS), so named because it pulled a number of functions together in one system, became available online.8
Fingerprint databases equivalent to AFIS in other nations had gone online a few years before the FBI’s IAFIS. Now, a number of systems around the world are moving to a second generation of automated fingerprint systems. The FBI is transitioning to a new system called Next Generation Identification (NGI).9 The Republic of Ireland just went to its second-generation system. The United Kingdom also has moved ahead with its new system, Ident1. The Netherlands is in the process of moving to a second-generation system as well. Along with increased speed and accuracy, the next generation of AFIS will see increased use of photos, palms, and improved latent functionality.
The law enforcement standard has always been 10 rolled fingerprints for enrollment. Known as the gold standard, 10 fingers rolled provide the most data. A second method of taking fingerprints is a flat impression from a single finger. This is sufficient for agencies that are solely concerned with verification of a specific person’s claimed identity. The flat impression used on some countries’ travel documents is a good example. A third method of fingerprint enrollment is called slaps. Slaps and flats are often confused, but they are not the same thing. Slaps are four fingers taken together.
The distinction among flats, slaps, and 10 rolled prints is where the policy considerations between verification and identification come into play. In the AFIS systems that many states and nations are currently upgrading, law enforcement is often being asked to incorporate the fingerprint records of an immigration service or a customs service or other agency that is concerned with simply verifying the identities of people entering and exiting a country.
But law enforcement is concerned with far more than mere verification. Law enforcement needs to collect as much data as possible. Law enforcement needs to have a 10 rolled print with all the data it contains. Some border control authorities would rather have a quick flat impression. One can see from their business models why they would want that fast verification. But this is something that law enforcement will have to resist because one characteristic on the side of somebody’s finger could very well turn out to be the one piece of evidence that winds up on a piece of paper or on a bomb fragment, suddenly becoming quite important.
The biometric traces left behind on a bomb fragment are called latent prints in the United States. In other countries, they may be referred to as tracing or markings. Latents are extremely important. Fingerprints are the only biometric that leaves something behind. And that is why, for the foreseeable future, fingerprints are going to remain law enforcement’s most important biometric.
Palm prints largely fell into disuse after the early days of Sir William Herschel, but their importance is being recognized anew in contemporary policing. About 30 percent of all the latents found at a crime scene are palm prints.10 One part of the palm that is particularly important is popularly called “the writer’s palm”—that is, the outside palm. In particular types of crimes, such as kidnappings, extortions, and bank robberies, the writer’s palm often makes the case. At a bank robbery scene, a single subject with a gun might stand in front of the teller with his gun hand resting with its outer palm on the counter. Especially in older banks with counters made of polished marble, metal, or glass, a clear writer’s palm print is often left behind.
Sometimes the bank robbery is what is called “a note job.” That is when the bank robber does not show a gun but passes a note to the teller claiming to have a bomb or a gun. The robber often takes a sheet of paper from the supply of deposit slips and writes the note right there in the bank. A good writer’s palm print left behind is often located where the note was written.
Some agencies only take the flat palm. For all of the above reasons, it is important to encourage the taking of the outer palm as well.
Palm prints are part of Australia’s National Automated Fingerprint Identification System, recognized as a model system.11 In the United Kingdom’s new Ident1 system, palms are an important component.
Facial recognition12 is a relatively new biometric that is garnering attention. Facial recognition has many advantages, a major one being that it is the only biometric that can routinely be obtained surreptitiously. Therefore, it has value for use in surveillances. Facial recognition is not as accurate as fingerprints—at least, not yet—but it is becoming increasingly more accurate as new advances in the technology are made.
There are a number of different theories of facial recognition; that is, different algorithms exist that scientists use to measure facial characteristics. Most of these are under copyright. Scientists at universities and in research labs of biometric companies continue to develop new approaches. Each approach works differently, looking at different parts of the face or looking at the face in different ways. Ear shape, for example, is the focus of some facial recognition systems. Ear shape changes very little as we age. Some researchers consider the shape of the ear a separate biometric.
None of these facial recognition systems are 100 percent accurate, but some of them are approaching that level.
As various facial recognition algorithms merge, the accuracy of facial recognition will increase. For instance, Cognetic, a company based in Dresden, Germany, has a good facial recognition algorithm. They have partnered with Morpho, a company based in Paris, France, which also has a variety of facial recognition algorithms. These two companies are merging their facial algorithms. Other providers and researchers around the world also will be doing this. As different methodologies merge, the result will be increased accuracy. Reliable applications in this field are not far off.
Iris recognition is another biometric of recent interest. The iris is the colored ring around the eye. Like fingerprints, the irises are formed in the womb after conception so that no two people, even twins, have the same iris.
Everyone has probably seen a very elderly person, such as a 100-year-old woman, photographed or interviewed on television. Sometimes in a close-up of her face you see her eyes sparkling, and they still look very beautiful. The reason is she has the same irises as she did when she was a 19-year-old young woman. Her skin may have wrinkled with age, but the iris has not changed at all.
The iris can be used for both verification and identification. It is not useful in surveillance because no method yet exists to get close enough to the subject’s iris without the subject’s consent or cooperation.
Confusion sometimes exists between retina and iris scans. The retina is the rear inner surface of the eyeball. It turns out—once again, thanks to chaotic morphogenesis—that the patterns of veins and nerves in human retinas are different in each individual. The pattern does not change over time, so the retina too can be used for identification. But it is difficult to work with the retina since it is so far back in the head—unlike the iris, which is externally visible. Working with retina identification requires significant cooperation from the subject. It has been used in some places for access control (that is, for verification). The Strategic Air Command of the U.S. Air Force and the U.S. Department of Defense has used it for this purpose, but it is not as easy to use as some other biometrics.
Further Biometric Frontiers
Hand geometry is another biometric. Not to be confused with palm prints or fingerprints, hand geometry deals with the shape of the hand. Like retina patterns, hand geometry is used mainly for access control. At the present time it has very little investigative value to law enforcement. There are many other biometrics being developed: skin texture, vein patterns, finger geometry, and ear shape as a separate biometric.
Speaker recognition is also a biometric. It should not to be confused with voice recognition, which is not a biometric. Voice recognition is used in devices such as dictating machines and translation machines. Voice recognition recognizes the words. Speaker recognition recognizes the person who is speaking.
Other biometrics are even further away in our future. These are being researched but are not near deployment. These include body odor, body salinity, lips, fingernails, and gait. They may have limited use, but each of these biometrics are currently being researched.
It may surprise some people to learn that the growing array of biometrics under study does not include DNA. This is because DNA is not a biometric.
Biometrics can be obtained through observation, albeit with the help of high-tech instruments in some cases, but DNA cannot be observed. In the case of DNA, unlike biometrics, an actual tissue sample is required, and, in contrast to biometrics, this sample cannot now be compared immediately and must be examined by an expert. And although DNA identifications are accurate, they cannot distinguish between identical twins as can true biometrics. As DNA patterns are formed at conception, twins will have the same DNA. But biometrics such as irises and fingerprints are formed in the womb, so even identical twins are distinguishable.13
Trends in Biometrics
As biometric technology expands, so do the methods by which a person can be identified and the accuracy with which an identification can be made. Here are five areas in which important trends have begun to emerge:
- Fingerprint identification—to include palm prints—will continue to be the main biometric that law enforcement depends on far into the foreseeable future. Law enforcement will continue to want all the data that can be gathered and must continue to resist pressure from other services for the simple “fast flat.”
- Facial recognition is going to have increased investigative use and intelligence use. Unlike fingerprints, however, law enforcement likely will not use facial recognition for evidence in the immediate future.
- Palm prints will see increased investigative use in the immediate future. As the national AFIS databases turn over to their second generation, law enforcement should seize the opportunity to make sure that palm prints are included in the new systems.
- Multimodule, sometimes called biofuse, cataloging is putting two or more biometrics together. The FBI will have multimodule capability in its NGI. In multimodule, for example, the digitized data from both a subject’s face and a subject’s fingers are stored together on the same disk. When there is a hit on either biometric, the results for both are displayed.
- International cooperation and the recognition of the need for it is a major current trend. Law enforcement agencies need to have the means, the modalities, and the shortcuts to share biometric data with one another. As J. Edgar Hoover once said, “Cooperation is the backbone of law enforcement.” That quote is now engraved on the courtyard wall of the FBI Building in Washington, D.C. He said it when he was concerned with federal, state, and local authorities in the United States. Now, all the nations of the free world must talk to one another and exchange data.
But what happened on the corner? The police quickly searched their database for a digitized arrest photo of the man they wanted. Using a surveillance van, the police photographed the young males on the corner and ran the photographs against the mug shot using facial recognition software. The on-scene commander received a message that confirmed, with 99 percent certainty, that one specific individual present was the intended suspect.
The team moved in and made an arrest. ■
1The first use of the word “biometrics” in this sense appears to have been in 1981 in the New York Times. Merriam-Webster’s New Collegiate Dictionary, 11th ed., traces the word back to 1902 in a different sense: referring to methods of statistical analysis used by biologists.
2“Chaotic morphogenesis” is such an important phrase in biometrics that a greater definition may be necessary. Morphogenesis comes from Greek words morpho (form or shape) and genesis (origin). Chaotic, in this sense, has a very specific mathematical definition: a pattern or state of order existing within apparent disorder—for example, the irregularities of a snowflake (Webster’s New World College Dictionary, 3rd ed.).
3Berthold Laufer, “History of the Finger-Print System,” Smithsonian Institution Annual Report (1912), reprinted in The Print 16 (March/April 2000): 1–13, http://www.scafo.org/The_Print/THE_PRINT_VOL_16_ISSUE_02.pdf (accessed February 11, 2011).
4William J. Herschel, The Origin of Finger-Printing (London, England: Oxford University Press, 1916), http://www.galton.org/fingerprints/books/herschel/herschel-1916-origins-1up.pdf (accessed February 11, 2011).
5Brian Innes, Body in Question: Exploring the Cutting Edge in Forensic Science (New York, N.Y.: Sterling, 2005), 32–33.
6G. S. Sodhi and Jasjeet Kaur, “The Forgotten Indian Pioneers of Fingerprint Science,” Current Science 88, no. 1 (2005): 185–191, http://www.ias.ac.in/currsci/jan102005/185.pdf (accessed February 11, 2011).
7“Homeland Security: Fingerprint Identification Systems,” GlobalSecurity, http://www.globalsecurity.org/security/systems/biometrics-fingerprint.htm (accessed February 11, 2011).
8“Integrated Automated Fingerprint Identification System,” Federal Bureau of Investigation, http://www.fbi.gov/about-us/cjis/fingerprints_biometrics/iafis/iafis (accessed February 11, 2011).
9“Fingerprints and Other Biometrics: Next Generation Identification,” Federal Bureau of Investigation, http://www.fbi.gov/about-us/cjis/fingerprints_biometrics/ngi/ngi2 (accessed February 11, 2011).
10Shaila K. Dewan, “Elementary, Watson: Scan a Palm, Find a Clue,” Technology, The New York Times, November 21, 2003, http://www.nytimes.com/2003/11/21/nyregion/elementary-watson-scan-a-palm-find-a-clue.html (accessed February 14, 2011).
11“Fingerprints,” CrimTrac, http://www.crimtrac.gov.au/systems_projects/Fingerprints.html (accessed February 11, 2011).
12The word recognition comes from the Latin words re (again) and cognoscere (to know) (Merriam-Webster’s Collegiate Dictionary, 11th ed.). It implies that to recognize somebody, that person must already be in one’s data bank.
13NAFIS, the Australian AFIS, maintains DNA files alongside its biometric files.
Please cite as:
Thomas J. Baker, "Biometrics for Intelligence-Led Policing: The Coming Trends," The Police Chief 78 (April 2011): 38–45.