Biometrics before machines
Biometrics means identifying people by measurable physical or behavioral traits. Long before any electronic device existed, people used the idea informally: recognizing a face, a voice or a handwriting style. The scientific version came in the nineteenth century, when police and governments wanted a dependable way to tell whether a person had been in custody before.
One early system was anthropometry, associated with the French police clerk Alphonse Bertillon, which relied on a set of body measurements and photographs. It was laborious and depended on careful, consistent measuring. Fingerprints offered a simpler alternative.
Fingerprints become a record system
Several people are involved in the history of fingerprint identification, and the credit is shared. In the late 1800s, Henry Faulds, a Scottish physician working in Japan, and William Herschel, a British administrator in India, each wrote about the use of fingerprints for identification. Francis Galton, an English scientist, studied them statistically and published a book on the subject in 1892, arguing that patterns persist through life and can be classified. In Argentina, Juan Vucetich developed his own classification system around the same time. Edward Henry, working in India and later in London, helped create a filing system that Scotland Yard adopted in 1901.
The common thread is that these were paper methods. A person pressed an inked finger onto a card, and trained examiners compared cards by eye. In the United States, the federal government began collecting fingerprint cards in a central bureau in the 1920s, and by the middle of the century the files were enormous. Searching them by hand was slow, which created demand for automation.
Computers meet fingerprints
The first major step toward machine reading was not a scanner at a door but a computer system for matching. From the 1960s, researchers, including those at law enforcement agencies and at the U.S. National Bureau of Standards, worked on ways to convert a fingerprint into a set of features that a computer could compare. The usual features are the places where ridges end or split, often called minutiae. Comparing a set of those points is much faster than comparing a whole image.
These early programs were built for police files, and they ran on large machines. Automated fingerprint identification systems entered use in the 1970s and 1980s, and the idea that a computer could shortlist candidates for a human examiner proved reliable enough to spread to many forces around the world.
The first electronic readers
A separate line of work asked whether a device could check one person’s finger against one stored record on the spot. That is a different task from searching a national file: it is verification rather than identification, and it needed a compact reader.
Reports from the 1960s and 1970s describe early commercial systems for access control, including at least one hand-geometry reader that measured the shape and size of a hand to decide whether to open a door. Hand-geometry devices were used in workplaces and some secure facilities. Early fingerprint readers followed in the 1980s, and their first customers were mostly institutions that could justify the cost: government agencies, banks and sites with sensitive areas.
These devices typically used one of a few methods:
- Optical sensors shine light on a finger resting on glass and capture the pattern of ridges as an image.
- Capacitive sensors use a small array of electrodes to detect the difference between ridges that touch the surface and valleys that do not. They rely on integrated circuits and became practical as chip manufacturing improved.
- Ultrasonic and thermal sensors appeared later, offering different trade-offs in cost and reliability.
Reaching ordinary people
Through the 1990s and 2000s, sensors became smaller and cheaper. Fingerprint readers appeared on laptops, door locks and time clocks. Governments also began introducing biometric passports, which include a chip that stores identifying data, with the face and sometimes fingerprints used for checks at borders.
The next change was in smartphones. Phone makers added a fingerprint sensor, usually to unlock the device or approve purchases, and this put a biometric scanner in the pockets of a very large number of people. Later phones increasingly turned to face recognition, which relies on smartphone cameras and software instead of a contact sensor, showing how the field has broadened beyond fingerprints to irises, faces and voices.
Limits and debates
Biometric systems are never perfect. A reader makes a statistical comparison and can falsely reject a legitimate user, or falsely accept someone else; designers tune this trade-off. Dry, worn or damaged fingers can be hard to read, and a sensor can sometimes be fooled by a copy of a fingerprint, which led to work on liveness detection, meaning checks that the finger is real.
There are also privacy and security questions. A password can be changed after a breach, but a fingerprint cannot, which is one reason that biometric data is generally stored in protected forms rather than as raw images. Debates about how and when governments and employers should collect biometric data are ongoing and differ from country to country. For readers interested in the security side, the broader field of cybersecurity has its own history.
A short summary
The development can be summarized in three stages. First came paper records and human examiners in the late nineteenth and early twentieth centuries. Second came computer matching of police files in the second half of the twentieth century. Third came compact, cheap electronic readers that checked one person at a time, and these spread from specialist buildings to consumer products. Each stage depended on the one before it, and each stage involved many contributors rather than a single inventor.