(FORTUNE Magazine) – It was personnel records that finally stopped the Mad Bomber of the Forties and Fifties, George Metesky. A longtime employee of Consolidated Edison, Metesky planted 32 bombs in New York City, wounding 22 people, and eluded authorities for 16 years. Then the Journal-American newspaper drew him into a fateful correspondence, ostensibly offering him a forum to air his grievances. Metesky wrote scores of letters to the paper, mostly concerned with Con Ed and what he described as its "dastardly deeds." He signed the letters "F.P.," which he later said stood for Fair Play. A search of the company's records turned up Metesky's file; it included angry letters he had written to superiors, using the phrase "dastardly deeds."

The serial bomber of the moment, the Unabomber, signs his letters "F.C.," which he says stands for Freedom Club (no one knows whether this is some kind of homage). He has entered into a correspondence with the New York Times and the Washington Post, demanding that a manuscript laying out his grievances be published. The FBI is combing the Unabomber's 35,000-word text, looking for telltale turns of phrase. Somewhere out there, the FBI thinks, is his personnel file, more than likely containing an angry letter or two.

"We feel that the Unabomber has a job, and probably has come to the attention of management," says Jim Freeman, the FBI's special agent in charge of the Unabomber task force. "He is quite specific in his targeting, say, of geneticists doing computer-assisted gene modification, or of psychologists engaged in behavior modification. He may have harped on those same themes to supervisors or co-workers. Maybe sometime he stated his opinions under his true name, in writing." The bomber's rasuma may also hold clues. Freeman says it probably shows a pattern, starting in the late 1970s, of movement from the Midwest to Utah to Northern California.

The FBI doesn't speculate publicly about the Unabomber's personality, in part because it wants to avoid planting preconceptions that could discourage potential tips. But Freeman spoke at length with FORTUNE about the manhunt, in hopes that the bomber has tipped his hand in the workplace through a memo, a fit of pique, a careless remark.

What kind of office mate might the Unabomber be? Some of his characteristics are clear. He has a mocking cast of mind: He has obviously chosen some of his return addresses for ironic effect. He once used as a return address the offices of Calgene, developers of a genetically altered tomato. He recently sent a letter to the San Francisco Chronicle with the return address "Frederick Benjamin Isaac Wood, 549 Wood Street, Woodlake, Ca. 93286"--which seems to poke fun simultaneously at the FBI (initials) and his own obsession with wood, which he uses in making his bombs.

He's ostentatiously literary and indulges in recherche puns of the sort only a graduate student would get, let alone find amusing. Since "wood" in medieval English means "mad" or "crazy," New York journalist William Monahan has gone so far as to argue that the bomber's use of wood in his bombs may itself be a complex joke. It may also be another form of homage to Mad Bomber Metesky.

Where news media are concerned, the Unabomber's taste is conservative. He prefers print over TV, substance over entertainment. He'd rather do business with the Times or the Post than with the networks. The information he uses in selecting targets seems to be print-based also, since it is sometimes out of date in a way that an online source probably would not be. The address on a bomb he sent to the California Forestry Association, for example, failed to reflect the fact that the organization had changed its name five years earlier. The president he listed on the label hadn't been president for a year. The Unabomber apparently has easy access not just to Eastern papers and standard reference works such as Who's Who, but also to a wide variety of scientific journals as well.

He is highly creative--a perverted artisan who shapes, polishes, and tests every facet of his handmade devices before taking them public. He eschews ready-made hardware, fabricating even screws from scratch. He has devoted 40 hours, by one estimate, to making a switch he easily could have bought for 89 cents.

An individual whose life he has threatened and who, not surprisingly, spends a lot of time trying to understand the man who wants to kill him, muses: "Oh, this guy's intelligent. I mean, anybody who does things well has to have some appreciation for him. There's nothing I like better than to hold in my hands something well-crafted in wood. He's skilled at what he does. He's, you know, perfected it. He has a style. He has an identity. He's imprinted his identity on everything he's doing. I mean, there's a lot of things about this guy that tell you there's a greater concept there than just some kook." Of course, the Unabomber has much in common with other serial bombers: he is a squeamish coward who murders unsuspecting people at a comfortable remove from his tidy, hobbyist world.

Right now the Unabomber appears to have entered a new phase. He has graduated from crude mixtures of gunpowder and match heads detonated by rubber bands to electrically triggered high explosives whose shock wave travels at 15,000 feet per second. Every day, more people learn of him and fear him. He has set a deadline: If his manuscript isn't published by September 28, he will resume killing. The assassin dreams of a pastoral utopia, and the universe of his potential targets includes all those who stand in the way: city dwellers, technocrats, financiers, and others who derive pleasure and profit from meddling with the natural order of things. As he put it in a recent letter to the New York Times, describing his motive for placing a bomb aboard an airliner in 1979 (one of his few duds): "The idea was to kill a lot of business people."

Fittingly, in all likelihood it will be software that catches the Luddite. The hunt for the Unabomber is by now the biggest, longest, most complex, and most costly in FBI history. At its heart is a network of computers, including a massively parallel processor, sifting lists. "These lists," says the FBI's Freeman, "were obtained by grand jury subpoena and allow us to compare big bites of data on everything from the geography of the case--where the Unabomber may have lived during various years--to which universities have been targeted." Most recently, agents have been studying the bomber's manifesto. The FBI won't confirm it, but sources say agents are trying to identify books from which he might have culled some of his ideas or word patterns. Library records can then be examined to see who, within certain dates, might have used those books.

Freeman believes the computer now contains enough information to catch the Unabomber; at this point, it's really a matter of collation. Each new datum the authorities gather is fed into the machine, adding grist, but the ghostly outline of the bomber is already in there, somewhere.

Still, the chase has already lasted 17 years. One reason is that the Unabomber is a do-it-yourselfer. Usually the only store-bought component in his bombs is the battery. Homemade parts are hard to trace.

Another problem with the case has been managerial. Until recently, the manhunt was split among three agencies: the FBI, the U.S. Postal Inspection Service, and the Bureau of Alcohol, Tobacco, and Firearms. Fragmentation meant three collections of evidence in three different labs; three sets of notes from three sets of interviews with witnesses and victims; three different manuals dictating procedure. The agencies cooperated, but no investigator knew everything his counterparts knew. Clues could go unnoticed. Facts could fall through cracks. At best, triplication was inefficient.

For example, in 1985 the Unabomber mailed a bomb to Boeing; it was spotted and disarmed. Later that year a booby trap disguised as wooden construction debris was left in the parking lot of a Sacramento computer store. The store's owner, Hugh Scrutton, leaving at lunchtime, picked it up to toss it into a dumpster and died on the spot, his right hand nearly severed and a hole blown in his chest. Pieces of the bomb were found 150 yards away. It was the Unabomber's first killing.

The Boeing bomb, because it had been mailed, became the Postal Inspector's responsibility. The other became the FBI's. Dennis Hagburg, the Postal Inspection Service's inspector in charge for San Francisco, thinks the two agencies may have interviewed some of the same individuals without knowing it.

The FBI says it's possible that it may have interviewed the Unabomber himself two years later. Perhaps emboldened by the murderous success of the Sacramento booby trap, the bomber in 1987 tried another, also made to look like construction debris. It was left outside a computer store in Salt Lake City. This time, however, the intended victim kicked the device out of his way and suffered only minor injuries. A woman had seen the bomber place the trap. The FBI interviewed scores of suspects but could establish the guilt of none. After that the Unabomber remained silent for six years. Had he been among the suspects interviewed? Had near-discovery scared him into hiding?

If so, he more than compensated for that setback with a dramatic return in 1993. At 4:30 on a June afternoon, sitting at the kitchen table of his house in the Marin County hills above San Francisco Bay, Dr. Charles Epstein, a distinguished geneticist at the University of California at San Francisco, sat opening his mail. He pulled the tab on a padded envelope measuring 8 by 11 by 3 inches.

Neighbors brought running by the blast found the kitchen a shambles, its windows shattered, the heavy wooden tabletop (six feet across and three inches thick) blown from its pedestal. Epstein, dazed and bleeding, had staggered to the street, where a carpet-cleaning crew working next door saw him, wrapped him in towels, and called 911. He survived, with damaged hearing, abdominal wounds, and the loss of three fingers.

Not two days later, at Yale University in Connecticut, a bomb inflicted similar injuries on historian and computer science professor David Gelernter. The Unabomber's mutilating touch had spanned the continent.

The attacks brought together senior officials of the FBI, Postal Inspection Service, and ATF. No longer, they agreed, could the government's efforts be fragmented. The agencies would pool money, manpower, and resources to create a single task force, with the FBI as lead agency. "That decision," says Freeman, "made a tremendous difference." He was transferred to San Francisco from the Bureau's Honolulu office, where he had worked on white-collar crime and drug-trafficking cases, and took up the hunt for the bomber.

Freeman, born in Clarksville, Arkansas, and raised in New Mexico, is a matter-of-fact man with a composed manner. He expresses emphasis only by gesture, poking the table with his finger, as if pressing a doorbell, to make a point. His office would gladden the heart of a GAO inspector. It is furnished in straight Bureau issue and reveals little of his personal life, with one or two exceptions (a ceremonial U.S. flag from his tenure in Hawaii; a picture of his grandson fishing in New Mexico).

When he took over the case in summer 1993, Freeman inherited a mountain of evidence accumulated by the agencies but had no good way to exploit it. Louis Freeh, the FBI's director, had told Freeman he had a blank check. Freeman cashed it: "We obtained computer support from the Department of Defense--a massively parallel processor, which is something we don't use ordinarily. It's for the comparison of huge amounts of data. This was the first time, to my knowledge, that one had been used in a criminal investigation."

A small problem arose. Not only were the agencies' existing computer databases different, there were three competing sets of good, old-fashioned forms and carbon copies stuffed into several groves of filing cabinets. "We tried to use that massive parallel processing early on, and it was ... it was not really possible to compare one set of data against another," says Freeman.

One set of lists would have people's names and dates of birth but not social security numbers. Another would have social security numbers but not dates of birth. Says Freeman: "You have to know if you're comparing John Doe One with John Doe Two--there has to be some common identifier so you know you're dealing with the same suspect." A daunting solution suggested itself: reinvestigate the entire case--14 crime scenes spread over 15 years and eight states. So they did that. "It took a year to get that data updated," sighs Freeman. "We sent task force agents out to revisit each crime scene, obtain photographs or videotape the area, reinterview every witness, reinterview the surviving victims. We devised a protocol of about 200 questions to ask each of them, so that all data could be put in a database and compared across the gamut for all victims."

Freeman's task force set up a telephone hotline for tips (800-701-BOMB) and is offering a reward of $1 million for information leading to the bomber's capture, half the money coming from the FBI, half from business groups that feel their members might be targets. So far, more than 15,000 tips have poured in.

Once 15 years of data had been reconciled, the task force found it knew more about the Unabomber than it had thought. There was, for example, such clear consistency among the devices--by now numbering 16--that forensic experts were finally positive the same person had made all of them. A definite geographic progression of crime scenes emerged: Activity starts in Illinois in the late 1970s, then in 1981 shifts west to Utah and California, where the bomber's focus has been Salt Lake, Sacramento, and the Bay Area.

While all this was going on, a little girl named Polly Klaas was kidnapped in Petaluma, north of San Francisco, within the jurisdiction of Freeman's office. Before she was found dead 60 days later, her case had influenced the hunt for the Unabomber in two ways.

The first was cosmetic. Only one witness--the woman in Salt Lake City--had ever seen the Unabomber, and Freeman discovered, in re-reporting the case, that she had never been entirely happy with the image drawn from her description. Freeman asked a new artist, the one he was working with on the Klaas case, to sit down with the woman and produce a second drawing. The witness said the new drawing looked much more like the person she'd seen: a white male then in his 40s, weighing about 160 pounds, 5-foot-10, with a ruddy complexion, a thin mustache, and reddish or sun-bleached blond hair.

The Klaas case also taught the task force how to use a relatively new management tool called Rapid Start, special computer software and hardware developed by the FBI after the World Trade Center bombing to keep track of huge amounts of crime-scene data. "Polly Klaas," says Freeman, "was a case that involved a lot of data, a lot of interviews being done on a daily basis. It boosted our learning curve with Rapid Start. We went back and revamped it big time for the Unabomber case."

The system allowed full-text retrieval of forensic data and information from witnesses. When new victims emerge as the result of additional bombings, the same 200-question protocol is applied to them. New forensic evidence is consolidated in a single task force lab in Washington, where it is equally accessible to FBI, ATF, and Post Office experts.

When the FBI discovered the Unabomber's cryptic message, "Call Nathan R., Wed. 7 p.m.," agents took up the challenge of identifying and examining 10,000 Nathan R.'s nationwide. When the bomber, in an aside to one of his letters to the New York Times, said he had grown weary of having to test bombs in the Sierra, agents, with the aid of Forest Service personnel, combed canyons and ravines. Stamps recovered from the bomber's packages have reportedly been checked for traces of saliva, which would reveal his DNA.

Investigators have stepped up the pace in recent months. Agents have fanned out to interview dealers of scrap metal from whom the bomber might have bought magnesium, lead-coated cable, or other parts of his bombs, including metal shavings sometimes used as shrapnel. The owner of Alco Iron & Metals Co.--a scrap yard in San Leandro, California--was asked, for example, which employees might have come from the Chicago or Salt Lake City areas.

Agents also have descended on--of all people--college art teachers, since it's possible the bomber learned metal fabrication not in private industry or from the military, but in art class. In the Bay Area alone, agents have paid visits to a long list of local colleges: Dominican, Sonoma, the College of Marin, U.C. at Berkeley, and Stanford.

At Dominican College in San Rafael, Sister Aquinas Nimitz, vice president for institutional research (and daughter of Admiral Chester Nimitz, commander of U.S. Naval forces in the Pacific during World War II), says agents asked her if Dominican had ever employed "anybody in the laboratory area or in maintenance who looked like the sketch." She said it hadn't. Agents also checked possible links between the bomber and the college's metal-art workshop.

Considered as an artisan, the Unabomber belongs to a rare species: those who produce works they know will be destroyed. But he is surely not performing for his victims. The Unabomber is playing to the only audience who can appreciate the nuances of his work: bomb experts.

Agent Freeman expects to lower the curtain. A lucky break, a careless remark overheard by a co-worker at the Unabomber's day job, could bring the case to a speedy end. Even if that doesn't happen, Freeman believes the task force can identify the bomber from information already stored in its databases. If tomorrow the Unabomber were to go down to his basement with 365 cans of Dinty Moore stew and stay there for as many days, Freeman thinks the task force would still nail him. How soon? Freeman answers slowly: "I think of it in months. At the most, a very small number of years."