DEEP in a freezer in a closet-sized room at the Armed Forces Institute of Pathology here are tiny vials of a clear liquid containing irreplaceable molecules. They are submicroscopic shards of one of the most deadly viruses the world has ever known, the influenza virus that swept the world in 1918.

The viral fragments come from the lungs of three victims of that epidemic: a soldier who died in South Carolina, one who died in New York and an Eskimo woman who died in a village on the northern coast of Alaska. They may be all that is left of that terrifying microbe.

But scientists treasure them for more than their historic value. If the virus strain ever emerges again, the fragments may be the crucial tools needed to save the world from a lethal epidemic.

Already scientists are analyzing the fragments, which may enable them to make a vaccine against the 1918 flu and to study whether current antiviral drugs would be effective against it, said Dr. Robert Webster, the head of the department of virology and molecular biology at St. Jude Children’s Research Hospital in Memphis. With luck, he said, scientists will learn why the 1918 flu was so deadly and where it came from. The answers will solve one of the great medical mysteries.

How the virus fragments were found is a detective story that began 50 years ago when a Swedish pathologist came to the United States and became obsessed with finding the virus that caused the 1918 epidemic, and ended last year because a young molecular biologist working for the Defense Department had discovered how to find viruses in the decayed tissue of dolphins.

The pathologist, Dr. Johan V. Hultin, now 72, an adventurer who thought nothing of working up to 18 hours a day in the Alaskan cold in the quest, and the biologist, Dr. Jeffrey K. Taubenberger, 36, whose fascination with the actual viral genes was so complete that he never took a 10-minute drive to see where the tissue specimens were stored, combined their efforts to identify the 1918 virus that killed 20 million to 40 million people. Dr. Hultin produced crucial evidence when he traveled to the remote Alaskan village to exhume the body of the Eskimo whose still-frozen lungs contained molecular footprints of the virus.

Not since the Black Death of the 14th century had the world seen an epidemic like the 1918 flu, which killed 675,000 people in this country, and 43,000 United States servicemen. About half of the deaths of United States servicemen in World War I were caused not by battle but by the flu. Young healthy people would fall ill and die within a few days.

Families were ravaged. James D. H. Reefer of Kansas City wrote to the flu researchers to tell them that when he was 4 years old and his brother was 6 that his 30-year-old father and his 27-year-old mother got the flu and died within a few days of each other, unable to breathe as the virus destroyed the air sacs of their lungs. ”Older members of the family later told me they simply drowned,” Mr. Reefer recalled.

Ever since that 1918 epidemic, medical experts have lived in dread that the virus would re-emerge and hoped that if it did, they would be able to recognize and fight it. But for that, they needed to know what made it so deadly.

Although the 1918 flu was the most dangerous in recorded history, influenza pandemics seem to sweep the world every 30 years or so. The last one was the Hong Kong flu in 1968, making the current effort particularly urgent, some say.

Dr. Hultin had been fascinated with the 1918 flu for most of his career. He came to the United States from his native Sweden in 1949 and later wrote his master’s thesis on influenza epidemics at the University of Iowa, where he studied microbiology. He was particularly horrified by the flu’s devastation of the Eskimos; at the epidemic’s height, entire villages were wiped out overnight.

In 1950, while at the university in Iowa, Dr. Hultin overheard a visiting virologist say that the only hope of understanding the 1918 flu would be to find the virus in the bodies of victims who might have been buried in permafrost, ground so cold it never thawed. Dr. Hultin saw an opportunity. The summer before, he had driven to Alaska for adventure and met a paleontologist, working with him in a search for bones of prehistoric horses. Dr. Hultin contacted the paleontologist and, through him, was put in contact with Lutheran missionaries who helped him get church records from Alaskan villages in 1918. The records, Dr. Hultin said, told him ”who died, what they died from, how fast they died, and where they were buried.”

Then he found the permafrost line on a map and found three villages that filled his criteria. In 1951, with some money from the university, he made his first trip to Alaska, looking for frozen flu victims. The first two gravesites he visited were unsuitable, he said. The ground had thawed in the 33 years since the 1918 flu. But the third site, in the village of Brevig on the Seward Peninsula, seemed ideal. Eighty people lived there in 1918, and when the flu struck, 72 died in one week. Missionaries buried them in a mass grave marked with two large wooden crosses.

In June 1951, Dr. Hultin got permission from village authorities to exhume the bodies. He began to dig, working 16 to 18 hours a day by himself, thawing the ground by building fires on it. When he had dug down six feet, he said, he found the body of a little girl.

She had ”beautiful black braided hair and a beautiful dress on,” Dr. Hultin said. He called three colleagues at Iowa State, who flew into Brevig to help. Together, they enlarged the hole Dr. Hultin had dug and removed lung tissue from five bodies. Then they closed the grave and returned to Iowa with their tissue. Dr. Hultin said that he and his colleagues wore gloves and masks but added that their protective gear ”today might be called primitive.”

”I worked a month and a half on those specimens, using all conceivable ways of trying to revive the virus.” Dr. Hultin said. ”But I failed. The virus was dead. He eventually moved to California and practiced as a pathologist, continuing to dream of tracking down the 1918 flu virus.

But it was only last year that he was able to act. Retired and living in San Francisco, Dr. Hultin read an article by Dr. Taubenberger in March in the journal Science that led him to believe that molecular biology had advanced enough so that it might be possible to get flu virus out of the people buried at Brevig.

Dr. Taubenberger, the chief of the division of molecular pathology at the Armed Forces Institute of Pathology in Washington, works in a windowless building, originally built as a bomb shelter in 1950, on the campus of the Walter Reed Medical Center. His job is to analyze tissue specimens for the Defense Department. A few years ago, he was given the difficult task of looking for viruses in severely rotted tissue taken from dead dolphins to determine why they died. Defense Department scientists suspected the dolphins had been killed by a measles-like virus rather than a red tide as had originally been suggested

Three years ago, Dr. Taubenberger discovered how to find the viruses in the decayed dolphin tissue, and that gave him an idea: he might be able to use the same technique to find the 1918 flu virus. There was a very slight chance, he realized, that traces of the virus might still be present in scraps of tissue saved from flu victims.

Ever since the Civil War, the pathology institute has been saving autopsy specimens, keeping them in ever-expanding warehouses. The collection is a Library of Congress of pathology — a record of every illness or death the military doctors have examined.

Dr. Taubenberger said he wondered if the warehouse contained lung tissue taken from soldiers who died in the 1918 epidemic and, if it did, if any of that lung tissue contained the genetic material of the flu virus that he could extract and analyze. He decided to search for the most likely victims, those who had died quickest, with the flu virus presumably abundant in their lungs.

He found 70 cases that seemed promising and requested their autopsy tissues. Within two days he got them, scraps of lung tissue soaked in formaldehyde and encased in chunks of paraffin about the size of a thumbnail and about a half inch thick.

Dr. Taubenberger and his colleague Ann Reid, took a razor blade and shaved off pieces of tissue from the paraffin blocks for molecular analysis. Within the first 30 samples they analyzed, they found one with footprints of the 1918 virus. The lung tissue was taken from Roscoe Vaughan, a 21-year-old private at Fort Jackson, S.C., who reported to sick call on Sept. 19, 1918, and died a week later.

Last year, Dr. Taubenberger and his colleagues published a report of their initial analysis of the virus in Private Vaughan’s lungs in the journal Science. And that was the article that caught the eye of Dr. Hultin.

In August, Dr. Hultin asked Dr. Taubenberger if he would like him to return to the Alaskan village and dig the gravesite again.

”I remember it was on a Monday that we had a conversation and we said we thought he should do it,” Dr. Taubenberger said. ”He said he didn’t think he could leave before Friday.”

Dr. Hultin said the reason for the three-day delay was that he had been finishing a replica of a 14th-century Norwegian log cabin in the Sierras. ”I didn’t want to tell Jeffrey that,” Dr. Hultin said. ”He might think he was dealing with a nut.” Dr. Hultin left that Friday, taking tools, camera and film. He did not take the time to seek financing from any foundation or government agency.

”I was so anxious to do the work that I wanted to remove all conceivable obstacles,” Dr. Hultin said. He said he did not want to repeat the experience of a proposed expedition to exhume Norwegian miners who died in the 1918 flu and were buried above the permafrost line. That expedition has been in the planning stages for more than three years and involves an international team.

Dr. Hultin arrived at Brevig last August, spoke to the village officials and got permission to dig, slept on an air mattress on the floor of the local school, and accepted the offer of four young men in the village to help him dig. On the third day, at 3 P.M., Dr. Hultin said, they came upon the first body, six feet down. But the tissue had rotted, which Dr. Hultin said he suspected was a consequences of his digging in 1951. The next day, at a depth of seven feet, they came upon four bodies with soft tissue left on them. One, the body of an obese woman, was particularly well preserved.

”I sat on this pail and saw this woman.” Dr. Hultin said. ”She was in a state of good preservation, her lungs were good. And I knew that this is where the virus would be.”

Dr. Hultin removed her lungs and also took tissue from the lungs of the three other victims, even though he doubted that their decaying tissue would contain virus particles. He put the tissues in a preservative and closed the grave. To make sure the tissues reached Dr. Taubenberger, Dr. Hultin made four packages, each containing tissue from all four victims. He mailed them on subsequent days using Federal Express, United Parcel Service and the Postal Service.

Dr. Hultin said he was not worried that he might be sending tissue containing live viruses. ”I tried to grow the virus under very good laboratory conditions in 1951 and it was not possible,” he said. ”The virus was dead. So I was not concerned.”

Dr. Hultin estimated that the entire trip, including paying his helpers, had cost him $3,200.

In the meantime, Dr. Taubenberger and his colleagues had found fragments of the flu virus in another of the pathology specimens that they had ordered from the warehouse. This was from a 30-year-old man, Pvt. James Downs, who died of the flu in Camp Upton, N.Y., on Sept. 26, 1918, three days after he became ill.

Now, Dr. Taubenberger and his colleagues have virtually completed analyzing 1 of the virus’s 10 genes, the hemagglutinin gene. The virus uses the gene to make a protein that allows it to stick to the surface of lung cells, Dr. Taubenberger said. One hypothesis, based on what happens with some flu viruses that strike birds, is that the 1918 flu became deadly because a particular area of the hemagglutinin gene, the cleavage site, was altered. But Dr. Taubenberger saw no such alteration.

”Our goal is to get the sequence of all of the genes,” Dr. Taubenberger said.

He says he thinks there will be evidence in the genes that will show why the virus was deadly. And he also hopes that by comparing the genetic sequence of the 1918 flu virus to other viruses that infect birds and pigs, scientists will be able to discover whether the virus came from birds, pigs or both. So far, he said, the virus does not look like a bird virus.

Other flu experts say they are delighted by Dr. Taubenberger’s work. Having three viruses from people who died in three different places strengthens the conviction that the viral genes are indeed those of the 1918 flu and not just an innocuous bystander virus that happened to be in the lungs of the first victim, said Dr. Edward D. Kilbourne, a flu researcher at New York Medical College in Valhalla.

The work, Dr. Kilbourne said, is ”a great combination of good epidemiology and very good molecular biology.”

Dr. Nancy Cox, chief of the influenza branch at the Centers for Disease Control and Prevention in Atlanta, said the work was ”extremely interesting and quite important.” And she said, ”We’re all delighted that he’s been able to get three viruses.”