Harvard University researches report that four samples of mothers' milk tested in two states contain traces of a super-toxic industrial chemical - a poison the Environmental Protection Agency describes as "perhaps the most toxic small molecule known to man."

The chemical is called dioxin. It's a manufacturing impurity that often contaminates certain industrial products, including the widely used herbicide 2,4,5-T. The herbicide has been used in this country since 1948 on pasturelands and in national forests, along highways and power line rights-of-way, and on rice crops.

Scientists have worried about the potential threat of dioxin since the early 1970s, when laboratory studies showed a single dose of the chemical in the several hundred parts per trillion range kills animals. This is the smallest fatal dose of any chemical ever tested.

Dr. Matthew Meselson, chairman of the department of biochemistry and molecular biology at Harvard, said that if confirmed by further testing the new findings by further testing the new findings would provide the first concrete evidence that dioxin from 2,4,5-T and possibly other sources may be accumulating in human tissues. If researches repeat Harvard's findings in a larger sample of women across the country, Meselson said "that would be cause for concern."

Meselson and other Harvard researches tested the breast milk from 18 women living in Missouri, Texas and Oregon, near areas where 2,4,5-T is routinely sprayed.

According to a new chemical analytic technique developed at Harvard, the breast milk from four women in Texas and Oregon appeared to contain dioxin at levels from 0.6 to 1.6 parts per trillion (ppt). The levels translate to 10 to 40 ppt of dioxin in the milkfat, the researches said.

Three of the breast milk samples, from Texas, were collected by La Leche League, a national group that advocates breast feeding. The Oregon sample was procided by a member of Citizens Against Toxic Sprays. CATS is one of numerous environmental groups across the nation that have sued the U.S. Forest Service in an effort to half 2,4,5-T spraying in national forests.

Five more women were tested in Boston, where the herbicide is not widely used. They did not have dioxin in their breast milk. Meselson said the Harvard findings have not been confirmed by other laboratories - researches used up all the milk samples - and he characterized them as "preliminary."

But the Harvard findings may add weight to previous research suggesting that dioxin from 2,4,5-T may contaminate part of the nation's food supply.

Researches at Harvard and Dow Chemical Co., of Midland, Mich., the major manufacturer of 2,4,5-T, reported last year they detected dioxin at levels up to 69 ppt in beef fat sliced from cattle which had grazed on pastures sprayed with 2,4,5-T.

"If there were levels of dioxin of even one ppt in the human diet, and if the material were highly accumulative, as monkey tests indicate," Meselson said in a recent phone interview, "then that would put us over a period of years in the region of having a body burden that would be lethal for a guinea pig."

"We don not know whether 10 or 40 ppt of dioxin in mothers' milkfat will be injurious to human babies," said Dr. James R. Allen, a researcher at the University of Wisconsin Medical School who is studying the effects of dioxin on monkeys. But he said the data have "some very serious implications."

Thomas Hollaway, an official in EPA's Office of Pesticides, says the agency will "reserve opinion" on Harvard's findings until they have been "confirmed by an outside laboratory."

William B. Seward, a spokesman for Dow, said, "We don't believe Harvard has a reliable method [for detecting dioxin] when dealing with levels below 10 ppt. We've analyzed mothers' milk, and we've never found any."

The Harvard mothers' milk findings open the latest round in the eight-year battle between environmentalists and the federal government over the herbicide 2,4,5-T. The herbicide first became the target of widespread public protests when U.S. warplanes sprayed it in massive quantities to defoliate Vietnam forests during the Indochina war.

When a government-sponsored study indicated in 1969 that 2,4,5-T caused birth defects in laboratory mice, the Pentagon bowed to public pressure and banned the herbicide from its Vietnam arsenal.

Later studies showed the chief culprit in causing the birth deformities and other toxic effects was the dioxin in 2,4,5-T. Soon after EPA was formed in 1971, the agency announced it would hold hearings to consider bannings all uses of 2,4,5-T from the market.

But those hearings were canceled and have never been held. EPA officials contend they can't ban the herbicide until - and unless - they gather conslusive evidence that dioxin actually threatens the public health by "bioaccumulating in man and the human food chain." For the past four years, researchers at EPA, Harvard, Dow and other research institutions have conducted a special "dioxin monitoring program" to find whether dioxin contaminates food and human tissues.

But their search has bogged down in a complex and frustrating technological battle. For even as researchers search for dioxin the environment, they are still trying to perfect the analytic techniques capable of measuring dioxin in the parts per trillion range - a "pioneering effort," according to EPA documents.

Since 1974, researches have reportedly detected dioxin residues in birds and fish in Oregon's Siuslaw National Forest, in beef fat and now mothers' milk. But scientists at EPQ, Dow and Harvard have consistently disagreed just how many parts per trillion of dioxon they're finding in just how many samples.

Hollaway said the data "certainly indicate from a qualitative standpoint that there is dioxin" in beef fat. And he says "I think Dr. meselson's studies are indicative of dioxin residues in humans." But until Harvard's methods are "recognized by scientists in the field as being a positive and valid method," the EPA won't know the precise levels of dioxin, Hoolaway said.

Without that information, the EPA won't be able to determine for sure whether dioxin in the environment threatens the public's health, he said. Hollaway said from the data he's seen so far, "we don't feel there is a significant risk of exposure to humans" from dioxin.

Some scientists and environmentalists reject EPA's view. For one thing, as Environmental Defense Fund general counsel William Butler says, "No scientist has ever been abel to define a 'no-effect' level for a chemical which causes birth defects."

Scientists such as Meselson warn that studies at the University of Wisconsin suggest primates "are rather surprisingly sensitive [to dioxin] and that it is highly accumulative in their bodies."

While the dioxin debate has focused on the herbicide 2,4,5-T, an EPA report warns that dioxin may contaminate other herbicides and numerous chemicals used in the paper, pulp, leather tanning, textile and other industries.