PICK A CHEMICAL, any chemical. Feed it in whopping quantities to rats, mice or both for a year or two. By then, if they haven't already died, most will have cancer. Right?

Wrong. Despite the popular belief that anything in sufficient amounts will cause cancer in laboratory animals, most compounds actually have been found innocent in this regard. Indeed, of some 7,000 that have been tested (some admittedly more adequately than others), all but about 500 have gotten a clean bill of health.

A finding that about 7 percent of the compounds cause cancer in animals does not exactly support those who think "everything seems to cause cancer" or those who would like to diminish our relianc on animal tests. That would, in fact, be a grave error.

As Dr. Marvin Schneiderman of the National Cancer Institute puts it, "Some historical perspective may be in order here. During the 19th century when germs were being discovered thick and fast, much as carcinogens are today, the same sort of skepticism prevailed as this or that microbe was reported as the cause of one or another disease. And there were the same kinds of arguments about whether, in the face of scientific uncertainties, it would pay to clean up the environment.

"Yet when the public water supplies of northern Europe were, in fact, cleaned up - and remember this was long before the advent of antibiotics - epidemics of cholera and other gastrrointestinal infections virtually disappeared."

If popular or highly profitable substances are found to be cancer culprits, of course, there is bound to be strong resistance to the findings, and researchers have not gone out of their way, as some seem to think, to pick on them.

When saccharin began to come under increasing suspicion, for example, some scientists thought orthotoluene sulphate, a frequent contaminant of the artificial sweetener, might be the villain, rather than saccharin itself. But it didn't work out that way. Animals fed large amounts of saccharin-free orthotoluene sulphate developed cancer no more often than similar, undosed control animals. Pure saccharin, by contrast, produced an excess of bladder cancer in male animals, particularly when their exposure to the sugar substitute began - via their mothers' diet - during fetal life.

But doesn't sugar itself cause cancer in laboratory animals if enough of it is given? Sure - but only if it is injected under the skin, which is not exactly how people ordinarily take sugar. That is why animal tests for cancer have to be physiologically appropriate.

Bad news in a rush

As Dr. Schneiderman suggests, the testing of chemicals for their cancer-causing potential is so recent an enterprise that the bad news is bound to come in a rush. There are now about 63,000 untested compounds in commerce, with 700 to 1,000 new ones introduced each year. But as the backlog diminishes and there is more pre-market testing, the pace of discovery and unpleasant surprises is likely to slow.

Meanwhile, there are two other reasons why "everything seems to cause cancer" (neither of them traceable, as one wag has put it, "to the failure of science to breed healthier laboratory rats.")

One lies with the media process, which naturally focuses on holes in the dike rather than reporting how well the rest of the dike is doing. On a recent Tuesday, for example, all three division television networks reported on their evening news shows that both reserpine, a drug used in the treatment of high blood pressure, and methapyrilene, an ingredient of, among other things, many non-prescription sleep-aids, had been found to be animal carcinogens. They did not report, of course, that malathion, an insecticide dear to the hearts of home gardeners, had at the same time come through the test with its reputation for safety unscathed.

The other reason stems from the fact that testting chemicals for cancer in animals is enormously expensive. Typically, it costs $450,000 to $5008000 per compound, up from $150,000 to $200,000 as recently as 1975 - and that only if rats and mice serve as the test animals. If larger species more closely related to man are enlisted, the price rises substantially because it costs more to house and feed them and they have longer life expectancies.

Given all this, it would be profligate to test chemicals at random. Instead, it makes sense to concentrate on compounds with suspicious molecular structures, such as the chlorinated hydrocarbon family to which dozens of solvents, pesticides, drugs, anesthetics and other products belong. It is also wise to focus on those produced in large quantities (although chemicals produced in less volume to which certain groups of people are extensively exposed are generally included, too.)

Even at that, educated guesswork is not always correct. A study performed for the National Cancer Institute by Litton Industries' Bionetics Laboratories of 120 common herbicides, pesticides and fungicides is illustrative.

Each of the 120 compounds was fed to mice from infancy through age 18 months in the largest doses that would not quickly will the animals outright, and all of the 120 could have been expected to be carcinogens. Yet only 11 of them - fewer than 10 per cent - resulted in a "significant" number of tumors - more cancers than occur by chance alone in treated animals than in undosed controls.

The cumulative effect

Nonetheless, many people see no reason why animal tests should have any relevance to humans when such heavy doses of chemicals are used. Again, the explanations are not as farfetched as is popularly supposed.

One is that people, unlike laboratory animals, are exposed to more than one carcinogen in their everyday lives and that - through the interactions of body chemistry - exposure to a variety of carcinogens can be cumulative. In fact, some agents "cooperate" to produce a greater harmful effect than either can produce alone. Both smoking and asbestos, for instance, can cause lung cancer independently. But the rate of lung cancer among smokers who are also exposed to asbestos is higher than mere double jeopardy would suggest.

Accordingly: 1) larger doses than man encounters in the environment are justified in the laboratory when the aim is to discover whether a particular substance is capable of causing cancer at all and 2) no dose of a carcinogen, no matter how small, can be relied on to be safe for every individual. Contrary to what much of industry would have people believe, the concept of a "threshold" or "no effects" doses has no practical validity.

Worth bearing in mind, too, is that, because of the costs, only a few animals - generally no more than 100 - can be used to study any one compound to which millions of people of varying sensitivity are exposed. An example having nothing directly to do with cancer may make the principle clear.

When trials of swine flu vaccine were conducted in 5,000 volunteers, there were no cases of Guillain-Barre paralysis. But when 42 million people later were inoculated, more than 200 became paralyzed. By the same token, each animal in these cancer experiments has to present just a fraction of a person, making it essential to use large doses if there is to be a prayer of detecting a compound's relatively rare effects.

"Relatively rare," however, is a term to be used with care. Even what scientists call a "weak" carcinogen can lead to many thousands of cancers if hundreds of millions of people are exposed. And the earlier the exposure begins and the longer it continues the greater will be the risk, whether the doses are large or small.

More fundamentally, it should be remembered that although rats and mice are not little people, they are like people in being collections of mammalian cells organized into tissues that undergo the same biological processes. By giving large doses of a chemical to a small creature with a rapid metabolic rate, a short life span and comparatively few cells, one can get a reasonable approximation of what will happen in a larger creature whose metabolic rate is slower and who has many times more cells and a longer life expectancy.

Of the 26 chemicals from aflatoxins (by-products of food molds), asbestos adn benzene to vinyl chloride gas (used in the manufacture of polyvinychloride plastics) tht are proven causes of human cancer, all - with the possible exception of arsenic, which has yet to be thoroughly tested - also causes cancer in animals. Is the reverse true? The National Toxicology Program, for instance, has so far completed tests of 194 compounds and identified 92 of them as animal carcinogens and 19 as borderline - this because scientists could not be 95 per cent or more confident that their findings were not just coincidence. Are all these compounds also cancer risks for man?

While it would be nice to have a clear-cut answer, that is currently impossible. An occasional chemical such as DES (diethystilbestrol) produces a type of cancer - in this case a cancer of the vagina in young women - which otherwise is almost unheard of and so enables scientists to track down its source (that proved to be DES use by expectant mothers to prevent threatened abortion). But the cancers that furnish such clues are few.

The more usual state of affairs is for a carcinogen to increase the rate of cancers that, for any number of reasons, commonly occur anyway and to take 10 to 30 years to do so besides. Thus it is well nigh impossible to roll back the clock those decades to trace an increase retrospectively to its origins.

Risk to humans

As if all of this weren't sufficiently complex, no one knows how to determine from animal tests the degree of risk to humans. For example, it was found after the thalidomide tragedy that humans are 60 times as sensitive to the baby-deforming tranquilizer as mice, 100 times as sensitive as rats and 700 times as sensitive as hamsters. But there is now no way to reverse the order of such calculations so that quantitative risk assessments can be made in advance. It simply has to be assumed for the sake of prudence that humans are at least as vulnerable to carcinogens as the rats and mice that serve as their surrogates.

In fact, both the causes of cancer and the frequency of certain kinds of cancer are subject to shifts that probably reflect previous changes in custom and technology. The widespread introduction of refrigeration and the consequent reduction in food spoilage during the World War I era, of instance, may well have been responsible for the decline in stomach cancer which began about 1930; at least so the evidence suggests.

In the same way, there are those who think that because cigarettes are not as strong as they used to be, something similar may be in the offing where lung cancer is concerned. If the scenario is correct, there will not only be less lung cancer attributable to cigarettes a decade or so down the road, but also more caused by such other influences as air pollution and occupational exposure to carcinogens.

No matter. What can be said with confidence is that animal tests have more often proved prophetic than not. The vinyl chloride gas already mentioned is a case in point. Granted, the tumors that were produced by the gas in animals before it was known to cause a rare form of liver cancer in humans occurred in the symbol glands of hamsters, facial structures that man does not have. Nonetheless, the premonitory evidence was there, confirming the principle that once a carcinogen circulates in the bloodstream it is free to attack whatever organ or organss may be most vulnerable.

Similarly, tests conducted on several species beginning in the 1930s accurately predicted that the chronic estrogen replacement therapy subsequently prescribed for millions of women to ease them through the menapause and retard the aging process would significantly increase their risk of getting uterine or breast cancer.

An equally dramatic example is provided by the chlorinated hydrocarbon Kepone. It is a decade or more too soon to know whether the employes of the now-defunct Life Sciences plant in Hopewell, Va., who were heavily exposed to this pesticide in the course of its manufacture will have more than their share of cancer, as rodent studies suggest. But it should have come as no surprise that many of them would be subject to tremors, other nervous symptoms and sterility as a consequence of that experience. Experiments conducted on quail in 1964 resulted in exactly those injuries.

When it comes right down to it, then, animal tests are meant to function as an early warning system, and their biggest drawback is that people fail to take them seriously. Given the fact that at least 70 percent of cancer is thought to come from exposures 10, 30 or even more years earlier, the really pertinent question is wht this is doing to our future selves. With one in five Americans dying of cancer, often prematurely, the public's scorn of animal tests should be seen for what it is: playing with fire.

Following are the 26 chemicals or industrial processes identified by the International Agency for Research on Cancer as associated with, or strongly suspected to be associated with, the occurence of cancer of humans:

Aflatoxins

Aminobiphenyl

Arsenic compounds

Asbestos

Auramine (manufacture of)

Benzene

Benzidine

Bis(chloromethyl)ehter

Cadmium-using industries (possibly cadmium oxide)

Chloramphenicol

Chloromethyl methyl ether

Chromium (chromate-producing industries)

Cyclophosphamide

Diethylstilbestrol

Hematite mining

Isopropyl oils

Melphalan

Mustard gas

2-Napthylamine

Nickel (nickel refining)

N,N-Bis(2 chloroethyl)-2-Naphthy-lamine

Oxymetholone

Phenacetin

Phenytoin

Soots, tars and oils

Vinyl chloride CAPTION: Illustration, By James K. Edwards - The Washington Post