THE SIGN IS a collector's item, the last of its kind, the mischief of an itinerant bureaucrat, an embarrassment to the National Park Service. It proclaims: "Polluted Water . . . No Swimming . . . Fish Contaminated." And the sign is a symbol.

Standing amid the fishermen in the shadow of Watergate, near Rock Creek's exit from forest to river, the sign advertises the confusion, misinformation and squabbling that attend the maligned Potomac.

"Fish contaminatedy?" Who says?

Not the District's Bureau of Air and Water Quality Control; not the Interstate Commission on the Potomac River Basin; not the Metropolitan Council of Governments, each an agency empowered to do so.

Only the National Park Service, which doesn't monitor river pollutants, doesn't spotcheck Potomac fish and doesn't arrest the fishermen who line the riverbank.In fact, through its concessions the Park Service rents rowboats to fishermen.

Given Kepone in the James, mercury in the Shenandoah, the questions demand answers: Are Potomac Fish safe to eat? If not, why not? If they pose no health hazard, why doesn't the District say so?

In its still-unpublished 1976 report to the City Council, the District's Department of Environmental Services stated: "No data (concerning Potomac River fish) are available on heavy metals, pesticides or other toxic substances." But there's a catch, call it "Catch 33," in the Department's blunt warning.

In November, 1976, the Potomac Bassmasters held their first annual fishing tournament near the 14t Street Bridge. The anglers hauled 33 pounds of largemouth bass from one of the least aesthetic sections of the river. The Department of Environmental Services saw significance in the tournament.

According to the report: "If fishing itself can be considered biomonitoring, the Potomac within the District of Columbia has become 'fishable' even for largemouth bass, which normally thrive only in water of reasonably good quality."

The U.S. Food and Drug Administration, the Environmental Protection Agency, the Commonwealth of Virginia and the Interstate Commission on the Potomac River Basin have accumulated solid evidence that the assessment is accurate.

Equally impressive evidence indicates the opposite, but a question should be answered before the inevitable conclusions: How does a river spread disease? The two-part answer deals first with water quality, then with the edibility of fish the water sustains.

Human feces are discharged daily into the river. The feces contain pathogens, viral and bacterial, which can enter the body via cuts, the mouth, or, less frequently, other orifices. Swim in the river and gulp the water, clean a fish and cut your finger, fry a catfish and serve it rare. Such activities encourage ingestion of pathogens that may be much too concentrated in Washington's Potomac.

Countermeasures are obvious. Chlorinate sewage effluents. Clean fish thoroughly. Cook fish well.

The fisherman's first line of defense, however, is the bodily functioning of the fish itself. Fish purge their guts in less than three days and remain purged until they re-enter a socalled "hot spot" on the river.

But while fish can purge themselves of fecal pathogens, they cannot get rid of heavy metals, industrial chemicals, pesticides and herbicides that "bioaccumulate" in otherwise edible tissues. In its 1976 "Quality Criteria for Water" handbook, the EPA lists more than 50 such metals, molecules and compounds and specifies safety criteria for each.

"MERCURY . . . CRITERIA . . . 20 ug/1 (micrograms per liter, or two-millionths of a gram) for domestic water supply; 0.05 ug/1 for freshwater aquatic life and wildlife; 0.10 ug/1 for marine aquatic life." The handbook continues: "For man, the fatal oral dose of mercuric salts ranges from 20 miligrams to 3.0 grams."

It seems obvious that mercury, even in infinitesimal quantities, is hazardous. Like Strontium 90 and Kepone, mercury bioaccumulates. Fish can ingest it directly from water and the 80th element climbs quickly to the top of the aquatic food chain, from minnows to bass, or from anchovies to swordfish.

A series of 200 related compounds, their toxicity far greater even than mercury's, caught the attention of aquatic biologist during the late 1960s. These are the polychlorinated biphenyls, or PCBs, a group of fire-resistant insulating liquids used mainly in electrical equipment. EPA water criteria for PCBs are especially stringent - .001 ug/1 for freshwater and marine aquatic life and for consumers thereof, plus the warning, "every reasonable effort should be made to minimize human exposure."

Their ill effects include skin discoloration, swelling of the eyes, liver disease, retarded infant growth and even fetal death. PCBs are known to be carcinogenic.

Now the Food and Drug Administration has proposed lowering its fish tissue standard from five PCB parts per million to two PCB parts per million, an action taken earlier by Canada.

Consider again that sentence in the District's Department of Environmental Services report: "No data are available on heavy metals, pesticides or other toxic substances."

It should not be construed as an admission of total ignorance by regulators and scientists who deal with Potomac chemicals and fish. It is, rather, an admission that no regular, systematic water analysis or fish tissue analysis programs exist today. It is also an admission that the available evidence is too stale and too scanty to act on, that data now in hand will not withstand scientific scrutiny.

What data is on hand?

William Mason is an aquatic biologist employed by the Interstate Commission on the Potomac River Basin, a man who speaks affectionately of the river.

During 1975 and 1976, Mason collected a variety of fish samples for tissue analysis. His gas chromatograph scans disclosed that all contained acceptably low levels of pesticides, and that all but one met FDA's current five parts per million PCB standard. That fish registered 9.7 PCB parts per million, and under federal law could not have been shipped in interstate commerce or sold abroad. That single fish also showed that the Potomach does not contain noxious chemicals.

But to put that one fish in perspective, consider that New York State tested 107 rockfish for PCB and found an average concentration of 8 parts per million. Now New York recommends that persons eat no more than one Hudson River rockfish per week and pregnant women and children under 12 abjure the fish entirely.

More evidence. In a report published in January, 1973, the authoritative American Chemical Society claimed that water samples collected between Key Bridge and Mt. Vernon contained a mix (more than 100 parts per million) of dry cleaning fluids, auto exhaust particles, plastic residues and industrial chemicals. At 10 parts per million, a dry cleaning fluid, tetrachlorethylene, was the worst offender. Which of the chemicals bioaccumulate in fish tissue waas not reported, nor was their toxic effort on man.

Detection of such chemicals is a little-noticed but expensive, never-ending war game between offender and defender. As soon as the defenders, the EPA, the FDA, the state authorities and laboratories, perfect new detection techniques, the offenders invent new chemicals that defy detection.

According to Curtis Coker of FDA's Bureau of Foods, a single test has been devised to detect the presence of more than 200 pesticides in fish tissue. But an entirely different test is required to uncover Kepone contamination. The gas Chromotography used to detect PCBs in a single tissue sample consumes about five manhours. Each sample costs about $100 to $125 to analyze, and tests required by other chemicals cost even more.

Yet none of these tests alerted the scientific community to the hazardous health effects of mercury, PCBs or Kepone. The presence of mercury in the upper Shenandoah was brought to the attention of Virginia authorities by the concerned company that apparently was responsible - responsible since 1928. The lethal PCBs attracted national attention because of a massive Great Lakes' fish kill. Ordinary physicians identified Kepone poisoning, and, in due course, the chemical that caused it.

It would thus seem likely, even probable, that if a hazardous new (or old) chemical is active in the Potomac, its presence will be detected by a young intern or a sports fisherman or a water policeman who sports a long slick of bluegills floating belly up, not by those whose mandate is to protect the public.

Despite the costs, despite the new legal liabilities, despite the mechanical problems, the FDA and the District Bureau of Air and Water Quality Control are drafting tissue analysis programs. The bureau is hiring a full-time biologist and intends to catch and analyze Potomac River fish before, not after, any Kepone-like problems arise in the river.

Besides making its laboratories available to local authorities, FDA has budgeted an independent program to conduct "surveillance scanning" of inland food fish. This program has not yet taken final shape, but FDA hopes to analyze at least 2,000 samples per year. Such surveillance scanning is now confined to fish wholesalers.

Beneath the Washington Ship Channel lies a carpet of silt 15 feet deep. Whether this slit, like the shenandoah's harbors puddles of mercury can be proven or disproven, but it hasn't been. Along Washington's waterfront, 14 pipes discharge industrial effluent. Each discharger has been awarded an EPA permit specifying kind and quantity of pollutants permissible. But new chemicals are constantly being developed and no one knows whether only permissible pollutants are contained in the discharges.

The overall problem's ultimate solution is patently absurd - constantly check the ambient water for all pollutants; drill daily through the silt and analyze the core; catch Potomac fish by the thousands and subject each to tissue analysis.

Only such frenzied activity would justify the National Park Service's sign: "Polluted Water . . . No Swimming . . . Fish Contaminated."

How the sign came to be in the first place is an interesting digression. About six years ago the National Park Service donated office space to a Public Health Service official charged with erecting signs along interstate rivers. After reading District Regulation 71-28, which prohibits swimming and commands cleansing of pets that contact the Potomac, the PHS official zealously ordered up the sturdy sign and other less sturdy ones which Hurricane Agnes blew away.

At its June 29 hearing on the Potomac, the City Council will study a recommendation from the District's Environmental Health Administration to revise water quality standards so as to specifically permit Potomac fishing. Passage would sweep away much of the legal confusion that fishermen should contend with but usually ignore. Still left to Mayor Washington would be the task of posting no-swimming notices along "every half-mile of (Potomac) shore, as near to the water as possible . . .

None of this seems to trouble biologist Mason, who fishes the Potomac and eats his catch.

"If you want to know whether the water's safe, just turn over a rock," he said. "If you see colonies of creatures milling about, don't worry. Midges are more sensitive to the poisons than we are."

Mason concluded that as rivers go "it's a pretty clean stream." And when Sec. 305A of the 1972 Water Pollution Control Act Amendments was passed, that, in his opinion, clinched it. The Potomac was ranked high among all waterways tested.