In tests 10 years ago, German scientists found that a chemical known as NDELA caused liver cancer in 20 of 20 rats that ate it.

Last month, the Food and Drug Administration unexpectly started tests intended to find out if application of the compound to the skin of laboratory animals also may cause cancer.

The reason for the FDA study: NDELA has been discovered in 25 of 27 skin creams, body and suntan lotions, hair shampoos and other cosmetic products that are typical of those most widely used.

These products are not eaten. But the scientists who made the discovery said this doesn't eliminate reason for concern.

They pointed out that an additive in some cosmetics, triethanolamine, is "a wetting agent and is used industrially to increase the penetration of organic liquids into wood." Consequently, they said, "It does not seem unreasonable to assume that a significant amount of NDELA applied to the skin may be absorbed."

But, they emphasize, "It is not possible at the present time to assess properly the potential hazards to man."

The scientists - Dr. David H. Fine and three colleagues at the Thermo Electron Research Center in Waltham, Mass., and two Massachusetts Institute of Technology chemists - disclosed their findings in a little-noticed report last month. A National Science Foundation grant aided the research.

An FDA official with "no doubts" about the validity of the research told a reporter that the Division of Cosmetics Technology has begun to evaluate its health significance and also will try to find out whether NDELA (N-nitrosodiethanolamine) is present in a large number of cosmetic products in addition to the 27.

Division director Heinz J. Eiremann estimated that it will take a minimum of several months to determine whether application of the suspect chemical to the skins of test animals will cause tumors.

As of now, he said, "We don't know what the health significance is. We must do whatever is possible to find out."

NDELA is a member of a chemical family called nitrosamines that are proved cancer-causing agents in laboratory animals and suspected carcinogens in humans.

Nitrosamines generally form when nitrites, which are created by bacterial effects on the nitrates that are present in every plant, combine with amines, which exist in a large variety of products.

The Thermo Electron report said that in the cosmetic products, NDELA probably forms in a chemical reaction between nitrites (which may be present as contaminants) and amines - including triethanolamine - sometimes added as detergents and emulsifying agents.

Eiermann said that the nitrosamine found in cooked bacon appears to be 200 times more potent a carcinogen than NDELA. Nitrites are added to bacon as a preservative.

Fine alerted the FDA to the research findings on March 22, a day before reporting them to a meeting of the American Chemical Society in New Orleans.

A few days later in Washington, Fine met with experts in the cosmetics industry at the request of the Cosmetic Toiletry and Fragrance Association, Inc.

As a result, a CFTA spokesman told a reporter, the trade association formed four separate scientific task forces. Their missions: to find independent laboratories to buy equipment such as Fine's, which is unique, and to use it to try to verify his study through duplication; to review the ingredients of all cosmetic products and try to isolate those that may participate in the production of NDELA; and the search for possible NDELA inhibitors.

Task force leaders will meet again Wednesday to discuss their progress, the spokesman said. He also said that manufacturers are investigating their production processes to try to eliminate the possibility of NDELA formation.

"The industry has a record of being responsive to recognizing new issues as they arise," the spokesman said.

The Thermo Electron report disclosed extreme variations of NDELA a samples of the cosmetic products.

The highest concentrations - 48,000 and 25,000 parts per billion (ppb) - were listed for two samples of Max Factor Ultralucent Whipped Creme Makeup.

The report identified none of the products, but the FDA named them all under a Freedom of Information Act request.

Two products, Diaprene Cradol shampoo and Nutraderm Dry Skin Lotion, were found to have either no NDELA or an amount so tiny that it could not be detected.

Ten samples with traces of less than 10 ppb were identified as: Max Factor Ultralucent Waterproof Makeup, one sample of Johnson's Baby Lotion (another had 100 ppb), Keri Lotion, Nivea Cream lotion, Gillette Deep Magic lotion, Sea & Ski Suntan Lotion, Almay Deep Mist Extra Rich Lotion, School Cocoa Butter Lotion, Breck Shampoo for Dry Hair, Mennen Baby Magic Shampoo and Wella Balsam shampoo.

The report listed these other products and concentrations: Relvon Moon Drops, 3,700 ppb; Helena Rubinstein Silk Fashion, 1,200; Clairol Herbal Essence Shampoo, 260; Scholl Rough Skin Remover, 140; Avon Topaze Cologne and PPP Baby Shampoo, both 100; Noxzema Skin Cream, 83; Head and Shoulders Shampoo, 70; Clairol Creme Formula Hair Color, 68; Bain de Soleil Suntan Creme, 47; Helene Curtis Everynight Exterbody Conditioner, 27; and Extra Strength Desitin shampoo, 22.

Routine use of Max Factor Ultralucent Whipped Creme Makeup could expose the user to 50 to 100 millionths of a gram of NDELA daily, the report estimated, adding: "Persons such as actresses and models may be exposed to even higher levels."

Johnson's Baby Lotion usually is used in larger amounts, giving an approximate daily exposure of 2 millionths a gram of NDELA daily. But if used as recommended at each diaper change, the exposure, in terms of relative body weight, could be "considerably larger," the report said.

Under an FDA order effective last Friday all labels affixed to cosmetics and toiletry products must list their ingredients in descending order of predominance.

The list is intended to enable shoppers to compare various cosmetics and toiletry brands and to help them avoid ingredients to which they may be allergic or sensitive.

In addition to Fine, the other Thermo Electron scientists who participated in the cosmetic research were Tsai Y. Fan, Ulku Goff and Leila Song. The MIT chemists in the study were G. T. Arsenault and Klaus Biemann.