Texas politicians and an entrepreneurial scientist announced yesterday that their at-sea test of oil-eating bacteria on a large spill last month significantly reduced oil in the water and showed no signs of harming marine life.

But the state laboratory that analyzed water samples from the Gulf of Mexico test site said it was scientifically impossible to prove the bacteria cleaned up the oil and that more statistically valid testing needs to be done.

Texas Land Commissioner Garry Mauro, a staunch supporter of the oil-hungry microbes, said the test on a 40-acre slick of oil from the Norwegian tanker Mega Borg was "an absolute success. The treatment has no downside and it appears to have an upside."

Analysis by the Environmental Protection Agency agreed that the bacteria were not toxic to marine life.

But state-run tests on the bacteria's effectiveness were inconclusive.

"From a statistically valid standpoint, you can't say there was decrease in oil concentration," said Buck Henderson of the Lower Colorado River Authority, the state facility that tested the treated water samples.

Still, Texas officials said the state would use oil-eating bacteria on some of the 1,000 small spills that plague the Texas coast every year. Bacteria will also be applied to any larger spill for a more valid test, they said.

Using a 55-gallon drum, a pump, hose and nozzle, researchers with Alpha Environmental, a Texas-based biotechnology company, last month sprayed about 110 pounds of bacteria and fertilizer on the Mega Borg spill.

According to the researchers with Alpha, the treated oil changed color, indicating that the bacteria were eating the oil and turning it into fatty acids that could be taken up by other marine animals. Aerial reconnaissance later revealed no oil in the area that was sprayed.

However, since the treated spill was not marked with buoys, researchers had no way of knowing whether the area observed contained the same water. Oil also deteriorates in water as natural bacteria consume it and waves break it up.

To help decide whether the bacteria worked, Alpha did a second test in which some oil patches were treated and some were not. But sampling problems made it difficult to show significant differences, Henderson said.

Alpha chief scientist Carl Oppenheimer said the results showed a trend. "It was not absolutely positive. But it was positive enough to give the state confidence to go ahead."