HEALTH officials believe that the rind of cantaloupes from south Texas was the source of the Salmonella poona bacteria in last summer's outbreak. "Bacteria are expected to be found on any raw agricultural product like a melon," the Food and Drug Administration says. Washing with water removes it.But through improper handling by public food servers, health officials say, the bacteria were allowed to get on the edible part of the melon and multiply. None of the melons were available for testing, so there's no direct evidence either way. Growers in Texas say that virtually all cantaloupes grown in or shipped through the state are washed with a brush in a chlorine solution that kills bacteria. The melons are stored and shipped under refrigeration. "The largest state to consume Texas cantaloupe is Texas," says William Weeks of the Texas Citrus and Vegetable Association. Yet that state didn't experience the outbreak. Still, when health officials elsewhere first associated cantaloupes and salmonella, the state agriculture department tested trucks, 100 melons, and packing houses without turning up Salmonella poona. As for epidemiological studies, "you can never be absolutely certain, ... but the results very clearly implicated cantaloupe as the source of the outbreak," says Craig Hedberg of the Minnesota Department of Health. Four separate investigations by state health officials came to that conclusion. BUT Mark Fow remains skeptical. Mr. Fow is a toxicologist and consumer safety officer at the Food and Drug Administration's division of emergency and epidemiological investigations. He coordinated the flow of outbreak-related information among state health agencies, the FDA, and the Centers for Disease Control. In investigations by other states and Canada, he says, "cantaloupes didn't fall out." Fow didn't name those states. "Epidemiological associations are not evidence of causality," Fow stresses. "It doesn't mean that it was really the cantaloupe that caused it." He offers another possibility. "If this is from salad bars primarily, it doesn't mean that cantaloupe necessarily had the salmonella on the surface," he argues. "Other fruit may have been carrying it and then it got on the cantaloupe." In New Jersey, for instance, cantaloupe was named although it was in a salad with five other kinds of fruit. Robert Tauxe of the CDC confirms that the Canadian study didn't pinpoint a source. But he is not aware of any state investigations that were completed and did not implicate cantaloupe. Dr. Tauxe remains certain that cantaloupe was associated with the outbreak. But in two out of four tests cited in the CDC report, the statistical "confidence intervals" describing the results were low enough that Tauxe says a statistician would call them "borderline significant" associations.