The chief federal regulator sent to the scene of the nuclear accident at Three Mile Island believes the accident was caused more by human than mechanical error.

"The biggest lesson of this whole affair," Harold C. Denton, director of the Office of Regulation for the Nuclear Regulatory Commission, said after 19 days at Three Mile Island, "is that we have not spent enough time on the human side of things to prevent accidents like this one."

The March 28 accident, triggered when a coolant system failed, caused the evacuation of thosands of people from the Harrisburg region for more than a week, closed down the plant at Three Mile Island for the foreseeable future and left a cloud of uncertainty over the future of nuclear power in the United States.

For as long as 13 hours after the shutdown, the loss of cooling water badly damaged the uranium core and caused larged amounts of radioactive fission products to escape into a containment building around the reactor and an auxiliary building outside the reactor. Technicians are still in the process of cooking the plant down, but so badly contaminated is the containment that it will be at least six months before a cleanup can be attempted.

In an interview the day he left Three Mile Island last week, Denton identified four major human mistakes that either caused or contributed to the accident. At least one of the mistakes violated NRC rules; all four involved poor judgment by the operators on duty for Metropolitan Edison Co., he said.

Denton did not absolve the NRC. He said it probably should have had an on-site inspector at the plant, and should have worked out beforehand with state and local governments a plan for evacuation of the region near the plant.

"We've dreamed up emergency planning back in Washington without fully involving the state people," Denton said, "and you get such questions as nursing homes in the area, the availability of doctors and nurses for the people being evacuated. What about farmers; are they going to leave their animals? Things that can only be anchored on a local scale."

Denton conceded that a resident NRC inspector would not have been able to stop the accident once it started, but insisted an inspector would have influenced conditions that helped bring on the accident, especially the decision to block the valves to the auxiliary cooling pumps in violation of NRC rules.

"Those valves might not have been blocked if there was an inspector in the plant," Denton said. "At least, the procedure for testing the valves might have been more adequate."

To Victor Stello, chief of NRC's Division of Operating Reactors, the valves to back up feed water systems should never again be blocked unless the plant is shut down.

"One way to stop such a practice is to require the valves to be locked open," Stello said in a separate interview at Three Mile Island."If they're doing maintenance on them like they were doing just before the accident, then they should be required never to be in a position where they would lock all of them out at the same time."

Denton and Stello agreed that "locking out" the auxiliary cooling system was the most serious human error. Mistake No.2, they agreed, was the turning off the emergency core cooling system's two high-pressure pumps, one at four minutes and 30 seconds into the accident and the second six minutes later.

"I'm not sure it was a violation of the NRC rules, but the operator should not have turned off those pumps when he did," Denton said.

"You can't put every possible scenario in a license. The Bible gets too thick if you do. At some point you have to rely on judgment and training."

Mistake No. 3 was in allowing the escape from the concrete containment of large amounts of radioactivity. When employes neglected to shut off a sump pump, so much contaminated water was pumped out of the containment that it overflowed waste tanks in the auxiliary building alongside.

"If they'd isolated the containment when they should have, the radioactivity would still be in the containment," Denton said."I'd say almost all of it, including the iodine that escaped. There were lots of signals to isolate the containment, but they didn't do it until they noticed radioactivity in the auxiliary building."

The fourth mistake, Denton and Stello said, was in allowing a pressurizer relief valve to remain stuck open long after the accident began. This allowed so much steam and water to escape that the reactor's uranium core was uncovered at least three times over 11 hours, damaging the core beyond repair and threatening explosion that would have made the accident even worse.

"They could have isolated it," Stello said. "There is another valve in that line which could have been and ultimately was closed even if the relief valve were left open."

Denton and Stello agreed that the operators at Three Mile Island did not violate rules in turning off the four main pumps driving cooling water when they noticed the pumps beginning to vibrate.

"Question," Stello said. "Had they left those pumps on, would the pumps have been damaged so bad they wouldn't be operable anymore? This whole question is one that needs a lot more thought."

"The operator didn't realize what was happening to the core when he turned off those pumps," Denton said. "Why didn't he realize it? I don't can't write rules-that say if the airplane is crashing, it's spiraling down, that you move the rudder to the left or right. What we tried to write are rules that never put you in that spiral. After that, you have to rely on a well-trained pilot to cope."