Four years after the Three Mile Island accident radiated fear across America and stunted the growth of the young nuclear power industry, technicians have probed gingerly into the reactor's damaged core to remove the first spoonful of radioactive debris.

The first entry mission, resembling a moon-walk as three men in air-cooled protective suits worked awkwardly inside the highly contaminated containment building, was cut short Friday after only one of three planned samples was retrieved.

From TV monitors in a nearby command center, controllers watched as one team member slumped down atop the service structure above the reactor, sat clutching his head and complained of feeling ill.

The command center immediately ordered the stricken man to leave, and a health technician escorted him from the structure.

As he emerged through an air lock that keeps radioactivity from escaping, another technician radioed: "He don't look so good." He appeared tired and pale, officials said, and complained of a "severe headache."

Officials refused to allow the two remaining team members to complete the mission and said a new group will try Monday to extract the other two samples. They said the original team would attempt to obtain three additional samples Tuesday.

The samples will be shipped to the Energy Department's Idaho National Engineering Laboratory, where they will be analyzed in search of a better understanding of what occurred in the reactor in March, 1979, during the nation's worst nuclear power plant accident.

The ill technician, officials said, was one of fewer than a dozen who have shown symptoms of "heat stress" over the past three years while working in the hot, humid confines of the containment building. He picked up a radiation dose of about 75 millirems during his entry--the equivalent of 2 1/2 chest X-rays.

The other members of the entry team received doses ranging up to 190 millirems during their 90-minute stay, officials said.

Officials of General Public Utilities, owner of the damaged TMI Unit 2 plant, say they have been "very successful" in minimizing radiation exposure to workers during the cleanup. They note that they have stayed well under the federal limit of 3,000 millirems per quarter or 5,000 per year for nuclear workers.

"We've never had anybody at anywhere near 3,000 millirems and we've only had a couple that have been over 2,000 millirems for an entire year," said spokesman Doug Bedell.

Since February, GPU has mounted an intensive "dose reduction program" designed to reduce the radiation "dose" to cleanup workers. Bedell said the program has cut radiation dose rates on the ground floor of the containment building from 350 millirems per hour to about 200 at present.

"That's one of the reasons the cleanup is so blasted expensive," said Bedell. "All the measures and protections and training and rehearsal that you need to reduce dose rates adds very much to the cost of the cleanup," which already has passed $380 million.

Even so, Jack DeVine, technical planning director for TMI 2, said cleanup workers thus far have received a collective dose of 3,000 rems (a rem is 1,000 millirems), making it increasingly clear that the 8,000-rem collective limit set by the Nuclear Regulatory Commission for completing the cleanup "is not a number that will let us complete the job."

"We haven't done the rem-intensive jobs yet, such as defueling the reactor and cleaning up the reactor building," DeVine said. "We've got some very detailed analyses and they show ranges of possible numbers for completing the cleanup, but they are all in the 20,000 to 25,000 rem range."

Bedell said the NRC already "has opened the process of revising the 8,000 limit, and there will be public meetings on it later this year."

Friday's one-cubic-inch sample of debris, taken from the bed of rubble that once constituted the top five feet of the damaged reactor's core, was fished out by the technicians using a stainless steel clamshell scoop attached to a 45-foot rod.

The debris was brought up through a control-rod guide tube into a heavy lead transfer cask, where an instrument inserted through a peephole to within six inches of the sample found it to be emitting 3 rems of radiation per hour.

Greg Eidam, GPU's manager of data management and analysis, said the radiation level suggested that the sample consisted of two elements--the zircaloy skin which contains the fuel and dust-like traces of uranium fuel called "uranium fines."

"We were expecting a worst-case condition through the peephole of 75 rems per hour," Eidam said. A reading that high would have indicated that the sample consisted entirely of uranium fuel.

"That's kind of a 'Go, no go' point," Eidam said. "If we got superhigh readings, we would have lowered it back down in the sampler and emptied it. But the cask and everything else was designed for worst-case conditions. Our number for turning it around and sending it back was 75 rems per hour."

The sample was raised through the transfer cask into a cylindrical handling cask having lead walls 2 to 3 inches thick. Radiation readings on the surface of this cask showed it emitting 60 millirems per hour from the top, 180 on the bottom and 400 on the side.

The handling casks will be loaded into more heavily shielded shipping casks before the samples are sent off for analysis.