A new study at the site of the nation's worst nuclear power accident shows that more of the Three Mile Island reactor's core was reduced to rubble than had been believed, officials said today.

Jack DeVine, technical planning director for Three Mile Island Unit 2, said sonar mapping of the interior of the damaged reactor revealed that the entire top of the core had collapsed into a bed of rubble.

Company officials discovered a void in the center of the core last year when they lowered a television camera into the reactor for the first time since the March, 1979, accident. But they said they did not believe it extended all the way to the edge. "The core void still will be shown as five feet deep," DeVine said today. But he said the mapping now shows the void extends "nearly to the periphery" of the core, which is 10 1/2 feet in diameter.

Company officials said they had inferred from last year's television examination "that perhaps the outer row or two of fuel assemblies" still were intact, and had expressed hope they ultimately could be removed from the reactor in one piece.

"I think now we see there isn't anything that can be handled in one piece," DeVine said.

Officials of General Public Utilities, owners of the damaged reactor, declined today to discuss what conclusions might be drawn from discovery that the void was larger than earlier believed.

"It may tell you that temperatures at a given point measured out from the center of the core were higher during the accident to cause the greater area of damage, but that's relative," DeVine said.

Officials said, however, that new information about what took place during the accident may emerge in the months ahead as a result of analysis of material from the core.

GPU officials said they expect to begin extracting the core samples this week and will send them off to be analyzed at the Energy Department's Idaho National Engineering Laboratory near Idaho Falls.

Greg Eidam, manager of Data Management and Analysis for GPU, said the company planned to take six samples from different parts of the core rubble, and these will be subjected to a variety of tests expected to take about nine months.

DeVine pointed out, however, that these samples will represent about "one millionth of the core's volume" so they will hardly answer all the remaining questions about the accident, and probably will not provide answers to such lingering questions as whether fuel actually melted during the accident.

DeVine said, however, tests would aid the clean-up effort by providing crucial data about the radioisotope content of the debris. "We don't know if that loose material still has a fair amount of cesium and strontium in it which will continue to dissolve and create a waste management problem, or whether it has released most of its fission product activity and is relatively benign," he said.