The design of a small pressure gauge did not take up much of anybody's time in the deep working levels of the Nuclear Regulatory Commission when it first arose a few months ago.

Engineers had noticed that the gauge could be misleading and had exchanged a couple of letters with the designers, Babcock & Wilcox Co. But nothing much was done. Like the small pressure-release valve that tended to stick open occasionally once it blew out, the problem was basically one of plumbing technology, the science of leaks and drips.

The Nuclear Regulatory Commission, in the four years of its existence, has not occupied itself with leaks and drips. There are several thousand gauges and at least 1,000 valves in the giant piece of plumbing that is a nuclear reactor. Constant corrosion from hot water and dissolved chemicals thins the walls of some pipes and plugs others with what engineers inelegantly call "crud." To foresee the possible impact of every leak or drip in combination with every possible stuck valve would be all but impossible.

Yet at Three Mile Island, when that particular gauge was misread and that particular valve stuck open the result was near-disaster. Misled by the gauge, the plant operator turned off pumps he should have left operating. When cooling water finally did flow around the reactor, the valve that was stuck open let all of it out again. The hot redioactive core of the reactor was exposed and severly damaged.

In the years since science began to tame the atom, regulators have not worried nearly as much about plumbing as they have about earthquakes, sabotage, theft, some vast cataclysm rupturing pipes 30 inches across. They have worked out "defense in depth" and "risk aversion" approaches that involve two or three backup systems to every one of the estimated 400 separate systems in the reactor.

Plumbing design has largely been left to the industry.

"There was no significant difference between what happened at Three Mile Island and all the other times" that valves have failed and small things have blown, said Robert Pollard, a former NRC engineer who now works with the Union of Concerned Scientists, the nuclear industry's major critic.

"This time they got caught . . . the industry never anticipated the sequence and the operators weren't trained to respond to it."

Yet the NRC has been more involved in the small decisions than regulators ever were before. Its predecessor, the Atomic Energy Commission, frankly promoted the development of nuclear power, soft-pedaling equipment failures and treating worried scientists as traitorss. The AEC often paid major suppliers like General Electric and Westinghouse to do safety tests of their own equipment, following a principle it called "limited industry self-regulation."

NRC Commissioner Peter A. Bradford once called this period one of "silenced concerns and rigged or suppressed studies." Congress created the NRC in 1974 in an effort to separate the regulatory function from the industry-promoter history.

Still, the first thing the new NRC did was to adopt nearly all its predecessor's regulations and procedures, keeping on most of the staff. "Clearly the NRCis still in transition from the pro-industry orientation of the AEC to being a genuine public interest body," commented Sen. Gary Hart (D-Colo.), chairman of the nuclear regulatory subcommtitee, in an interview last week.

In a computerized age, and overseeing a computerized industry, the NRC is still largely chained to records that are piles of paper and rolls of micro-film. No one is sure, for example, whether a directive might have been issued a few years ago on the misleading pressurizer gauge, or whether all the reported instances of pressurizer valve failure among the 68 operating reactors, if any, could be counted without weeks of paper-shuffling.

The NRC directives and regulations are written down in various places and forms,a collection of rules, guides, formal technical specifications, design requirements and performance criteria. To check all the edicts applying to a specific part is far easier at the industry level than it is at the NRC. "There's no reason why they can't computer-analyze their reports" on unexpected events. "We do it constantly," said a noted industry consultant.

Charged with guaranteeing the public safety, issuing construction permits and operating licenses and defining safeguards, the NRC really has only two choices in enforcing its edicts. It can fine a utility a maximum of $25,000, small potatoes in the industry, or it can shut the reactor down. That, said Pollard, is like choosing between a slap on the wrist and capital punishment.

The agency also gets involved in the issuing of licenses to export nuclear materials. Chairman Joesph Hendrie has said he does not think the NRC ought to be saddled with the political aspects of that kind of decision, but only with the technical parts. Until Three Mile Island, the NRC had been controversial chiefly for its role in debating such exports to India, Brazil and other countries.

Many industry supporters complain that the NRC is too involved in politics and defending itself before Congress to properly organize and streamline its regulatory job.

"They should be focusing on reducing the paperwork so the operators can spend more time watching the indicators... they should concentrate on requiring monitors on a reactor's vital signs and not issue this blizzard of paper," the consultant said.

Nuclear power's critics, on the other hand, think the NRC has not been responsive enough. "They're still generally fairly protective of the industry,"said Peter Franchot of the Union of Concerned Scientists. "Things are usually hashed out with the industry before any technical decisions are arrived at, and then those decisions are applied mostly to new plants rather than retroactively to all the old ones."

an example, he said, was the closing of five nuclear power stations last month because a computer program that had been changed in 1972, after the plants were designed, left them vulnerable to earthquakes. Old plants were not required to check themselves when the computer formula was changed, Franchot noted, and the potentially dangerous situation lasted nearly seven years.

There remains the question of whether the nuclear industry with all its valves and gauges is so complex as to defy regulation from Washington.

"Events that occur too frequently suggest deep-seated problems that may not be readily resolved," said Rep. Morris Udall (D-Ariz.), chairman of another nuclear regulatory sub-committee. "Nuclear may be one of those technoligies that get so complex that they fall of their own weight."