One night in the spring of 1977 Russell Murray II, then an assistant secretary of defense, settled into his Pentagon office to analyze key excerpts from about 50 videotapes.

The tapes had recorded tests of a new second-generation Maverick antitank missile the Air Force wants to buy at an eventual cost of $5 billion. From the sound tracks he could hear what the pilots testing the new Maverick were saying over their radios.

Murray says he heard "the pilots screaming and yelling, saying the 'birds' missiles were breaking lock . . . . It was a disaster . . . crystal clear before my eyes right now. It made you think, 'My God, people are lying about the test.' "

What those pilots were describing was the sophisticated weapons system's failure to track properly, or hold a "lock on" to its target.

Also in some cases the pilots saw a series of images on a five-inch TV screen in the cockpit, a screen on which tanks were supposed to be distinguishable from trees, rocks or the smoking hulks of simulated battlefield debris.

One pilot who flew in the tests was asked at an April, 1978, congressional hearing: "Why did you lock on to a bush?"

"I thought it was the target," he replied.

The new Maverick missile the Air Force wants to buy as its primary antitank weapon has a long history of troubles, delay and failure, according to thousands of pages of documents and congressional testimony. Yet Richard D. DeLauer, the undersecretary of defense responsible for weapons development, was certain that the Air Force would get approval from the Pentagon soon to go ahead with production of the missile.

On Tuesday, because of problems with the missile and recent unsatisfactory tests, the Air Force put off a full production decision until 1984.

There is no alternative anywhere in the long weapons development process, and the Air Force says it needs a missile for use against ground targets, primarily tanks.

How can the Air Force and the Pentagon consider buying 61,000 new Mavericks over the next decade when the missile has such a history?

The old Maverick with the television guidance system was a failure in Vietnam combat conditions, mainly because its chief target, tanks, especially camouflaged tanks, blended in with the surrounding jungle, thereby blinding the television sensor system that depends on visual contrast to work.

In addition, the old TV Maverick could not work at all in darkness or in low-visibility conditions such as rain, fog or smoke from the battlefield.

For years the military had been developing and improving heat-seeking sensors that could detect infrared heat from targets such as tanks at distances of four to six miles or more, even in darkness and through smoke.

In late 1975 the Air Force decided to push for full-scale development of such an advanced, second-generation, heat-seeking Maverick.

"The operational commander has very little say so about the new systems," said Gen. John W. Vogt, one Air Force commander who retired after extensive Vietnam and European experience.

Vogt had found the first-generation TV Maverick a failure in Vietnam. "The things I have been screaming for, both as a commander in Vietnam and a commander in Europe, were not the things that were high on the list of priorities in Washington for development," Vogt told Air Force historians in a post-retirement debriefing, portions of which have been declassified.

While others, "particularly in the R&D research and development area," were pushing the Maverick, he said, "I am trying to get area unguided weapons. And who is winning? They are. We are buying Mavericks, that sort of thing. So, how do you get the voice of the commander into the hardware business? Well, you've got to change the system somehow . . . . "

In November, 1976, Deputy Secretary of Defense William P. Clements, who is now governor of Texas, approved full-scale development of the new heat-seeking Maverick on the condition that it be subjected to "real-world" testing and evaluation not only by the Air Force but with the Army, whose troops are supposed to be protected in close air support of ground operations.

Clements' ordered the Army and Air Force to participate in tests under simulated battle conditions, and that the results be judged independently.

In February, 1977, these new test were set up at Fort Polk, La. A small 1.5-by-2 kilometer area was set aside at the test range. In the Maverick's nose is the sensor, which detects an object that is noticeably warmer or cooler than its surroundings.

This thermal contrast, as it is called, generally in the shape of the object, then appears on a five-inch video screen in the cockpit of specially equipped tactical airplanes.

Some of the tests showed, however, that the image created on the tiny scope by a sun-warmed rock or tree, or a bonfire, hot smoke, flare grenades or burning armor hulks, can resemble potential enemy targets. Such thermal clutter can be extensive on a battlefield.

In these conditions, said Thomas S. Hahn, a physicist on the staff of the House Armed Services Committee until last April, "what you get on your display is a bunch of bright spots . . . . You don't know what you are looking at . . . . If you can't determine that there's a gun barrel, all you can determine is that there's a blob and it's hot."

On the small test range, the target tanks traveled over the same few corridors in each test. One official familiar with the Fort Polk tests said, "After the first couple of nights, the pilots learned the layout . . . how to beat the system by learning where the tanks are, where they go back and forth. This was not dishonesty. They simply wanted to look as good as they could."

The pilots made 123 runs with simulated launches, according to the declassified version of a secret 1979 General Accounting Office report. Practice runs that might possibly be fatal in battle were not counted.

"That's how you get excellent results, only counting the successes and throwing out the unsuccessful passes," said a retired Air Force combat pilot.

Still, fewer than 60 percent of the counted passes produced lock-ons on true targets, meaning that the missile "recognized" the target and homed in on it without losing sight of it.

The rest of the tests produced no lock-ons at all, or lock-ons on false targets, such as fires or sun-warmed objects: the look-alike images that can appear on the five-square-inch videoscreen in the cockpit. At times even hot snow, the result of a tank passing over snow-covered ground, deceived the missile sensor.

To make the required independent evaluation, the Pentagon hired System Planning Corp., a private consulting firm in Arlington. SPC confirmed that pilots would have to use their eyes to find tanks--certainly to ensure that they were enemy rather than friendly--and, in the process, would become highly vulnerable to air defenses.

The analysis also showed that the effectiveness of the new Maverick was impaired sharply by thermal clutter presented by other hot objects on the battlefield that for up to half an hour are hot enough to show up on the scope.

The Air Force generally disputed the analysis, insisting that pilots can launch Mavericks successfully with reasonable chances of surviving.

Videotapes of the test showed that the firing of a simulated enemy tank on the ground at times caused the sensor to break its tracking.

These were the tapes reviewed in the spring of 1977 by Assistant Secretary of Defense Murray.

At about this time, in April, 1977, the Air Force was gearing up to award a $38 million development contract to Hughes Aircraft Co., the producer of both the old TV-guided Maverick and the new heat-seeking version.

But Congress, learning of the Fort Polk results, rejected the budget request. It was troubled, the GAO said, by "the deficiencies in target discrimination, acquisition, and lock-on experienced during operational testing . . . . " Congress did, however, provide some funds to keep the program alive.

The next tests were conducted at Baumholder Air Force Base in Germany in January and February, 1978. Anthony R. Battista, a member of the professional staff of the House Armed Services Committee, interviewed pilots who flew the in both the 1977 and 1978 tests.

He said he was convinced that practice runs in familiar settings had taught the pilots where the targets were and that what appeared on the cockpit scope had not guided them, as would be necessary in an unfamiliar combat situation.

Using the tapes of what appeared on the screen, Battista said, "I asked about the quality of the tapes. They said it was about as good as what they had seen in the cockpit."

At the 1978 German tests, the tanks traveled on the same six prepared lanes every night, a symmetry not found in combat. The record shows that 317 practice runs were not counted. Of the 215 passes the Air Force did score there were lock-ons to targets 113 times, or 53 percent.

Overall the record was worse than the 1977 Fort Polk tests.

William J. Perry, director of defense research and engineering in the Carter administration and a supporter of the new Maverick, told a House subcommittee in April, 1978, that when a Maverick launch is to be made at considerable distance, "The pilot will have to be cued into the area where he will operate. He will have to have some idea what it is that he is looking for."

Yet in a recent written reply to questions about the new Maverick the Pentagon said, "It is very practical for a pilot to acquire a target without eyeballing it."

The new heat-seeking Maverick is not an all-weather missile, and it will not work in heavy rain. What defeats or causes problems for the new Maverick is humidity, which absorbs infrared heat. And during the winter, when the 1977 and 1978 tests were conducted, the cooler air does not hold as much moisture as it would in summer, a time when the significant tests have not yet been conducted.

During a House research and development hearing in April, 1978, on the tests, Rep. Thomas J. Downey (D-N.Y.) asked Maj. Carleton Whitley, a pilot who flew in the tests:

"How do you distinguish burning bushes from tanks if you lock on to them . . . ?"

"I made a mistake," Whitley answered. "Three other times throughout the test, or two other times, I made a mistake."

Defense research and engineering chief William Perry acknowledged at the hearing that it was "extremely difficult to differentiate between the targets and other objects with similar shape and contrast . . . . "

"I am falling considerably short of giving the program a clear bill of health. I concede that there are still problems . . . which we need to work on."

After more tests were conducted, the Air Force Test and Evaluation Center said there was a high success rate. Full development funds were released by the House Armed Services Committee over objections from the staff, which recommended withholding the money.

"We produce our best technical engineering judgment," staff member Battista said, "But the decision-making has been, is and always will be the prerogative of the members."

He added: "Programs are like freight trains. Once they get started, it's very hard to turn them off, even if they don't make sense."

Even if Congress was satisfied, the GAO, its watchdog agency, remained dissatisfied with the operational testing done through the end of 1979. In a February, 1980, report, the GAO once again urged that the new Maverick be tested in "a realistic battlefield environment," and said that the problems needed to be "corrected or reduced to an acceptable level" before going ahead with production.

The Air Force scheduled five operational live firings between last October and the end of this January.

One aborted, two failed and the other two have been postponed. The Air Force halted the live firings and persuaded Undersecretary DeLauer to delay the meeting on the pilot production decision for 490 missiles that had been set for this Jan. 28.

That meeting was moved to March 2, next Tuesday, but on Monday of this week, as The Washington Post was about to begin this series of articles on the Maverick, the Air Force announced that it would request only about 200 missiles for continuing tests, not the initial production of 490 as expected.

Then on Tuesday the Pentagon announced that the full-production decision for 61,000 new Mavericks, originally scheduled to be made this summer, was being put off nearly two years, until March, 1984.

Maj. Gen. James H. Marshall, director of weapons development and production for the Air Force, said he sees a problem with any delay of the new Maverick. "You see, there's a manpower problem, realistically, that you've got at Hughes. You got a team together. You've been building missiles for the tests , and you'd like to have the same guys who got the learning go on and start building your production birds. And you got to make a trade-off when you start talking about slipping production, because those guys are going to disappear and you're going to lose something in that process . . . . "

Last month, DeLauer, the senior official who for all practical purposes will determine the future of the new Maverick, summed up why he was so confident that the Pentagon would decide to go to full production on the Maverick.

"What gives me the confidence? I've looked at the test data so far. I've looked at the design. I've had the guys in who designed it. One of the guys who designed it sat in that chair," he said, motioning in the direction of his interviewer. "I have high confidence in his technical ability and his honesty to me . . . . "

Asked to identify the person who instilled in him such confidence about the Maverick, DeLauer said he was referring to Malcolm R. Currie, a vice president for missiles at Hughes Aircraft Co., Maverick's manufacturer, which stands to get the major portion of the $5 billion contract if the Pentagon puts the missile into production.

Currie was also a promoter of the Maverick when he held DeLauer's job at the Pentagon from 1973 to 1977.

And, DeLauer said, "Mal Currie. You know, he's got a string of successes." CAPTION: Picture, RUSSEL MURRAY II. . . "My God, people are lying. . . "; Picture 2, GEN. JAMES H. MARSHALL.; Picture 3, The Maverick antitank missile the Air Force wants to buy--for $5 billion.; Picture 4, Richard D. DeLauer; Picture 5, MALCOLM R. CURRIER