In 1976, Congress set aside $1 million to study illegal aliens in the United States. By last spring, however, about 18 months and $500,000 later, not a single interview had been conducted, and now, with every cent of the million dollars gone, the government is struggling to find out what, if anything, was learned.

The heart of the project was supposed to be a single, vast survey of illegal immigrants across the country. Pollsters were supposed to fan out through the barrios and inner cities, suburbs and apartment buildings where the immigrants are believed to live, asking 100,000 of them more than 60 questions each.

The contract for this undertaking went to J. A. Reyes Associates, a Washington-based consulting firm owned by a Mexican-American who emphasized his understanding of the people in question and the the 14-year record of his business.

"Our company prides itself on: fulfilling the contracts which result in products of higher quality than required; performing contracts in accordance with time limits; and performing contracts without cost overruns," Joseph A. Reyes wrote at the time.

Yesterday he refused to comment on what happened to the survey. "Anything I've got to say would be twisted around to make me look like a jerk... I'm in a legal situation. I know what's right and I have to keep my mouth shut. All my life I've been chastised and beat up. That's way it goes."

Whiel not so reluctant to talk, officials of the U.S. Immigration and Naturalization Service were at a loss yesterday to explain just where all the money went. "That's what we're trying to find out." said one. That's why they served notice on Reyes at his 16th Street NW town house office Friday that he has to turn ove within the week all the raw material he has gathered.

The controversy over this case, however, has been brewing for a long time.

"This particular project has been plagued from start to finish," INS Commissioner Leonel J. Castillo told the House Immigration Subcommittee last March.

He began his attempt at an explanation by nothing that the project originally had been undertaken by his predecessor. "At the time that I reported to work, a quarter of a million dollars had already been obligated on this study, and they were already behind schedule." Nevertheless, he said, the decision was made to continue.

The entire time, it appears, the project was sliding further and further into a quagmire of bureaucratic indecision and contractual confusions.

The first long delays, Castillo said, came when the INS tried to get other federal agencies to take part in the project, and of course, foot part of the bill. The Department of Health, Education and Welfare, the Labor Department, and the Department of Housing and Urban Development were all asked to join through a series of lengthy consultations in which INS tried to persuade them the information gathered would be to their benefit.

Only HEW agreed to the point of spending money, however, adding $80.000 to the $810,000 committed by the INS.

Another $147,000 was spent on a separate contract for management and technical assistance to oversee the project. The remainder of the $1 million, about $20,000, went for studies of fraudulent documents and the little electronic devices used to detect people crossing the border.

There were problems getting hold of relevant bits of information from other parts of the government -- names and addresses and other detailed census information protected by the Privacy Act, Castillo explained.

Then, too, the INS had trouble getting information from itself. Material was supposed to be supplied to Reyes from the registration cards aliens are expected to file each year. That, said Castillo, was late as well.

Meanwhile, Reyes and various subcontractors were trying to develop some sort of methodology for finding, identifying and questioning illegal aliens who have absolutely no desire to be found, identified, and questioned. Additional people were put on the payroll (though Reyes would not say yesterday how many), and the money was going out at a rate close to $30,000 a month. "J. A. Reyes claims," Castillo said, "that waiting was very expensive for them."

HEW and INS, as resources were dwindling, also started changing their ideas of what the survey was to show, according to Castillo's testimony.

By March it was obvious that the money was going to run out, and with the way things were going, nothing was going to get done. To try to salvage something from the project, the proposed sample of 100,000 was cut to 10,000.

Congress was getting suspicious. "I think we're being had by this person," Rep. Sam B. Hall Jr. (D-Tex.) told Castillo at the March congressional hearing. "And I think I can see that he's going to come in and ask for additional funds."

"He already has," said Castillo. The figure under discussion was another $1 million.

Castillo told Reyes to finish the study with whatever money was left and do it in less than five months. At the beginning of the summer, interviews finally got under way. In July they were finished -- but so was the money. "Down to the penny," according to INS contracting officer Jack Keller.

Reyes wanted another $130,000 to complete his work, said Keller.

"I decidee that these cost overruns have a way of just running out of sight. So I felt it was simply time to cut bait and take what we had," Castillo recently told a reporter.

That might have seemed like a reasonable idea, but so far the INS has been unable to get whatever it is that Reyes has, which is thought to be mainly raw data from the interviews and computer programs for analyzing it.

"He hasn't said he won't hand it over," said Keller. "But he hasn't handed it over, either."

An INS spokesman said yesterday that if Reyes continues to hold on to the data after Friday the Attorney general could impound it.

Unfortunately, once the INS does get the material, officials of the agency seem uncertain just what they will do with it. "We can't very well [analyze] it in house because we don't have those kinds of people," siad Keller.

It may be given to another government agency to sort out, such as the Census Bureau; it may go to an academic institution for analysis, or another consulting firm, which would, of course, cost more money.

"We underestimated this thing. Very much so," said Keller.... But this is not to say we made a mistake."