CENSUS FIGURES SET OFF NEW CONTROVERSY

----WASHINGTON It may be possible to find someone in America who believes the nation's population as of April 1 was 249,632,692, but it will not be easy.

Statisticians do not believe the punctilious precision of that figure. Neither do the number crunchers in the ever-growing information industry. Neither do the mayors and planners in city halls around the country. Neither, for that matter, do the officials of the Census Bureau and its parent agency, the Commerce Department.

Pressed recently as to whether he would call the bureau's new figure ''reality,'' an exact accounting of the population, Commerce Department Undersecretary Michael R. Darby dodged, weaved and finally said, ''It's auditable.''

For perhaps the first time since 1790, most Americans believe what the numbers buffs have known: When the Census Bureau tells you that there were 249,632,692 people in America on April 1, 1990, it just ain't so. But it is not clear what Congress and the courts will do.

For the moment, Congress and the courts have only two choices. They could accept the census figures as a fair reflection of the relative distribution of the population, though not a precise count of every person in every state.

Or they could say that some people and places are shortchanged more than others, and alter the figure with statistical formulas intended to improve the relative standing of historically undercounted groups such as blacks, Hispanics and inner-city residents.

With the announcement of the new census figures, the battle lines between adherents of the current method, the old-fashioned head count, and adherents of using statistical formulas to adjust that head count have become more focused.

''The only people who are happy with the number are the lawyers,'' said Kimball W. Brace, president of Election Data Services, a Washington consulting firm that evaluates census data.

The battles will take place in a variety of places, among them Federal District Court at Brooklyn, where New York and other cities have filed suit to force the Commerce Department to adjust the figures, and Congress, which could require an adjustment but has avoided doing so.

There will also be arguments in the 50 state capitals, where congressional and legislative districts will be drawn up next year, and in the statistics departments of universities, where academics argue about how to count.

The next major phase of the fight will come on or shortly before July 15, when the Commerce Department announces whether it will adjust the figures.

Conventional wisdom says adjustment helps Democrats in urban areas and that the status quo favors Republicans and the suburbs. But nothing is certain, except that the census numbers have never been what most people thought they were.

In 1940, the Census Bureau figured it missed nearly 8 million Americans, about 5.6 percent of the population. In 1950, it did better, missing only 7 million, by its estimates. In 1960, the undercount was 6.2 million; in 1970, 6.1 million, and by 1980 the estimated undercount was 3.2 million, or 1.4 percent.

But the bureau has never felt that it could correct the undercount by statistical formulas and still preserve an accurate portrait of the population.

In recent years, more and more prominent statisticians, some from the bureau's own ranks, have begun to disagree.

Statisticians argue bitterly concerning whether the statistical adjustment sought by New York and other cities will make the census more right, or more wrong.

''The Census Bureau could have done a better job,'' said William F. O'Hare, director of the Population Reference Bureau, a Washington-based consulting firm.

''But I give them good marks,'' he said. ''There was bound to be an undercount. There are changes in society, people moving around, homelessness, people resisting government instrusion.''

Because of those factors, he said, ''I lean toward adjustment as a matter of equity, if it makes the census more accurate. ''

But David Freeman, a professor of statistics at the University of California at Berkeley, questions whether an adjustment would do that. ''We've had missing data in the past,'' Freeman said, ''we're going to have missing data again.

''That's not a good basis for adjustment. I want to see some solid evidence that it makes things better.''

Adjustment would certainly make things better for some states and their officials. California, Florida and Texas, already big winners in the population sweepstakes during the past decade, also have a disproportionate share of the minority populations whose numbers would be enhanced by adjustment. Adjustment would probably mean these winners win more.

For New York, Illinois and Michigan, all of which saw only small population gains and all of which are losing congressional seats, ''the losses would be reduced,'' if the final figures were adjusted, O'Hare said.

But the political power and federal aid being divided would not expand, so the gains by those three states would be somebody else's loss. The most likely losers would be states that have already lost population and power in the past decade, like Iowa, Kentucky, North Dakota and Wyoming.

If the Commerce Department rejects the adjustment option, New York and the other cities and organizations representing minority groups will go back to court, and members of Congress from the losing states and cities will sign on to bill after bill to give them back some measure of what they have lost.