Boston — Two years ago Cornell economist Thomas Poleman suspected something fishy about commonly accepted estimates of how many people in the world go to bed hungry at night.
He began poking around in estimates made back in 1974 when the UN Food and Agriculture Organization (FAO) was preparing for its World Food Conference.
To his amazement, Dr. Poleman discovered that the FAO had jacked up its estimate just in time for the conference. The earlier estimate of 360 million was hiked to 434 million - a jump from exactly 20 percent of the poor world's population to exactly 25 percent.
''One suspects . . . a certain measure of arbitrariness,'' Dr. Poleman concluded with dry understatement in a report critical of various United Nations estimates.
Since then a growing cadre of more wary, independent analysts have also been taking aim at the hunger statistics of organizations like the FAO and the World Bank.
They agree that there are many millions who are hungry and who lead day-in, day-out lives of misery. But the numbers of seriously malnourished people, they say, may be far more modest than the impression given in recent years by widely publicized high figures.
Most quoted has been the figure of 500 million -- a rounded-off version of the FAO's latest estimate that 440 million people do not get the minimum of food they need.
World Bank analysts go by a figure more than twice that high - upward of a quarter of mankind.
But university researchers at Cornell, Harvard, and Stanford now think that a more balanced picture looks like this:
* The number of people at the very ''bottom'' -- facing the severest forms of hunger -- may actually range from 100 to 200 million.
* The figure might, indeed, be as high as 500 million if one includes people so preoccupied with the daily search for food that they cannot make much more out of their lives.
* Between 500 million and 1 billion probably face occasional malnutrition, have a constantly boring diet, and could have their condition greatly improved by economic reforms in their countries.
Researchers who gravitate to that kind of picture make no claim that they have more precise data than the FAO or World Bank.
''Precise data'' on hunger simply does not exist. No one is out in the field taking head counts. Hunger is exceedingly hard to define or measure, and the perception of it varies from country to country.
In the end, all researchers must make educated guesses of the extent of hunger based on food availability in poor countries, patterns of its distribution, infant mortality rates, and/or the numbers of people thought most vulnerable - preschool children, pregnant women, and the elderly.
The FAO and the World Bank, for their part, are the first to admit the tentativeness of their own conclusions.
In fact, when coming up with its 1 billion figure, the bank's researchers were not even trying to determine how many people are seriously hungry, but rather how many do not get enough food to lead active lives - which is why their estimate is so much higher than the rest.
But for many analysts, a more precise understanding of the extent of the number at the very bottom of the hunger ladder could have important implications.
If it is, indeed, in the 100- or 200-million range, it could mean that the task of reaching the most needy is more manageable, argues Nick Eberstadt of Harvard's Center for Population Studies (who gravitates to the 100-million figure).
A more modest estimate could also require a shift in the focus of international aid efforts, says Walter Falcon, director of Stanford University's Food Reseach Institute (who gravitates to the 150- or 200-million figure).
''If you take the number of most severely affected to be 100 or 200 million, Africa would take on relatively more prominence,'' he says. ''Although the largest number of malnourished live in Asia, they are lower down in the scale of need.''
Despite the debate over numbers, all the analysts seem agreed that the hunger problem should not be taken lightly.
''Even if the number is only 50 million, that's enough to bother me,'' says Harry Walters of the World Bank. ''Fighting over numbers should not become a justification for thinking there's nothing to worry about.''