Suppose that in 1908 President Theodore Roosevelt had been asked to predict every major development in the American economy over the next seven decades and to lay out economic policies to steer the nation through every crisis until 1983.

Would he have foreseen the Great Depression, two world wars and the war in Vietnam? Would he have predicted the baby boom and double-digit inflation?

Surely not. Yet that kind of stargazing is precisely what the actuaries of the Social Security System are asked to do. They must give Congress an estimate of how much Social Security will gather in taxes and how much it will spend for benefits over the next 75 years.

The uncertainty of such projections is the reason Social Security may again face bankruptcy somewhere down the road, despite the massive financial rescue package approved this year to keep the system healthy for the next 75 years.

Very slight changes in long-range fertility rates, death rates, inflation and productivity could alter the system's financial picture radically.

One of the most disputed issues is the long-range fertility rate of American women. Social Security long assumed American women would have an average of 2.1 children.

Last year the Census Bureau predicted that the long-range fertility rate probably will be 1.9 children per woman. But the bureau, conceding that it had failed to foresee either the post-World War II baby boom or the "baby bust" of the 1970s, hedged its prediction and said the rate could be as high as 2.3 or as low as 1.6.

The fertility rate has an enormous impact on Social Security. A higher rate would mean more future workers paying into the system and a better financial picture.

"It's very difficult to project," said Harry Ballantyne, the system's chief actuary.

After considerable discussion, Ballantyne and his colleagues decided this year to use a fertility rate of 2.0 children per woman as one of the central assumptions of their intermediate projections. (The intermediate scenario is the one considered the most likely to occur and most often used as a guide for policy.)

Assuming a fertility rate of 2.0, the intermediate scenario shows the old-age, survivor and disability trust funds having a tiny surplus over the next 75 years.

However, Robert J. Myers, chief actuary of the system from 1947 to 1970 and staff director of the presidential commission that devised the Social Security rescue package, said the figure may be a bit pessimistic. He said he would have stayed with 2.1 for a while longer "to see more experience." That rate would produce a slightly larger surplus.

But Haeworth Robertson, chief actuary of Social Security from 1975 to 1978 and now managing director of a large actuarial firm, said, "I find it hard to believe the fertility rate will rise" even as high as 2.0.

Ballantyne said his office also calculated projections using all the intermediate assumptions, except the 2.0 fertility rate. It used a 1.6 fertility rate instead. With that fertility rate, instead of a tiny surplus over 75 years, the old-age, survivor and disability trust funds would have a long-term deficit of 0.9 percent of taxable payroll, about $14 billion a year in current terms.

Mortality rates are another crucial unknown.

The Social Security intermediate scenario now assumes that life expectancy for women, which was 77.8 years in 1981, will rise to 84.4 by the year 2060; and for men, from 70.3 to 76.3 during that period.

However, many old-age specialists, such as Eileen Crimmins of the University of Southern California, have said they believe life expectancy will rise more rapidly. That would mean people would live to collect more Social Security benefits, so the financial position of the system would be worse than now projected.

Ballantyne said that if all other factors in the intermediate projections were kept constant, but life span increased as much as forecast by Crimmins, the projected long-range surplus would turn into a substantial deficit, about 0.9 percent of taxable payroll.

Another variable is the growth of fringe benefits, which are not now subject to Social Security payroll taxes.

Wages as a percentage of total compensation have been declining rapidly for many years with the growth of health and retirement plans for workers. In 1982, Ballantyne said, wages constituted 84 percent of total compensation.

Ballantyne is assuming that figure will not continue declining as rapidly as it has in the past. Even so, the intermediate scenario assumes that it will drop to 66 pecent by the year 2060. If it instead drops to 61 percent by then, and all other factors in the intermediate scenario were unchanged, the long-range surplus would disappear.

Another crucial factor inflation. As a rule of thumb, the present system can stay solvent only if wages increase faster than prices.

The intermediate scenario assumes a long-term real-wage growth of 1.5 percent a year. This is slower than in some periods of the past, but is considered a prudent figure.

Even so, Myers and Robertson said they fear that it may be high. Robertson would like to see a 1 percent rate used. If that turns out to be the correct figure and no other changes occur in the intermediate scenario, the projected surplus once again would turn into a substantial long-range deficit.

On the other hand, if real wages rise at an annual rate of 2.5 percent, the long-term surplus would be huge, about $21 billion a year.

What all this means is that there is no way to insulate Social Security totally from unpredictable events. Despite assurances that Social Security is safe for the next 75 years, it is likely that Congress will have to reopen the subject more than once between now and then.