AS THE ANNUAL struggle over the federal budget gets under way, it is worth pondering what will happen to money for science. American science and technology are still the best in the world. But there are enough signs of strain to suggest that our accustomed preeminence--on which a large part of U.S. security and economic power depends--is fragile, even endangered.

The trouble starts with education. For more than a decade, secondary school curriculum requirements and achievement have fallen sharply in science and mathematics, while an opposite trend has been present in most other developed countries. The result is already evident. A NASA official reported recently, for example, that the space agency's cost overruns come in part from delays that are, in turn, the result of a lack of technically skilled workers.

Federal support for graduate education is in doubt for the first time in 30 years. In many fields, engineers are in short supply, but engineering schools cannot take in more students because they cannot find trained faculty to teach them. Shortage of faculty means heavier teaching loads and therefore less research. Schools do not have enough money to pay more professors even if these could be found, nor can they replace badly obsolete laboratories. Troubles that now afflict engineering are beginning to be seen in the sciences as well.

Money for basic research in this country has been essentially constant for 10 years. To the extent that scientific advance is linked to money--there is a close, but not rigid relationship--that means a decade without real growth. Meanwhile increases in research funds in Japan, West Germany, France and elsewhere have paid off with growth in both scientific and industrial productivity. And now federal research budgets face severe cuts.

Basic research, a long-range investment for the benefit of all of society, is properly and necessarily the responsibility of the federal government. Industry can be asked to expand its support of applied research and of development projects, but it is not industry's role, nor is the industrial setting the best environment for basic research. Yet less than 15 percent of federal research and development funds currently goes to basic research. Too much federal money supports development projects that are the proper province of industry. Allen Bromley, president of the American Association for the Advancement of Science, recently made the valuable suggestion that the ubiquitous but misleading term "R&D" be dropped, in order to separate the financing of these two very different activities.

If serious damage is to be avoided as the federal budget is cut, Congress and the administration should not only protect, but in some fields increase, basic research funds. Ways should also be explored to assure more continuity in the amount of support such research is given. It takes nine years to produce a PhD scientist, and years to assemble a research team and carry a project to fruition. When the money disappears for a few years, the people disappear too, and can seldom be brought back. Abrupt changes like those that took place during last year's budget cycle can therefore wipe out years of past investment and future productivity.