Paul Smith slips into the office of a well-known scientist, has a brief conversation and slips out again. He takes a walk around the block and returns only when the scientist places a rubber band on the office doorknob, signaling that the material Smith seeks is ready.

Such intrigue may befit a spy, but Smith is a geneticist. He has been leaving the offices of some of the finest scientists and mathematicians in the country carrying not their latest formulae, but their sperm. With it he hopes--through the cooperation of increasing numbers of interested American women--to conceive unusually intelligent babies who will help curtail what Smith sees as the genetic degradation of the human race.

For the last two years, Smith has been living his undercover life as the research officer for the Repository for Germinal Choice--sometimes called the "Nobel sperm bank"--a unique and increasingly controversial institution located on a wealthy pharmacist's ranch in Escondido, Calif.

In an era shaken by genetic engineering and test-tube babies, the repository has become a lightning rod for fears that the social consequences of modern biology have gotten completely out of hand. Already the unique sperm bank has produced three pregnancies and received calls and letters from more than 200 willing women, all because of a theory of race still widely disputed in scientific circles.

Smith and his millionaire employer, repository founder Robert Graham, argue that normal rates of undesirable mutation in human beings and the ability of modern medicine to keep severely handicapped people alive threatens to degrade the health and intelligence of all future generations. Smith said he and Graham decided to heed the words of the late Nobel laureate Hermann Joseph Muller that "if the human species was to keep from regressing, natural selection had to be replaced with artificial selection."

Their unique scheme has since been denounced as a harebrained plot to create a master race by collecting and freezing sperm from a few demonstrably brilliant scientists, including three Nobel prize winners. This view gained greater currency recently when the first woman to give birth to a child fathered from repository sperm was discovered to have been charged with abusing her two older children to make them smarter.

Others, including the many women who have sought help from the repository in the last few months, praise the institution as a godsend for wives who want bright and healthy children but whose husbands are infertile. Some also laud Smith for his willingness to let them select their own donor from detailed profiles, including photographs, something few of the nation's more than 17 other sperm banks will allow.

There are approximately 20,000 to 30,000 artificial inseminations in the United States each year, of which between 2,000 and 4,000 result in pregnancy, according to Dr. Cappy Rothman, head of the Southern California Cryobank, a sperm bank.

But in the scientific community, the questions raised by the Nobel repository plunge very quickly into a central debate of the 20th Century: is man formed by his heredity or his environment? At this stage of scientific inquiry, the answer appears to be: inherited intelligence is important, but it can be wiped out by bad upbringing.

"We know from adoptive studies that the children of more intelligent natural parents, those children in good adoptive families tend to be more intelligent than the children of less intelligent parents," said Sandra Scarr, professor of psychology at Yale University. But she said she agrees with Richard Snow, professor of education and psychology at Stanford College, that unhappy home life with little intellectual stimulation can keep a child from ever using the intelligence he carries in his genes, and institutions like the repository can do little about how the children who result from their work are raised.

Lee Cronbach, emeritus professor of education at Stanford, said studies have been unable to determine how much of intelligence can be attributed to heredity, and how much to environment, since the children of intelligent parents are so often raised in confortable, stimulating homes. The only direct, provable links between intelligence and heredity, he said, involve inherited mental deficiencies.

Scarr said it was "unfortunate" for a sperm bank like the repository to try to single out a small group--in this case brilliant scientists--as sole donors of sperm since there were many other good sources. It is "sensible," she said, for infertile parents to seek donors with good brains and bodies, but she and other childhood intelligence experts said there is no evidence that the human race is going downhill intellectually, the fear that inspired the California repository in the first place.

Cronbach and Scarr said British scientists had been studying the possibility of genetic degradation for many decades, because of an early concern that the British working class was producing far more children than the British upper class. The studies appear, however, to show no decline in the average intelligence of British children in the most recent generation, and the birth rates at all strata of British society have in the meantime fallen to about the same level.

Smith, 41, with a degree in medical genetics from the University of California at Berkeley, insists this scientific view confirms the value of his efforts. The children produced from the repository's sperm "are not going to be put into sensory deprivation," he said. He tends to shrug off the controversy a series of one-liners. A problem that does concern him, he said, is wives who veto their brilliant husbands' generous desires to contribute sperm to the repository. One wife who agreed to let her husband go ahead said: "If he is going to inseminate other women, it is best that he do it artificially."

Despite the publicity won by the Escondido bank because of its Nobel donors--including transistor inventor William Shockley--more than 90 percent of the repository's female applicants request other donors, who are not Nobel laureates, Smith said.

Smith said the repository has sperm donations from about 12 men, although only six of those are in quantities large enough at present to be useful. The supply has built up slowly, which has until now proved no problem because for at least a year after Smith went to work for Graham in 1980, no women applied. Smith said the repository has provided sperm for 20 to 30 inseminations, which so far have produced one known birth, two additional pregnancies and one miscarriage.

Graham, 76, made his fortune developing a popular form of hard plastic lens for eyeglasses. With Smith's help, he began collecting sperm from men known to be leaders in their scientific fields and one unusually successful financier. He keeps it frozen in liquid nitrogen in a radiation-proof container inside a small storeroom on his ranch near San Diego.

Graham's money makes it possible to provide "the only non-commercial sperm bank I know of," Smith said. Donors are not paid and female recipients are charged no more than the transportation costs of the sperm. Donors fill in a long questionnaire so that Smith can screen out those from families with histories of heart trouble and other diseases: "We certainly do not want to hand out a pacemaker with each baby," Smith said.