Girls and boys are not equal in mathematical thinking, according to a study of 10,000 gifted students tested over a period of eight years. Boys as a group show a large superiority as early as age 9.

The work, published today in Science magazine, is a major volley in the continuing academic battle over the extent to which boys and girls differ naturally, and the extent to which they become different at the hands of society.

This battle has focused recently on why boys in general do better than girls on tests of mathematical ability. Some reseachers have contended that these apparent differences result wholly or in great part from sex-biased training girls get at home and school which leads them to think math is a "boy's subject" and not very feminine.

The new research challenges that view. Camilla Benbow and Julian Stanley, Johns Hopkins University coauthors of the study, concluded from their data that "sex differences in achievement in and attitude toward mathematical ability. . . ."

They said that this superiority could be caused by a number of things and is not necessarily genetic. It probably results from a combination of genetic and social factors, though social factors may be less important than previously believed.

However, Dr. Elizabeth Fennema of the University of Wisconsin, an education researcher who has also studied the subject, said the whole question -- while it may make interesting research -- has no practical importance.

"the overlap in girls' abilities and boys' abilities is very great, so almost all females can learn almost as well as all males. And few of us come near our potentials anyway. So in practical terms, genes are not the things determining whether women will succedd in engineering or some other subject," Fennema said.

But Sheila Tobias, author of the book "Math Anxiety," a study of girls and math ability, countered that perceptions are important because they influence the way girls are treated in school. "It is a political issue. This will influence the behavior of the educational community. . . . I am in this research to get at the truth, but also for a political motive," Tobias said. "I want to get the country to accept this premise: that potentially girls can and boys can perform equally."

The authors of the Johns Hopkins study said they undertook the research in order to make clear that there is a large difference in the math performance of girls and boys and it ought not to be ignored.

"It is like the parable of the boy who wants to play on the basketball team but who is very short," said coauthor Benbow. "The boy argues that he should make the bakektball team because he would have been tall except for a childhood disease he had. And being potentially as tall as anyone else, he should make the team. This same kind of argument has been used to ignore the differences between girls and boys, and that doesn't help the girls."

At age 12, boys in the new study were found superior to girls by an average of about 40 points, out of an average of 475 scored. Among the top-scoring girls and boys there were even greater differences -- as high as 190 points. The test used was one designed to measure reasoning ability in mathematics -- the Scholastic Aptitude Test in Mathematics -- and not to measure mathematics knowledge or computational ability. Gilrs have proved equally good at both those other tasks, the researchers said.

Critics have questioned the new research since the study was done among gifted boys and girls, and they speculate that it may not apply to other children. They also ask whether "mathematical reasoning" can be defined or tested.

The roots of this skirmish began some years ago when tests of high school students were used to show that girls were inferior mathematical thinkers. Fennema, among others, attacked that conclusion.

She wrote a paper suggesting that since boys take more math, they would naturally show up better in comparison to girls who hadn't had so much. She said that if one took away the differences in the courses taken, girls would be equal or at least more nearly equal.

Today's report, according to author Stanley, is intended to refute Fennema's hypothesis and similar views of feminist educators who, he said, have ascribed the whole boy-girl difference to social factors.

That hypothesis in 1976, we have found that even if girls do take as much math, some differences between boys' and girls' scores remain. Problem-solving differences between boys and girls have shown up in other studies as well. I'm very troubled by that.

"But the genetic question is really not important enough even to answer; you may beat your head against that question and never come to a conclusion," Fennema said. She suggested that more important is how well boys and girls are taught math, so that both might succeed.