With at least four apologies on four successive days, Harvard President Lawrence Summers has begun incrementally removing his foot from his mouth. Last week, Mr. Summers theorized at a conference that the lack of women in high positions in science and engineering was due partly to innate differences.

No transcription is available of Mr. Summers' exact remarks, but two pieces of his speech have been cited repeatedly. Mr. Summers argued that some innate differences between men and women were evidenced by his daughter's inclination to treat her toy trucks as if they were dolls. While a charming anecdote, unless his daughter had a problem counting her trucks, it's irrelevant to science or math.

Mr. Summers also referenced the relative lack of females scoring at the top of college entrance exams. He's right about the data: According to a 2004 report from the College Board, about twice as many males as females scored between 750 and 800 on the math section of the SAT. But he's wrong about assuming innate differences as the cause. By the time students reach late high school age, surely the nature-nurture continuum has advanced to the point that the word "innate" has grown a little stale.

Data printed elsewhere in the same report demonstrates the danger in Mr. Summers' line of thinking. One statistic, for example, indicates that 62 percent of females have an A average in high school, versus 38 percent of men. Would Mr. Summers take that to mean that boys are innately bad at school? And venturing beyond gender differences, the report shows that African Americans average about 200 points less than white students on their total SAT score. What conclusions would Mr. Summers draw about racial intelligence?

Mr. Summers initially defended himself by saying that his remarks were intended to stir debate. Instead, with a half-brained theory, he only stirred up controversy. When the president of a prestigious university presents a theory at a conference, he (or she) should rely on valid scientific evidence to back it up.