EMPIRICAL studies casting doubt on the IQ as an adequate summary of an individual's intellectual capacities1,2 have lent support to the increasing use of tests of “divergent thinking”, and to attempts to show that such tests are usefully related to academic achievement, especially in science3,4. These tests are distinguished from more conventional tests of intellectual capacity by the fact that they are designed to measure originality and flexibility. The older tests, by contrast, give greater emphasis to the finding of strictly logical, “correct” solutions. Two chief points about the newer tests can be emphasized: the first concerns their doubtful long-range validity, because it has not been shown that they correlate with later achievement in the way that ordinary IQ tests do5, and the second centres on Hudson's report6 that sixth form science specialists in British schools are usually convergent in their intellectual bias.