Applies the confluence model to a longitudinal intact family data set. Also tests the data with simple linear models. Results suggest that the confluence model should be rejected as a useful model for these data based on principles of parsimony and predictability. (Author/AS)

Persons representing the extremes of intelligence cannot be included in the same study unless they are approximately equated on mental age, in which case the relative performances of the extreme groups can provide useful information about the nature of intelligence. (Author/RD)