Twins and other kinship studies
N. J. MACKINTOSH
BURT'S PAPER 'The genetic determination of differences in intelligence: a study of monozygotic twins reared together and apart', published in 1966, provided the most systematic account of his data on kinship correlations for IQ, including what was at the time much the largest sample of separated identical (MZ) twins ever reported. Earlier papers had mentioned these data in passing, in the context of other arguments. The 1966 paper was the first whose central purpose was to present these data in their entirety with new, larger samples, to provide information about how they had been collected and analysed, and in effect to summarize his life's work on this issue. Although Burt continued to publish further papers on the heritability of IQ, he published no new data. The 1966 paper may be taken as his final, magisterial summary. At the same time, a reading of the paper makes it rather clear that it was written, at least in part, as a rebuttal of the arguments of various armchair critics who claimed, according to Burt in the absence of any evidence, that environmental factors were largely responsible for individual differences in IQ. Burt took evident pleasure in presenting data to refute their position. As if in revenge, his twin data have provided Burt's later critics with the basis for some of their most dramatic, and widely publicized, accusations of fraud.
The first charge was that although, over the years, the number of separated MZ twins in Burt's sample more than doubled, the correlations in their IQ scores often remained constant to the third decimal place (this was equally true for some other kinship categories). This was first noticed by Leon Kamin, who publicized his observations in several lectures given during 1972, and wrote to Arthur Jensen in that year to draw his attention to this remarkable feature of Burt's data. Acknowledging Kamin's role, Jensen himself published a systematic survey of all Burt's kinship correlations for IQ, and reported that he had found some 20 correlations that