The turn of the century saw the introduction of psychometric tests of intelligence to the United States, mostly due to the efforts of H. Goddard and L. Terman. Both of these psychologists were hereditarians, who believed that the genetic composition of individuals was the primary determinant of performance on these tests. A major thrust of work in those dark days was directed toward showing that individuals from certain racial groups (primarily Southern and Eastern Europeans) were of inferior intelligence; together with the assumption that this was due to their genes, certain social policies such as the restriction of immigration were effected (see Gould 1981).There was, however, a major stumbling block in reaching these conclusions from the data that existed (mostly results of massive IQ testing of World War I recruits): many of the testees were not native speakers of English, yet they were tested in English. Thus, whether these individuals were suffering from a "language handicap" in testing became a major issue. Hereditarians, such as Terman's student Kimball Young, argued that there was no language handicap. Those in favor of environmental determinants of IQ stressed the language handicap.The social context in which this early research on bilingualism and intelligence was conducted is radically different from the context of the more recent studies of the matter, basically modeled after Peal and Lambert's (1962) well-known study. These different contexts are primarily responsible for the differences in the focus of the research and the methodologies adopted. From our current perspective, the methodology of the earlier research was wildly deficient, failing to control for a variety of obvious variables such as socio-economic status (SES). But the earlier researchers assumed that low SES was the result, not the cause, of inferior performance on IQ tests, so this questionable methodology passed the inspection lines of science (a review of this history will appear in Hakuta, forthcoming).We contemporary researchers are not, however, so objective either. I think 679