bivalent and, since four potential gametes result from each meiosis, these strands pass to different gametes and may, genetically, be dealt with separately. The only concern, therefore, is with the proportioris of each type of strand resulting from crossing-over and their possible combinations in the zygotes. Allopolyploids, such as wheat, show the same general type of genetical behavior since multivalent formation is rare. In autopolyploids, on the other hand, more than four chromatids are concerned in the formation of the various configurations which occur at the first division of meiosis, and, since there are still only four gametes formed, each gamete receives more than one strand from each configuration. Hence with autopolyploids there is the additional consideration of the ways in which the chromatids may be combined in the gametes themselves. This leads to a number of complications in the analysis of linkage in autopolyploids. MULLER (1914) calculated the monofactorial segregations to be expected from autotetraploids if the two chromatids of each chromosome acted as one at the first meiotic division, and HALDANE (1930) extended the work to include several higher autopolyploids. This mode of inheritance is called chromosome segregation. HALDANE (1930) also established the segregations to be expected if the two chromatids from each chromosome acted entirely independently of one another. Random chromatid segregation is the term applied to this method. SOMME (1930) reported some linkage studies in autotetraploid Primula sinensis and DE WINTON and HALDANE (1931) in a more comprehensive investigation of the subject, using the same plant, gave the linkage formulae applicable to the chromosome type of segregation. SANSOME (1933), working with the tomato, presented the corresponding linkage formulae on the basis of random chromatid segregation. In addition to these cases of linkage studies in autopolyploid plants, investigations on crossing-over in triploid Drosophila melanogaster have been reported by BRIDGES and ANDERSON (1925) and REDFIELD (1930, 1932). Later workers ( MATHER 1936) have recognized that these segregation and linkage expectations are in the nature of limiting types, and that the true segregations to be expected TF irst division of meiosis four chromatids take part in the formation of each