Interaction of sex, shape, and weight genes in watermelon.

Several workers on inheritance in watermelon {Citrullus vulgaris Sclirad.) have independently reported relation between fruit shape and sex and between fruit shape and weight. The reported relation between shape and weight, if consistently true, will have some economic importance in view of the present demand for smaller highquality watermeloas. Weetman (IS)^ found that shape inheritance when represented quantitatively by an index computed by dividing equatorial by polar diameter may be classified qualitatively to show a phenotypic relation of one pair of alíeles, round vs. elongate, withFi phenotype intermediate and F2 segregating into 1 round to 2 intermediate to 1 elongate. The same author found a highly significant negative correlation of low magnitude between fruit shape and weight, suggesting that long fruits are heavier than round ones. Rosa (9) reported that the monoecious sex habit was completely dominant over the andromonoecious by one pair of alíeles and that andromonoecious plants produced fruits that were more nearly round. F2 populations classified for weight frequently show positive skewness such as is found in geometric progressions. Many reported investigations of this kind are reviewed by Powers (8), MacArthur and Butler (^), and Charles and Smith {2)\ Powers (8) has developed formulas to calculate the genetic and envhonmental variances for segregating populations from parental and Fi variances. Thus far little attention seems to have been given the possibility of estimating the number of genes involved in such segregations. The ideal analysis is to separate weight into its component parts, as Powers (6, 7) and Sinnott (10) have done in comparatively simple cases. With complete information, however, any given case may disclose some groups of genes operating with geometric effects and others with arithmetic effects. Student {12) presented a gene analysis of Winter's data (14) on the results of mass selection for high and low oil content in an openpollinated maize variety. By using a genetic standard deviation derived from the given gross uncorrected deviations for 28 years, Student estimated that the number of genes segregating for oil content was probably of the order 20 to 40. Goodale (3) studied weight in the albino mouse after positive selection for about 7 years and estimated that at least 32 genes were segregating. Butler (Í ) studied ovary weight in the tomato and estimated the segregation of 6 genes because he was able to obtain a good-fitting binomial with the exponent 6. Goodale and Butler used the uncor-