Multiple Factors and Variation

MULTIPLE FACTORS AND VARIATION When a variable species is adequately analysed on Mendelian lines a very large number of genes are found to affect any char acter considered. Thus in Primula sinensis twelve genes affect the flower colour, apart from those determining the size and shape of the eye; in Drosophila melanogaster 103 genes are known which have marked effects on the wings. When almost all the variations in a population are due to one or two genes and the environment, analysis is generally easy. When a large number are present it becomes quite impossible except as the result of prolonged experiment.

Multiple Factors.

While in plants single genes often de termine the difference in size between races (as in the case of the tall and short sweet peas) this is rarely so in domestic ani mals. Thus the difference in size between the Hamburgh and Sebright bantam fowls appears to be due to four genes, of which the Hamburgh possesses three, A, B and C, and the bantam one, D, all in the homozygous form. They are all incompletely domi nant. The F, is intermediate between the parents, and uniform. The is very variable and includes the genotypes AABBCCDD and aabbccdd, which are larger and smaller respectively than either parent race. When two pure-breeding races are crossed, something of this kind usually occurs.

Correlation.—In a species such as man where inbreeding is rare, variation in stature is apparently continuous, though the atomic constitution of matter makes it fairly clear that the number of possible phenotypes is finite. Provided that mating in such a population is at random, or alternatively that the intensity of assortative mating is known, a great deal may be found out regarding heredity in human populations by statistical methods. The resemblance between parents and children, brothers and sis ters, and other relatives, can thus be estimated. Great care must be taken to exclude the effect of similar environments in increasing such correlations. But for example among the richer classes in England, where rickets and underfeeding are rare, the environment is relatively unimportant in determining stature. In the case of a number of human characters, such as stature, span, forearm length, the coefficient of correlation between parent and child ranges from .51 to .42, between sibs from .54 to .46. If we consider the variance (see VARIATION AND SELECTION) of a population if the parents (and hence all ancestors) are given, the variance is diminished to about 46%, which is mainly due to segregation, and not to differences in the environment. In other words differences in ancestry and segregation within families have about equal effects in producing variance. While the pre Mendelian genetics do not account for segregation, Mendel's prin ciples do so. It has been shown that the coefficients of correlation found for human stature are quantitatively predictable on the hypothesis that it is the resultant of very numerous genes. On the other hand until these genes can be determined, statistical methods are the only ones available for the analysis of this kind of in heritance.