Abstract
1. Selection for long or short wings was carried out in three pairs of lines of which two were started from the same basic population. At the end of the selection experiment the frequencies of the chromosomal arrangements were determined in each line and compared with the corresponding original frequencies to see if the chromosomal polymorphism of Drosophila subobscura is related to the genetic variation of size in a regular way.2. The three pairs of lines show an asymmetrical response to selection. Selection for long wings is less effective in changing the mean of the population than selection for short wings.3. Heritability is lower in the lines selected for long wings, especially in the later generations of selection.4. Viability shows a very small decrease during selection in most selected lines.5. In no case does the coefficient of variability show statistically significant changes, but there is a general tendency for it to decrease.6. Selection for long wings favours combinations heterozygous for the standard chromosome orders and specific complex inversion orders. Selection for short wings generally fixes in homozygous combination specific complex inversion orders.7. The heterozygous combinations of the standard chromosomal orders, which are most frequent in northern populations, when combined with the complex inversion orders which are most frequent in southern populations, have a heterotic effect on size.8. The complex inversion orders usually fixed in homozygous combinations in the lines selected for short wings are those most frequent in southern populations which have a smaller mean size than northern populations.9. The blocks of genes integrated in ‘supergenes’ in the inverted segments of the chromosomes of D. subobscura differ in genetic composition with respect to the control of wing length.