On Germinal Selection as a Source of Definite Variation