The concepts of dominance and recessivity as applied to human genetical traits are losing their original connotations. It is better to speak of genes unconditionally deleterious in heterozygote form and those for which such disadvantage is doubtful. The classical dominant traits come into the first category and in the large majority they give rise to grossly pathological conditions and are ipso facto harmful-many in fact are lethal. In the case of these genes it is difficult to conceive that any agency other than recurrent mutation is instrumental in keeping them at their present frequency.

Assuming that such mutation rates are of the same order of magnitude over the entire spectrum, the frequency of these traits will vary inversely with their severity and this is found to be approximately true-from an incidence of 3/million for an almost invariably lethal trait such as acrocephalosyndactyly to 1/1000 for a relatively innocuous trait such as ptosis.

We would not expect genes even at the higher of these frequencies to be often seen and recognised in homozygote form with the low degree of inbreeding prevalent in human populations. This is in fact confirmed in practice and examples of possibly homozygous forms or dominant conditions are very rare. To be recognized as causing a recessive trait, with a few exceptions in inbred communities, a gene must be more common and often cannot therefore be maintained by recurrent mutation alone. Very wide geographical variations in the incidence of such traits also militate against such an interpretation. Phenylketonuria is an excellent example. The abnormal gene reaches a frequency of as much as 1% in some populations and it is unknown in others.