I have mentioned the irregularity of heredity of nervous diseases and tried to account for it. The apparent complexity of human pathogenic gene is caused by considering a number of mutations which have different origins as disease unit. Generally speaking, pathogenic gene may be regarded as normal gene whose functions are degenerated or lost through mutation. So instead of Holmes's formula c=f(G. E), I have found my own, c=f[(P. N. gM)E](P: pathogenic gene, N: normal gene, gM. genotypic milieu). Don't P and M stand rather in the relation of compensation than in that of dominance? Thus in the same disease we find some gene dominant and some ressesive, Expressivity of skipping or hetero and homo may be looked upon as mere difference of quantity. The individual difference of the extent of the site of the pathologic changes is vast, ranging from "formes frustes" to "homo lethal". About these phenomena I have tried to give a systematic and consecutive explanation. As disease examples, I have mentioned, including ones of my own examination, Huntington's chorea, Friedreich's ataxia, athetose double, dystrophia progressiva, etc.