Another supposed piece of evidence which has deceived a great many students is the investigation of Bezzola into the distribution of the birth-rate of imbeciles in Switzerland. He announced that in wine-growing districts the number of idiots conceived at the time of the vintage and carnival is very large, while at other periods it is almost _nil_. The conclusion was that excesses of drunkenness occurring in connection with the vintage and carnival caused this production of imbeciles. But aside from the unjustified assumptions involved in his reasoning, Professor Pearson has recently gone over the data and shown the faulty statistical method; that, in fact, the number of imbeciles conceived at vintage-time, in excess of the average monthly number, was only three in spite of the large numbers! Bezzola's testimony, which has long been cited as proof of the disastrous results of the use of alcohol at the time of conception, must be discarded. Demme's plausible investigation is also widely quoted to support the belief that alcohol poisons the germ-plasm. He studied the offspring of 10 drunken and 10 sober pairs of parents, and found that of the 61 children of the latter, 50 were normal, while of the 57 progeny of the drunkards, only nine were normal. This is a good specimen of much of the evidence cited to prove that alcohol impairs the germ-plasm; it has been widely circulated by propagandists in America during recent years. Of course, its value depends wholly on whether the 20 pairs of parents were of sound, comparable stock. Karl Pearson has pointed out that this is not the case. Demme selected his children of drunkards by selecting children who came to his hospital on account of imperfect development of speech, mental defect, imbecility or idiocy. When he found families in which such defective children occurred, he then inquired as to their ancestry. Many of these children, he found, were reduced to a condition approaching epilepsy, or actually epileptic, because they themselves were alcoholic. Obviously such material can not legitimately be used to prove that the use of alcohol by parents injures the heredity of their children. The figures do not at all give the proof we are seeking, that alcohol can so affect sound germ-plasm as to lead to the production of defective children.