While it is true that "like clings to like," still this does not imply kinship, but it very often implies likeness in mental characteristics. This tendency has been shown to be present in the neurotic by Roller,[122] De Monteyel,[123] Kiernan,[124] Bannister,[125] and Manning,[126] so far as Germany, France, the United States, and Australia are concerned. Bannister puts the statistical proof of this tendency very forcibly as follows: "There are in Illinois, according to the most recent estimates, in round numbers, about 6,000 insane, or one to a little over 500 of the population. Even if we double, treble, or quadruple this frequency to include all that have been or are to be insane, as well as those insane at the present time, it would not appear that there was much probability of two insane persons being married according to any ordinary law of chances. In fact, we find four out of the 104 with insane heredity had both father and mother insane. In one of these cases the insane heredity involved both parents and both grandparents on each side, though in the case of the latter the histories show it only as collateral. Besides these, three patients had direct paternal and collateral maternal heredity; two had direct maternal and collateral paternal heredity, and in one case there was collateral heredity of insanity on both sides. This makes altogether nearly 10 per cent. of those with insane heredity with it on both sides, maternal and paternal, and thus favoured with a double opportunity to inherit mental disease. If we add to this the instances where, with insanity of one parent, there is reported either epilepsy, hysteria, or drunkenness, 'brain disease,' nervousness, &c., of the other, the ratio of double inheritance rises to over 20 per cent." The beneficial effects which may result from atavism are, it will be obvious, offset by this tendency of the neurotic to intermarry, thus perverting the principle of atavism to the assistance of degeneracy.