This work conclusively demonstrates that crime runs in families. The probable value of the correlation between conviction for crime on the part of the father and son was found to be .60, while the correlation between mother and son was only slightly less. The tendency to resemble brothers in criminality was shown by the probable fraternal correlations of .45. Whether this family resemblance is mainly through nature or nurture is the problem. In analyzing the influence of the home he uses partial correlation and finds that the correlation between age at first conviction and the number of convictions for a constant period of time after the first conviction is -.243. “From the value and sign of this coefficient, we see that the earlier in life a child commits a criminal offence, and is consequently removed from his home, the worse criminal does he become; and, accordingly, we conclude that criminal proclivities are more bred in the home than inoculated there” (_119_, p. 368). This argues against the predominant influence of the home training or example as explaining _family resemblance_ in criminality. Nevertheless, it would seem that the result might also be interpreted as meaning that the contact with other delinquents and official discipline outside the home at a more impressionable age notably increases the tendency to recidivism. Besides the argument as to the earlier removal from home, we have a test of the question whether those kinds of crime that are most influenced by contagion show closer correlation within the family. His statement of the results is as follows: “Our table 177, above, starting with crimes of fraud, passes to stealing and burglary—professional crimes, where the influence of criminal contagion should be the most intense; and then progressively to violence, arson and sexual offenses, in which last it is difficult to understand how the influence of example could have any effect at all. We can understand the influence of parental training in the original moulding of a professional burglar or thief, and, to a certain extent, it is conceivable that the constant spectacle of the lack of control in parents might lead their offspring to emulate them in acts of unlawful violence. But, that parental example could play any part of importance in the perpetration by their offspring of crimes such as arson and wilful damage to property, and, particularly, of sexual offenses, is not reasonably to be supposed. As seen in the above table, 177, the parental correlation for sexual crimes, and crimes for wilful damage to property is from .45 to .5; for stealing, it is from .48 to .58. We would assume then, from this evidence, that the tendency of the inherited factor in criminality is from .45 to .5, and the intensity of criminal contagion is anything between .05 and .1” (_20_, p. 367).