Although it is a fact that the birth-rate in drunkard's families may be and often is larger than that of the general population,[20] it is none the less a fact that many of the worst drunkards leave no or few offspring. They die of their own excesses at an early age; or their conduct makes them unattractive as mates; or they give so little care to their children that the latter die from neglect, exposure or accident. As these drunkards would tend to hand down their own inborn peculiarity, or weakness for alcohol, to their children, it must be obvious that their death results in a smaller proportion of such persons in the next generation. In other words, natural selection is at work again here, with alcohol as its agent. By killing off the worst drunkards in each generation, nature provides that the following generation shall contain fewer people who lack the power to resist the attraction of the effect of alcohol, or who have a tendency to use it to such an extent as to injure their minds and bodies. And it must be obvious that the speed and efficacy of this ruthless temperance reform movement are proportionate to the abundance and accessibility of the supply of alcohol. Where the supply is ample and available, there is certain to be a relatively high death-rate among those who find it too attractive, and the average of the race therefore is certain to become stronger in this respect with each generation. Such a conclusion can be abundantly justified by an appeal to the history of the Teutonic nations, the nations around the Mediterranean, the Jews, or any race which has been submitted to the test. There seems hardly room for dispute on the reality of this phase of natural selection. But there is another way in which the process of strengthening the race against the attraction and effect of alcohol may be going on at the same time. If the drug does actually injure the germ-plasm, and set up a deterioriation, it is obvious that natural selection is given another point at which to work. The more deteriorated would be eliminated in each generation in competition with the less deteriorated or normal; and the process of racial purification would then go on the more rapidly. The fact that races long submitted to the action of alcohol have become relatively resistant to it, therefore, does not in itself answer the question of whether alcohol injures the human germ-plasm. The possible racial effect of alcoholization is, in short, a much more complicated problem than it appears at first sight to be. It involves the action of natural selection in several important ways, and this action might easily mask the direct action of alcohol on the germ-plasm, if there be any measurable direct result.