Next without reference to Darwin's authority, let us see for ourselves where the inconsistency really lies. To allow that generic characters may be useless, while denying that specific characters can ever be so (unless correlated with others that are useful), involves an appeal to the argument from ignorance touching the ancestral habits, life-conditions, &c., of a parent species now extinct. Well, even upon this assumption of utility as obsolete, there remains to be explained the "stability" of useless characters now distinctive of genera, families, orders, and the rest. We know that specific characters which have owed their origin to utility and have afterwards ceased to present utility, degenerate, become variable, inconstant, "rudimentary," and finally disappear. Why, then, should these things not happen with regard to useless generic distinctions? Still more, why should they not happen with regard to family, ordinal, and class distinctions? On the lines against which I am arguing it would appear impossible that any answer to this question can be suggested. For what explanation can be given of the contrast thus presented between the obsolescence of specific characters where previous utility is demonstrable, and the permanence of higher characters whose previous utility is assumed? As we have already seen, Mr. Wallace himself employs this consideration of permanence and constancy against the view that any cause other than natural selection can have been concerned in the origin and maintenance of _specific_ characters. But he does not seem to see that the consideration cuts two ways--and much more forcibly against his views than in favour of them. For while, as already shown in the chapter before last, it is sufficiently easy to dispose of the consideration as Wallace uses it (by simply pointing out with Darwin that any causes other than natural selection which may have been concerned in the genesis of _specific_ characters, must, if equally uniform in their operation, equally give rise to permanence and constancy in their results); on the other hand, it becomes impossible to explain the stability of useless _generic_ characters, if, as Wallace's use of the argument requires, natural selection is the only possible cause of stability. The argument is one that cannot be played with fast and loose. Either utility is the sole condition to the stability of _any_ diagnostic character (in which case it is not open to Mr. Wallace to assume that all _generic_ or higher characters which are now useless have owed their origin to a past utility); or else utility is not the sole condition to stability (in which case his use of the present argument in relation to _specific_ characters collapses). We have seen, indeed, in the chapter before last, that his use of the argument collapses anyhow, or quite irrespective of his inconsistent attitude towards generic characters, with which we were not then concerned. But the point now is that, as a mere matter of