These are, to some extent, an argument against the cumulative inheritance of such effects. When a nerve atrophies from disuse, or a duct shrivels, or bone is absorbed, or a muscle becomes small or flabby, it proves, so far, that the average effect of use through enormous ages is _not_ transmitted. When the fibula of a dog's leg thickens by 400 per cent. to a size "equal to or greater than" that of the removed tibia which previously did the work,[69] it shows that in spite of disuse for countless generations, the "almost filiform" bone has retained a potentiality of development which is fully equal to that possessed by the larger one which has been constantly used. When, after being reared on the ailanthus, the caterpillars of the _Bombyx hesperus_ die of hunger rather than return to their natural food, the inherited effect of ancestral habit does not seem to be particularly strong. Neither is there any strongly-inherited effect of long-continued ancestral wildness in many animals which are easily tamed. WOULD NATURAL SELECTION FAVOUR USE-INHERITANCE? If use-inheritance is really one of the factors of evolution, it is certainly a subordinate one, and an utterly helpless one, whenever it comes into conflict with the great ruling principle of Selection. Would this dominant cause of evolution have favoured a tendency to use-inheritance if such had appeared, or would it have discouraged and destroyed it? We have already seen that use-inheritance is unnecessary, since natural selection will be far more effective in bringing about advantageous modifications; and if it can be shown that use-inheritance would often be an evil, it then becomes probable that on the whole natural selection would more strongly discourage and eliminate it as a hostile factor than it might occasionally favour such a tendency as a totally unnecessary aid. USE-INHERITANCE AN EVIL.