_pari passu_, but if the sexes of the mimic differ, as in this case, we must suppose the starting-point to have been the condition exhibited by the male of the mimicking species. But _Argynnis hyperbius_ is also a species in which the female mimics _D. plexippus_; and by using the same argument as that just detailed for _Elymnias undularis_ we can shew that the Danaine model, _D. plexippus_, must also have been like the male of _Argynnis hyperbius_. And if the resemblance of _A. hyperbius_ was developed subsequently to that of _E. undularis_, then both _D. plexippus_ and _E. undularis_ must at one time have been like the male of _A. hyperbius_, a proposition to which few entomologists are likely to assent. Further, since the female of _H. misippus_ also comes into the _plexippus-chrysippus_ combine we must suppose that these species must at some time or another have passed through a pattern stage like that of the _misippus_ male. It is scarcely necessary to pursue this argument {70} further, for even the most devoted adherents of the theory of mimicry as brought about by the operation of natural selection on small variations are hardly likely to subscribe to the phylogenetic consequences which it must entail in cases where a model is mimicked by the females of several species whose males are widely dissimilar in appearance. Even if we suppose the two Danaines to have been originally like the male of one of the three mimics, we must still suppose that the females of the other two originated as "sports," sufficiently near to Danaines to be confused with them. But if such sports can be produced suddenly by some mutational process not at present understood, why should not these sports be the females of the three mimicking species as we see them at present? Why need we suppose that there were intermediate stages between the mimicking female and the original hypothetical female which was like the male? If a sport occurred which was sufficiently similar to an unpalatable species to be confused with it, it is theoretically demonstrable that, although relatively scarce to start with, it would rapidly increase at the expense of the unprotected male-like female until the latter was eliminated. We shall, however, return in a later chapter (p. 96) to the argument by which this view can be supported. So far we have discussed what we called the two extreme views as to the way in which a mimetic resemblance may be supposed to have originated. Of the two that which assumes the resemblance to have {71} been brought about by a succession of slight variations must also assume that model and mimic were closely alike to start with, and this certainly cannot be true in many cases. On the other hand, there is so far no reason against the idea of supposing the resemblance to have originated suddenly except what to most minds will probably appear its inherent improbability.