Thus a difficulty presents itself as to the explanation of the three following relationships:--( 1) That of the Pterodactyles with carinate birds; (2) that of the Dinosauria with struthious birds; (3) that of the carinate and struthious birds with each other.
Either birds must have had two distinct origins whence they grew to their present conformity, or the very same skeletal, and probably cerebral characters must have spontaneously and independently arisen.

Here is a dilemma, either horn of which bears a threatening aspect to the exclusive supporter of "Natural Selection," and between which it seems somewhat {72} difficult to choose.
It has been suggested to me that this difficulty may be evaded by considering pterodactyles and carinate birds as independent branches from one side of an ancient common trunk, while similarly the Dinosauria and struthious birds are taken to be independent branches from the other side of the same common trunk; the two kinds of birds resembling each other so much on account of their later development from that trunk as compared with the development of the reptilian forms.

But to this it may be replied that the ancient common stock could not have had at one and the same time a shoulder structure of _both kinds_.

It must have been that of the struthious birds or that of the carinate birds, or something different from both.

If it was that of the struthious birds, how did the pterodactyles and carinate birds independently arrive at the very same divergent structure?
If it was that of the carinate birds, how did the struthious birds and Dinosauria independently agree to differ?
Finally, if it was something different from either, how did the carinate birds and pterodactyles take on independently one special common structure when disagreeing in so many; while the struthious birds, agreeing in many points with the Dinosauria, agree yet more with the carinate birds?
Indeed by no arrangement of branches from a stem can the difficulty be evaded.
Professor Huxley seems inclined[55] to cut the Gordian knot by considering the shoulder structure of the pterodactyle as independently educed, and having relation to physiology only.