possible that they are not in the direct line of descent at all. The evidence is entirely circumstantial and indefinite, and it is impossible to show that any of the reptile-like birds which have been discovered have any descendants at the present day. All we can say is that, if our birds are not their descendants, it is very probable that they are the descendants of some other unknown form very much like them.
At the best it is simply a question of probability, not of direct proof, and paleontological evidence is never definite enough to enable us to reach a specific conclusion which may not possibly be wrong, and, this being the state of the case, we may fairly ask whether such speculations upon probability, in the absence of direct evidence, are entitled to be called science. In order to answer this question, and to show that phylogenetic speculation may be strictly within the legitimate scope of science, we will make use of an imaginary illustration.
Suppose that a large continental area, which is inhabited by a homogeneous human race, is invaded by a band of settlers from another country, about as the first European settlers forced themselves upon the homogeneous inhabitants of the United States.
Suppose that these settlers, increasing in numbers, gradually spread over the whole country, interbreeding with the autochthones, until, in later generations, the population comes to consist of two equally distributed races, represented by individuals of pure descent, with strongly marked race-characteristics; and, in addition to these, a great number of hybrids, presenting the characteristics of the two pure races in all degrees and manners of union.
As time goes on, imagine this latter class to increase at the expense of the others, until few persons of pure blood are left; and meanwhile suppose that a number of persons of a third race are introduced, about as the negroes were introduced into this country, and, after this immigration has lasted for a time, suppose it to stop, and let this third race spread and increase, and, after a time, gradually mix with the other two. Let the same process take place again, until the population comes to be made up of four quite dissimilar races with well-marked race characteristics, crossed in such a way that no individuals of the original race or of the first immigration are of perfectly pure blood, while there are a few nearly pure types of the third race, and still more of the fourth. Suppose, now, that an anthropologist undertakes to study the inhabitants of the country in order to learn what he can of their origin and history, and let him begin by attempting to classify them. Any attempt to divide them up into groups will fail, on account of the complexity of their relationships; and, although there are traces of four types, it is not possible to arrange them in four classes, since most of them have resemblances to more than one type. After long study of their relationships, and an enumeration of all the forms which are distinguishable, we may suppose him to hit upon some such expe-