or of a class or order are exhibited in their simplicity, and uncomplicated by the presence of the characteristics of any of the minor divisions of the group: thus amphioxus is a generalized or diagrammatic vertebrate, with structural features which are common to the whole group, and none of the distinctive marks of any of the great subdivisions of the group. In a phylogenetic tree such a form would be represented by a line running almost directly upward from a point where great branches diverge from a common stem; and the fact that these generalized forms are much more often found among fossil than among recent animals is very suggestive.
This short analysis is sufficient to show that the essential similarity between a system of classification based on homology and a phylogenetic tree is so great that all objections to the one method of generalizing from the facts must apply to the other. If phylogenetic speculations retard science, speculations upon homology must do the same thing, and the only way to avoid danger will be to stick to facts, and, stripping our science of all that renders it worthy of thinking men, to become mere observing machines.
As it is plain that the strictest construction of the proper scope and limits of scientific thought does not make any such demands as this, we may feel at liberty to speculate upon phylogeny with such a basis as is furnished by the comparative study of the systematic relationship of living animals, but we are not able to go very far in this direction, for nearly all living animals fall into a few great well defined groups, and generalized types are the exception. Our conclusions must, therefore, be very vague and general, and we turn to the paleontological record for more exact data; but here, again, we soon meet with limitations which prevent any very exact and definite generalizations. One of these limitations, the imperfection of the record, we have already examined, and another is due to the fact that most of the main stems of the phylogenetic tree, and the minor branches of many of them, were established before the time of the oldest fossiliferous rocks, and we can not hope to find fossil remains of the unspecialized ancestors from which they were derived. This renders it hopeless for us to attempt actual proof of the more deep-seated phylogenetic relationships; and another consideration, which we will now examine, renders the discovery of the exact relationship of smaller groups of genera and species almost equally hopeless, even when the most ample series of fossils is discovered.
All that we know of variation indicates that it is not induced in the adult, but that it is congenital, and the effect of some force which has acted upon or through the parents. Hence it happens that, when a new variety is produced, it is not usually by the addition of something new to the characteristics of a mature animal, but by a divergence which shows itself before maturity is reached. When we compare two closely related species or varieties of birds, we do not find