Notwithstanding these great sources of difficulty, I apprehend every one would admit, that if it were possible, a genealogical classification of our domestic varieties would be the most satisfactory one; and as far as varieties were concerned would be the natural system: in some cases it has been followed. In attempting to follow out this object a person would have to class a variety, whose parentage he did not know, by its external characters; but he would have a distinct ulterior object in view, namely, its descent in the same manner as a regular systematist seems also to have an ulterior but undefined end in all his classifications. Like the regular systematist he would not care whether his characters were drawn from more or less important organs as long as he found in the tribe which he was examining that the characters from such parts were persistent; thus amongst cattle he does value a character drawn from the form of the horns more than from the proportions of the limbs and whole body, for he finds that the shape of the horns is to a considerable degree persistent amongst cattle{447}, whilst the bones of the limbs and body vary. No doubt as a frequent rule the more important the organ, as being less related to external influences, the less liable it is to variation; but he would expect that according to the object for which the races had been selected, parts more or less important might differ; so that characters drawn from parts generally most liable to vary, as colour, might in some instances be highly serviceable--as is the case. He would admit that general resemblances scarcely definable by language might sometimes serve to allocate a species by its nearest relation. He would be able to assign a clear reason why the close similarity of the fruit in two varieties of pine-apple, and of the so-called root in the common and Swedish turnips, and why the similar gracefulness of form in the greyhound and racehorse, are characters of little value in classification; namely, because they are the result, not of community of descent, but either of selection for a common end, or of the effects of similar external conditions. {447} _Origin_, Ed. i. p. 423, vi. p. 579. In the margin Marshall is given as the authority. _Classification of "races" and species similar._