Races and reputed species agree in some respects, although differing from causes which, we have seen, we can in some degree understand, in the fertility and "trueness" of their offspring. In the first place, there is no clear sign by which to distinguish races from species, as is evident from the great difficulty experienced by naturalists in attempting to discriminate them. As far as external characters are concerned, many of the races which are descended from the same stock differ far more than true species of the same genus; look at the willow-wrens, some of which skilful ornithologists can hardly distinguish from each other except by their nests; look at the wild swans, and compare the distinct species of these genera with the races of domestic ducks, poultry, and pigeons; and so again with plants, compare the cabbages, almonds, peaches and nectarines, &c. with the species of many genera. St Hilaire has even remarked that there is a greater difference in size between races, as in dogs (for he believes all have descended from one stock), than between the species of any one genus; nor is this surprising, considering that amount of food and consequently of growth is the element of change over which man has most power. I may refer to a former statement, that breeders believe the growth of one part or strong action of one function causes a decrease in other parts; for this seems in some degree analogous to the law of "organic compensation{260}," which many naturalists believe holds good. To give an instance of this law of compensation,--those species of Carnivora which have the canine teeth greatly developed have certain molar teeth deficient; or again, in that division of the Crustaceans in which the tail is much developed, the thorax is little so, and the converse. The points of difference between different races is often strikingly analogous to that between species of the same genus: trifling spots or marks of colour{261} (as the bars on pigeons' wings) are often preserved in races of plants and animals, precisely in the same manner as similar trifling characters often pervade all the species of a genus, and even of a family. Flowers in varying their colours often become veined and spotted and the leaves become divided like true species: it is known that the varieties of the same plant never have red, blue and yellow flowers, though the hyacinth makes a very near approach to an exception{262}; and different species of the same genus seldom, though sometimes they have flowers of these three colours. Dun-coloured horses having a dark stripe down their backs, and certain domestic asses having transverse bars on their legs, afford striking examples of a variation analogous in character to the distinctive marks of other species of the same genus. {260} The law of compensation is discussed in the _Origin_, Ed. i. p. 147, vi. p. 182.