What now is the mental process by which classification is effected? Manifestly it is a recognition of the _likeness_ or _unlikeness_ of things, either in respect of their sizes, colours, forms, weights, textures, tastes, &c., or in respect of their modes of action. By some special mark, sound, or motion, the savage identifies a certain four-legged creature he sees, as one that is good for food, and to be caught in a particular way; or as one that is dangerous; and acts accordingly. He has classed together all the creatures that are _alike_ in this particular. And manifestly in choosing the wood out of which to form his bow, the plant with which to poison his arrows, the bone from which to make his fish-hooks, he identifies them through their chief sensible properties as belonging to the general classes, wood, plant, and bone, but distinguishes them as belonging to sub-classes by virtue of certain properties in which they are _unlike_ the rest of the general classes they belong to; and so forms genera and species. And here it becomes manifest that not only is classification carried on by grouping together in the mind things that are _like_; but that classes and sub-classes are formed and arranged according to the _degrees of unlikeness_. Things widely contrasted are alone distinguished in the lower stages of mental evolution; as may be any day observed in an infant. And gradually as the powers of discrimination increase, the widely contrasted classes at first distinguished, come to be each divided into sub-classes, differing from each other less than the classes differ; and these sub-classes are again divided after the same manner. By the continuance of which process, things are gradually arranged into groups, the members of which are less and less _unlike_; ending, finally, in groups whose members differ only as individuals, and not specifically. And thus there tends ultimately to arise the notion of _complete likeness_. For manifestly, it is impossible that groups should continue to be sub-divided in virtue of smaller and smaller differences, without there being a simultaneous approximation to the notion of _no difference_.