Hence this grouping of the like and separation of the unlike which constitutes Classification, can reach its complete form only by slow steps. I have shown (_Essays_, Vol. II., pp. 145-7) that, other things equal, the relations among phenomena are recognized in the order of their conspicuousness; and that, other things equal, they are recognized in the order of their simplicity. The first classifications are sure, therefore, to be groupings of objects which resemble one another in external or easily-perceived attributes, and attributes that are not of complex characters. Those likenesses among things which are due to their possession in common of simple obvious properties, may or may not coexist with further likenesses among them. When geometrical figures are classed as curvilinear and rectilinear, or when the rectilinear are divided into trilateral, quadrilateral, &c., the distinctions made connote various other distinctions with which they are necessarily bound up; but if liquids be classed according to their visible characters--if water, alcohol, sulphuret of carbon, &c., be grouped as colourless and transparent, we have things placed together which are unlike in their essential natures. Thus, where the objects classed have numerous attributes, the probabilities are that the early classifications, based on simple and manifest attributes, unite under the same head many objects that have no resemblance in the majority of their attributes. As the knowledge of objects increases, it becomes possible to make groups of which the members have more numerous properties in common; and to ascertain what property, or combination of properties, is most characteristic of each group. And the classification eventually arrived at is of such kind that the objects in each group have more attributes in common with one another than they have in common with any excluded objects; one in which the groups of such groups are integrated on the same principle; and one in which the degrees of differentiation and integration are proportioned to the degrees of intrinsic unlikeness and likeness. And this ultimate classification, while it serves to identify the things completely, serves also to express the greatest amount of knowledge concerning the things--enables us to predicate the greatest number of facts about each thing; and by so doing implies the most precise correspondence between our conceptions and the realities. § 99. Biological classifications illustrate well these phases through which classifications in general pass. In early attempts to arrange organisms in some systematic manner, we see at first a guidance by conspicuous and simple characters, and a tendency towards arrangement in linear order. In successively later attempts, we see more regard paid to combinations of characters which are essential but often inconspicuous, and an abandonment of a linear arrangement for an arrangement in divergent groups and re-divergent sub-groups.