II. SUBSTANCES. Logicians have endeavoured to define Substance and Attribute; but their definitions are not so much attempts to draw a distinction between the things themselves, as instructions what difference it is customary to make in the grammatical structure of the sentence, according as we are speaking of substances or of attributes. Such definitions are rather lessons of English, or of Greek, Latin, or German, than of mental philosophy. An attribute, say the school logicians, must be the attribute _of_ something; colour, for example, must be the colour _of_ something; goodness must be the goodness _of_ something: and if this something should cease to exist, or should cease to be connected with the attribute, the existence of the attribute would be at an end. A substance, on the contrary, is self-existent; in speaking about it, we need not put _of_ after its name. A stone is not the stone _of_ anything; the moon is not the moon _of_ anything, but simply the moon. Unless, indeed, the name which we choose to give to the substance be a relative name; if so, it must be followed either by _of_, or by some other particle, implying, as that preposition does, a reference to something else: but then the other characteristic peculiarity of an attribute would fail; the _something_ might be destroyed, and the substance might still subsist. Thus, a father must be the father _of_ something, and so far resembles an attribute, in being referred to something besides himself: if there were no child, there would be no father: but this, when we look into the matter, only means that we should not call him father. The man called father might still exist though there were no child, as he existed before there was a child: and there would be no contradiction in supposing him to exist, though the whole universe except himself were destroyed. But destroy all white substances, and where would be the attribute whiteness? Whiteness, without any white thing, is a contradiction in terms. This is the nearest approach to a solution of the difficulty, that will be found in the common treatises on logic. It will scarcely be thought to be a satisfactory one. If an attribute is distinguished from a substance by being the attribute _of_ something, it seems highly necessary to understand what is meant by _of_; a particle which needs explanation too much itself, to be placed in front of the explanation of anything else. And as for the self-existence of substance, it is very true that a substance may be conceived to exist without any other substance, but so also may an attribute without any other attribute: and we can no more imagine a substance without attributes than we can imagine attributes without a substance.