This is nearer the truth. The essence of an object is that part of it, which being present, a given sentimental result follows, or may be expected to follow, or may be made to follow. A certain experience of things is necessary before we can know what is the objective minimum consistent with some sentimental utility. If things are classified with due regard to their utilities, the essence will be the same as the general idea. It is however not true that the essence or any other part of the object causes the sentimental effect (VII). A common form of the generalistic superstition is to suppose that a thing is explained or sufficiently accounted for by classifying it. In all philosophies of Greek derivation--the Asiatic seem to be free from this defect--reason is considered to be 'the bringing of a thing under a class-notion,' and when this is done we are supposed to know the thing completely. An elaborate and utterly false dialectic has been erected on this foundation. No doubt our first attempt at explaining a thing is to refer it to a general idea--to classify it. This usually suggests something to add to the bare phenomenon by way of explanation or hypothesis. But only if we have a prior knowledge of the general idea, derived from things better known than the present phenomenon. The general idea is simply a short formula of that prior knowledge. Suppose we thoroughly know a body of similar things _a_, _b_, _c_, and also reduce them to the general image X; then on seeing _d_ and noticing that it is like _a_, _b_, _c_, we briefly think, 'Oh, it is X,' which excuses us from studying it further. We at once transfer to _d_ our whole knowledge of _a_, _b_, _c_, and in this ideal transfer the explanation consists--not in the classification. The transfer is often tacit--if explicit it is an 'argument.' If there has been no better known _a_, _b_, _c_, it is evident that the mere generalisation of new facts _d_, _e_, _f_, will not add anything to our knowledge of _them_. In deduction we should only return to them the knowledge just extracted from them. We should be explaining things by themselves--reasoning in a circle[12]. The _unity_, which explains is not the general idea. It is a unity of function or service, and may include things utterly heterogeneous, and therefore incapable of being reduced to a common idea. The pen in my hand consists of wood and metal; if I generalise them into Matter--the nearest class that includes both--I do not thereby explain the pen. But it is explained by the unity of service: the wood and metal contribute to form one instrument for writing.