270. It is very plain that in this passage Hume is riding two horses at once. He is trying so to combine the notion of the constitution of space by impressions with that of its constitution by points, as to disguise the real meaning of each. In what lies the difference between the feelings of colour, of which we have shown that they cannot without contradiction be supposed to ‘make up extension,’ and ‘coloured points or corpuscles’? Unless the points, as points, mean something, the substitution of coloured points for colours means nothing. But according to Hume the point is nothing except as an impression of sight or touch. If then we refuse his words the benefit of an interpretation which his doctrine excludes, we find that there remains simply the impossible supposition that space consists of feelings. This result cannot be avoided, unless in speaking of space as composed of points, we understand by the point that which is definitely other than an impression. Thus the question which Hume puts--If extension is made up of parts, and these, being indivisible, are unextended, what are they?--really remains untouched by his ostensible answer. Such a question indeed to a philosophy like Locke’s, which, ignoring the constitution of reality by relations, supposed real things to be first found and then relations to be superinduced by the mind--much more to one like Hume’s, which left no mind to superinduce them--was necessarily unanswerable. True way of dealing with the question. 271. In truth, extension is the relation of mutual externality. The constituents of this relation have not, as such, any nature but what is given by the relation. If in Hume’s language we ‘separate each from the others and, considering it apart, form a judgment of its nature and qualities,’ by the very way we put the problem we render it insoluble or, more properly, destroy it; for, thus separated, they have no nature. It is this that we express by the proposition which would otherwise be tautological, that extension is a relation between extended points. The ‘points’ are the simplest expression for those coefficients to the relation of mutual externality, which, as determined by that relation and no otherwise, have themselves the attribute of being extended and that only. If it is asked whether the points, being extended, are therefore divisible, the answer must be twofold. _Separately_ they are not divisible, for separately they are nothing. Whether, as determined by mutual relation, they are divisible or no, depends on whether they are treated as forming a quantum or no. If they are not so treated, we cannot with propriety pronounce them to be either further divisible or not so, for the question of divisibility has no application to them. But being perfectly homogeneous with each other and with that which together they constitute, they are susceptible of being so treated, and are so treated when, with Hume in the passage before us, we speak of