406 B. BOSANQUET : together in pairs or threes, as " father and son," &c. We get off the track of a process or relation when it becomes general in its scope. It is especially within the limits above assigned to comparison in the psychological sense, that the idea of reciprocal subsumption or apperception of each datum under or by the other seem* to apply. I wish to raise the question how far this subsumption, and especially its reciprocal character, is anything but a characteristic of sense-perception and of imperfect knowledge. The logical purpose of the whole process, which must ultimately govern the comparing activity even as known to psychology, is surely to make the identity and diversity of the data explicit by subsuming them not under each other, but under some standard, quantitative perhaps, or else furnished by the notion of Kind or of Purpose. The subsumption of one under the other, or rather under an element in the other, is really the beginning of the subsumption of both under a principle. But, of course, in the effort to light on the principle we try elements first out of one datum and then out of the other, and we look or listen alternately, chiefly because our sense is subject to time and space. I will work out one or two examples. I have to match a particular fish-hook. I take my pattern, hold the one I am testing close beside it, and try if they coincide in length and curve ; supposing that they do. then I judge the second the same as the first in these respects : I might say I subsume the length and curve of the second under those of the first better, I equate the two. Then I recur to the first and examine the thickness of the steel, and again subsume the corresponding property of the second under this property. With reference to reciprocal sub- sumption, a remark must be made here. We are apt, in comparison, after one such subsumption as described above, to continue dwelling on the second object, to pass on to another property in it, and to make this a starting-point from which we return to the first to subsume the corresponding property of this first under this character in the second like a ship which unloads one cargo and then takes another on board at the same port, so as not to make an empty return journey. This kind of alter- nation is, I think, a bad practice, and the source of confusion. It is better not to pass on to a further quality in the second datum, but to make an empty return journey and start again from a further quality in the first object ; otherwise we risk confusing the first with the second object. But of course, at the end of the process, the second may still have superfluous qualities, and we must then take these as a datum and, finding nothing in the first to correspond with them, register them as a difference between the two. Apart from the erroneous tendency mentioned above, this is the only trace that I can find of nlti'i-n>tt<> subsumption. But in as far as mastering the identity and diversity of the contents before us is a process in time, of course there is successive subsumption. I will now take a couple of instances in which comparison in the psychological sense passes with unusual facility into logical
