249. So far we have been considering the composition of impressions generally, without special reference to extension. The contradiction pointed out arises from the confusion between impressions as felt and impressions as thought of; colour, between feelings as they are in themselves, presented successively in time, and feelings as determined by relation to the thinking subject, which takes them out of the flux of time and converts them into members of a permanent whole. It is in this form that the confusion is most apt to elude us. When the conceived object is one of which the qualities can really be felt, _e.g._ colour, we readily forget that a felt quality is no longer simply a feeling. But the case is different when the object is one, like extension, which forces on us the question whether its qualities can be felt, or presented in feeling, at all. A compound of impressions of colour, to adopt Hume’s phraseology, even if such composition were possible, would still be nothing else than an impression of colour. In more accurate language, the conception, which results from the action of thought upon feelings of colour, can only be a conception of colour. Is extension, then, the same as colour? To say that it was would imply that geometry was a science of colour; and Hume, though ready enough to outrage ‘Metaphysics and School Divinity,’ always stops reverently short of direct offence to the mathematical sciences. As has been said above, of the three main questions about the idea of extension which his doctrine raises--Is it itself extended? Is it possible without reference to something other than a possible impression? Is it the same as the idea of colour or tangibility?--the last is the only one which he can scarcely even profess to answer in the affirmative. [1] Even when he has gone so far as to speak of the parts of a perception, a sound instinct compels him, instead of identifying the perception directly with extension, to speak of it as ‘affording through the situation of its parts the notion of’ extension. [2] In like manner, when he has asserted extension to be a compound of impressions, he avoids the proper consequence of the assertion by speaking of the component impressions as those, not of colour but, of coloured points, ‘atoms or corpuscles endowed with colour and solidity;’ and, again, does not call extension the compound of these simply, but the compound of them as ‘disposed in a certain manner.’ When the idea which is a copy of this impression has to be spoken of, the expression is varied again. It is an ‘idea of the coloured points _and of the manner of their appearance_,’ or of their ‘disposition.’ The disposition of the parts having been thus virtually distinguished from their colour, it is easy to suppose that, finding a likeness in the disposition of points under every unlikeness of their colour, ‘we omit the peculiarities of colour, as far as possible, and found an abstract idea merely on