They are no more capable of being described than sensations are; and, as it appears to me, they are as little susceptible of analysis into simpler elements.

Like simple tastes and smells, or feelings of pleasure and pain, they are ultimate irresolvable facts of conscious experience; and, if we follow the principle of Hume's nomenclature, they must be called _impressions of relation_.

But it must be remembered, that they differ from the other impressions, in requiring the pre-existence of at least two of the latter.

Though devoid of the slightest resemblance to the other impressions, they are, in a manner, generated by them.

In fact, we may regard them as a kind of impressions of impressions; or as the sensations of an inner sense, which takes cognizance of the materials furnished to it by the outer senses.
Hume failed as completely as his predecessors had done to recognise the elementary character of impressions of relation; and, when he discusses relations, he falls into a chaos of confusion and self-contradiction.
In the _Treatise_, for example, (Book I., Sec.