Sarvadarshanasangraha 1.38

1.38

This is mere building of castles in the air. For an inferential sign possessed of pervasion and presence in the subject of inference is accepted as conveying knowledge by those who advocate the epistemic validity of inference. And pervasion is a relation devoid of both kinds of extrinsic condition. And that [relation] does not function as an operative factor through its mere existence, like the eye and other organs, but rather by being known. What, indeed, could be the means of knowing it? First, it cannot be perception. And is that [perception] held to be external or internal? Not the first: for, since it causes cognition only of an object with which it is in contact, even though its operation is possible with respect to the present, it is impossible with respect to past and future, and thus pervasion, which encompasses all cases, is impossible to know. Nor should it be thought that cognition of pervasion has a universal as its object, for that would entail the absence of invariable concomitance between the two individual instances. Nor is it the latter: for, since the internal organ is dependent upon the external senses, it cannot operate independently with respect to an external object. So it has been said: