Then again _comparison_,[19] &c., must be utterly rejected as the means of the knowledge of the universal proposition, since it is impossible that they can produce the knowledge of the unconditioned connection [_i.e._, the universal proposition], because their end is to produce the knowledge of quite another connection, viz., the relation of a name to something so named. Again, this same absence of a condition,[20] which has been given as the definition of an invariable connection [_i.e._, a universal proposition], can itself never be known; since it is impossible to establish that all conditions must be objects of perception; and therefore, although the absence of perceptible things may be itself perceptible, the absence of non-perceptible things must be itself non-perceptible; and thus, since we must here too have recourse to inference, &c., we cannot leap over the obstacle which has already been planted to bar them. Again, we must accept as the definition of the condition, "it is that which is reciprocal or equipollent in extension[21] with the major term though not constantly accompanying the middle." These three distinguishing clauses, "not constantly accompanying the middle term," "constantly accompanying the major term," and "being constantly accompanied by it" [_i.e._, reciprocal], are needed in the full definition to stop respectively three such fallacious conditions, in the argument to prove the non-eternity of sound, as "being produced," "the nature of a jar," and "the not causing audition;"[22] wherefore the definition holds,--and again it is established by the śloka of the great Doctor beginning _samásama_.[23] But since the knowledge of the condition must here precede the knowledge of the condition's absence, it is only when there is the knowledge of the condition, that the knowledge of the universality of the proposition is possible, _i.e._, a knowledge in the form of such a connection between the middle term and major term as is distinguished by the absence of any such condition; and on the other hand, the knowledge of the condition depends upon the knowledge of the invariable connection. Thus we fasten on our opponents as with adamantine glue the thunderbolt-like fallacy of reasoning in a circle. Hence by the impossibility of knowing the universality of a proposition it becomes impossible to establish inference, &c.[24] The step which the mind takes from the knowledge of smoke, &c., to the knowledge of fire, &c., can be accounted for by its being based on a former perception or by its being an error; and that in some cases this step is justified by the result, is accidental just like the coincidence of effects observed in the employment of gems, charms, drugs, &c. From this it follows that fate, &c.,[25] do not exist, since these can only be proved by inference. But an opponent will say, if you thus do not allow adṛishṭa, the various phenomena of the world become destitute of any cause.