________________
538
Appendix to Volume I island who do not know how to make fire would have been able to infer fire from smoke. Some say that the invariable association of the hetu with the sādhya is perceived by mental perception (mānasapratyakşa). They hold that in perceiving the association of smoke with fire and the absence of the former when the latter is absent, the mind understands the invariable association of smoke with fire. It is not necessary in order to come to such a generalization that one should perceive the agreement of smoke and fire in all the infinite number of cases in which they exist together, for the agreement observed in the mind is not between smoke and fire but between smoke-ness and fire-ness (jralanatrā-di-šāmānya-puraħsaratayā z'yāpti-grahaņāt). The objection against this view would be the denial of class-concepts as held by the Cārvākas, Buddhists, and others. There are others, again, who say that even if universals are admitted, it is impossible that there should be universals of all cases of absence of fire as associated with the absence of smoke, and under the circumstances unless all positive and negative instances could be perceived the inductive generalization would be impossible. They, therefore, hold that there is some kind of mystic intuition like that of a yogin (yogi-pratyakşa-kalpam) by which the invariable relation (pratibandha) is realized. Others hold that an experience of a large number of positive instances unaccompanied by any experience of any case of failure produces the notion of concomitance. But the Nyāya insists on the necessity of an experience of a large number of instances of agreement in presence and absence for arriving at any inductive generalization of concomitance!. The Cārvākas, of course, say to this that in determining the unconditional invariable agreement of every case of a hetu with its sādhya the absence of visible conditions may be realized by perception; but the possibility of the existence of invisible conditions cannot be eliminated even by the widest experience of agreement in presence, and thus there would always be the fear that the invariable concomitance of the hetu with the sādhya may be conditional, and thus all inference has the value of more or less probability but not of certainty, and it is only through perceptual corroboration that the inferences come to be regarded as valida. The reply of Nyāya to this is that the assertion that in
1 Nvāja-manjarī, p. 122.
2 athā-numānam na pramanam yogjo-pādhinam yogj'a-nupalabdhyā'bhura-niscaye py' ayogjo-padhi-sarkajā ajabhicăra-samsawit satasah sahacaritajor api tyabhicāro-palabdheś ca loke dhūmā-di-dursană-ntaram rahnya'di- 'yurahāras ca
