________________
DHARMOTTARA'S THEORY OF KNOWLEDGE
263
with the case of an incorrect superimposition (āropa) {24,7-13} it is clearly implicit that it should be counted as invalid.
Besides the kinds of doubt explained above which are related to two things, there are also some which determine only one part (ekaniscaya). These are cognitions which may come about either (a) without having seen either the object directly or something necessarily concomitant (avinābhāvin) that could serve as reason in an inference, or (b) based on a logical mark or reason (linga), the nature of which is uncertain.
Type (a) is exemplified by a cognition phrased in the following form: “In the well (in front of the house) there is water.” From the additional material gained from other texts21 it must be understood · that this cognition takes place when one is standing inside a house.
Therefore neither can one see the water itself nor is there any inferential mark from which the fact that there is water could be inferred. And because one is aware of this fact this cognition remains doubtful {18,11-18).
Type (b) is illustrated by a cognition where a fire is inferred on the basis of a cognition in which from far away a swarm of mosquitoes has been determined as being smoke. In this case, Dharmottara explains, the nature of smoke is not determined definitely (res par rtogs pa ma yin; ni rūp-). And as a cognition of a fire that presupposes smoke, the nature of which is not ascertained, is merely a doubt, it does not cause a person to obtain its object. For, he goes on, inference causes a person to obtain the object because it determines an existent or non-existent object definitely (res par 'jog pa) (18,19–19,8).
At first glance it is not clear why this sort of cognition is classified as samsaya and not as mithyājñāna, like the cognition which erroneously takes a mirage to be water, for it merely seems to be an incorrect determining of the logical mark, linga or hetu. But to understand Dharmottara's intention here we can enlist help from a longer passage in the Pramanaviniscayatikā and from Durvekamiśra's Dharmottarapradipa, where the same problem is treated more extensively.22 According to these texts it must be understood that before the present inference takes place, one either did not recognize the nature of smoke which is restricted (niyata) to being the effect of a fire (vahnikārya) as being different from the nature