[Footnote 1: See also pp. 340 and 409. It is unfortunate that, excepting the _Nyâyabindu, Nyâyabindu@tîkâ, Nyâyabindu@tîkâ@tippanî_ (St Petersburg, 1909), no other works dealing with this interesting doctrine of perception are available to us. _Nyâyabindu_ is probably one of the earliest works in which we hear of the doctrine of _arthakriyâkâritva_ (practical fulfilment of our desire as a criterion of right knowledge). Later on it was regarded as a criterion of existence, as Ratnakîrtti's works and the profuse references by Hindu writers to the Buddhistic doctrines prove. The word _arthakriyâ_ is found in Candrakîrtti's commentary on Nâgârjuna and also in such early works as _Lalitavistara_ (pointed out to me by Dr E.J. Thomas of the Cambridge University Library) but the word has no philosophical significance there.] 155 Sautrântika theory of Inference [Footnote ref 1]. According to the Sautrântika doctrine of Buddhism as described by Dharmakîrtti and Dharmmottara which is probably the only account of systematic Buddhist logic that is now available to us in Sanskrit, inference (_anumâna_) is divided into two classes, called svârthânumâna (inferential knowledge attained by a person arguing in his own mind or judgments), and parârthânumâna (inference through the help of articulated propositions for convincing others in a debate). The validity of inference depended, like the validity of perception, on copying the actually existing facts of the external world. Inference copied external realities as much as perception did; just as the validity of the immediate perception of blue depends upon its similarity to the external blue thing perceived, so the validity of the inference of a blue thing also, so far as it is knowledge, depends upon its resemblance to the external fact thus inferred (_sârûpyavas'âddhi tannîlapratîtirûpam sidhyati_). The reason by which an inference is made should be such that it may be present only in those cases where the thing to be inferred exists, and absent in every case where it does not exist. It is only when the reason is tested by both these joint conditions that an unfailing connection (_pratibandha_) between the reason and the thing to be inferred can be established. It is not enough that the reason should be present in all cases where the thing to be inferred exists and absent where it does not exist, but it is necessary that it should be present only in the above case. This law (_niyama_) is essential for establishing the unfailing condition necessary for inference [Footnote ref 2]. This unfailing natural connection (_svabhâvapratibandha_) is found in two types ___________________________________________________________________ [Footnote 1: As the _Pramâ@nasamuccaya_ of Diñnâga is not available in Sanskrit, we can hardly know anything of developed Buddhist logic except what can be got from the _Nyâyabindu@tîkâ_ of Dharmmottara.]