does  not "teach"  in this  sense  because  like  all
　　   deductions　　 its   conclusion   does   not   contain
　　   information  nor already found in the premises, Thus,
　　   from  the  standpoint  of an upaaya  the retroductive
　　   inference  is  enough,  or,  as  the  author  of  the
　　   Nyaayaprave`sa  put it, "...these  three members make
　　   the [retroductive] syllogism and no more."

　　　　  A further  point in favour of reading the Nyaaya
　　   inference schema as a retroduction  is that it makes
　　   the  remainder  of the  manual  on logical  methods,
　　   especially   the  detailed   sections  on  kinds  of
　　   fallacies, more intelligible and enljghtening.  More
　　   than  two thirds  of the text covers  identification
　　   and classification  of fallacies, but none bear  any
　　   resemblance  to the  formal  fallacies  of deduction
　　   such  as affirming  the  consequent  or denying  the
　　   antecedent, nor  does  the system  resemble  Western
　　   notions  of  an  informal  fallacy.   Fallacies   of
　　   irrelevance  such  as the  ad  hominem  or post  hoc
　　   propter  hoc call  attention  to the lack of support
　　   between  premises   and  putative   conclusion.   In
　　   Buddhist logic the classification  of fallacies does
　　   not attempt to circumscribe the ways premises can be
　　   irrelevant;  on the contrary  it fives criteria  for
　　   grading the strength or weakness  of the explanatory
　　   hypotheses.  This is precisely  what is required for
　　   retroductive  accuracy.  Weak hypotheses  emerge  in
　　   three circumstances: (1) the hetu is unrecognized by
　　   proponent or opponent, (2) the hetu is inconclusive,
　　   or (3) it is contradicted.  Inconclusive  hetus  are
　　   those  which are not supported  by further  evidence
　　   from   the  similarity   and  dissimilarity   cases;
　　   contradicted   hetus  are  those  which   prove  the
　　   opposite  of the  pak.sa.  Such  a contradiction  is
　　   established  by deducing the opposite property-locus
　　   assertion.  A hetu  can fail  to be recognized, that
　　   is,  it can fail as a teaching  device by not making
　　   the  auditor  (or speaker) aware  of the  connection
　　   between the assertion  statement  and its warranting
　　   hetu.  Thus, when hypotheses  fail to be understood,
　　   they engender fallacies of recognition, but when they
　　   fail in evidential  support they engender  fallacies
　　   of contradiction  or inconclusivity.  On the  whole,
　　   this   classification   of  fallacies   reflects   a