On 'patra/pattra' again!

[Madhav Deshpande]

John Smith raises the following question:
"Could you make clear just how far you are going here, Madhav? I have no
problems with your remarks on put(t)ra, but how many other -tra words are
suspect in your view? Surely not astra, zastra, zAstra, pAtra, chat(t)ra,
vaktra, etc., etc.? If not all cases, then which ones? And what is the
basis for discriminating between them (so that pat(t)ra is suspect but --
presumably -- zastra is not)?"

My general response is that the orthography of Sanskrit words is
independent of their etymology.  While the orthography and pronunciation
are guided by other factors, the etymology is often an after-the-fact
device.  By 'etymology' I do not mean to take away the prerogative of
modern scholars to find appropriate etymologies.  But our modern
etymologies may not account for the ancient orthographic practices.
        As to the specific words ending in -tra, a glance at the
U.naadisuutras shows what I mean.  Panini's rules 3.2.181 (dha.h karma.ni
.z.tran) - P.3.2.186 (kartari car.zidevatayo.h) prescribe -tra and -itra
affixes after some specific verb roots.  This gives us a list of -tra
words for which we have Paninian etymologies.  However, -tra has then been
extended to all verb-roots by the U.naadisuutra 4.158 (sarvadhaatubhya.h
.s.tran) with a number of subsequent rules prescribing specific additional
changes in the case of specific verb roots.  Thus, we can make a basic
distinction between -tra words which Panini treated as derived, and those
which he treated as underived, but were derived by others including the
author of the U.naadisuutras.  Among the words specifically derived by the
U.naadisuutras, we have the following:  bhraa.s.tra, gaantra, naantra,
hantra, vai.s.tra, aa.s.tra, dautra, u.s.tra, khaatra, suutra, muutra,
antra, citra, mitra, zastra, putra, strii, gotra, dhartra, vetra, paktra,
vaktra, yantra, satra, hotra, yaatraa, maatraa, zrotra, bhastraa, gaatra,
daatra, k.setra, bhaavitra, vaaditra, gaaritra, caaritra, azitra, vahitra,
varutra, vadhitra, ka.titra, lotra, amitra.  These formations are derived
by specific U.naadisuutras.
        The following are listed as being derived by the general U.naadi-
rule which adds -tra to any verb root.  This list is from Mahaadeva
Vedantin's U.naadikoza (ed. Kunjunni Raja, U of Madras, 1956, p. 100:
        vastra, astra, zaastra, chatra, paatra, patra, gantra,
da.m.s.traa, raa.s.tra
        At least these words are traditionally treated as underived by
Panini (avyutpanna), and they are treated as derived (vyutpanna) by the
specific U.naadisuutras.  It is the residue of such non-derived
(avyutpanna) words which forces Panini to offer two complementary
definitions for the nominal stem, arthavad adhaatur apratyaya.h
praatipadikam (P.1.2.45) for the underived nouns, and
k.rt-taddhita-samaasaaz ca (P.1.2.46) for the derived nouns.
        With such wide differences on whether a word is a derived word or
not, and if it is derived, how it is derived, the traditional grounds for
determining proper orthography of words on the basis of etymology are
indeed very shaky.
        Best,
                                        Madhav Deshpande