m/anusvAra & s/visarga

[Jacob Baltuch]

I wrote yesterday:

>Since I do a lot of ASCII input and typing <H> and <M>
>is a pain in the neck. I want to use in <s> for <H> (visarga)
>(or if it comes before a sibilant, that sibilant)
>and <m> for <M>. But before going ahead I want to be
>fairly sure I will always be able to restaure things
>from the context when I need to.
>
>For anusvAra I am 100% sure but I'd thought I'd mention
>it so as to make 101% sure.

Well maybe I _shouldn't_ have been so sure! The problem was
to make certain that I couldn't find the following sequences
in any Sanskrit word:

                     mk, mkh, mg, mgh, mG
                     mc, mch, mj, mjh, mJ
                     mT, mTh, mD, mDh, mN
                     mt, mth, md, mdh, mn
                     my, mr,  ml, mv
                     mz, mS,  ms,
                     mh.

Today, with my eyesight somewhat improved, I checked a list
of conjuncts, and sure enough I find, <gasp>, conjuncts for
sequences mna, mya, mra, mla, mva!

So now I'd like to ask, _are_ there any Sanskrit words where
             mna,     mya, mra, mla, mva
are pronounced with a full, real m, and are opposed to other
words with sequences
             Mna/nna, Mra, Mya, Mla, Mva
or are those conjuncts only used for orthographic variants?

In other words,
when one sees                mna,     mya, mra, mla, mva
may one always spell it      Mna/nna, Mya, Mra, Mla, Mva
or not? (And if not, are forms with mn, my, ... nevertheless
fairly rare?)

(Well, I'm not optimistic, but you see it's not easy to give
up such a convenient idea :)