Pa.n.natti/pa~n~natti

Dear Eisel and Ong, etc,

Have you noticed that there is an ~n/.n alternation in the word
pa~n~natti/pa.n.natti (manifestation)? See PED. I did a quick search of
the CSCD and pa.n.natti is commonly found in commentaries, .tiikaas,
other late Paali works, and even in the Cullaniddesa and the Yamaka. As
it is found in the Burmese CS edition, the form is not confined to the
Siamese tradition unlike .nh/~nh.
Again, it appears to be a Prakitism or a survival from an earlier
stratum of the Pali language. Geiger, Pali Grammar para 48.2 mentions
that rarely pa.n.narasa (15), is found instead of pannarasa & pa~cadasa,
and that pa.n.naasa (50) is found besides pa~n~aasa. He notes that in
Pkt pa.n.narasa and pa.n.naasa.m are used. Also, in footnote 2 to para
53.1, he mentions that, instead of the usual j~n > ~n~n development in
Pali (described in para 53.1), the Pkt j~n usually becomes .n.n. An
exception in Pali is aa.naa = Skt aaj~na, where a retroflex appears for
a palatal like in Pkt aa.naa; see para 63.2. Geiger does not mention
pa.n.natti.
Are there any rules in the Kaccaayana or  Saddaniiti with regards the
assimilation of ~n and  .n?
Best wishes,
                   NT