KR Norman on by/vy alternation

Dear Lance, Jim, etc,

I asked Dr K.R. Norman about the initial by/by alternation and got the
reply reproduced below.
Bh. Nyanatusita

"There is a dialect variation of b/v, which can be seen as early as the
A''sokan inscriptions, e.g. dbaadasa for dvaada''sa. It is presumably
based upon a change of pronunciation, with the labio-dental v becoming
the labial b. (We also find v becoming p in spagra < svarga). In A''soka
this change is very limited, and seems not to occur in the vy group,
which in the future passive participle, for example, either remains as
vy or becomes vv or viy. In Paali we find bb in tabba, presumably via
*tabya.

"We should therefore expect  initial vy in Paali to become (b)b, by way
of by, but it seems not to work that way. I do not know if the VRI
edition is typical of the Burmese tradition as a whole, but that prints
by in initial position, and also in words like habya < havya, but I have
not found tabya for tavya.

"Again, I do not know if the BJT edition is typical of the Sinhalese
tradition, but I have just looked at the edition of S I, and that has a
mixture of vy and by in initial position, even in the same word, e.g.
vyaakar- and byaakar-, and also has habya.

"Sometimes in Paali we find the expected  initial (v)v < vy, and it is
also clear that some of the forms are wrong back-formations based upon
this development, e.g. vyappatha is < *vaak-patha, and should really be
*vappatha. Somewhere along the line someone has assumed that initial v
was from vy and has "restored" it.

"It is clear then that the situation  has became stylised, and there is
no simple answer to the question. In fact, Ven. Nyanatusita  has
answered his own question. The solution(s) lie in something earlier than
the Paali language as we have it now. It is sometimes a Sanskritisation,
sometimes a MIA form, and sometimes a matter of pronunciation. Since
Paali is an artificial (in OvH's sense of the word) language, with
features taken over from a number of predecessors, it is not possible to
say which is more 'original', although we might dare to say that  vyap-
< vaak- is probably a late feature, relatively  speaking."