>Had it been "_-th_ = Q _-t_ dual" instead, I would have understood
>that those who said that Gn. _-th_ does not represent Q. _-r_,
>believed that its origin was the same as Qenya dual _-t_
>(both < _-tt@_).

I think that's exactly the intended meaning, the correspondence of
_th_ and _tt_ probably being so trivial for Tolkien that he doesn't
explicitly mention it.

>But what follows is also somewhat ambiguous: "_-tt@_ a dual ending =
>_-nt@_". Both _-tt@_ and _-nt@_ seem to be old dual endings, but what is
>their relation, and why is _-nt@_ introduced in the discussion?

It might be a bit of an overinterpretation, but I propose the following:

In Quenya, a dual ending _-t_ is seen, but it may come from _-tt@_
with loss of the schwa just as well as from _*-t@_. The parallel
existence of _-nt_ suggests that there were different modifications of
a more simple suffix _*-t@_, one with reduplication of the consonant,
another with nasalization.

Note that there is a suggested alternative reading _-tta_, _-nta_; in
either case both suffixes are most probably derived from the dual root
ATA (PE12:33, beside WI/U).

So basically the existence of _-nt_ supports _-tta/-tt@_ which on its
turn might be the source of Goldogrin pl. _-th_. Or else Tolkien
simply mentions it for the sake of completeness.

The suffix _*-t@/*-ta_ seems to appear in _-wid_ (PE11:11). Note also
the allative endings _-nta_, _-tta_ in EQG (PE14:46).

Roman Rausch