From: "Patrick Dunn" <[log in to unmask]>
> >     Sing            Dual       Plural
> > Nom -s, 0           -e, -/-i  -es
> > Voc 0               -e, -/-i  -es
> > Acc -m/-m=          -e, -/-i  -ns/-n=s
> > Gen -es/-os/-s      -ous? -s? -om/-m
> > Abl -es/-os/-s;     -bhy, -m -bh(y)os, -mos
> >     -ed/-od
> > Dat -ei             -bhy, -m -bh(y)os, -mos
> > Loc -i              -ou        -su
> > Ins -e/-o, -bhi/-mi -bhy, -m -bhis/-mis, -is
> > Copied exactly, except that macrons are rendered with circumflexes, and
> > = should be a circle under the letter.
> >
> > And goes on to say: "These endings represent a composite set of
> > possibilities for the Proto-Indo-European noun; no single form reflects
> > them all."
> That'll come in handy.  I want to make a pseudo-Indo-European language
> from the roots, something that *could* have evolved.  I'll have to
> simplify, probably.
> When one is offered several choices above, are those variations within a
> single word class, or examples of different declensions?

I don't know about the others, but from what I understand, the -bh(y) / -m
alternation in the abl/dat/ins lines is from a dialect difference/change:
(Germano?)-Balto-Slavic had -mos where others had -bh(y).

The -m/-m=, -ns/-n=s differences probably depend on whether the stem is
vowel or consonant.

And wherever I see e/o/0 alternations, I file that under 'ablaut', the use
of which I entirely do not understand.