Loans Back and Forth: 'bow'

[Koontz John E]

Thinking about loans back and forth, it occurred to me that I've explained
the -ku in some Siouan 'bow' terms as a reflex of the ...(e)kw- part of
the Algonquian form, but several Siouan languages form third person
inalienable possessives with a suffix -ku (Dakotan) or a prefix ko-
(Mandan).  I wonder if a hypothetical Proto-Siouan *maN(aN)t-e-ku (or
*-ko) 'his bow', if borrowed into Algonquian, wouldn't, if rendered
animate, come out me:ntekw-a.  Then I'd have to wonder if the IO forms
maN(aN)hdu and the Wi one maNaNc^gu might not result from borrowing a form
like that back, while cases like Omaha-Ponca maNaNde might occur, not by
truncating the Algonquian loan, but lacking the -ku Px3 marker in an
original Siouan form.

Or maybe the IO and Wi forms reflect a variant morphology *maN(aN)t-ku
instead of the *maN(aN)t-e-ku suggested in the Algonquian forms and the
loans were all one way, Siouan to Algonquian?  In that case Dhegiha would
simply have a third variant *maN(aN)t-e.

On the other hand, the variation between aN and iN in Siouan forms, cf.
Da ita(-zipa) or Ks (?) miN(iN)j^e, has been explained as the result of
handling -e- in the Algonquian forms.

Which came first, the Siouan or the Algonquian?  How much passing back and
forth might be reasonable?