The suggestion to post on R-sig-ME was a good one.
The problem turned out to be a bug in lmer parsing, which is now fixed.
S Ellison writes:
> > Is there a difference between I(x*y) and I(y*x) ?
> > I have a call to lmer that results in this complaint:
> > Error in is.alpha2.subordinate * ~z.min.co.res :
>
> A reproducible example would help ...