> Any way, I think that it all depends on the common case. Of course > there > is Real.==, so it isn't that the operation isn't available. OK, thanks, that helps. So ML does have the nonreflexive Real.==; the question seems to be whether it thinks it's a bad idea to define a reflexive = (like Scheme eqv?, I think now) and make Real an equality type. And if so, why? Brad