> But things like sort([x,y, not x, not y]) will be different. Should this change cause a problem with to_poly_solve (unlikely), I'll fix it: (%i17) z %and a; (%o17) a %and z (%i18) z %and a %and not z; (%o18) false My test showed no problems with the to_poly_solve testsuite. --Barton