One final comment on this tonight. A careful reading of the Dictionary suggests that one cannot rely on side effects in u^:n y where n is a list.
"Finally, u^:n y for an array n is produced by assembling u^:a y (for all the atoms a in n) into an overall result." Since the order in which a verb is applied to a list of argument cells is undefined, it would be legitimate for J to evaluate all the powers up to >./n and then use n to select, rather than doing it in the order, which I had been assuming. This may also be the cause (although I can't see how) of the extra evaluation. Best wishes, John ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
