Ah, see I thought that *your* solution wasn't transparent because it hides the list behind a junction. If the person wants to read the list then s/he would have to query the result's eigenvalues (to borrow from Q::S's function. I don't know the p6ism for that). I regard that as distinctly unfriendly. I'm mulling over whether PDCawley's context idea would be a good idea - in scalar return a junction, in list return the list. Under p5 I'd rather return an array reference but that may just be habit.