I bumped my own loop in the subs to 1_000_000, so I got my multiplier, but not the way you did. I figured it didn't matter where the looping occurred, and that if it was inside the sub, then my test would include more concatenation and less call dispatch. The time machine on the other hand, won't compile strict yet.

I still have this problem, though, in that I don't trust my own results. I believe there to be an inherent hit taken when using $old="$old$new" over $old.=$new because of the necessary runtime scanning. Does anyone have anything to say on that?