So -- however meaningless and silly an excercise this was, in and of itself -- at least it didn't violate the whole idea of a RAID array; it appears this intrepid experimenter *did* actually experience a speed increase.
As opposed to the case with a "cheap'n'cheerful software RAID" array consisting of separate *partitions* (be they ordinary partitions, or "Windows Dynamic" ones) of the *same physical* drive, where there isn't even a theoretical chance of an improvement.
RAID-ing parts of the same actual disk hardware is an oxymoron, a symptom of a deep misunderstandidng (which this iMac tinkerer did *not* labour under) of the whole idea: However many partitions you split a disk up into, you've still only got the same number of heads and platters that you started with.