A quick follow-up:
I decreased the powder charge by .3 grains (from 6.0 to 5.7) and the groups tightened up. Further decreases didn't seem to help accuracy so I think I've found the load I'll use. I haven't chronoed these yet.
I'm surprised I missed this two years ago: I hope you are a believer in "better late than never."
What you have done is to identify one of the "all other things being equal" factors that is not always equal.
It is well known that a firearm barrel begins vibrate as the slug travels through the bore. For analytical purposes, folks tend to model barrels as fixed in position at the breech end and untethered (except by internal flexural rigidity) at the muzzel end -- this convention models free-floated target rifles but isn't all that accurate for auto pistols. In any event, the result of the vibration is that, if we plotted the location of the exact center of the muzzle during the time of bullet travel through the bore in a two-dimensional mode, we'd get a circle.
Two propositions follow. One: the smaller the circle, the more accurate the firearm. This, for instance, is why thick wall barrels and no-taper barrels tend to be more accurate than thin or tapered barrels. Two: regardless of the radius of muzzle deviation, if the bullet always leaves the muzzle at the same instant, and therefore at the same point of muzzle deviation, the more accurate the firearm will be. This is the underlying reason why consistency of velocity is important to accuracy, and it is more important than the effect of velocity consistency on exterior ballistics. (Note, for instance, that differential velocity in terms of exterior ballistics should open our groups only in the vertical spread, but we observe that inaccurate groups still tend to be sort of round.)
However, there is a third factor at work, and that is that the frequency of barrel vibration is also a function of the velocity with which the bullet travels the bore, and at certain frequencies the muzzle deviation will be smaller than at others. In addition, this relationship is not linear, but rather harmonic. This is why, as a very broad generality (and derived mainly from rifle experience than pistol experience): the fastest loads are seldom the most accurate. It also explains why folks often experience: a) an accuracy increase with a slight reduction in velocity, and then b) an accuracy decrease with a further velocity reduction. What you are doing is swinging through a nodal "sweet spot."
And the problem with all of the foregoing is that, even for two nominally identical firearms, no two barrels are alike. Which is why load development for accuracy is firearm specific.
Sorry if this is TMI.