It is normal to see split groups when you're using a non-optimal powder charge.
Go to my main OCW webpage at
www.clik.to/optimalchargeweight and read especially the OCW versus the Ladder Test. Ladder tests are ineffective about 1/3 of the time, based on what we now know is going on with the acoustic shock wave (not the harmonic vibration of the barrel, which is another thing entirely).
When the shock wave is at the muzzle, you'll get what I call the "scatter node" and what you'll see is a widely spread out group, for no readily apparent reason (this is assuming that the rifle shoots consistently well with other proven load recipes).
Here is an OCW test target I shot a while back...
You shoot these tests in round robin fashion, for reasons explained at my site. You'll see that the two higher graduations both split (the shot marked "pulled" is not to indicate a called flyer, just to point out that the rifle is favoring two points of impact at that powder charge, as well as the one below it).
But look at the first two charges. Same POI for both graduations... so I chose as the OCW (optimal charge weight) a charge of 40.7 grains. At the 40.7 grain level, there is room for pressure varations without causing a debilitating change in point of impact (no flyers). /ubbthreads/images/graemlins/wink.gif So that's what you're looking for, a load that is pressure tolerant.
This isn't to necessarily just allow you to throw charges with less concern as to the exact tenth of a grain--but rather also to accommodate the odd primer, or odd brass case that might creep into the shell box. With an OCW load you'll see flyers far less often when these anomalies come to call.
So what I'd do is go back to the drawing board and shoot an OCW test, and then let's analyze the target. It looks like at 28.6 grains the 4895 load had come together. If 28.6 grains is the Optimal Charge Weight, it would follow that somewhere around the 28.0 grain mark would be the scatter node, i.e. the shock wave is at the muzzle, causing too much obturation of the bore during bullet release. Since this shock wave turns and goes back to the chamber (it does this at the speed of sound in steel, at around 18,000 fps, back and forth three or four times before the bullet exits the muzzle), all you must do to have the bullet exit when the shock wave is back at the chamber end (the best possible time, when the muzzle is at its calmest), is to add a bit more powder to accelerate the bullet a bit faster... then voila, the group tightens.
Again, this is not about harmonic muzzle whip. That's another matter which can be dealt with by adjusting seating depth of the bullet.
We've found that once you see where the scatter node charge is (seemingly 28.0 grains in the 4895 load you show), once you add about 1.5% to the powder charge, you're in the OCW zone and accuracy comes around.
I would be curious to see how 28.8 grains looks with that load recipe also.
Dan