Following up on this, I finally went and shot this 162 ELDM test I loaded up.
First test was same IMR7828SSC powder and charge as above but switching primer to a FED210. Mean radius 0.98"!! If I look at each sub-group independently, I get 0.43" and 0.47". Overall group size is 2.5" for 10 shots, vs 1.75" for 6 in the initial test above. I don't know what happened here, either some change in my shooting input, I made a mistake in the reloading room, or this is just a real, weirdly shaped group.
View attachment 851619
What's really really weird is I shot this full test of 30 round robin style, in sets of 3/3/4 of each load, and the scatter into two groups manifested across multiple strings for this load. I would also expect whatever systematic error I had here to manifest in the other two loads, but here's the H4831SC/CCI200 group:
View attachment 851624
MR 0.33" for 10 - awesome, and a 40% improvement over the initial load.
And the H4350/CCI200 group:
View attachment 851626
MR 0.45", not great, but a 18% improvement over the initial load.
So, putting aside the very strange double group, my conclusion so far is that powder can make a HUGE difference in accuracy, but primer not so much. I'd like to hear the "nodes and harmonics don't exist" crowd explain that! No seriously, someone explain it to me - I thought precision was just proportional to recoil. Even though these are all different powders and speeds, the difference in precision is way bigger than the difference in recoil, and also not correlated.
Next up, I want to pick one of these loads and make a change in seating depth and a change in powder charge. I will also try and pick one load that I can keep constant and switch bullet, to really round out the test matrix.