You're seeing statistical noise. Like ghosts in fog bank.
You might search for the Hornady Podcast episode "Your groups are too small".
They ran a bunch of experiments with groups up to 500 rounds. Short summary.
- Group sizes generally don't get bigger after 50 shots but may not be entirely circular.
- Groups are usually circular by 100 rounds.
- The data suggests you need groups of 35 to predict which load is better (statistical significance).
- Groups of 20 are usually predictive but not always.
- Generally speaking, stick powders are usually more accurate than ball powders.
- Different barrels like different bullets and different powder.
- GROUP SIZE (dispersion) INCREASES WITH VELOCITY!! THEY COULD NOT FIND "VELOCITY NODES" OR "ACCURACY NODES" WITH GROUPS OF 35 OR MORE SHOTS.
-Bottom line, for hunting, pick a bullet type and stick powder with a good reputation for the cartridge. Select the slowest safe load that will deliver the bullet at your target distance at a speed sufficient for terminal performance (eg, 1800 fps for many traditional cup and core bullets). Shoot 20 (or more) shots in a manner that replicated your use. For hunting with a light barrel, 7 or 8 3-shot strings on thr same target allowing the barrel to cool, will give you 21-24 shot replicating a common hunting scenario. If the result is sufficiently good, don't keep looking for something better. If it's not, try a different powder or a different bullet.
Like many people here, I've stared at 3 shot groups like a medium reading tea leaves. In a former life, I taught stats and am convinced the Hornady approach is sound.
I no longer chase accuracy nodes and simply load for just enough velocity.
I listened to the podcast a couple of times then went out and tried this with my Model 70 Classic 30-06. The barrel has been lapped and it has been bedded into a Bansner stock. Scope was a Nightforce SHV 3-10 set to 10x, and I made two each 20-shot groups at 100 yards using Norma 150-grain Whitetail ammo. I used a sandbag front rest and no rear rest.
I was somewhat sloppy with my hold, so the extreme spread for the first three shots in the first group was about 1.5 MOA. Two more did not expand the ES but landed within the spread of the first three. The second five-shot string didn’t increase ES, but clustered together near the first shot. The third and fourth strings gradually opened up the group, with the final ES being 2.209 MOA. Mean radius for this group was 0.655 MOA.
I was more careful with hold on the second group. The first three shots were nearly touching. Two more opened the ES to about 0.75 MOA. The second five-shot string opened the ES to about 1 MOA. That’s 10 in 1 MOA, which was the gold standard for Townsend Whelen back when he said that only accurate rifles are interesting. The third string opened up ES to about 1.5 MOA. All five shots in the fourth string hit to the left of the group and increased ES to 2.326 MOA. I take the rap for that: I had just fired 40 rounds of 30-06 through a sporter-weight rifle from the bench and I could tell that my hold was loosening up again. Mean radius for this group was 0.658 MOA.
Then I chronographed 10 rounds with my MagnetoSpeed chronograph. I had been using 2,749 fps as a working figure based on checking just three rounds. But 10 shots averaged 2,818 fps with an extreme spread of 58 and a standard deviation of 18, so that’s probably closer to the truth.
What did I learn?
• This rifle may be better than I thought, but I need to shoot more groups with at least 10 shots to really understand what I’m seeing.
• I need to improve my bench technique, and should look at using a rear bag for this kind of work.
• A different type of target might help make groups smaller.
• 40 rounds of 30-06 in one day is probably too much.
• If I’ve been thinking that this load did 2,749 fps from this rifle when it’s really closer to 2,818 fps, what else have I been getting wrong?
Okie John