I'd like to kick off this blog series with a walk-through of my load testing experience at the range last Sunday. While future articles will dive into the details of a specific concept, I think it will be helpful to first offer some context and demonstrate how objective thinking leads to practical results.
I fired a match at 600 meters last Saturday and wasn't happy with the grouping. I was using roughly the same load I had been last year which performed well, but whether it was the warm weather or a new lot of Varget I started in December, something just seemed off.
Many times I have gone to the range with a plan, left with a conclusion, but by the next morning after running some numbers, there were still open questions. This time I decided not to leave until I was happy, period. I need confidence in a load that will take me through the summer, and I'll do absolutely anything for it.
I brought 110 cases primed with BR-2, and everything I need to measure powder and seat bullets. It's essential to bring what you need to load at the range. It only takes a few shots to identify a bad group, but you'll need many, many more to prove a good one.
You may notice the Autotrickler is missing. For range days where I only load 5 at a time, it's faster to use a manual trickler than to automate. I always use my scale, but screw a Hornady plastic trickler to the wind panel base which is convenient and light.
Today I set up two Two-Box Chronos and a LabRadar. My secondary objective was to collect data from all three and try to measure the precision of the LabRadar. Unfortunately the LabRadar only recorded 5 shots until it gave an error that the "transmitter frequency is out of range" and would no longer arm.
For the Two-Box at 15-foot spacing, I just needed a pillow on one end to get them level. With crude but reasonable alignment they reported as expected, with a consistent difference of 9-11 fps for almost all shots.
Let's see what we're working with.
I use Berger 185 Juggernauts with Varget. For my targets, I aim at the circle and the bullets land on the cross, so my sight picture is always clear. The distance is about 115 yards, so 1 moa is a bit larger than the distance from the circle to the cross.
First I fired some leftover ammo from last match, where I had loaded charges of 43.6, 44.0, and 44.4 at 13 jump. This result confirms my suspicion that the groups are less than impressive.
My interpretation of these groups is that they all belong to the same population, meaning I believe the 44.0 group is coincidentally small. The group just so happens to be low. I've already shot two matches at 44.0 and the result was mediocre, so I believe it was just chance that 7 shots fell in the bottom part of the group.
Velocities of these groups look reasonable. I usually have an SD around 6, and no shots show anything outside of expectation. For now, I will simply collect velocity data until I have anything statistical to say, and keep an eye out for outliers.
Where to go from here?
At this point I should explain my underlying theory of how rifles work. I have working hypotheses that, within reasonable limits for pressure and jump:
This model allows you to separate and measure the variables independently, rather than having to test every load combination for both SD and group size. It drastically reduces the amount of data you need to have statistical confidence in a test result.
This theory requires a heavy, target barrel. Barrel whip is significant factor for hunting rifles and AR-15s, but changing velocity will not move your POI around with barrels typically used in F-Class.
I plan to elaborate on this in a future post, but for now, just think of it as a heuristic to aid in decision making. Whether it's actually correct or not doesn't matter at the moment, as we are just at the range trying to find something that works, and any theory that gets us there faster is welcome.
Back to the challenge at hand, clearly changing jump is prudent, because the group size needs to improve. I haven't tried jamming these bullets, and many people say, in general, touching the lands is a good place to start. For charge, I felt last week that 44.4 held the tightest elevation at distance, so we might as well try going in that direction and see what happens.
Now we seem to be getting somewhere. Here are two 10-shot groups at 5 jump and 2 jam. Velocities are falling in line between 2791 and 2817, having an SD of 6.
This is encouraging, but what happens if we keep going? Increase both jam and powder charge, find where things fall apart and then we know the limit to stay away from.
So I increased the charge to 45 grains. Four shots are tight, but one spiked up to 2843 fps and flew way out of the group! The primer had some distinct cratering, so I think I found the pressure limit, and now I know what not to do.
Now that I found the upper limit, I was curious: what happens if I still touch the lands but bring the charge way down to 43.0 and 43.5? It's a good idea to test all sorts of ideas to learn how your rifle behaves, even if you think you've found something you like.
Well that's quite unexpected. Velocities are consistent, but the groups are ugly despite being near the lands. Remember, I assume powder charge should not affect group size, and being near the lands seemed to produce tight groups. One of these assumptions must be incorrect.
To get to the bottom of this, we can go back up in charge, say 44.3. If the groups are still good, then maybe we have to accept that powder charge caused the large groups. If they are bad, then we must add this data to the earlier result and perhaps touching the lands is not so special.
This looks a lot like it did before. Not bad, 20 shots around 1/2 moa. Unfortunately my theory about the nature of rifle accuracy takes a beating, as this seems to confirm that 44.3 produces a different group than 43.5.
Oh, one more thing...
I could pack up and leave now, having fired 40 shots between 44.3 and 44.5 at +/- 5 jump. I'd leave with what appears to be a 1/2 moa group and question that my theory about powder charge and seating depth is wrong. I wouldn't say I'm over the moon with this result but it's surely acceptable.
At this point I asked myself a question. If I fired one more group, do I feel confident that it would be just as good? Or am I scared that more testing just means more chance of finding out it's not actually good at all?
Intuitively, we don't want to "push our luck". But that kind of thinking is counter productive. A load is what it is and more testing is always a good thing.
So, with plenty of primed cases left and not much else to do today, I fired 20 more shots into three groups:
Throughout this string, I was thinking, "okay, this isn't great but I can live with it". Until fate grabbed onto the very last bullet and yanked it so far outside the group that I could barely comprehend what had happened.
I walked down to the targets and just sat there on the ground, thinking, and staring at the paper. I labeled each group and racked my brain for an interpretation that explains the result.
We have 59 shots in a group I was willing to accept, plus one that speaks a different language. I don't believe in fliers. Fliers are statistical outliers which are an expected part of a normal distribution. When you have a shot like this in 60, it screams at you that something is wrong. I can only make so many excuses for a so-so load before I have to face the fact that this is not actually "good enough".
Here's what those 60 shots look like, overlayed into a single group:
Now we see the truth plain as day. This load sucks. That "flier" on the right simply joins a low shot as well as the top right corner shot that we didn't think about without the context of all the data together.
Calling that one shot a flier and accepting that result with false confidence would have been a huge mistake. We are only human after all.
Back to the drawing board.
Still sitting on the ground at the targets, I looked for something new I hadn't tried yet. Realizing that 'at the lands' is actually bad, every group so far is ruled out, and the only thing left to try is more jump. If that doesn't work, I'm in serious trouble.
In other news, my theory about charge and seating depth being independent is brought back to life! Disproving the theory required that 44.3 was good at the lands while 43.5 was not. The world is beginning to make sense again.
Calling on some superstition, I loaded my trusty charge of 44.0 that got me through 2016, and pushed the bullets out to 23 and 28 jump.
Aha! This feels good. Now I will change to 44.3 to take advantage of all the velocity data collected so far, and keep going out to 33 jump to see if I'm at a limit:
From 23 to 33 we have amazing 5-shot groups. At this point I only have 10 primed cases left, so I loaded 44.3 and 25 jump for verification. I felt a lot more confident this time around that these last tests would confirm the load rather than ruin it.
Success! The last group (9) is wider than the others, but there was quite a bit of wind later in the afternoon and it seemed to be pulling the first four shots in that group progressively to the right. Then, for the fifth shot, suddenly mirage appeared, the wind dropped, and the last bullet popped back into the middle (which is the left of this group). If I had more cases I would have kept shooting, but I'm pretty confident this was the case here.
Bringing it all together.
Overlaying all the shots for jammed, 13 jump, and 23+ jump, we have these combined plots. The hash marks are 1/4 moa.
I came to the range with 13 jump, and settled on 25. The elevation spread is about 3/8 moa, and that's through some mirage. That's about as good a result as I could ever hope for.
I settled on 44.3 grains after confirming consistency up to 44.5, with problems appearing at 45. More velocity is better so I'd rather shoot 44.3 than 44.0, all else equal. I should be in the 2780 fps range, which is great for these bullets.
Have we statistically proven an improvement? Based on the group size from 13 and 25+ jump, there is a 99.8% chance that they are from different populations, meaning the observed difference is extremely unlikely to be random chance. That calculation uses the F-test, which I will cover in detail later.
This graph groups the velocity SD across powder charges, and displays an 80% confidence interval (based on number of shots). This is something I plan to dive into on a future post, but at first glance we can see the SD is fairly consistent across the full range, with the exception of the 5 shots at 45.0.
For 104 shots from 44.0 to 44.5, the SD is 6.7, with high confidence. With this load I can expect an 95% extreme spread of 0.87 moa at 900 meters, which is pretty good. The SD is a tad high, which limits the performance at long range, but I can work with that. Now that I know my SD is insensitive to powder charge, I can move past fine tuning charge and try weighing primers or newer cases to see if there is a minor improvement to be made (which will require a very precise chrono).
This exercise helped me settle on the load I will likely use at the F-Class World Championships this year, and I hope it inspires you to run a similar test yourself. Your rifle might be capable of a lot more than you realize.
Many people worry over the details of reloading that don't actually matter that much. As long as you are consistent and careful in your reloading methods, you will find the most improvement from efficiently sampling powder charge and seating depth, and analyzing data for statistical conclusions.
If you have some chronograph data and photos of groups, send me your data and I'll have a look. I'll post what I think in the comments. Thanks for reading!