I'm a bit new, but here's what I do - reload, shoot, analyze deeply...

[ELR LVR]

With all due respect this looks like a classic example of over complicating a very simple process.

Was also thinking same

 

[ELR LVR]

I could, but AR barrels are relatively inexpensive. And I'm shooting very light loads. Don't AR barrels last for about 10,000 or so rounds?

This one only has about 500 through it.

Depends, 4140, 4150, Chrome lined? Melonited? Not?
There's a few variables
I wore one out in approx 3000
Was a tack driver with Sierra 69 SMK's
After about 3000 rounds it began spitting them out sideways onto the target

 

[Jager]

Your first 3-shot group was 1.3 MOA. You coulda stopped right there.

 

[joshb]

“ This set was shot in an AR with a 12" CF barrel.”
With that barrel, you’re just wasting your time and components. It can “maybe” be a fun exercise in statistics, if you like that kind of thing.
Personally, I run up thru the charge window and look at my targets. If I get a good group, I THEN dive into the “details” around that charge weight.
Welcome to the forum!
Josh

 

[ShaunBrady]

This thread took a predictable arc. It doesn't seem to happen as often as it used to, but the plot is some Statistics Karen shows up and announces that 3 or 5 shot groups are not enough to completely quantify the ammo or gun capabilities. They're not, but if you have more than a superficial understanding of statistics, you'll know that large groups are not practical and we're proceeding at risk. Some are better at managing that risk, and some are just lucky. Attacking the problem with larger counts frequently introduces systemic effects that further degrades the signal to noise ratio.

The bit of statistics I get the most milage from is the RMS error budget. The first lesson from it is there are many things that scatter the poi of a given shot and we don't know enough of them. The second is there is no point on working on the little things until the larger problems are addressed. I don't think there are really that many BR guys that can ag below 0.2" on demand, even in good conditions, but am awed by the attention to detail it must take. The last bit from the RMS calc is something else the Statistics Karens don't seem to get. If the correlation is only 50%, that doesn't mean the test didn't prove a relationship. It actually shows a decent sized player in a mix of other similar sized players and it needs to be addressed.

The ES is only 2 points regardless of the number of samples thing is also beat to death. The practical application of statistics here would be knock the tails off the sample. Hand loaders call this culling and sorting. ES is a simple, pragmatic approach that has been the standard for heading on 100 years. You don't get to start the accuracy game by changing the rules to satisfy your extremely simplistic view of statistics.

A 12" AR is not an accuracy test fixture. It's a machine that's trying to tie itself in a knot as the bullet moves down the barrel. The pistol length gas port puts a huge, highly variable, kick on the barrel before the bullet leaves. The use of a carbon fiber barrel reduces stiffness and weight, both of which compound the problem. The bullet leaves with an extremely high muzzle pressure squirting around it. A boat tail bullet that intensifies the uncorking problem. A heavy bullet that requires a fast twist amplifying the cast off problem started at the throat. The gas has a huge amount of unburned powder in it that further randomly destabilizes the bullet.

I've used the ShotMarker in a way similar to what the OP proposed. My approach is still evolving, but here are some samples. First, a 300 PRC RPR. Again, not exactly a precision test fixture.



These were all shot into the same hole in two 10 shot groups and then pulled back apart in a spreadsheet. Group 1 is shots 19, 20, and 1. "Observed" is what would have been seen if I used separate aiming points. The "95%" values are plus and minus 2 SD from the average. It's a check on if the round count caught a fair representation of the likely largest and smallest samples.

What started this exercise was the first 3 shot group of 1/4". This gun and load usually shoots 5 shot groups between 3/4" and 1" with 10 shot velocity spreads in the low 20s. There is a ammo temperature test buried in here as well. I'm using Hornady brass, purchased once fired during the pandemic. The theme of this project was to test the feasibility of using a near factory gun in local ELR matches. The ammo was a similar test. Weight sorted, neck turned, pressure suitable for 8ish firings.

We can do other tests, like checking if the zero shifts during or between strings. This is a concern for ELR guns. "Shot" is 5 shot group ending with shot #. Again, shot 1 is in a group with shots 17, 18, 19 and 20. Yea, it's math and not real but for checking shifts in the zero it's useful. For perspective, this is a tenth mil scope with 0.34" adjustments.









On the old school precision end of things; I was intrigued by the introduction of the RPR but not so thrilled by the barrel. Initial ladder shot with lapua 243 brass, 105 Hybrids, H4350 and 210M.



I didn't think there would be much point in calling Ruger and telling them about my self image as an excellent shooter and handloader. These were the first and last 20 shots fired from this barrel. It was replaced with a Bartlein I chambered in 6XC.



First group after 3 to zero. Pretty much the same components except Norma 6XC brass. Midrange first guess pressure load borrowed from another gun that used the same reamer.

The old way worked just fine here.

 

[seymour fish]

Actually, 30+ shots is where the most reliable results are. Three to ten shot are considered a small sample size in statistics and are not expected to be reliable or repeatable. Some even say 50-100 samples are required for reliable, repeatable, predictive indicators.

Hornady did an extensive test, shooting thousands of rounds and carefully measuring everything. Ond of the things they found is that even shooting a precision rifle, mounted in a pedestal, in an indoor range, when they shot many 3 shot groups, like 30+, the group sizes varied by as much as 70%! That's a lot!

For example, if the tightest group was 0.5 MOA, a 70% variation would put the largest group at 1.67 MOA. When they shot large sample sizes, i.e. 30-50 shots per group, the group size variation went waaay down.

With the large sample sizes, again, 30+ per charge, they found that all the powder charges shot essentially the same group size. I find the same thing with my 30 shot groups.

It's a long watch, but well worth the time.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

List all variables controlled for. Ex: wind, bbl condition

 

[Lynn Brase]

wow, what a way to sell more components. Shoot more test groups. you are not shooting enough shots per group! 30 per group may not be enough??

 

[ELR LVR]

I personally, and this is just my personal opinion but I like to test more for real world scenarios
I rarely will be firing 30 shots at something, let alone an animal
maybe 4 shots max if I'm far enough away and the report didnt scare it away.

I obtain the best results with multiple 3 shot groups and compile those into my log book and review them every year
I break out that particular rifle, go test to ensure that load still shoots like it did last time
sometimes as often as every 3 months to ensure that load holds up for each differing season.
and if so, put it back in the safe until needed
if not, adjust the load until it is back to required accuracy for either Comp or Hunting.
So if every single time I break out that certain caliber, and it does as expected
I see no reason to keep wearing the barrel out to prove it can do so for 30 shots
which to me, is unrealistic, even if it is a more accurate test bed statistically
in reality it will never happen
---
the 30 shot test I would agree may be good for an AR and testing accuracy under sustained fire
(seeing how it holds for a full mag)

 

[Tucker 7br]

Three shot groups never get smaller by putting more shots into the group. A lot of effort from the OP to prove that. Actually, I was kinda impressed that 10 consecutive 3 shot groups from a 12" AR on a bipod averaged 1.1 MOA. I don't think the original author looked at it that way. I would've taken the win and called it a day.

 

[kelbro]

Statistics are so much easier on my one shot groups.

 

[jelenko]

This thread took a predictable arc. It doesn't seem to happen as often as it used to, but the plot is some Statistics Karen shows up and announces that 3 or 5 shot groups are not enough to completely quantify the ammo or gun capabilities. They're not, but if you have more than a superficial understanding of statistics, you'll know that large groups are not practical and we're proceeding at risk. Some are better at managing that risk, and some are just lucky. Attacking the problem with larger counts frequently introduces systemic effects that further degrades the signal to noise ratio.

The bit of statistics I get the most milage from is the RMS error budget. The first lesson from it is there are many things that scatter the poi of a given shot and we don't know enough of them. The second is there is no point on working on the little things until the larger problems are addressed. I don't think there are really that many BR guys that can ag below 0.2" on demand, even in good conditions, but am awed by the attention to detail it must take. The last bit from the RMS calc is something else the Statistics Karens don't seem to get. If the correlation is only 50%, that doesn't mean the test didn't prove a relationship. It actually shows a decent sized player in a mix of other similar sized players and it needs to be addressed.

The ES is only 2 points regardless of the number of samples thing is also beat to death. The practical application of statistics here would be knock the tails off the sample. Hand loaders call this culling and sorting. ES is a simple, pragmatic approach that has been the standard for heading on 100 years. You don't get to start the accuracy game by changing the rules to satisfy your extremely simplistic view of statistics.

A 12" AR is not an accuracy test fixture. It's a machine that's trying to tie itself in a knot as the bullet moves down the barrel. The pistol length gas port puts a huge, highly variable, kick on the barrel before the bullet leaves. The use of a carbon fiber barrel reduces stiffness and weight, both of which compound the problem. The bullet leaves with an extremely high muzzle pressure squirting around it. A boat tail bullet that intensifies the uncorking problem. A heavy bullet that requires a fast twist amplifying the cast off problem started at the throat. The gas has a huge amount of unburned powder in it that further randomly destabilizes the bullet.

I've used the ShotMarker in a way similar to what the OP proposed. My approach is still evolving, but here are some samples. First, a 300 PRC RPR. Again, not exactly a precision test fixture.

View attachment 1736584

These were all shot into the same hole in two 10 shot groups and then pulled back apart in a spreadsheet. Group 1 is shots 19, 20, and 1. "Observed" is what would have been seen if I used separate aiming points. The "95%" values are plus and minus 2 SD from the average. It's a check on if the round count caught a fair representation of the likely largest and smallest samples.

What started this exercise was the first 3 shot group of 1/4". This gun and load usually shoots 5 shot groups between 3/4" and 1" with 10 shot velocity spreads in the low 20s. There is a ammo temperature test buried in here as well. I'm using Hornady brass, purchased once fired during the pandemic. The theme of this project was to test the feasibility of using a near factory gun in local ELR matches. The ammo was a similar test. Weight sorted, neck turned, pressure suitable for 8ish firings.

We can do other tests, like checking if the zero shifts during or between strings. This is a concern for ELR guns. "Shot" is 5 shot group ending with shot #. Again, shot 1 is in a group with shots 17, 18, 19 and 20. Yea, it's math and not real but for checking shifts in the zero it's useful. For perspective, this is a tenth mil scope with 0.34" adjustments.

View attachment 1736592




View attachment 1736593


On the old school precision end of things; I was intrigued by the introduction of the RPR but not so thrilled by the barrel. Initial ladder shot with lapua 243 brass, 105 Hybrids, H4350 and 210M.

View attachment 1736595

I didn't think there would be much point in calling Ruger and telling them about my self image as an excellent shooter and handloader. These were the first and last 20 shots fired from this barrel.

View attachment 1736596

First group after 3 to zero. Pretty much the same components except Norma 6XC brass. Midrange first guess pressure load borrowed from another gun that used the same reamer.

The old way worked just fine here.

Just to confirm, the difference between the first and second targets was just a change in brass? Of, was the powder charge also different?
Thanks

 

[jlshaw]

Actually, 30+ shots is where the most reliable results are. Three to ten shot are considered a small sample size in statistics and are not expected to be reliable or repeatable. Some even say 50-100 samples are required for reliable, repeatable, predictive indicators.

Hornady did an extensive test, shooting thousands of rounds and carefully measuring everything. Ond of the things they found is that even shooting a precision rifle, mounted in a pedestal, in an indoor range, when they shot many 3 shot groups, like 30+, the group sizes varied by as much as 70%! That's a lot!

For example, if the tightest group was 0.5 MOA, a 70% variation would put the largest group at 1.67 MOA. When they shot large sample sizes, i.e. 30-50 shots per group, the group size variation went waaay down.

With the large sample sizes, again, 30+ per charge, they found that all the powder charges shot essentially the same group size. I find the same thing with my 30 shot groups.

It's a long watch, but well worth the time.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

A big enough sample size to produce meaningful results?
Statistical rigor?
Tosh!
This is the internet. There's no place here for actual designed experimentation!

 

[LHSmith]

I skimmed through the posts, but did not find any mention of windflags. Without use of a reliable wind indicator and experience reading them -you are just plinking, making it impossible to correlate target results.

 

[joshb]

I think we have a new fGuffey on our hands! Maybe not? He might have left because we questioned his data, procedure and purpose. Engineers have always been a little over involved with “data”.

 

[bozo699]

I think we have a new fGuffey on our hands! Maybe not? He might have left because we questioned his data, procedure and purpose. Engineers have always been a little over involved with “data”.

^^^^^^^^
Good one buddy !….
Wayne

 

[joshb]

I skimmed through the posts, but did not find any mention of windflags. Without use of a reliable wind indicator and experience reading them -you are just plinking, making it impossible to correlate target results.

I went back and read it again. He purports that he took wind out of his equations by shooting at 50 yards. That explains his groups from a 12” barreled AR. I was initially impressed. My 10” barreled AR can barely hit a pie plate at 50.

 

[ShaunBrady]

Just to confirm, the difference between the first and second targets was just a change in brass? Of, was the powder charge also different?
Thanks

That was a barrel change. Factory Ruger 243 Win to custom chambered Bartlein in 6XC. Nothing magical about the cartridge switch, I was just set up for the 6XC with reamer, components and load data.

Premium components, carefully assembled, nominal 1.5" groups at 4 different charges. I can chamber a barrel in about 6 hours. Bartleins were about $350 at the time. There really wasn't much of a decision to make about the fate of that Ruger barrel.

It was intended as an example of no amount of fiddling with components, charges or seating depth was going to fix that barrel. At least not to the level the barrel change would or with nearly as certain an outcome. The final group is another example of no amount of fiddling with components, charges, or seating depth was going to improve the group size. It was now floating on my shooting ability. Once in the ballpark, changes to the ammo would be lost in larger random variations. Conventional hand loading mental models would describe it as parked in the middle of the low node.

What I'm getting out of the analysis on the ShotMarker data is more or less, kinda sorta:

The SD on group size is about 1/4 the average group size.
The Zero floats around about 1 SD.
20 shots substantially capture the group size and its variation
I use 2 SDs on either side as "competition grade" accuracy. It should cover 19/20 cases. This produces a potential variation in group sizes for a given set of components of about 3:1 with 1.5:1 being relatively common.

I'm still playing with it but think it'll go like my concentricity gauge. That is, I'll use it to learn what affects concentricity, implement those changes, test for a while to confirm, and then not use it very often after that. On the other hand, I honestly don't like shooting groups at 100 yards and anything to make it more interesting is likely to stay in my routine.

 

[6MMsteve]

I think we have a new fGuffey on our hands! Maybe not? He might have left because we questioned his data, procedure and purpose. Engineers have always been a little over involved with “dat

Gufferization

 

[JFrank]

Ya’ll understand that the OP didn’t post a ladder test, it was just a bunch of shots using the same random powder charge from an AR ..

I'm loading 223/556 ammo with a light load, 22.0 gr of Winchester StaBall Match. I will load 30 rounds as identical as I can get them and go shoot them. The next day I may use the same components but increase the charge to 22.3 gr,

For any evaluation of consistency, wouldn't you actually need to find the best/repeatable load to compare against itself over several groups rather than just grabbing a load out of wherever ?

 

[joshb]

Here’s my 30 round test! A cousin is getting light primer strikes so I ran some rounds thru it. I got 4 light strikes out of 30 rounds. I put them aside to measure and shot 4 more.
The first splatter target was to check zero on his 3X9 scope. I think this was the weak link. The right splatter target is factory PMC. The left are my handloads.
Then I shot the 30 rounds into the next target.
My analysis is:
The gun shoots pretty good.
Factory ammo sucks.
The gun might perform better with a higher power scope.
Whew! My head hurts! That took 2 minutes to analyze!