The things we do that effect accuracy, or don't

[Blue68f100]

There are so many variables it's going to be a hard test. As we know what works in 1 gun does not necessarily mean it will work in another.
Then there are bolt vs simi-auto, which is a very interesting challenge.

A week or so ago I shot the best group I ever had at 300 yrds, 1" out of my AR 6.5 CM. And only the 2nd time I've shot at 300 yrds. I did check the run out on these just for kicks before I shot them, all < 0.001", avg at 0.0004" . You ran into this early on the reason you went custom dies. I have several cases where I played with OAL. In most cases I can shrink a group but not always. Depending on jump to lands it may have no impact what so ever.

This neck tension discussion (for a better word) has so many variables it's going to be hard to run. I anneal every cycle like you, others only do it every 3-5 time. Those who neck size having the micrometer head allows them to adj the neck to 0.001". How much impact that has is guessing game. I don't even think measuring them with a force gauge would show up over the base number. Then are they annealing every time or, progressively increasing the tension due to work hardening of the brass?

I worked in a R&D center for 22 yrs. The money you have to spend to do test right is outrageously expensive. The company got bought and the lab closed down and sold. So I no longer have access to some very good equipment that could detect the hardness change in the brass. On top of that there are ~10 different alloy of brass which all behave a little different. I had some engineers that could not run a test and duplicate it a second time. It takes someone with a OCD condition to become a very good test technician. All the little things make a difference. Then you have to have a means in which to test. Not all guns shoot the same, particularly with us humans operating it. Then there the issue of having a base line that is consistent day to day, with all the changing environmental variables.

I really think this is one of those exercise that each one need to do for them self to prove there is a change, good or bad.

 

[murf]

What would you guys think about taking a poll for these accuracy variables and let the results guide us as to which to one to tackle first? The criteria would be "which variable affects accuracy the most"

We'd try to keep it simple and standardizes as much as we can by establishing a protocol to perform the test. If the results from that first test (hopefully a few of us trying it) prove to be inconclusive or not repeatable then we would know the exercise as a whole is probably not worth continuing because there are too many other confounding variables

since you are the op, why don't you set the general parameters for this endeavor?

i'm assuming: bolt gun, rifle weight restriction, no tight-neck chambers, no new cases, commercially available components (especially the bullet) only, dollar limit on custom rifle, standardized test: shot quantity, distance, environmental limits. everyone should have to shoot a group baseline using premium factory ammo so everyone can compare group accuracy in relation to a standard in spite of using disparate firearms.

it is going to be hard to get everyone to standardize on a certain process, but you can probably come close. i think if you have us choose the five most important variables and make your list based on that, a decent list should result.

murf

 

[taliv]

no problem. i asked again, pretty much, to bump the thread. it is important, imo.

i consider 30 shots minimum to verify the accuracy of a particular firearm/load combo. would be interested in your number. i think you already expressed that in another thread, but thought you could share that info here.

good brass, like you are using, eliminates a lot of brass prep. the only thing i would add to your list is: make sure all primer pockets are of a uniform depth. i think lock-time consistency is important when going for accuracy.

murf

Sorry I missed this earlier. I think the number varies by application. This has been the subject of much debate. If you are hunting and only need 1-3 rounds you don’t really need to test 30 rounds to know. And your super lightweight hunting bbl May warm up and start throwing rounds all over by the tenth round but really all that tells you is you shouldn’t use that rifle for fclass matches.

If you’re shooting 20 round strings with unlimited sightsee in fclass then yeah I’d want to test with 30 rounds.

I have some rifles that may only have 1000 round barrel life. Eg 300nm. And the rounds cost a few bucks each. 30 round accuracy tests get pretty dang expensive.

I shot 15 rounds in a group in another thread. I really didn’t pick that for an accuracy test. Iirc I loaded 25 or so and had spent the other ten trying to zero and only had one shootnsee dot left so I dumped my last 15 rounds into it. If I were less lazy I’d prob have put a few more dots and made it three 5 round groups which would prob have looked better since I wouldn’t have chewed up my POA which was the right edge of the dot

 

[murf]

Sorry I missed this earlier. I think the number varies by application. This has been the subject of much debate. If you are hunting and only need 1-3 rounds you don’t really need to test 30 rounds to know. And your super lightweight hunting bbl May warm up and start throwing rounds all over by the tenth round but really all that tells you is you shouldn’t use that rifle for fclass matches.

If you’re shooting 20 round strings with unlimited sightsee in fclass then yeah I’d want to test with 30 rounds.

I have some rifles that may only have 1000 round barrel life. Eg 300nm. And the rounds cost a few bucks each. 30 round accuracy tests get pretty dang expensive.

I shot 15 rounds in a group in another thread. I really didn’t pick that for an accuracy test. Iirc I loaded 25 or so and had spent the other ten trying to zero and only had one shootnsee dot left so I dumped my last 15 rounds into it. If I were less lazy I’d prob have put a few more dots and made it three 5 round groups which would prob have looked better since I wouldn’t have chewed up my POA which was the right edge of the dot

thanks for the reply. you could crank up your scope four clicks, shoot your group, then click back down. a good way to check the ranging on the scope and keep your poa at the same time.

i try to use 30 shots (whether ten three shot groups, six five shot, etc.) to find out if i'm going to get fliers, or any other oddities in a particular load. just wanted your opinion here.

murf

 

[taliv]

Yep I do that when I run out of dots usually.

Honestly sample size is prob a question better answered by a six sigma guy than specific to shooting.

 

[taliv]

If I were testing the op theories I’d want a much bigger sample size than if I were testing to see if one rifle was “accurate”

 

[murf]

found a handloader #257 article by gary d. sciuchetti called "developing an accurate .308 load", subtitle: "handloading for accuracy". a great article about this guy testing all the major variables on his way to a sub .5 moa (avg of four 5-shot groups) accuracy. he actually made it with one load. it took a second rifle, a berger bullet, varget powder loaded just under max and individually weighed, a neck sized case (nosler (not new)), a mag primer (cci 250), a precision bullet seater, and a lot of research for a starting load. he did a ton of work to get there. a good read for anyone wanting to work up an accurate load.

if you don't have that issue, you can buy it from wolfe publishing for ten dollars. i think it is well worth the money to read about what it took for mr. sciuchetti to achieve his goal. it might help with deciding whether to continue this quest, or how to continue.

i won't post any of the article here due to copyright laws, but can answer specific questions as i have that issue.

murf

 

[Varminterror]

Honestly sample size is prob a question better answered by a six sigma guy than specific to shooting.

Nah - just a statistician. Us “six sigma guys” have to focus way more on the process of CI than the actual stats, the lowly yellow belts (affectionate tongue in cheek) are the ones running the stats.

I’ll offer an example - we have a thread here with single digit ES on 5 shots. Using only a 95% confidence interval, his actual ES is larger than his statistically predicted confidence range, which is an indicator (a flashing red light with sirens) the sample size is too small for the inherent variation in the product. That’s about 3min in Excel anybody with a working stats knowledge can turn out.

#notmyonlyblackbelt

 

[denton]

Required sample size depends on how small a change you want to detect, the natural variation of the data, and the type of data you have. As a general rule, measures of location (mean, median, group center) require smaller samples than measures of dispersion (standard deviation, range, "extreme spread", or group size).

There is a common incorrect notion among shooters that if you're only going to shoot three shots at a target, then you don't need more than three shots to check the group size. The real question is, how well does a group with X shots predict the long term performance of the system? As I mentioned earlier, the average of three five-shot groups will get you an estimate of that within about plus or minus 25% of "truth". If you want to cut the estimate to about plus or minus 12.5%, you'll need the average of 12 groups. Also note that a firearm that, on average, prints 1" groups will routinely print groups as small as 1/2" and as large as 1 1/2" with no change whatever in the rifle, cartridge, shooter, or conditions.

If your sample size is too small, you can easily end up with a result that is not much better than a coin flip.

 

[Nature Boy]

found a handloader #257 article by gary d. sciuchetti called "developing an accurate .308 load", subtitle: "handloading for accuracy". a great article about this guy testing all the major variables on his way to a sub .5 moa (avg of four 5-shot groups) accuracy. he actually made it with one load. it took a second rifle, a berger bullet, varget powder loaded just under max and individually weighed, a neck sized case (nosler (not new)), a mag primer (cci 250), a precision bullet seater, and a lot of research for a starting load. he did a ton of work to get there. a good read for anyone wanting to work up an accurate load.

if you don't have that issue, you can buy it from wolfe publishing for ten dollars. i think it is well worth the money to read about what it took for mr. sciuchetti to achieve his goal. it might help with deciding whether to continue this quest, or how to continue.

i won't post any of the article here due to copyright laws, but can answer specific questions as i have that issue.

murf

@Blue68f100 sent me a copy of that a while back. It’s a really good example of an analytical attempt at finding the best load by looking at a lot of the variables we have discussed. I need to go re-read that again.

 

[Legionnaire]

Really interesting thread. And thanks for the Sciuchetti reference; I just ordered the back issue of Handloader.

 

[taliv]

i never read "handloader" but man, "precision shooting" was awesome

 

[fotheringill]

Re: Seating depth. Definitely a factor. Once you figure it out by using a Berger like seating depth test, refine it and then hope to hell you don't need to start all over again as your barrel erodes and then again when you finally rebarrel.

 

[Walkalong]

i never read "handloader" but man, "precision shooting" was awesome

I got them both, and both were good for what they were targeting. Precision Shooting was much more specialized.

 

[murf]

here's another article that may help with the "neck tension" debate: http://www.6mmbr.com/jgcaseprep.html

murf

 

[Nature Boy]

There are big things and little things. There should be a way to get at these "things" in a controlled fashion that will benefit the average reloading enthusiast, however, I concede that this all might be a bridge to far. If so, I do my thing, he does his thing, the other guy does something else and we debate about what matters and what doesn't, over and over. I wish there was a way to verify it empirically

You have to have a way of isolating the variable you're trying to test for or randomizing out all the other influencing variables. I'm still not sure how to accomplish that. Frankly, with work and personal life intruding I haven't had the bandwidth to dedicate much time and energy to it. I'm not giving up though

Here's what makes this exercise all the more challenging. I have to be consistent enough in my shooting technique for any test of reloading variables to show up. The interface of my shooting hand with the rifle and where I put my thumb (across the grip/on top/or riding along the side) likely has more influence on consistency than many of the reloading variables on this list.

 

[Walkalong]

Here's what makes this exercise all the more challenging. I have to be consistent enough in my shooting technique for any test of reloading variables to show up.

That is everyone's dilemma.

 

[Varminterror]

The number of degrees of freedom are simply too wide. As you discussed above - when one guy does one thing with his rifle, if another guy repeats it with his, there’s at LEAST three variables: time of execution, the rifle used, and the experimenter themself. And of course you then also throw in age of barrel, age of brass, gear they loaded on, etc.

Even if the same observer conducts comparative testing with the same set of brass in the same rifle and the same loading gear, you’re still subject to the variability within the shooter, the offset for barrel age, and environmental conditions for the two components of the test.

So there’s no way to eliminate variables.

The ONLY way then, in experimental method, to determine the contribution of different variables is to perform an exhaustive study of multivariable combinations.

The best example we have at hand of this data set can be seen by studying competition data and trends. There are subjective bias trends in these games, such as sponsorship influence, “keeping up with the joneses” pressures, and passing fashionable fads, but by and large, we see competitors using what has been proven to be capable of winning. How do we know a $300 factory rifle won’t shoot better than a $3000+ Custom? Because competitors wouldn’t spend money on high dollar custom rifles if a handful of guys came in over the years and won matches with factory rifles at a fraction of the price. How do we know Lapua brass yields greater precision than R-P? Because guys aren’t winning matches on cheap brass.

The guys with the years of experience, dozens of rifles, and trophies on the wall have tried a lot of things, and have learned from the guys who came before them who tried other things.

I’ve been watching this evolution in the relatively YOUNG game of PRS. As the game evolved in the early 20-teens, we saw the variability of cartridges making the finals year over year shrink dramatically. Then Within a couple of years we saw that list of 6.5mm cartridges shift over to a majority of 6mm’s. For that application, it’s fair now to say a 30caliber cartridge can’t compete in that format. There are a lot of reasons why that game might not be a true test of raw precision potential, BUT, there are games which are a more true test.

So when you look at unlimited classes at different ranges, what do we see? 6-6.5mm cartridges, Lapua brass, Berger bullets. Sierra and Nosler bullets do win some, as do Nosler and Norma brass. Brand of action, barrel, trigger, and stock, and the smith who built them all change almost at random, but on average, there’s just not much deviation in the multivariable combination of ammo components which win matches.

It’s NOT as simple of “different rifles act differently, but there ARE more than one answer to the question of precision.

 

[murf]

i think the best thing to recommend here is what gary sciuchetti said in his article: "go with the best". meaning: try different types of a specific component for a specific load and "go with the best". e.g. try a bunch of primers and go with the primer that consistently results in the smallest groups.

pretty much what varminterror is saying above.

murf