Shooters Forum banner
61 - 80 of 99 Posts

·
Super Moderator
Joined
·
13,751 Posts
This could be a case of a bit of statistical illusion. Generally speaking, the standard formula for sample standard deviation doesn't work well for sample sizes smaller than 8. Since your samples are just 6, you're more likely to get a valid SD by dividing the extreme spread by zye of n, which is 2.534 for n=6. So for your first load your first extreme spread of 26 has an SD of 10.3 by that method, and your ES of 21 likely corresponds to an SD of 8.3. 9.3 is your average for that pair and that is very good.

It is important to remember that SD is an estimate of what your population standard deviation (sigma) will turn out to be, even if it is comprised of an infinite number of samples. Small samples offer less foundation to base your guess at the final population standard deviation on, so they vary more from one to the next than a larger sample would do.
 
  • Like
Reactions: Darkker

·
Registered
Joined
·
258 Posts
Discussion Starter · #62 ·
Thank you for that clarification unclenick. Math formulas were never my strong suite and the numbers I used are what was provided by the Magneto Speed. Do understand the sample size ratios. Though on a much simpler note I was more looking at ES and again hoping to get it below 50FPS which mostly seems to have happened.
The one ES of 51 I am not quite certain what happened there. I am suspecting a weak powder charge as that one round out of 36 was that low. Then when looked at as a whole of 6 shots that spread became even greater at 59. The other two loads when looked at as a whole of 6 were a ES of 38 and 39.
 

·
The Hog Whisperer (Administrator)
Joined
·
37,408 Posts
You really need larger sample sizes, before you hang your hat on those ES numbers.....

What were the groups like?
 

·
Registered
Joined
·
258 Posts
Discussion Starter · #64 ·
You really need larger sample sizes, before you hang your hat on those ES numbers.....

What were the groups like?
Yes I understand that and mostly just using these numbers as a reference. Sampling changes. What this last test today show was that there is significant improvement with the changes I made. That what I did actually worked! Also these last few tests were all done with the same loads while making one significant change so that I could compare this test to the previous ones.

Actually the groups are good and if it was some one other than me working the trigger they would probably be even better! Also this group of loads has also produced the best groups overall which is why I am using then to test changes.
 

·
The Shadow (Moderator)
Joined
·
8,891 Posts
If the sample size is no good, you can't use the numbers as a reference for noting improvement; it's statistical noise.

Cheers
 

·
Registered
Joined
·
114 Posts
Kevin,

I’m going to be more blunt. You are searching for answers in numbers that have very little meaning. The SDs and ESs that you generate are for a small statistically insignificant number of samples. While your tested numbers may have improved you really don't know much at all. for the 23 gr group with a 14 SD the 95% confiodence on the SD is between 7.3 and 88FPS. For the second it between 5.2 and 62.8 FPS. To make matters worse, you are shooting 6 rounds, supposedly the same but analyzing in groups of three. They should be analyzed together. At least your confidence interval would be smaller.

Generating test data is a waste of time unless time is taken to analyze what it means. While it involves some math you should read the articles posted on precisionrifleblog. It is a three-part series and is probably the best article I have seen on the subject.


In the graphs below group 1 was a ten shot sample of Blazer 22LR ammo I shot over a chronograph to demonstrate just what we have been discussing. The error bars represent the 95% confidence interval for mean and standard deviation. What that means is there is a 95% chance that if I shot all 500 rounds in that box over the chronograph, the average velocity and standard deviation would fall within those extremes. You can see when the ten shots are analyzed they yield a reasonable confidence interval for the mean velocity and a fair range for standard deviation. Groups 2,3, and 4 are random 3 shot groups of shots from the ten shot group. You can see that if only those shots are analyzed (or were shot) the prediction can be quite different. Groups 5 and 6 are based on random 5 shot groups of the 10 group 1 shots.




102376
 

·
Registered
Joined
·
36 Posts
I used a chrono to develop loads for many years but have pretty much stopped. I don't own a chrono today. It seemed to be overly tedious and of less and less value to me. I am not a benchrest shooter nor paper puncher other than that necessary to develop an accurate load . I have many heavy barreled varmint rifles. Too many. But my two favorites are a twenty pound 6MM Imp and a fifteen pound 6BR. Both are super, super accurate varmint rifles I use to hunt sage rats and coyotes. I never chrono any of the loads I use in them. I don't care about trying to get the fastest load possible. I load for best accuracy, not speed. I start with a safe mid-range load and work up, always recording the accuracy results. When I start to see accuracy fall off, I go back to the last accurate load and use it. In terms of those caliber's, I am always significantly below the max for a given powder, etc. I am not saying the chrono is not useful here, only that I think it can become a task that is not always necessary. What matters is accuracy, not the velocity. I get that accuracy is a function of velocity and other factors. Many times I use AmmoGuide to find loads that seem to work for a majority of shooters and use them as a basis for my load testing. Although this observation is largely antidotal, I rarely see or hear anyone say that their rifle has its best accuracy node at or near max velocity. If you are a varmint hunter like I am, the goal is different than shooting competition benchrest. Many times these forum discussions are a mixture of benchrest and varmint hunters and game hunters and the arguments about load testing kind of all run together and the focus is often confused. The sage rat getting hit by a 55 grain VMAX at 3,000 fps really doesn't care what the speed of the bullet was. The point is he got hit. Varmint hunting for me is a fun sport and I would rather hunt than spend time recording chrono data that is of no value to me. Not trying to step on toes here or commit some "process hearsay". If you enjoy crunching the numbers and looking for the absolute perfect load, then do so and enjoy it. Thanks for reading my two cents worth.
 

·
Registered
Joined
·
258 Posts
Discussion Starter · #68 ·
If the sample size is no good, you can't use the numbers as a reference for noting improvement; it's statistical noise.

Cheers
So then I have 3 loads all the same, 23.0gr, 23.2gr, 23.4gr all loaded with all the same components and all shot in the same rifle but at different times for a total of 48 recordings each though not fired at the same time does that make a sample pool for each load? Or do they all need to be loaded and fired consecutively to be a sample?
 

·
The Hog Whisperer (Administrator)
Joined
·
37,408 Posts
You can pool all your data, but unless the loads were prepared identically.... then you have some variables. If they were as identical as you can make them, then I don't see a problem. It's just more data points.
 

·
Registered
Joined
·
114 Posts
Unless they are loaded at the same time it not a good idea to pool data. That said, if you are consistent, the error introduced should be small. Being shot at different times can introduce additional factors, most typically powder temperature, the magnitude of which varies with powder choice.
 

·
Registered
Joined
·
258 Posts
Discussion Starter · #71 ·
So then what I an understanding is that unless all rounds are loaded Identically at the same time in large samples and then fired at the same time then all samples are useless. NOISE. And prove nothing and cannot be trusted.
 

·
The Hog Whisperer (Administrator)
Joined
·
37,408 Posts
Well.... I don't know about "useless." Yes, you want to control variables as best you can. But I have shot the same load over the chronograph at different dates, sometimes years apart, and usually the numbers are within the expected range. So, have to use your judgment here.

I've also plotted out charge weight increases vs. velocity .... the expectation being that as charge weights go up, so should velocity. Here's an example of both: some surplus WC844 data in the .35 Rem, two different batches of ammo loaded and shot at different times, yet remarkably consistent. And then another series with H335. Since WC844 is a surplus powder, kind of on my own to figure out what the data should be. The H335 data was pretty consistent with published data.

Never mind this doesn't have anything to do with ES or SD numbers, but it shows that sometimes data pooled together can be useful. The straight line is a least squares fit to the data points (the data points are averages of several shots, I forget how many exactly).
102392
102393
 

·
Registered
Joined
·
258 Posts
Discussion Starter · #73 ·
So the recent testing I did was with those 3 loads that I have loaded many times exactly the same to determine repeatability. which it did. Then made a change to the loading by using annealed brass as I wanted again to determine if this procedure made a difference. What I found that even with this Small Sample size, I was able to determine that yes it did make a difference. I did not need 100 samples of each load to determine this. All three loads improved significantly over the loads that were loaded with brass that was not annealed.

So now someone tell me this test was invalid or meaningless.
 

·
The Hog Whisperer (Administrator)
Joined
·
37,408 Posts
It isn't meaningless, but it isn't entirely definitive, either. The odds of getting a false conclusion go up as the sample size goes down. Dig into "Student's t-test" if you want to learn more about small sample sizes.

Generally, I think 30 samples is considered a pretty solid basis to draw firm conclusions on.... but you could do that with ten each of three loadings and feel reasonably good.

There is a chance, however small, that the new sample just gave some low numbers by random chance. And that's the rub with statistics; it's not only about "chance" but the "chance of a chance." Without the higher sample numbers, the probability of a correct conclusion starts to go down.
 

·
Registered
Joined
·
114 Posts
Well folks I am still relatively new at this rifle stuff. Almost a year now and just started with the chrono about 6 weeks ago.
Before the chrono I was just going by how they were grouping and looking for that accuracy node and not a speed node. But of course all the bench shooters insisted that I needed to know the numbers. Now I'm not so sure.

Part of the reason for all this is because I will be shooting with our range regulars in an informal winter league. They shoot 10 weeks and vary between 100, 200 and 300yds and use a variety of target sizes and shapes and this year they are adding a week to shoot weak side. A number of these guys are shooting quarter size groups at 300yds. so I figure I will be a donator to the wolves!

So when I started with the chrono and tracking numbers it was mostly out of curiosity as to whether I was close to book numbers and if my loads were consistent. Well this has added a whole new perspective to consistent!

Anyways I appreciate everyone insight and contribution.
Best I can do at our range is 300 yards and I am still chicken to do that. Have shot it a few times but just now getting comfortable and confidence at 200. I was going to set up a target at 300 last time but it started raining.

I went through this thread again and picked out the above relative to this entire thread. First you need to have realistic expections for precision and accuracy. You do not have the equipment to compete with guys shooting 1/4” groups at 300yds. That kind of precision is not going to come from a Savage 10 regardless of ammunition. That is 3 shots in one ragged hole at 100 yds. I doubt the rifle is capable of much better that 2” 5 shot groups at 300yds.

1/4” groups at 300 yds? If they can shoot it, that kind of ability takes years of experience and requires wind reading skills that only come with trigger time. The first quote leads me to believe that you are a fairly new shooter. Instead of shooting and worrying about speed and SD or ES, you would be better served by careful reloding and working on shooting skills. 100 fps ES is about 1” total deviation at 300 yds. That’s .3” at 100 yds and. .7”at 200.

The consensus of the recommendations here has been to put the chronograph away. It is not going to help you achieve what you are trying to achieve. When you can load and shoot better tha 1” five shot groups consistently at 100yds then you may want to look at your loads. Until then, load your rounds in quality, consistent brass, consistently sized, either neck only or full length. Use quality bullets like the SMK. For 300yds you may want to stick to flat based bullets like the 52gr Berger or 53gr SMK.

If you don’t have one, purchase a bullet maker’s reloading manual. It really doesn’t matter which one, and follow there reloading procedures. Be careful when taking and implementing reloading advice at the range.

For load development I would recommend using Dan Newberry’s OCW method. It doesn’t depend on a chronograph and yields consistently accurate ammunition.

I hope this helps. Shooting and reloading should be fun and enjoyable. Chasing statistics and probability usually is neither of these.
 

·
The Hog Whisperer (Administrator)
Joined
·
37,408 Posts
Pretty good summary Doom. I do use my chronograph working up loads.... I want to ensure that the velocities are realistic (ie. neither too high which might indicate a pressure problem, nor too low which might impede bullet performance). Personally I do like the OCW method; the chronograph helps target specific barrel times although I can't say definitively that it is the end-all and be-all of accuracy methods. But it helps to have a starting point, and a plan.

Especially with surplus powders or working with wildcats / obscure cartridges, the chronograph is a vital tool for keeping out of the ditch, plus common sense. But I loaded a lot of good ammunition before I owned one.

Seat bullets straight, once all the (careful) brass prep is done, and you'd be surprised how much more accurate all your loads will be no matter the powder/primer and other variables. One of the concentricity gages (I have the RCBS tool but they're all basically the same) would be vital for anyone who is looking for accurate loads, and if I had to give up either that tool or the chronograph, I'd give up the chronograph. It's just difficult to shoot good groups with crooked ammunition, and off-the-shelf loading dies can give some surprisingly bad results at times. Either that or go with (expensive) in-line, bench rest quality dies.

Till you are seeing more vertical spread than horizontal spread in the groups.... wind is more of a concern than ES, and the chronograph won't help read wind.

Get some trigger time before your head starts hurting with all of the statistical stuff ;)
 

·
Registered
Joined
·
258 Posts
Discussion Starter · #78 ·
It isn't meaningless, but it isn't entirely definitive, either. The odds of getting a false conclusion go up as the sample size goes down. Dig into "Student's t-test" if you want to learn more about small sample sizes.

There is a chance, however small, that the new sample just gave some low numbers by random chance. And that's the rub with statistics; it's not only about "chance" but the "chance of a chance." Without the higher sample numbers, the probability of a correct conclusion starts to go down.
Small Sample Size, I understand that and am not disputing that. A sample of 2 isn't really a sample. Unless maybe you are comparing for taste!

Also understand "Random Chance".
Back a couple weeks ago I received a three shot group that was 2810, 2810, 2807 chrono told me SD1, ES3. Great numbers for certain but I put zero faith into it. Anomaly, out-layer. Just happened to be one were everything came together perfectly as you suggest. The very next group of that same loading was 2818, 2839, 2798. Back to normal!


I went through this thread again and picked out the above relative to this entire thread. First you need to have realistic expections for precision and accuracy. You do not have the equipment to compete with guys shooting 1/4” groups at 300yds.

I hope this helps. Shooting and reloading should be fun and enjoyable. Chasing statistics and probability usually is neither of these.
Thanks, but you seem to omit or pass over the number of times I reference my understanding of the Limitations I have with, the gun, the components, experience in both shooting rifles and reloading rifle cartridges. I also mention many times tat I know those that I will be shooting against outclass me by a whole nother level or 3!

I'm not looking for a Speed or a Perfect load, Speed means nothing to me and I even mentioned that a couple times. I was even told by a Long range Bench shooter that yes numbers are great and have tremendous value at 1kyds. At 200/300 not so much. And I trust him. I have watched him shoot.

What I am looking for is consistency in my loads. Having a shot in each group that is consistently 100fps higher or lower in not consistency by anyones imagination. Most all of this recent testing has been with the same loads, same components as a control group, Base Line, and then looking to see if some of the changes I've made have had an effect. Not looking for a number in this just changes.

Again I stated this before and will do it again here. Yes I am having fun, I'm learning much and I am getting much needed trigger time. Through all this my ammo is getting better, my groups are getting better and my shooting is getting better.

So you can all disagree with my method and tell me it's a waste of time or I'm doing it all wrong and there are Better ways. That may be true but what I am doing is working for me.
OH and BTW I do have an assortment of different manufactures manuals. But then there is Book Learning and then there is Practical Learning.
 

·
Elk Whisperer (Super Moderator)
Joined
·
11,593 Posts
Best load for my .243. Velocity? SD, EV, PTO, PTA, RPM?



Yeah, I can chronograph the load but will it make it less or more accurate 🤔

Yes I do have and use a chronograph but not as much as I used to. I guess I got over my curiosity.

You guys go ahead and do it however you like though. I'll watch.

RJ
 
  • Like
Reactions: KMW1954

·
Super Moderator
Joined
·
13,751 Posts
Mike,

Your two velocity ladders show the other common use, which is to find a place where a change in load has minimal effect on velocity, and between shots 6 and 7 on both your graphs, you appear to have the makings of a flat spot that might be more apparent with finer load increments near the charge values that produced them. I don't know if those two plots represent the same string of charge weights, but WC844 and H335 are the same powder in bulk and canister grade, respectively, so it wouldn't be too surprising to find that flat spot at about the same place, assuming your lot of WC844 doesn't have one of the more extreme variations in burn rate that bulk grade powders can have.

Part of the value of 300-yard shooting is just being able to distinguish velocity and barrel vibration flat spots clearly by how they affect vertical stringing. If you shoot, say, a 168-grain MatchKing at 2650 fps (Federal GM308M test barrel velocity), a 25 fps velocity change only makes about 0.05" difference in elevation at 100 yards, about 0.2" at 200, but makes a clearer 0.5" difference at 300, so it is just easier to see. If the wind is moving your shots around a little at that range, you can shoot a string of same-charge loads in different wind gusts to see how much vertical aerodynamic jump is affecting elevation. Run a diagonal trendline through them all to find its angle. Then when you shoot the ladder, draw a line at that same angle line through every hole, then put a single vertical line through the poa and use the place where those diagonal lines meet the vertical as you shot placement locations. Extra bother, but neutralizes the affect of wind on the ladder.
 
61 - 80 of 99 Posts
Top