Extreme Spread, Standard Deviation

  • 512 Views
  • Last Post 2 weeks ago
Wm Cook posted this 19 March 2022

Does standard deviation and extreme spread make you lean one way or the other when working up a load.

This past while I've been working on finding a competitive load for a new production .308 bolt gun to use in Production class.  This will be limited to 100 and 200 yards.  Right now I think I have two powders and two bullets that may hold their own if I hold up my end of the deal.

Just for the heck of it, last night I went back and looked at the ES, SD to see how the two powders compared to the other powders I tried. 

Note: I was shooting powders that I already had on the shelf and would have liked to have had a few others but it is what it is.  All groups were three and 5 shot. The two powders that produced the best accuracy are shaded.  I'd appreciate your thoughts, Bill.

Attached Files

  • Liked by
  • John Alexander
Order By: Standard | Newest | Votes
Larry Gibson posted this 19 March 2022

To be honest, the ES/SD of single 3 or even 5 shot tests is just about as meaningful as the group sizes if single 3 or 5 shot tests.  Under certain test parameters they may tell us we are headed in the right direction.....maybe.   A minimal test of ten shots can provide a meaningful ES/SD.  Even then, with the 10 shot test another two or three 10 shot tests is best to confirm the validity of the first test results. 

To answer the question; Yes i do use ES/SD of several 10 shot tests along with the group sizes at 100 and 200 yards as load selection criteria.

LMG

Concealment is not cover.........

Attached Files

MP1886 posted this 19 March 2022

Larry we've been told that the lowest ES and SD are the most accurate load. This is true SOMETIMES, but not all the time. More then often I've seen the worse ES and SD give the smallest group.  I too like you use ES and SD.  One area I use it in is to see how efficiently the powder is buring and my consistancy in assembling the componants. 

Attached Files

  • Liked by
  • RicinYakima
  • Bud Hyett
John Alexander posted this 20 March 2022

I agree that the large ED and SD sometimes gives good groups.  Conversely, the very lowest ED may not shoot well

However, I can't remember anybody except maybe something in the high circulation, low quality gunmags. claiming that lowest ES and SD gives the most accurate load.  I certainly don't believe it is a wide spread belief among CBA shooters.

John

Attached Files

Larry Gibson posted this 20 March 2022

MP1886

I agree for the most part.  Chasing the ES/SD god can be very disappointing.  I think it is because the idea/myth is put forth the lowest ES/SD is necessary for good groups.  That is indeed not necessarily true.  First what is needed, yet is most often not had, is a sufficient sample to give a 95+/- % surety of what the true ES and SD are.  [Note; the only way to know the real ES/SD of a lot of ammunition is to test/shoot the entire lot.  That is not practical by any means.]  Thus, is why I mention to Wm Cook that 3 or even 5 shot tests are not valid. 

Second is the bullet must match the barrel twist an velocity.  If it does not then you can have an excellent ES/SD yet have very poor accuracy.  As an example shooting Federal .308W Match with a 175 gr MK out of a 14" twist barrel will get you very good ES/SD yet poor accuracy.  That is simply because the twist is too slow for the MK bullet. 

Third, the "lowest" ES/SD is not always a reasonable expectation. A reasonable expectation of ES/SD is usually higher than many assume.  With a sample size of 10 shots I have chronographed and pressure tested enough match 30-06, 308W/7.62 and 223/5,56 factory/arsenal loads to realize an ES of 50 fps with an SD of 15 to 25 fps is an excellent load and, given quality match bullets and an accurate rifle, are indeed "match" loads. Many excellent 3 and 5 shot groups are fired with what is assumed to be a poor ES/SD when actually the ES?SD is well within a reasonable good ES/SD.  How many times have you or someone you know gone to the range and worked up a load with 3 or 5 shot test increments.  They find the one that gives them a clover leaf group along with what they think is a great ES/SD of double/single digits and yell "Ureka!".  They go home and load up a whole box of 50 of that load then next trip to the range the group size is twice as big if not more and the ES/SD is three times as large.....  They/you are befuddled and don't understand why.  The why is simply the first sample size of 3 or 5 rounds was not large enough to give the real picture of the probable ES/SD or the group size.

Lastly, if the cast bullet is over it's RPM threshold it can have super excellent internal ballistics (ES/SD) yet be miserably inaccurate.

I also have chronographed and pressure tested thousands of cast bullet loads.  I have seen some very low ES/SDs and with quality bullets and an accurate rifle they all shot well at 100 yards with most also at 200+ yards.  Conversely, I have also seen, with quality bullets and an accurate rifle, many loads with not so good ES/SDs that seemed to be accurate at 100 yards but when tested at 200 or 300 yards the bottom falls out.  To be sure I have seen some pretty lousy loads seemingly perform extremely well at 50 yards and even 100 yards they can be "okay".  However, given quality bullets jacketed or cast and an accurate rifle those loads giving the best ES/SDs from a reasonable test sample will prove to be the most accurate in the long run and, particularly, at long ranges.  

When I test rifle loads, I consider the ES/SD ratio.  For a good load I consider the lowest reasonable ES/SD giving the best groups.  The SD to ES ratio should be, given a minimal 10 shot test, between 30 and 45% of the ES.  If the ratio of SD to ES is smaller or larger than that 30 - 45% If the ration is not within that i look for anomalies in the group and in the individual shot data.  

If one is only shooting 50 yards or even just 100 yards chasing the lowest ES/SD god can indeed be misleading.

LMG

Concealment is not cover.........

Attached Files

RicinYakima posted this 20 March 2022

Lyman still annotates their reloading manuals with low ES/SD as the "most accurate" load. Even they know it isn't true, but they have publish that information for over 60 years, and can't give it up, according to Tom Griffin the editor. 

Attached Files

  • Liked by
  • Bud Hyett
Wm Cook posted this 20 March 2022

Personally, if it’s a jump ball between two loads that seem to shoot decent groups I’ll start with the one that has the lowest ES, SD and use the backup load if things start to get rocky as the day goes along.

I think tiny standard deviations are like shiny objects. Sometimes they’re a diamond laying loose on the ground, other times it’s a headlight coming at you.

Before you buy a particular powder in bulk you need more than a few tiny groups that have tiny SD’s. Been there, done that one. Thanks, Bill C.

Attached Files

  • Liked by
  • John Alexander
  • Maven
RicinYakima posted this 20 March 2022

Bill C., I like 10 ten-shot groups before I pull the trigger on buy powder by the keg. FWIW

Attached Files

  • Liked by
  • Bud Hyett
  • John Alexander
MP1886 posted this 20 March 2022

MP1886

I agree for the most part.  Chasing the ES/SD god can be very disappointing.  I think it is because the idea/myth is put forth the lowest ES/SD is necessary for good groups.  That is indeed not necessarily true.  First what is needed, yet is most often not had, is a sufficient sample to give a 95+/- % surety of what the true ES and SD are.  [Note; the only way to know the real ES/SD of a lot of ammunition is to test/shoot the entire lot.  That is not practical by any means.]  Thus, is why I mention to Wm Cook that 3 or even 5 shot tests are not valid. 

Second is the bullet must match the barrel twist an velocity.  If it does not then you can have an excellent ES/SD yet have very poor accuracy.  As an example shooting Federal .308W Match with a 175 gr MK out of a 14" twist barrel will get you very good ES/SD yet poor accuracy.  That is simply because the twist is too slow for the MK bullet. 

Third, the "lowest" ES/SD is not always a reasonable expectation. A reasonable expectation of ES/SD is usually higher than many assume.  With a sample size of 10 shots I have chronographed and pressure tested enough match 30-06, 308W/7.62 and 223/5,56 factory/arsenal loads to realize an ES of 50 fps with an SD of 15 to 25 fps is an excellent load and, given quality match bullets and an accurate rifle, are indeed "match" loads. Many excellent 3 and 5 shot groups are fired with what is assumed to be a poor ES/SD when actually the ES?SD is well within a reasonable good ES/SD.  How many times have you or someone you know gone to the range and worked up a load with 3 or 5 shot test increments.  They find the one that gives them a clover leaf group along with what they think is a great ES/SD of double/single digits and yell "Ureka!".  They go home and load up a whole box of 50 of that load then next trip to the range the group size is twice as big if not more and the ES/SD is three times as large.....  They/you are befuddled and don't understand why.  The why is simply the first sample size of 3 or 5 rounds was not large enough to give the real picture of the probable ES/SD or the group size.

Lastly, if the cast bullet is over it's RPM threshold it can have super excellent internal ballistics (ES/SD) yet be miserably inaccurate.

I also have chronographed and pressure tested thousands of cast bullet loads.  I have seen some very low ES/SDs and with quality bullets and an accurate rifle they all shot well at 100 yards with most also at 200+ yards.  Conversely, I have also seen, with quality bullets and an accurate rifle, many loads with not so good ES/SDs that seemed to be accurate at 100 yards but when tested at 200 or 300 yards the bottom falls out.  To be sure I have seen some pretty lousy loads seemingly perform extremely well at 50 yards and even 100 yards they can be "okay".  However, given quality bullets jacketed or cast and an accurate rifle those loads giving the best ES/SDs from a reasonable test sample will prove to be the most accurate in the long run and, particularly, at long ranges.  

When I test rifle loads, I consider the ES/SD ratio.  For a good load I consider the lowest reasonable ES/SD giving the best groups.  The SD to ES ratio should be, given a minimal 10 shot test, between 30 and 45% of the ES.  If the ratio of SD to ES is smaller or larger than that 30 - 45% If the ration is not within that i look for anomalies in the group and in the individual shot data.  

If one is only shooting 50 yards or even just 100 yards chasing the lowest ES/SD god can indeed be misleading.

LMG

 

I see that a lot when you go back out and the load isn't so great.  You recorded the first out in your book and when you refer back it to load more and you go out and it's terrible. 

I have an 8x57 Mauser that shoots smaller moa's at 465 yards then it does at 100 yards. Although I haven't placed target in increments out to the target, I would love to know the bullet is going to sleep.  BTW it's the Lyman Loverin 213 grain Spitzer. Has a really high BC.  I have it in both the original Lyman and NOE and they both shoot pretty much the same. 

Attached Files

joeb33050 posted this 20 March 2022

Extreme Spread is called the “Range” in Stats talk, largest # - smallest#.

Chronometers tell us “Standard Deviation”, SD.

Both are ESTIMATES of the true Standard Deviation.

Like Kilometers and Miles.

The ratio of SD/Range varies as n, the sample number: number of shots.

 

n          SD/Range

2          1.28

3          1.69

4          2.06

5          2.33

6          2.53

7          2.70

8          2.85

9          2.97

10        3.08

 

Measure a zillion samples of n = 10, calculate SD/Range for each sample if 10, and average SD/Range = 3.08.  

 

If the ratio of chronograph numbers is not around the ratio above, probably more samples/shots

are needed.

 

If 5 shots are chronographed, and SD/ES is < 2.06 or > 2.53, something is amiss, and more shots are probably needed.

 

This is true for groups and bullet weights and all continuous distributions that are sorta Normal.

I have more info on this if needed.

Attached Files

Larry Gibson posted this 20 March 2022

joeb33050's excellent post on calculated SD/range matches very close to, if not precisely with, the 30-45 % ratio range i found by actual testing of many, many chronographed loads over several years.  

An interesting observation I've made is the group distribution of shots vs the SD ratio.  This observation is not a 100% given but falls into the most often observed category.  The more groups (10 shot) of a given load that is fired the more likely the shot dispersion I'll mention will be observed.

If we observe a group (10 shots) that has an SD ratio less than 25% of the ES we most often see a small cluster of 7 to 9 shots with 1 to 3 shots "out" of that cluster.

If we observe a group with an SD ratio of 25 - 35% the group, given an accurate load in an accurate rifle under the RPM threshold, we will see a nice tight group with all shots relatively close together.  How close is dependent on the accuracy potential (cone of fire) of the rifle.  This is the range of SD/ES ratio we should strive for.

If we observe a group with an SD ration of 35 - 45% the group, given an accurate load in an accurate rifle under the RPM threshold, we will see a group that is not the best for accuracy potential (cone of fire) of the rifle such as a 2" group in a 1.25" capable rifle.  However, the shots will be pretty evenly distributed about the group and while not particularly "accurate" it may still be useful.  I'm coming to think some powders are more prone to this than others.  

Again, this may be observed with just one group (10 shots) fired and maybe not.  Fire several 10 shot groups with the same load in the same rifle and more than likely the pattern commensurate with the SD/ES ratio will become observable.  Of course, the adverse effects of wind and called flyers must be taken into consideration.  

In my opinion the reason for such distribution patterns is the closer the SD/ES ratio falls in the 25 - 35% range the more likely the bullet exit from the muzzle is at the same barrel node location.  Of course, the heavier, stiffer barrels have less barrel "node" movement so the shot distribution may be less.  The above is most noticeable in thinner sporter barrels using heavier cast bullets with faster velocities because they cause a larger barrel "node".  Just my opinion on the "why" this is observable.

LMG

Concealment is not cover.........

Attached Files

lotech posted this 20 March 2022

With cast or jacketed bullets, low SD and ES numbers are certainly an indication of consistency but not necessarily accuracy. Lots of other factors influence accuracy. I concentrate on group size, paying little attention to the SD and ES figures. However, I suspect the SD and ES numbers may take on real meaning at longer ranges, beyond 200 yards, but I have not verified this. Perhaps others have. 

If you want to see very low SD and ES numbers, try compressed loads. I've noticed this with H4831 and jacketed bullets, but don't recall such loads being the most accurate. 

Attached Files

  • Liked by
  • Bud Hyett
  • John Alexander
Larry Gibson posted this 20 March 2022

Lotech

 

Absolutely correct.  There are many other factors that can contribute to good or bad accuracy.  Still the SD/ES ratio is something to pay attention to, especially at longer ranges as you mention.  

As an example, I would not have been able to shoot this 11 shot 5.9" group at 600 yards with a load not having a low SD/ES and excellent SD/ES ratio.  Rifle was my 30x60 XCB with the NOE 310-165-FN XCB bullet at 2900 fps.  The last 10 shot test this load gave a SD/ES of 8/26 fps which has a ratio of .307....excellent.

 

 

Concealment is not cover.........

Attached Files

  • Liked by
  • RicinYakima
Bud Hyett posted this 20 March 2022

Sometimes I think we are on a technical bend that is fruitless in determinant evaluations. We work too hard to justify the methodology without regard for the results. The final arbiter is the results on the target for the competitor or in the field for hunting. 

I liked the old reloading manuals for jacketed bullets, they gave the maximum velocity for each powder. Assuming this was the most efficient load, I worked up while watching the muzzle to see when the load "smoked" with telltale wisps of powder smoke even if this was not the maximum load. If this was not the precision I wanted, I went to the next fastest powder. If this did not work, I went to the the next slowest powder. After several powders on each side, I accurized by trading.

This will not work for cast as we can have the lubricant smoking if we use too much. I like a light lube star on the muzzle to assure me that the lube is working and that enough lube remains in the barrel to give the same bore conditions for the next shot. Beyond that, I believe in the lube purge theory of Tom Gray. Enough lubricant is just enough.

And there are my peccadilloes. I shot over 500 test loads in a .22 Hornet Ruger #1B simply because I would not give up. Ended up with a compressed load of W-W 680 and a Berger 35 grain bullet. I was afraid I was going to wear the barrel out before getting a good load. 

Aerospace has a saying, "At some point, we must shoot the engineers and begin production." 

Farm boy from Illinois, living in the magical Pacific Northwest

Attached Files

John Alexander posted this 20 March 2022

Thanks Joe.

 Another Joe B. useful stat table to help me have a clearer picture about how sorta normal distributions manifest on the chronograph screen or the target.

This one goes up in the reloading room along several of your others.

Thanks,

John

Attached Files

John Alexander posted this 20 March 2022

Larry sez: "How many times have you or someone you know gone to the range and worked up a load with 3 or 5 shot test increments.  They find the one that gives them a clover leaf group along with what they think is a great ES/SD of double/single digits and yell "Ureka!".  They go home and load up a whole box of 50 of that load then next trip to the range the group size is twice as big if not more and the ES/SD is three times as large.....  They/you are befuddled and don't understand why.  The why is simply the first sample size of 3 or 5 rounds was not large enough to give the real picture of the probable ES/SD or the group size."

===================

WOW! THE TRUTH IN PLAIN ENGLISH.THANKS LARRY.  I believe most of us are smart enough to understand this. Yet we don't act like it.

Many keep on using a single pair of five shot groups for what they call load development, or using single thee shot or five shots in a so called ladder test.

However, CB shooters don't believe in magic as much as JB benchkresters who mostly have bought into a "fine tuning" method that relies entirely on a single three shot group telling the truth about what that load will do in a match -- a procedure that is only one step better than predicting the future by studying chicken entrails which also used to be widely believed and popular.

Homo sapiens are a very peculiar animal.

John

 

 

 

Attached Files

  • Liked by
  • RicinYakima
4and1 posted this 20 March 2022

I am biting my tongue........

Attached Files

Wm Cook posted this 21 March 2022

What do you think would be a fair expectation for how many rounds need to be fired to have a load dialed in where it's ready for competition?

Lets assume the shooter if reputable, the rig has proven itself and you've just screwed on a new barrel.  Thanks, Bill.

Attached Files

RicinYakima posted this 21 March 2022

I have come down to one 25 shot group, no time limit. That should capture all of the "outliers". Rifle and pistol. 

Attached Files

  • Liked by
  • John Alexander
  • Maven
Wm Cook posted this 21 March 2022

Well, I'm impressed. I'd like to bend your ear about that for a half hour or so.  Thanks, Bill.

 

Attached Files

Larry Gibson posted this 21 March 2022

With proven cast bullet loads for the cartridge, I pretty much do what RicinYakima does.  I shoot 5 foulers then 20 shots for group with no time limit but shooting fast enough to keep the fouling "soft".  That's usually a shot every 2 - 3 minutes.  That 20 shot group will tell if that load is good in that rifle.  If that group is acceptable then, depending on how acceptable, I might try another test of that same load with a test on each side with a .5 gr + and - difference.

If developing or attempting to "dial in" a load with a different powder unproven powder or with a different bullet, then you can shoot an awful lot of rounds.  I prefer to go with a couple, maybe three, known performing powder(?) if testing a new bullet.  Probably will expend 50 to 100 rounds with a powder "dialing in" a load. 

Actually, I am trying a new powder in my .308W match rifle.  I have 50 rounds (5 different loads) loaded up in 1/2 gr incremental charges to chronograph and shoot for group. If the groups and SD/ES are acceptable i will then do the 25 round test of what appears to be the best one.

Bottom line answer is I will expend maybe about 100 rounds with a proven load testing a new rifle/barrel.  However, many times known proven loads will perform as good as it gets and thus only the 25 rounds with each load would be expended testing.  If working/dialing up a load perhaps 150 - 200 or less rounds may be expended depending on the performance.    

Frankly, I am not one to shoot 2, 3 or 5 shots out of a new barrel and then jump into a competition with it.  

I have seen numerous BR competitors at a BR matches (jacketed bullet BR matches) put a new barrel on, shoot 2 - 3 shots then shoot the match.  Never saw a winner do that but have heard many moan and groan at the "wailing wall" (score board) because the new barrel shot "lousy"...... 

LMG

Concealment is not cover.........

Attached Files

  • Liked by
  • RicinYakima
Show More Posts
Close