• Watch Out for Scammers!

    We've now added a color code for all accounts. Orange accounts are new members, Blue are full members, and Green are Supporters. If you get a message about a sale from an orange account, make sure you pay attention before sending any money!

Ignoring ES/SD numbers.....

Re: Ignoring ES/SD numbers.....

Lovely, you reference a single shot and the numbers one could pull from it. Then, in previous posts, you discuss what is "good enough" for tactical shooters versus what a Prometheus is capable of and how it isn't needed. Yet, you haven't actually shot using one, citing you know the use of a dildo and you don't want a dildo either.

Now I am ignorant because I have seen the capabilities of the machine and you haven't bothered with it.

So, how many rounds was my tests group and what where my numbers?
 
Re: Ignoring ES/SD numbers.....

Well, it's obvious that complexity and subtlety are lost on you.

Carry on with what you are doing. It's obviously working well for you.
 
Re: Ignoring ES/SD numbers.....

Lindy, you can't look at an SD and derive an ES. That is why people shouldn't use ES in the first place.

Again, the cutoff for "range" is completely arbitrary. Is it where 90% of occurrences fall, 95%? 97.5%? 99%? 95% is the most commonly used in science, unless you are a climate scientist, in which case, you use 90% to overstate your case.

I agree with the point of the article, which is that you should use SD, not ES. I always disregard ES whenever people post them... there is no way to derive the ES someone got from their SD's. The ES is random and based on 2 shots (that statement coming straight from the article).

As for the range being + infinity to - infinity, you are right, the distribution probably is not gaussian. As the old saying goes, we use gaussian distributions because we can solve for them. My guess is that a log normal distribution would more closely approximate what we should find... but why jump through all those hoops? Just calculate a SD and be done with it.

By the way... all of these statistics for MV apply to groups as well. It is much better to figure out the SD of the distances to center than the groups size. Group size is just an ES of shots.
 
Re: Ignoring ES/SD numbers.....

CKA wanted to send flowers to you on Valentines Lindy....but then thought better of it and went to his local Sex shop for a B.O.B.

he has asked me to ask you what color you want and and if 110volt or batteries....lmk

also.....will this fit in your PO box or will we have to send to some other location in time for Saturday.....?
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Carter Mayfield</div><div class="ubbcode-body">Lindy, you can't look at an SD and derive an ES. That is why people shouldn't use ES in the first place.</div></div>

Carter, I'm curious here... are you saying that its *not* generally possible to take an SD value and extrapolate that the ES values should fall somewhere within +/- 2 SD approximately 95% of the time, etc.? Would it be fair to say that the key word there is 'within'... the ES numbers may *not* always be equal to the 'worst case' prediction of +/- 2 SD (95%) or +/- 3 SD (99.7%), that they *could* be some thing less than the maximum spread? In my limited experience, as the sample size goes up, ES (or range) usually does approach the theoretical values described above - but again there is that weasel word 'usually'...

<div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">Again, the cutoff for "range" is completely arbitrary. Is it where 90% of occurrences fall, 95%? 97.5%? 99%? 95% is the most commonly used in science, unless you are a climate scientist, in which case, you use 90% to overstate your case.</div></div>

That is an interesting question, one which I've thought about from time to time. Many people (by which I mean shooters/reloaders) want 99.999% but don't seem to like the wide ugly spread that is usually necessary to achieve that level of certainty. Is 95% adequate for our purposes, given that there is always going to be a little variation due to factors outside our control? For the sake of maintaining one's sanity I would say yes... I like fiddling with some of this stuff more than most, but after a point all that really matters is what shows on the target down range
wink.gif


FWIW, I do agree with you about the 'value' of using extreme spread for MV and group size. My way of looking at it is that the 'extreme spread', or more formally 'range', only accounts for two data points - which by their very definition are the two most extreme examples i.e. 'outliers' and the least reliable in terms of use in predicting future results.


Interesting discussion, despite some distractions (boltripper...)
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: JRose</div><div class="ubbcode-body">I cannot get good ES/SD numbers on any of my loads, yet they continue to be very accurate, even at range. When doing load development, sometimes I will get good ES/SD numbers on a particular load, but then it seems the most accurate load might not have the best numbers.

So I've said to hell with it and started ignoring them, is this a mistake?

Examples:

My CM load, it is 44.0 gr of RL-17 with a 139 hBN Scenar, is EXTREMELY accurate, but ES/SD numbers are not real desirable.

My .243, 47.5 gr RL-25 with a 115 hBNDTAC, developed by George, is a hammer... Shitty ES/SD numbers....


:? </div></div>

JRose,

I've got some thoughts on tightening your SD.

When you first got your 6.5CM It looked to me like you were disappointed with the max velocity you were getting.What was it,I can't remember 2650-2700 FPS with 139's ???

Then you ensued upon achieving faster velocities with R-17 and ended up with almost 2900 FPS yes/no ???

I use allot of R-17 in my 6x47L.I noticed that the SD went way up in the teens when I pushed the velocities.It didn't take much more powder to cause this.

I bet whats happening is that you have found a higher velocity node in your barrel causing good accuracy but also high SD's.

So I suggest backing the load off until you find a lower accuracy node.Doing this brought my SD's into the 7's.Remember you are only going to loose a few clicks of the turrets at distance.

Also some other things I can think of that might help.

Experiment with neck tension.Some believe more neck tension reduces SD.

Keep your brass trimmed.

When seating bullets,cull any brass which has obviously more or less neck tension than the others.

I know weighing cases is a PITA but it helps to identify the extra heavy or light cases.I believe it does roughly effect case volume,especially if there's a wide spread in weight.

Or just don't worry about all this.When I saw your hits on your steel at 1000Y I didn't see any vertical problems
smile.gif


Steve
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: memilanuk</div><div class="ubbcode-body">

Carter, I'm curious here... are you saying that its *not* generally possible to take an SD value and extrapolate that the ES values should fall somewhere within +/- 2 SD approximately 95% of the time, etc.? Would it be fair to say that the key word there is 'within'... the ES numbers may *not* always be equal to the 'worst case' prediction of +/- 2 SD (95%) or +/- 3 SD (99.7%), that they *could* be some thing less than the maximum spread? In my limited experience, as the sample size goes up, ES (or range) usually does approach the theoretical values described above - but again there is that weasel word 'usually'...
</div></div>

My point is that you cannot mathematically derive what ES "should" be. When we calculate SD, that tells us what shape the bell curve should look like, then we can talk about the probability of a given shot falling within a range. But how do I know that in a given sample whether I have an incident that is within the 90th percentile or the 60th? I don't know. That is why we don't use ranges.

Now Lindy is right that because of the way that we calculate SD in the first place, ES cannot be less than 2 SD's. CKA's ES is greater than 2 SD's, but just barely. I can't say that it is impossible that he got those figures... is it suspiciously close to the limit? Yes... but we derive a percent of incidences number because we just don't know. ES is highly sensitive, in fact, completely dependent upon the outliers. SD is not, or at least is less so, which is why it makes more sense to use SD.

Before I go any further, I need to note that we are dealing with two things: the sample, which is what we are taking measurements on, and the population, which includes all shots, measured, not measured, and yet to be fired. The SD of the sample is analagous to the SD of the population, and in fact, after N=30 for a given sample, the SD of the sample should very closely approximate the SD of the population. For N<30, you can use a correction factor to make the SD larger (the SD of a sample tends to underestimate the SD of the population up to a certain number of observations).

The SD of the sample tells us something about the SD of the population. The range of a sample tells us nothing about the range of the population.

Now, because of the way that we calculate SD, it is impossible for the ES to be less than 2 SD. CKA's ES is suspiciously small compared to SD, but I can't call BS on it, because maybe his sample falls in the 99th percentile of samples... in statistics, you can't really rule something out, you can only call it unlikely or improbable.

Anyway, the big takeaway is that you should use SD because it can apply the measurements that you have taken to the population of all shots, including the ones you will be shooting in the future. ES of the sample cannot be compred to ES of the population... nor can it down-convert to SD. ES is what it is.

..........

As for 95% vs. 90%, etc., this has limited usefulness for us, though it does have soem usefulness. For instance, in the Smack the Smiley Contest, I know based on SD's for my 22LR at 50 yards, that I can hit a 1/8" target about 75% of the time. The chances of making that shot twice in a row is about 56%. So if I get 4 targets, my chances of hitting are 1 - (1-.56)*(1-.56)*1-.56)*(1-.56) = 96%. So if I burn 4 targets, I am likely to hit my confidence shot 96% of the time... I'll take that shot. If I had a less accurate stick and could hit that 1/8" target 40% of the time, then the chance of hitting twice in a row is about 16% and the chance of doing it with 4 targets is 50.2%. I have a 50% chance of losing the match with that strategy... of course, I can get into expected values and all of that to figure out whether I am better off taking the confidence shot at the small smiley or the next bigger target, but I am not going to get into it. (BTW... this is also condition dependent... I burned 2 targets this time around before figuring out that the wind was playing me).

In the field, knowing the percentages can tell you whether you should take the shot or stalk closer at a given distance.

MV is different... we use SD to make a set of shots analogous to another set of shots. You can do something in the reloading room and then figure out if it makes your SD smaller or not. SD is only helpful in respect to your goals and/or for comparison to see whether something you are doing in the reloading room makes a difference in your shooting. Confidence intervals don't serve a lot of purpose here. The main point of using SD's in this case is to know if you are moving in the right direction. If you are using ES, you are likely to be thrown off because you are using a less perfect measure. Basically, one errant shot might tell you that you have done something wrong where you really haven't. Or the lack of an errant shot might tell you that you are doing something right when you really aren't.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: steve123</div><div class="ubbcode-body"><div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: JRose</div><div class="ubbcode-body">I cannot get good ES/SD numbers on any of my loads, yet they continue to be very accurate, even at range. When doing load development, sometimes I will get good ES/SD numbers on a particular load, but then it seems the most accurate load might not have the best numbers.

So I've said to hell with it and started ignoring them, is this a mistake?

Examples:

My CM load, it is 44.0 gr of RL-17 with a 139 hBN Scenar, is EXTREMELY accurate, but ES/SD numbers are not real desirable.

My .243, 47.5 gr RL-25 with a 115 hBNDTAC, developed by George, is a hammer... Shitty ES/SD numbers....


:? </div></div>

JRose,

I've got some thoughts on tightening your SD.

When you first got your 6.5CM It looked to me like you were disappointed with the max velocity you were getting.What was it,I can't remember 2650-2700 FPS with 139's ???

Then you ensued upon achieving faster velocities with R-17 and ended up with almost 2900 FPS yes/no ???

I use allot of R-17 in my 6x47L.I noticed that the SD went way up in the teens when I pushed the velocities.It didn't take much more powder to cause this.

I bet whats happening is that you have found a higher velocity node in your barrel causing good accuracy but also high SD's.

So I suggest backing the load off until you find a lower accuracy node.Doing this brought my SD's into the 7's.Remember you are only going to loose a few clicks of the turrets at distance.

Also some other things I can think of that might help.

Experiment with neck tension.Some believe more neck tension reduces SD.

Keep your brass trimmed.

When seating bullets,cull any brass which has obviously more or less neck tension than the others.

I know weighing cases is a PITA but it helps to identify the extra heavy or light cases.I believe it does roughly effect case volume,especially if there's a wide spread in weight.

Or just don't worry about all this.When I saw your hits on your steel at 1000Y I didn't see any vertical problems
smile.gif


Steve </div></div>

Steve,

Thanks for the tips.

I laddered both H4350, and RL17 loads from all the way down to the mid 2600fps velocity up to 2900. I saw about the same ES/SD numbers on all, even when I would find an accuracy node at various speeds, they'd still have mediocre numbers.

I'm by no means unhappy about the performance of the load/rifle, I was just curious as to why something that is shooting so well, wouldn't have good numbers.

I haven't trimmed my brass yet for two reasons, 1: It was new brass, or 1x fired. and 2: My CM brass won't fit in the .308 collet for my Wilson trimmer, I have to get one made.

I took another guy's advice (thanks DocB) and cleaned the inside of the case necks with steel wool.

I have noticed varying neck tensions, but to be honest, I'm too cheap to cull out brass that works... LOL

Right now my neck tension is way to light, but I have a tighter bushing on order to tighten that up a bit.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">Lindy, you can't look at an SD and derive an ES. That is why people shouldn't use ES in the first place.</div></div>

Agree completely. However, one thing you can know about any normally distibuted data set is that the ES is <span style="font-weight: bold">very</span> unlikely to be less than 4 times the SD, and more likely to be 6 times it. But I agree that it's a completely pointless number.

<div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">It is much better to figure out the SD of the distances to center than the groups size.</div></div>

That is much better. But when I want to see how a rifle and load are shooting, what I want to know is the greatest distance from one shot to the next shot. I don't try to measure that shooting groups. Instead, I shoot 5 shots on 5 aimpoints, and measure the distance from the center of the aimpoint of each shot.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Carter Mayfield</div><div class="ubbcode-body">
The SD of the sample tells us something about the SD of the population. The range of a sample tells us nothing about the range of the population.</div></div>

Amen. If there was one point I wish more people understood about ES vs. SD, that'd be it. Low ES in any given string is nice, but doesn't mean much for future strings, which for me is the whole point of running rounds across a chronograph in the first place - to get a reasonable prediction of what to expect from the rounds yet to be fired.

I'll just say that in my limited experience, about the only time I see ES/SD numbers such as the ones CKA posted are when I'm using a fairly small sample size, in the 5-7rd, *maybe* 10rd, range. Generally by the time I get to a 10rd string, and almost definitely by the time I get to a 20rd string, the ES or range value is much, much closer to +/- 2 SD or more. Anything is possible, however improbable.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Lindy</div><div class="ubbcode-body">
That is much better. But when I want to see how a rifle and load are shooting, what I want to know is the greatest distance from one shot to the next shot. I don't try to measure that shooting groups. Instead, I shoot 5 shots on 5 aimpoints, and measure the distance from the center of the aimpoint of each shot.
</div></div>

That's a damned good way to do it... BUT a) it assumes you are zeroed and b) you will be changing your position shot-to-shot, so there is a little more skill involved there. I prefer to take the shooter out of it... it is much easier to shoot a group of 5 than at 5 different aimpoints. That said, if you are doing this at 100 yards, those shots will be right on top of each other, making your method superior at close range.

Skill-wise, I am sure you have the aimpoint thing knocked. I just see it as a variable we can eliminate by firing 5 shots without breaking cheek weld or moving position. The only thing you have to worry about at that point
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: memilanuk</div><div class="ubbcode-body">
I'll just say that in my limited experience, about the only time I see ES/SD numbers such as the ones CKA posted are when I'm using a fairly small sample size, in the 5-7rd, *maybe* 10rd, range. Generally by the time I get to a 10rd string, and almost definitely by the time I get to a 20rd string, the ES or range value is much, much closer to +/- 2 SD or more. Anything is possible, however improbable.</div></div>

This was the point that Lindy was trying to make, but I felt the need to correct him for saying that the "actual ES" is larger. The ES for the population is not knowable (or even really estimable), and the ES of the sample is what it is. If CKA said that his SD was 5 and his ES was 6... yeah, that is pretty much impossible because of the way that SD is calculated.

In CKA's case, likely, the low relative ES is from not sampling a lot of shots... and if the selection of shots is small, the SD is likely underestimated as well and needs either to be corrected upwards to reflect the small sample size or more shots need to be fired.

If this is based on 20 shots, it is still possible... but who cares. ES is meaningless for this very reason. Anything is possible.
 
Re: Ignoring ES/SD numbers.....

I feel that doing the things that help are as important as the numbers themselves. I mean, once you've done what you can, the numbers aren't telling you anything you can do something about anyway.

Greg
 
Re: Ignoring ES/SD numbers.....

This has been a very, very interesting discussion. It's also made me remember how much I disliked my statistics class at TCU.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: JRose</div><div class="ubbcode-body"><div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: steve123</div><div class="ubbcode-body"><div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: JRose</div><div class="ubbcode-body">I cannot get good ES/SD numbers on any of my loads, yet they continue to be very accurate, even at range. When doing load development, sometimes I will get good ES/SD numbers on a particular load, but then it seems the most accurate load might not have the best numbers.

So I've said to hell with it and started ignoring them, is this a mistake?

Examples:

My CM load, it is 44.0 gr of RL-17 with a 139 hBN Scenar, is EXTREMELY accurate, but ES/SD numbers are not real desirable.

My .243, 47.5 gr RL-25 with a 115 hBNDTAC, developed by George, is a hammer... Shitty ES/SD numbers....


:? </div></div>

JRose,

I've got some thoughts on tightening your SD.

When you first got your 6.5CM It looked to me like you were disappointed with the max velocity you were getting.What was it,I can't remember 2650-2700 FPS with 139's ???

Then you ensued upon achieving faster velocities with R-17 and ended up with almost 2900 FPS yes/no ???

I use allot of R-17 in my 6x47L.I noticed that the SD went way up in the teens when I pushed the velocities.It didn't take much more powder to cause this.

I bet whats happening is that you have found a higher velocity node in your barrel causing good accuracy but also high SD's.

So I suggest backing the load off until you find a lower accuracy node.Doing this brought my SD's into the 7's.Remember you are only going to loose a few clicks of the turrets at distance.

Also some other things I can think of that might help.

Experiment with neck tension.Some believe more neck tension reduces SD.

Keep your brass trimmed.

When seating bullets,cull any brass which has obviously more or less neck tension than the others.

I know weighing cases is a PITA but it helps to identify the extra heavy or light cases.I believe it does roughly effect case volume,especially if there's a wide spread in weight.

Or just don't worry about all this.When I saw your hits on your steel at 1000Y I didn't see any vertical problems
smile.gif


Steve </div></div>

Steve,

Thanks for the tips.

I laddered both H4350, and RL17 loads from all the way down to the mid 2600fps velocity up to 2900. I saw about the same ES/SD numbers on all, even when I would find an accuracy node at various speeds, they'd still have mediocre numbers.

I'm by no means unhappy about the performance of the load/rifle, I was just curious as to why something that is shooting so well, wouldn't have good numbers.

I haven't trimmed my brass yet for two reasons, 1: It was new brass, or 1x fired. and 2: My CM brass won't fit in the .308 collet for my Wilson trimmer, I have to get one made.

I took another guy's advice (thanks DocB) and cleaned the inside of the case necks with steel wool.

I have noticed varying neck tensions, but to be honest, I'm too cheap to cull out brass that works... LOL

Right now my neck tension is way to light, but I have a tighter bushing on order to tighten that up a bit.

</div></div>

Yeah,No problem.Let us know if the new bushings help.

Steve
 
Re: Ignoring ES/SD numbers.....

Alright, I fear the culprit may be my neck tensions. This Hornady brass has no where near consistent neck walls. I got the proper bushing, the way I always do, I mic a loaded round, then subtract .002". I'm getting all kind of varying neck tensions when I push in bullets. Out of 60 pieces, 10 couldn't even hold a bullet, the other 50 where either very light, or just enough. I mic'd the inside of the necks, and they were off a few thou from one another, even though my OD's were pretty damn consistent to my bushing.

Guess I'll be turning necks now...
frown.gif
 
Re: Ignoring ES/SD numbers.....

Geez J...What a hassle.

Oh well,by the time you're done you'll have really consistent neck tension and very uniform neck wall thickness.

Steve
 
Re: Ignoring ES/SD numbers.....

Ok, I fear that i'm slightly threadjacking, but yet it's been mentioned already, however I don't see an answer. Is light, or heavy neck tension conducive to better SD's?

I've long thought that light neck tension was the key, please correct me if i'm wrong. I'm about to finally start working on varying neck tension here soon. I'm currently in the fire forming stage on new brass.

Branden
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Lindy</div><div class="ubbcode-body"><div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">I've been curious if you have chrono data for your load over a bunch of rounds, what is the preferable way to choose the velocity value that you use for your ballistics software (ignoring temperature factors and such)?</div></div>

Yeah, use the average - bearing in mind that about two-thirds of the rounds are going to be within one SD plus or minus of the average, and the rest will be outside that. There's just no way around that.

And see:

Sources of Ballistic Program Inaccuracies
</div></div>

Thanks Lindy.
 
Re: Ignoring ES/SD numbers.....

Well, seeing as my neck tensions are stupid light right now, I'm not seeing s correlation between light neck tensions and low SD's. Now that doesn't mean it's not true, I'm just saying what I'm experiencing. Light or not though, since my neck tensions, albeit light, are not consistant, and I think until I get my ammo consistent, any data will be pulled from a dark place.
 
Re: Ignoring ES/SD numbers.....

Reasonable neck tension and loads near the upper node generally give decent numbers.

If your a serious Long Range competitor/reloader and do not use the ladder test your missing a real competitive advantage. Jason Baney has an excellent article over on 6mmbr.

Maybe CKA just got the numbers backwards.........
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: JRose</div><div class="ubbcode-body">Well, seeing as my neck tensions are stupid light right now, I'm not seeing s correlation between light neck tensions and low SD's. Now that doesn't mean it's not true, I'm just saying what I'm experiencing. Light or not though, since my neck tensions, albeit light, are not consistant, and I think until I get my ammo consistent, any data will be pulled from a dark place.

</div></div>

Try using a Lee Neck die instead of a bushing die. It uses a mandrel to size neck tension based on the ID rather than the OD of the neck so differences in wall thickness don't impact your neck tension nearly as much.
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: criver</div><div class="ubbcode-body">

Maybe CKA just got the numbers backwards......... </div></div>

You mean like he meant to write SD 5, ES 30, shots fired 11 but it came out SD 5, ES 11, shots fired 30?
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">...my SD is 5fps and ES is 11fps. </div></div>

Didn't see *any* mention of # shots fired in that... which is normally included to help readers assess how much weight to give said numbers. Definitely not anything I'd be upset at seeing on my chrono, even for short strings as it'd look pretty promising and worthy of further testing.
 
Re: Ignoring ES/SD numbers.....

If you really want consistent neck tension you need to at least mic your case neck thicknesses. After that it might be a good idea to *LIGHTLY* turn them to a relatively consistent state. Some folks prefer inside neck turn to remove the Doughnut, which in itself can cause all manner of goofey things to happen.

Another thing most folks never think of is always prep ALL cases at the same time, and prefferably get into a habbit of always letting roughly the same amount of time to pass ( 1 day, 3 days, 1 week and so on ) between when you prep them and when you assemble them. Laugh if you want but this has been proven, and you can prove it yourself, to mess with neck tensions quite a bit.

Another thing too many folks do not think about is using 91% alcohol, after you prep the cases preferably, to measure the case volume. Weighing the cases themselves directly is *NOT* a very accurate way to get consistent cases. Though, after you volume sort them, you could weight sort them if you Really want to get all techy and hardcore.

Of course you know about getting a Meplat tool and "pointing" them right? I mean, getting them exactly the same length from the lands does make a difference.

These are the biggest things you can do for brass prep and ES/SD. The site 6mmbr has some awesome articles on all of this. Again all of it time proven.

Good Shooting,
Gary
P.S. if you take GOOD care of your High quality brass you do realize that you only rarely have to do all this work right? Neck annealing your brass every 3 to 5 firings ( depends on a few different factors ) can help make your cases last a long time. Also neck annealing ever so often *IS* important to your ES/SD numbers.
 
Re: Ignoring ES/SD numbers.....

When I was checking ES/SD, I got the best numbers with lighter neck tensons. Then I went to ridiculously low neck tensions, and all sorts of OAL shift problems cropped up.

Nowadays, I seldom use the chrono, and set my neck tensions according to one criterion. I adjust the neck tension until it's light enough that I can rotate the bullet in the neck with thumb and forefinger. I then readjust it until it's just tight enough to resist rotation.

When I measure case volume, I use the propellant I'll be loading with, a full case charge filled with a drop tube. It's just more 'real' than interpolating volumes computed using another medium.

Greg
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: memilanuk Carter, I'm curious here... are you saying that its *not* generally possible to take an SD value and extrapolate that the ES values should fall somewhere within +/- 2 SD approximately 95% of the time, etc.? Would it be fair to say that the key word there is 'within'... the ES numbers may *not* always be equal to the 'worst case' prediction of +/- 2 SD (95%) or +/- 3 SD (99.7%), that they *could* be some thing less than the maximum spread? In my limited experience, as the sample size goes up, ES (or range) usually does approach the theoretical values described above - but again there is that weasel word 'usually'...

[quote</div><div class="ubbcode-body">Again, the cutoff for "range" is completely arbitrary. Is it where 90% of occurrences fall, 95%? 97.5%? 99%? 95% is the most commonly used in science, unless you are a climate scientist, in which case, you use 90% to overstate your case.</div></div>[/quote]

This is actually a very complicated question, and I think none of us knows the correct answer. Those percentages are averages for Normally distrbuted data. A Normal distribution probably describes velocity distributions for values close to the mean, but it can't describe the long tails because at some point physical laws, such as Energy Conservation, impose limits on the possible velocities.

IF velocities were Normally distributed then the SD is a well-defined quantity, but the ES is not. By this I mean that including more shots in the data set only improves the determination of the SD. But the ES will keep growing with the number of shots, theoretically without bounds.

IF the distribution is Normal-ish over a range of a few SD's about the mean, but eventually has a hard cutoff (e.g. 6000fps rounds violate Conservation of Energy), then knowledge of the ES would eventually converge if enough shots were measured. One can eventually fire enough shots to map out the hard cutoff values. But that cutoff value will probably have no relationship to the SD value. It no longer arises from random influences. The cutoff could be different for different rifles, barrel lengths, etc. So one man's ES would be another's "BS". This is probably the more realistic picture.

SD is a well-defined quantity which can be determined with arbitrary precision given enough data. ES is a much more nebulous quantity. Since either is easily determined from chronograph data, I think it better to quote the more interpretable SD value.
 
Re: Ignoring ES/SD numbers.....

JRose,

To answer your original question, yes I think it is a mistake to forget SD and ES data. While I know it is tough to get SD <10 and ES <30, I think this is a sign that you need to go back through your reloading routine and assess where you might be imparting a bit of variance in your loads.

When I experienced a similar issue, I went back and found that I was doing a few things wrong that once corrected, quickly improved my results. In my case, I wasn't getting truly accurate powder dumps from my Chargemaster after much investigation. The cause was that I wasn't letting the Chargemaster warm up and this was causing the scale to vary over my loading session. Once I changed that (by leaving the Chargemaster on constantly) I found my SD and ES numbers plummeted with my loads.

Once that change was made, I also invested in a VIC-123 scale and started measuring my most accurate loads to .01 grains. I also leave the VIC-123 scale on to alleviate any scale expansion while it warms up and now show consistent powder charges.

The net result of my introspection is SD less than 10 and ES less than 30 consistently. On my fire-formed loads that receive quite a bit of care, I've seen SD numbers around 5-6 and ES around 15-18.

I hope that helps....
 
Re: Ignoring ES/SD numbers.....

And BTW, since we are talking statistics here (I'm a bit of a studied statistician), the above data is using a sample size of 10 rounds for my data. 10 is significant enough to get reliable data although 20 provides further granularity (beyond that is simply a waste).

For those that forget how the statistics work, your sample size really matters. I'm sure if I had more than 10 shots in my sample, I may open up the ES spread even further but it becomes less likely so with the more shots produced from essentially identical cartridges all things considered.

From there the ES, while of lesser importance, is one data point for analysis amongst many. I wouldn't say ignore it as some here have because it give you an absolute statistic to understand the outer boundaries of your loads.

That said, you can derive an ES simply by taking ~3 standard deviations away from the mean. What I mean by this is that if you truly want to look into "predictable" values based on loading data and sample sizes, 1 SD encompasses 68.2% of your total normal distribution, meaning that there is a 68% chance that your next velocity will fall between +34.1% and -34.1% of the mean (average). 2 SDs encompass 94.5% and 3 SD encompass 99.7% for a standard distribution (which is the best to use in this case for extrapolating ES).

So ES can be helpful in understanding the absolute outer boundaries for predictability in velocities but SD can give you a better picture in predicting where your next reported velocity will fall.

If you need the formulas, I'd be happy to oblige.
 
Re: Ignoring ES/SD numbers.....

And one last data point that helped improve my velocity readings. If you are using the CED M2, get the infrared panels. I found that by using the infrared panels, the M2 was a much more accurate device. Not only did it remove a number of the "did not read" errors but provided much more accurate readings in total. It's a pain to charge and carry around the battery and wire setup for this but the improvement in data collection was well worth it.

My home range is north facing and has a roofline that would cause a significant shadow across the 10 foot boundary from the shooting bench. Glare, shadow, and angled light seemed to provide quite a bit of variance throughout the day as I was chrono'ing my loads.

Once I added the infrared panels, things seemed to clean up well. I'd recommend them.
 
Re: Ignoring ES/SD numbers.....

Chasing the same demon...check out the accuracy of the chrono. Mine says (IIRC) +-2% Competition electronics....On 2900 FPS, thats huge...
 
Re: Ignoring ES/SD numbers.....

Very interesting discussion dispite a few disturbances. I really enjoy when we bring it back to the math because the math doesnt lie. Now, misinterpreting math (especially statistics) can lead you to some very faulty conclusions. But I am supprised that it took so long for someone to point out systemic problems like instrument error. Thanks User#1
 
Re: Ignoring ES/SD numbers.....

Actually, all of the discussion of SD and ES is based upon the assumption that muzzle velocities are normally distributed. That assumption is unproven - and I'm not going to try to establish whether it's true. I wish <span style="font-style: italic">someone</span> would, though.
laugh.gif

 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Lindy</div><div class="ubbcode-body">Actually, all of the discussion of SD and ES is based upon the assumption that muzzle velocities are normally distributed. That assumption is unproven - and I'm not going to try to establish whether it's true. I wish <span style="font-style: italic">someone</span> would, though.
laugh.gif

</div></div>

You would think one of the benchrest guys would have by now. I mean, there's no reason for them not to fire EVERY round though a chrono.
 
Re: Ignoring ES/SD numbers.....

That test would require considerable experimental rigor.

Factors which would have to be controlled for, off the top of my head, without too much thought:

Chronograph error
Barrel wear
MV variations as a function of at least the following: primer lot, power lot, bullet weight, temperature

Too much work. I'm supposedly retired. OK, semi-retired. I'd have to charge a lot of money for my time for that study, plus, of course, expenses - which would be considerable.

I think I'll just work on the fundamentals of marksmanship, which have a lot more to do with my missing the target than uncertainties about muzzle velocity.
laugh.gif
 
Re: Ignoring ES/SD numbers.....

Pretty much, you need to run an ANOVA on your range of velocities taken on different days within say .05. You can run an ANCOVA if you want to throw in wind variances, etc. Then run a regression analysis on it and MAYBE your data will mean something. You can do T-tests too but they don't work too well when you are dealing with mulitple factors, generally speaking.

Chad
 
Re: Ignoring ES/SD numbers.....

<div class="ubbcode-block"><div class="ubbcode-header">Originally Posted By: Lindy</div><div class="ubbcode-body">Actually, all of the discussion of SD and ES is based upon the assumption that muzzle velocities are normally distributed. That assumption is unproven - and I'm not going to try to establish whether it's true. I wish <span style="font-style: italic">someone</span> would, though.
laugh.gif

</div></div>

Lindy, yes a good point but frankly, even if you introduce a larger sample size over the total useful life of a barrel from beginning to end, and environmental differences from both extremes (temperature, pressure, etc.), it will probably tell you what you already know from shooting 20 rounds through a chronograph to a certain degree. The standard distribution may flatten and widen (increasing variance, i.e. sigma squared) but you would further solidify the mean and maintain a predictability from the mean through standard deviation, albeit with more of the area under the curve falling outside of a single SD (but not by much would be my guess).

Of course my language of "all things being equal" was the insurance to say that there is still a bit of error in a sample size of 20, 100, 1000, etc as there is with any order drawn from somewhat chaotic natural models.

Thinking aloud, the best simplified test would be to develop 40 identical loads and shoot 20 of them at the very beginning of a barrel's life and store the other 20 away properly. Then toward the end of the barrel's life, perform the same test preferably at the same temperature as originally performed. Take the two, model the means and SD and draw a histogram to view the two sets for comparison. If you wanted to get fancy, you could then use multiple regression models and comparative statistics to evaluate correlation coefficients, coefficient of determination, and p values. Of course the latter would be for fun......
grin.gif
 
You must log in or register to reply here.

Similar threads

Gd33333
SD and ES inconsistencies
Replies
76
Views
1K
secondofangle2
secondofangle2
Sogan
Cant get good SD/ES with Lite barrel
Replies
22
Views
833
MarkyMark007
MarkyMark007
B
Lower SD and ES but larger group size at 100 yards
Replies
24
Views
2K
jasent
jasent
RMB
2023 South Dakota PD Shoot Report
Replies
3
Views
502
Dakotaman
D
M
Sudden Speed and ES/SD Change
Replies
14
Views
461
Supersubes
Supersubes
Top Bottom
Back