ES and SD

[DenisA]

Thank you David and Dave.

My confusion is clearer to me now. There are 2 topics of SD in this thread that I'm mixing prematurely which is overwhelming and confusing.

The first topic is how SD is useful and does have a predictable effect on POI but with large samples. So SD will become more predictable as the sample size increases and the bell starts to take shape.

The second topic is how to best apply that to developing a load practically and economically.

I understand now why Dave Mc is breaking it down in to multiple threads.

Looking forward to what's in store.

Thank you both again for your time, persistence and sharing.

[AlanF]

...The more sampling we do the more we are convinced most groups follow a normal distribution. Sometimes they are skewed, have longer tails etc...

One thing I've found is that when simulating a very high numbers of shots, X counts are higher than in practice i.e. the ratio of Xes to points is greater than what happens in the real world of Queens shoots etc. My current thoughts are that we focus our loading and tuning (and shooting) methods on hitting the 6 ring, and not so much on the X, so this affects the shape of the bell curve. I tend not to bother doing some of the advanced accuracy practices that short range BR shooters do. Also I will occasionally aim on the downwind side of the X (in case of drop-offs) or high/low in the X (in anticipation of a known pattern of vertical behaviour). I suspect my approach is a general characteristic of F-Class.

[Old Trev-39]

Something else to throw into these calculations beside E.S. & S.D is projectile imbalance. Those lucky enough to have a Juenke machine can eliminate most of this problem. The rest of us just have to put up with it, but it is definitely another to contend with. Maybe some of the wizz kids out there can tell us how much this effects us. Harold Vaughn in "Rifle Accuracy Facts" writes on this and it is another factor affecting our accuracy especially over long ranges.
Trevor.

[williada]

Trev, you are talking about projectile imbalance from poor cores or jacket run out that react like a wheel out of balance. I have an old Juenke machine and used it on most brands. I can say with confidence the modern manufacturers do a pretty good job with run out and annealing of jackets. That's why we pay for top brands. Even experimented with slower twist barrels to minimise the perceived projectile wobble. In the old days we used tight barrels to iron out the cores and jackets too. Found out the air density was a far bigger factor and you need a certain amount of gyroscopic stability with a correct node velocity at the longs. Batching from base to ogive correlated with the Juenke results. Do that and you're mostly there if you include weighing the critters too. It might help with the psychology and add a little to super centre count if you can batch with a Juenke. But I can tell you that they drift like a bastard and that can destroy your confidence about what you have already batched.

[Old Trev-39]

All projectiles I use in my 1000yd. b/r shoots are measured base to ogive, minimum metplat on each batch, bearing surface length batched, then weighed to 1/10 grain. If I had a juenke machine they would be run over it first, instead of wasting time on un-balanced projectiles.
Trevor.
P.S. I know some will say I have way to much spare time, but it keeps me out of the pub.

[jasmay]

And then of course, you just get utter freaks Like Mark Fairbairn that just make you wonder what the hell we are trying to do, when he rocks up with a 28" barrel he cut down the day before from 32", habit done zero load development and hits a 60.6 & 65.4 @ 1000yds.

The bloke could shoot bullets out of a 1" Gal pipe propelled by s rubber band a still hit 60's all day

[johnk]

There's just a minor possibility that that guy knows his $#@t.

[williada]

Trev, if you owned a machine you would understand they are very finicky, fiddly and will show the slightest variance. That is not that significant in itself but If you batched in the fine movements with them, you would find very few bullets. Your cost per bullet would go through the roof. The bullets actually come in ranges left and right of the centre reading when the machine is a adjusted rather than finding lots of unbalanced bullets per se. The reading time of spinning is not very quick either and measuring base to ogive is faster. So that tends to show up the outliers anyway. That is because the jacket walls thin towards the metplat. I think the machine picks up that variance too. Gyroscopic stability of varying ogives can be enhanced by velocity changes to match the bullet batch. The approach many use with machines is to batch bullets by ogive and weigh first, then super batch them with the Juenke. Varying bullet length has a significant effect on yaw and nodal performance. A super batch may be used at 1000 yards. Doing all that may send your partner to the pub.

Yes, Jason he is world class. His technique and rest support would be honed with a lot of practice. He is a skilled gunsmith and would know how to set up his gear. Mark would have the parameters in his head for many setups and be able to connect the dots.

[Old Trev-39]

Fortunately I do not have a partner so I can do what I like when I like. No fear of cold shoulder or dirty looks Been down that path and life is much easier with out them.
Cheers,
Trevor.

[DaveMc]

...The more sampling we do the more we are convinced most groups follow a normal distribution. Sometimes they are skewed, have longer tails etc...

One thing I've found is that when simulating a very high numbers of shots, X counts are higher than in practice i.e. the ratio of Xes to points is greater than what happens in the real world of Queens shoots etc. My current thoughts are that we focus our loading and tuning (and shooting) methods on hitting the 6 ring, and not so much on the X, so this affects the shape of the bell curve. I tend not to bother doing some of the advanced accuracy practices that short range BR shooters do. Also I will occasionally aim on the downwind side of the X (in case of drop-offs) or high/low in the X (in anticipation of a known pattern of vertical behaviour). I suspect my approach is a general characteristic of F-Class.

Sorry about being late in this response but just reading back through the postings.

One of the reasons for this Alan is the models assume a perfect zero whereas in reality we rarely start on a perfect elevation and wind reading. We "adjust" towards the centre. This has the result of skewing the distribution and lowering x counts. Peters software unwinds the sight settings and can centre the groups. Once this is done we bring the skewed group back towards a centred normal distribution again. But as you say we also put a human influence on this in many ways that can skew the groups in a scoring situation. What the rifle is doing and what we see on the paper need to be pulled apart and compartmentalized.

[GSells]

Another great old thread for newbies and maybe revision for others .

[williada]

Gman I don't intend to get into discussion. Suffice it to say I go by ES initially because my barrel has limited rounds it can fire before it wears out. I look for the trends by overlaying a sequence of tests up to and past the speed hump to get a handle on SD for when the barrel is cherry ripe for a big event.

Statistically we can call things random error or an outlier but in reality there is a reason why a shot went where it did. Always believe the paper it prints the hole in. Vibrational analysis is another tool to do that but that is too complex for the uninitiated and I use that to confirm difficult problems.

A good starting point is polynomial curve fit of data in Excel of an incremental load test. https://www.youtube.com/watch?v=-pD3MCdWnPA

[bruce moulds]

well if you are only going by 3 shot groups, e.s. is all you have meaningful in any way.
which goes to show the value of 3 shot groups (not).
bruce.