Satellite flyers, soooo frustrating!

[Utectok]

superimposed they’ll look the same I can guarantee you that. Just an off the shelf Schultz Larsen Hunter in 25-06 pinging 117g hornady factory ammo eldx precision hunter.
Being 1:10, it’s not even supposed to like 117g bullets

That group was 200yds offhand of course

Well that’s nice but I’ll believe it when I see it mate

 

[HonestJohn]

Well that’s nice but I’ll believe it when I see it mate

You know, there are good shooting rifles and good shooters out there you should think about that

 

[Mungo]

Here’s a three shot group. I’d argue it tells a relatively good story about grouping, esp since when i shoot abother 3 shot group on another day it looks the same

If that’s the tunnel at Braidwood, your actual group could easily be 1.5”. I am profoundly suspicious of that thing. What were the x and y values for those shots?

 

[Utectok]

You know, there are good shooting rifles and good shooters out there you should think about that

Ok post the 30 shot group mate I’m afraid it’s that simple

 

[ejg]

I think 10 could be quite good,.... 10 barrels shot out and statistics established. Only issue is that you don't have a gun to shoot anymore. So one has to be realistic and look several points. I know a few rifle builders that after years of arsing around with statistics have again settled on 3 shot groups. If that group is 2" it won't get better with 10 or 50, or 2000 shots.
I prefer several 3 shot groups, also on several days or years to check if the rifle is still ok. Mainly however check at longer ranges.
We zeroed a new Strasser RS 700 in 6.5CM yesterday. The first shot from the new barrel was 2650fps, second 2672fps. After that it went up while zero and was at around 2750fps after maybe fifteen rounds. Accuracy testing in this stage is useless but it will give an indication. We tried 4 different ammunitions in that CM. Sako TRG136gr 14mm centre centre, 143 ELDX 14mm, 123gr Nosler BTHP 11.3mm, 140ELDM 10.2mm. All 3 shot groups. Guss our PRS guy shot the ELDX group, I shot the others off the hood of the Landy. This is a good help to the customer who knows the rifle doesn't seem to be fussy. Serious testing is only worth it after the barrel has started to settle in velocity, customer will do that.
Back to the 300wm some seem to get much better accuracy with normal LR primers vs Magnum.
edi

 

[caorach]

Ok post the 30 shot group mate I’m afraid it’s that simple

You are, of course, correct. The posts on which you are commenting provide an excellent illustration of the matters being discussed.

 

[caorach]

If that group is 2" it won't get better with 10 or 50, or 2000 shots.

I think that is precisely the point - what you can say however is that if you shoot another 3 shot group there is a good chance that it will be better, indeed there is even a small chance that it will be a very tiny group indeed and at that point you can point out the efficacy of some arcane process or belief. Should you then shoot a larger 3 shot group you can always put that down to bad wind, or Saturn being in Uranus, or something. At that point you've convinced yourself that the rifle can shoot tiny groups on the basis that it shot one tiny group.

 

[Smellydog]

Yeah but where's the link?

 

[ejg]

I think that is precisely the point - what you can say however is that if you shoot another 3 shot group there is a good chance that it will be better, indeed there is even a small chance that it will be a very tiny group indeed and at that point you can point out the efficacy of some arcane process or belief. Should you then shoot a larger 3 shot group you can always put that down to bad wind, or Saturn being in Uranus, or something. At that point you've convinced yourself that the rifle can shoot tiny groups on the basis that it shot one tiny group.

Why should I shoot another 3 shot group if a rifle/ammo has proven to be inaccurate? If you have a great 3 shot group then you have a reason to keep going and try another later or keep developing in that direction. Experience will tell you if conditions (including shooter) are right or wrong to try and evaluate accuracy of a rifle.
edi

 

[Mungo]

Right - as promised.

I have simulated 9 groups, all using the exact same parameters. X and Y values randomly drawn from a normal distribution with a mean of 0 and a standard deviation of 0.2 (see justification for these parameters below). In each plot, there are 3 shots in red to represent what you would see if you only shoot a group of 3. There are then another 6 shots in grey, showing what you would see if you carried on shooting to a a total of 9. What should be clear is that the classic 'two together and one off' pattern is very common with 3 shot groups. They then start to blend into the wider group as you shoot more.

The take home is that fliers are not fliers - they are an entirely expected result of random variation, and if you shoot larger groups, they become informative about the true accuracy of you and your set up.




I chose these parameters for the simulations because this approximates a gun that, across a lifetime of 2000 shots will put 95% of its shots in a 1 inch circle. This is something most people would be happy to agree is comfortably under a 1MOA gun. Though even this will generate 'fliers', with a lifetime extreme spread of 1.5" (ie. if you shot all 2000 shots into the same target, the resulting group would measure 1.5" at it's widest point, measured centre to centre for the relevant shots.

 

[Mungo]

Even more vividly. Exactly the same distribution parameters as above, but this time with 3 shots in the initial group (red), then 15 in the successive group (to a total of 18). The 'two together and one off' pattern in the 3 shot groups is very clear.

 

[Mungo]

Several people have asked what it looks like when you increase the sample even more. Here is what it looks like when you shoot 103 (red 3 are the initial group, plus another 100, all the same parameters as above). The red 3 are now very clearly part of the overall group. One other thing should be really obvious from these: if you were doing load development based on 3 shot groups, it would be completely impossible to tell the difference between the effect of random variation and any effect of changing the powder charge. I imagine I could post the red groups below, and claim they were generated by a loads increasing on 0.2gr steps. People would confidently tell me there were 'nodes' at 2 and 8...

 

[Utectok]

I love these @Mungo so clear and concise a really good lesson in group size distribution anomalies

 

[Utectok]

Right - as promised.

I have simulated 9 groups, all using the exact same parameters. X and Y values randomly drawn from a normal distribution with a mean of 0 and a standard deviation of 0.2 (see justification for these parameters below). In each plot, there are 3 shots in red to represent what you would see if you only shoot a group of 3. There are then another 6 shots in grey, showing what you would see if you carried on shooting to a a total of 9. What should be clear is that the classic 'two together and one off' pattern is very common with 3 shot groups. They then start to blend into the wider group as you shoot more.

The take home is that fliers are not fliers - they are an entirely expected result of random variation, and if you shoot larger groups, they become informative about the true accuracy of you and your set up.


View attachment 401354

I chose these parameters for the simulations because this approximates a gun that, across a lifetime of 2000 shots will put 95% of its shots in a 1 inch circle. This is something most people would be happy to agree is comfortably under a 1MOA gun. Though even this will generate 'fliers', with a lifetime extreme spread of 1.5" (ie. if you shot all 2000 shots into the same target, the resulting group would measure 1.5" at it's widest point, measured centre to centre for the relevant shots.

Makes me think about barrel tuners testing methodology which controversially I believe is simply witchcraft.

 

[hw101]

Several people have asked what it looks like when you increase the sample even more. Here is what it looks like when you shoot 103 (red 3 are the initial group, plus another 100, all the same parameters as above). The red 3 are now very clearly part of the overall group. One other thing should be really obvious from these: if you were doing load development based on 3 shot groups, it would be completely impossible to tell the difference between the effect of random variation and any effect of changing the powder charge. I imagine I could post the red groups below, and claim they were generated by a loads increasing on 0.2gr steps. People would confidently tell me there were 'nodes' at 2 and 8...

View attachment 401368

It’s another brilliant statistics example - many thanks.

 

[caorach]

Makes me think about barrel tuners testing methodology which controversially I believe is simply witchcraft.

The physics of barrel harmonics was well described by Mallock in a paper presented to the Royal Society by Lord Rayleigh. Mallock was undoubtedly a great mind and Lord Rayleigh was one of those very few people who was on a different level to normal humans.

Despite the paper being freely available it isn't discussed when the internet subjects us to various claims about barrel vibrations and when people make wild claims I've never once seen them adjust Mallock's maths to account for the claims they are making. I find that a bit odd in view of how confident some people are in the claims they make - as Mallock demonstrated it is an engineering problem that can be clearly described in engineering terms and so anyone capable of coming up with a new or novel solution should be capable of producing a mathematical treatment to make their solution clear.

Then we get to the point of this thread which is that most of the stuff people "know" from shooting at a target is purely a function of the normal distribution. When you put that on top of various black magic you have to conclude that people who are fooled by these various weird procedures and who ignore how the stats work must know less and less about how their rifle shoots with each change in reloading or the rifle itself.

The Mallock paper is here:

https://royalsocietypublishing.org/doi/epdf/10.1098/rspl.1901.0052

 

[caorach]

Why should I shoot another 3 shot group if a rifle/ammo has proven to be inaccurate?

The next group you shoot with that rifle, indeed the next 100 groups you shoot, might all be much smaller.

 

[Mungo]

The physics of barrel harmonics was well described by Mallock in a paper presented to the Royal Society by Lord Rayleigh. Mallock was undoubtedly a great mind and Lord Rayleigh was one of those very few people who was on a different level to normal humans.

Despite the paper being freely available it isn't discussed when the internet subjects us to various claims about barrel vibrations and when people make wild claims I've never once seen them adjust Mallock's maths to account for the claims they are making. I find that a bit odd in view of how confident some people are in the claims they make - as Mallock demonstrated it is an engineering problem that can be clearly described in engineering terms and so anyone capable of coming up with a new or novel solution should be capable of producing a mathematical treatment to make their solution clear.

Then we get to the point of this thread which is that most of the stuff people "know" from shooting at a target is purely a function of the normal distribution. When you put that on top of various black magic you have to conclude that people who are fooled by these various weird procedures and who ignore how the stats work must know less and less about how their rifle shoots with each change in reloading or the rifle itself.

The Mallock paper is here:

https://royalsocietypublishing.org/doi/epdf/10.1098/rspl.1901.0052

I was extremely reluctant to start reloading because I couldn't understand how you could possibly find an accurate load without firing hundreds of rounds. It seemed utterly daunting to try to optimise all those variables, and then systematically work through them all and all the possible combinations.

I then realised it was effectively impossible for a recreational shooter - and that it didn't matter. Good enough is good enough.

 

[Mungo]

The next group you shoot with that rifle, indeed the next 100 groups you shoot, might all be much smaller.

Possibly when phrased the way he did ('if you shoot a 2 inch group, you don't have a 1MOA gun'), he has a point.

If you take the parameters I've used above (95% of shots taken across a lifetime fall within a 1" circle), then a 2" group is very unlikely (p<0.00001).

However, the opposite is not true: shooting a single 0.5" group does NOT tell you as much. If I change the parameters to draw x and y from a normal distribution with a mean of 0 and an SD of 0.4, around 30% of shots fall within a 0.5" circle, so a 0.5" group can happen roughly every 40 groups (ie. each shot has p= 0.3 of being inside 0.5MOA, then for three shots, p=0.3^3, which is 0.027). This is for a gun where 95% of shots would go into a 2" circle, and lifetime extreme spread is just over 3".

However, this also comes down to what people want to define as a 1MOA gun. When I asked this question a few months back, there was not a huge amount of consensus. The definition I have been using for the simulations (95% of shots over a 2000 shot lifetime go inside a 1inch circle) is too strict for some and not strict enough for others. Though anyone who says a 1MOA gun can ONLY be defined as one that never shoots outside a 1" circle across its entire lifetime is probably in for a nasty surprise as to just how unrealistic that is!

 

[ejg]

The next group you shoot with that rifle, indeed the next 100 groups you shoot, might all be much smaller.

The thing is, statistics on fired groups is a bit like statistics on a broken wheel bearing, so many factors. So many things that might or might not repeat. Not even knowing what one is trying to "statistic"... the rifle? the hold? the shooter? shooter fatigue? the weather? the ammo? primer/bullet etc.
On the new rifle we tested this weekend we had alone over 100fps difference between the first and ~tenth shot. That will probably increase for a while, then settle, then drop again after x. One thing was consistent though, the ammo that my friend has been using for two world championships PRS and the one that I have been using for hunting in the last three years has shot the same group size as in every other rifle we used it in. There are ways of making rifles more consistent as well as bullets just easier to group. I don't see the PRS guys doing much 100m grouping or statistics. Just zero then see what she dose down range. Trying to not shoot out the barrel before you know if or if not she shoots well enough.
edi