“Mind blow” I mean it’s all made up self deception!!!??

[bogtrotter]

Me I just played around till I found one that suited me and my rifle, stuck with it and have been using the same.load for over thirty years .
Don't see any need to change, would hardly call it load development but it works for me.

 

[Precision123]

What bugs me, what really grinds my gears, are target turrets.
Why? Because 'you' don't really know where your group average is, where the POA and POI reside. Then said deer is not at 100m but 110 so click, click, click...
Your scope is either zero'd or it isn't. And you can't account for variation if you don't know what it means or what it is. And for the vast majority, PBR is all you need. Some may take 300 or 400m+ shots but compared to the scopes being sold, this reality is out of kilter.

I'll put my soap box away.

 

[Mungo]

Do you reload @Mungo ?

If so how do you conduct load development?

Use Quickload to tell me the max for my chosen powder and components, create a series of charges 3 in 0.4 gr increments starting 1.8gr below max. Fire 3 at each charge to test for pressure. Pick the resulting safe charge that looks prettiest. I really don’t worry about it - I know that I have zero information on accuracy from this, save that provided by the overall group generated by all the shots. Since in all the rifles I’ve done this for, the overall group was never more than 1.25”, I really didn’t care.

Look - you can rail against the stats, and argue that they somehow don’t represent ‘real life’, but that just completely misunderstands statistics. Yes, stats are boring and annoying, and very often appear to violate intuition and ‘common sense’ - but they really are the only robust way to do things like this. Otherwise it’s just voodoo.

 

[Mungo]

Me I just played around till I found one that suited me and my rifle, stuck with it and have been using the same.load for over thirty years .
Don't see any need to change, would hardly call it load development but it works for me.

Is probably every bit as effective (quite likely more so) and vastly more efficient than spending forever obsessing about small details and fooling yourself that you’re actually measuring anything.

 

[HonestJohn]

I studied statistics at mba level. My professor was a wise old elf - the first thing he said on the first day of class was “students, let’s go to the pub and play darts, we have more chance of learning something meaningful, than starting to talk about it here”.

I don’t think we learned much about spread and deviation or correlation coefficients from the pub/darts, except that the longer we stayed, the more we played and more we drank, the worse the results were on many levels. Maybe that was the lesson..

 

[Heym SR20]

I have done multiple stats courses. Firstly as part of O the A level maths, then as part of BSC Agriculture where you are trying to make sense of Biological experiments such as crop trials. Then as part of an MPhil in Management where the real question was not so much about statistics, but more why management and economists are trying to use statistical analysis of complex behaviours. And really the conclusion at this on multivariate analysis was much more it is good for making you look clever when trying to apply science to a subject much more about the human condition, and an artistic approach is probably just as helpful.

At a scientific / engineering level stats are good at helping understand probabilities of failure and success of components.

With firearms and ammunition, its an engineering rather than a biological challenge, so there is or should be much less variation than you would get in 1,000 different wheat plants in a trial plot.

So for my load development I load one cartridge at each increments from start up to max load from book or quickload data. Shoot them slowly - each from a cold barrel- at the same target and carefully record point of impact.

I then select the middle of three loads that are nicely close together, especially if its a nice round number and use that as my load.

I have always that pretty much all the rounds go into a sub 2” group at 100m. And I usually find little clusters that are within 1/2”. I tend to find my end result is MoA, and more than adequate for hunting purposes.

And to be honest the biggest variability in any ammo and rifle combinations is the shooter and his / her technique.

And pretty much any riflesmith, gunshop, coach will have plenty of stories about Mr Nimrod coming back with a rifle and ammo that “doesn’t shoot”, only to take it to the range and they can shoot multiple bughole groups.

I am very very aware that rifle shooting is a real skill. As a boy I learnt with a 22 shooting 10 bull targets. I could keep the bullets in the black, but keeping them in the centre to get high 90 scores - well there was one lad who using the exact same rifle was consistently much better than I.

And in the rifle club there many guys and girls who are very good shots. They have the skill, but also the time and patience to develop really accurate loads and rifles.

Ditto shooting clays. I use a range and clays as a means of developing and keeping my shooting skills at a level where I am confident shooting wild game within my limits.

And reloading means that I can keep a consistent load without having to search far and wide for a particular brand of ammo.

Saying that, a good friend uses 80gn in his 243. He just buys whatever 80gn load is available and shoots it. He is shooting 50 odd Roe a year and rarely has a problem. But then for him 150 yards is a very long shot.

 

[308tikka]

Statistical significance. They have a point esp in relation to group sizes, from group sample for the average club shooter or stalker. Most stalkers shoot nowhere near enough deer or in practice to worry about all of the variables. You dont have a 0.25 moa group off 3 shots...

Pre 2020 when shooting thousands of rounds a year, I was making effort to make excellent competition ammo. In the main this is (with prepped cases) using a carefully worked up load (by results) and trickled powder to v fine tolerances. I dont weigh trimmed cases but I should.

Group sizes: This summer I had a 75.15 in competition at long range (iron sights, sling, not string shooting, 308, 155gn). Lets just say this is rare at any range.

12-13 out of 15 shots were around 3/4 moa. Highly acceptable on this range which moves bullets up and down when the wind blows due to ground undulation, and changing light. The other shots pushed it over 1 MOA, likely 1.25 MOA. Well prepped ammo - weighed carefully - but nothing else in caseprep to the nth degree.

For stalking, I throw charges. Its a waste of time to do anything else..... still better than factory.

 

[Mungo]

With firearms and ammunition, its an engineering rather than a biological challenge, so there is or should be much less variation than you would get in 1,000 different wheat plants in a trial plot.

And that's the key point.

As variation declines, the sample size needed to detect differences increases - markedly. With a modern well made rifle, shooting consistently loaded ammo with high quality components, being shot by a good shot, the variance will be small. A load development procedure that only shoots 5 shots at each charge level has no chance at all of detecting the 'best' load. Really - none.

So for my load development I load one cartridge at each increments from start up to max load from book or quickload data. Shoot them slowly - each from a cold barrel- at the same target and carefully record point of impact.

I then select the middle of three loads that are nicely close together, especially if its a nice round number and use that as my load.

Which is fine, but all you're doing is choosing a load based on random clustering. Fire a sequence of shots, and by chance you will get 3 in succession that cluster. To actually be certain that cluster was a real 'node', you'd need to repeat the sequence several times to detect whether the same cluster reappeared.

 

[Heym SR20]

And that's the key point.

As variation declines, the sample size needed to detect differences increases - markedly. With a modern well made rifle, shooting consistently loaded ammo with high quality components, being shot by a good shot, the variance will be small. A load development procedure that only shoots 5 shots at each charge level has no chance at all of detecting the 'best' load. Really - none.



Which is fine, but all you're doing is choosing a load based on random clustering. Fire a sequence of shots, and by chance you will get 3 in succession that cluster. To actually be certain that cluster was a real 'node', you'd need to repeat the sequence several times to detect whether the same cluster reappeared.

Yes you are correct, especially given your skills in statistics - admittedly mine are at a level where I understand that they are important and whether or not results are statistically significant, but I have yet to understand why. Perhaps you could try again sometime whilst dining on a roast woodcock. Mind you a good claret will soon cloud things.

What I do know is ammo manufacturers will typically test many hundreds if not thousands of rounds when developing a new load or introduce a new bullet type. And its pretty amazing how good most factory ammo is in most factory rifles these days.

 

[NigelM]

Which is fine, but all you're doing is choosing a load based on random clustering. Fire a sequence of shots, and by chance you will get 3 in succession that cluster. To actually be certain that cluster was a real 'node', you'd need to repeat the sequence several times to detect whether the same cluster reappeared.

I think I'm starting to get it now. Trying to work out why I get to little groups when the science says I'm doing it all wrong.

I load 3 per charge weight (1% of case capacity increments) over 5 charge weights, bracketing a GRT/OBT node. Chronograph them all and check velocity matches the prediction, if not recalibrate and go again. If no obvious node detected I repeat so I have 6 of each. Very rare that this doesn't hint at something. If I see what looks like an obvious node I repeat with 5 shots to check. I then go out to 200m, 300m and 400m to check drops with 5 shots each. By the time this has been done I have 23 shots down the road with my chosen load. If they are all sub 1/2 minute I'm happy and that's my load banked. If not its back to the drawing board.

It might still not be "statistically relevant" but it tends to work in real life and without putting 250 bullets down the range.

 

[Mungo]

What I do know is ammo manufacturers will typically test many hundreds if not thousands of rounds when developing a new load or introduce a new bullet type. And its pretty amazing how good most factory ammo is in most factory rifles these days.

Exactly. They are testing at the volumes needed.

And the very fact that most factory ammo shoots well to very well in most rifles tells you that load development is unnecessary. An 'average' load, made consistently, using good components, shot in a well made rifle, will be more or less impossible to distinguish from the 'best' load unless you shoot tens of groups.

 

[NullMac]

Or just kept shooting 6 shot groups with the same ammunition till I got an sd I liked lol

Which is exactly what people do with 3 shot groups then post "my rifle shoots 0.5 moa if I do my part". A 2 moa rifle will shoot 0,5 moa if you shoot enough groups.

 

[Mungo]

I think I'm starting to get it now. Trying to work out why I get to little groups when the science says I'm doing it all wrong.

Almost certainly because you have a very well made custom rifle, where even the 'worst' group is unlikely to be over 1".

What you describe means you are almost certainly chasing a statistical artifact. But that doesn't actually matter - you're getting accuracy you're happy with.

If you ever wanted to put this to the test, make 6 rounds of your chosen load. Make 6 with one of your other charge weights. Get someone to load them into the mag without telling you which is which. Shoot 2 groups. See if they actually differ.

 

[Mungo]

Which is exactly what people do with 3 shot groups then post "my rifle shoots 0.5 moa if I do my part". A 2 moa rifle will shoot 0,5 moa if you shoot enough groups.

In fact, will do so surprisingly often!

 

[Heym SR20]

In fact, will do so surprisingly often!

And that also explains why the 0.5 MoA rifle will sometimes throw a bullet a bit wide.

And of course the idiot behind the butt has feck all to do with it.

 

[Heym SR20]

@Mungo i now completely understand. To be sure that a good red wine is truly better than another you need to drink it by the case, not by the bottle!

 

[caorach]

Sorry, I beg to differ. This is maths and statistics getting in the way of real life.

I appreciate that stats get a "bad press" but the fundamentals, and that's pretty much what we are dealing with here, are pretty much like any other "law" of physics or similar - the basic stats are as fundamental as gravity. So the truth is that the stats really are fundamental to our real lives and we have no choice but to work with them. For the first while that I reloaded I got carried away with all the various processes etc. that the internet detailed and it took me some time for it to dawn on me that, actually, I wasn't achieving anything other than fooling myself. On a positive note the extra practice shooting at targets was a good thing so there was a benefit but the only actual information I gained was that I'd been an idiot.

 

[Pedro]

As a boy/young man I lived next door to a psychiatrist. He was a very intelligent man and quite fascinating. But his mechanical know-how wasn't up there with his understanding of Freud. At the time (1970s) he had acquired a Datsun Bluebird. A family car, which for the period was very advanced in that it had pretty much electric everything. But this was the time when electronic ignition was yet to be commonplace and the car had points. He'd bought the car because my Dad had one and my Dad was an engineer. Anyway, his car started to pop and miss a bit, so he thought it must be the points. So he went out and bought some and changed them. The car was back to it's reliable self. He came around, pleased as punch to tell my Dad of his achievement. My Dad said: "Very good. What gap did you set them to?" The psychiatrist looked nonplussed, not understanding the concept of the "gap". They went around to check and my Dad found that, totally randomly, he had installed the points with the correct gap.

I think for many, the concept of reloading is a satisfying activity in itself. You might manage to get better performance without doing a great lot of testing, but if you reload and get some satisfaction out of the process and obtain ammo that works in your rifle at least as good or maybe a tad better than factory rounds, even if it is somewhat fortuitous, there's pleasure in that. Just don't tell my Dad (which would be difficult since he's passed away).

 

[NullMac]

There is no requirement for load development to be done in a statistically significant way. Indeed it is a waste of time for most of us.

There are common practices that are exposed by top tier competitive shooters that are probably not statistically valid but their common usage by those seeking best precision point the way to processes that will be very successful for the normal shooter.

By adopting these best practices and some very limited sample testing we can be confident that we have a load that will generally perform as we require provided we maintain the quality of the reloading process and shoot straight.

By controlling the commonly accepted variables and with limited testing the shooter can have some confidence that the load will perform as expected. No need to shoot heaps of rounds wasting time, components and barrel life.

Having said all that, these are not the practices I see discussed as routine on this forum. Neither are the variables often discussed the (currently accepted) best understanding of the key variables in reloading. The comment above about the relationship between powder weight and group size being a case in point.

I only ever shoot 2 shot groups, and not many of them either. I cannot shoot a group tighter than the first 2 shots, and it will either shoot to my requirements or not. The statistically valid grouping will be a bit different but so what? With control of my shooting and loading, it's not going to be hugely different. My total load development runs to a total of 10 rounds, unless I have a snafu. Is this statistically valid. Nope, but with one graph and a 2 shot group I can tell if I am using the correct powder and if the rifle will "shoot". If it doesn't I switch powder looking for something more reliable. (Generally Viht !)

To do this though, you need to invest in some gear and minimise the shooter in the testing.

Yer best resource on here is Laurie. I would hesitate to call him statistically significant but if you follow his advice then results will follow

 

[Mungo]

I think I'm starting to get it now. Trying to work out why I get to little groups when the science says I'm doing it all wrong.

I load 3 per charge weight (1% of case capacity increments) over 5 charge weights, bracketing a GRT/OBT node. Chronograph them all and check velocity matches the prediction, if not recalibrate and go again. If no obvious node detected I repeat so I have 6 of each. Very rare that this doesn't hint at something. If I see what looks like an obvious node I repeat with 5 shots to check. I then go out to 200m, 300m and 400m to check drops with 5 shots each. By the time this has been done I have 23 shots down the road with my chosen load. If they are all sub 1/2 minute I'm happy and that's my load banked. If not its back to the drawing board.

It might still not be "statistically relevant" but it tends to work in real life and without putting 250 bullets down the range.

So - here's an example.

Take a look at the image below:


Let's say I told you that was what I'd got when load developing a 6.5 Creedmoor, shooting 3 rounds at each charge. Most people would quickly see that loads between 42.8 and 43.4 and say - there's your node. Pick one of those and work with that.

But I say no! We need a bigger sample size. So we shoot again, and get:



Aha! A pattern emerges. Looks even more like something is going on around 43 - 43.2. But! We're extra careful, so we shoot the whole lot one more time, and get:


And you think - right, that's it! Definitely somewhere around 43. I'll go with that. And surely this is enough - I've now shot 90 rounds!

But what if I now tell you that's simulated data. Every one of those 'groups' was simply a sample of 3 drawn (completely at random) from THE EXACT SAME underlying distribution, with a mean of 0 and a Standard Deviation of 1, approximating a 1.5 MOA gun.

It is extremely easy to fool ourselves into seeing patterns that aren't there.