How much "Extreme Spread" is acceptable in competition?

[Mulligan]

A little background,

This is in regards to a thread I read on this forum regarding weight sorting primers.

I have attached a short article describing a quick test I performed to see if a real test is warranted.

In the attached "quick test"- The average MV from the lightest 5 primers (actually 3 as I had an issue or two) was 26 fps slower than the heaviest 5 primers in the quick "test" attached.

I will be having surgery soon that will keep me at the house for a while and I will have the time to weight sort primers and prep all the stuff for a "real" test.

Here is where I need your help.

How many rounds of the lightest and heaviest primers would it take for you to consider this a valid test? Now keep in mind, I cannot afford to eat up a barrel during this test.

Here is what I am thinking;
Sort a brick of primers by weight
separate the 10 heaviest primers and 10 lightest primers
10 shots across the chronograph of each loaded at the range with the same piece of brass.
Sound good?

If more shots are required, how many rounds is a piece of brass good for?????

My ask is this, can some of you help me frame this up to make the test valid, without breaking the bank?

Thanks for your input

CW

EDIT
should I use my 6PPC or 6 Dasher for the test?

 

[Prose]

20 loadings on one piece of brass may stretch the limit and you'd end up testing a reduced charge to ensure the brass made it. How about using two pieces of brass, 2 sets of five lightest primer, 2 sets of 5 heavy primer. You could graph each case and have four separate lines of data represented on one chart.

 

[Joe Salt]

CW I'd like to see it done with a Dasher! I think you'll find if you run the test a few times and throw in light or heavy one, once you find a good load you will see why you should be weighing primers. Then reweigh the spent primers and see if they all weigh the same!
Good Luck

Joe Salt

 

[chkunz]

You probably need about 30 data points (degrees of freedom) to have a 95% confidence level that heavy is different from light. Is there a statistician in the house?

 

[Mulligan]

20 loadings on one piece of brass may stretch the limit and you'd end up testing a reduced charge to ensure the brass made it. How about using two pieces of brass, 2 sets of five lightest primer, 2 sets of 5 heavy primer. You could graph each case and have four separate lines of data represented on one chart.

Just make it clear for me, are you suggesting one set of 5 pieces of brass will be used for one light primer test and one heavy test? With each primer test being 5 primers either light or heavy.
CW

 

[Mulligan]

CW I'd like to see it done with a Dasher! I think you'll find if you run the test a few times and throw in light or heavy one, once you find a good load you will see why you should be weighing primers. Then reweigh the spent primers and see if they all weigh the same!
Good Luck

Joe Salt

Joe,
I have a forester press which has a contained primer catcher on it. I may be able (assuming I don't forget between now and then) to separate out each primer or groups of primers. I am sure there will be some barnacles or debris deposited in the primer catcher each time (maybe not?), how should I account for that? Not sure it will be significant?
CW

 

[dcali]

The best way to do this test is a a little work.

Take 30 or so primers, and weigh them before *and after* firing. The difference is the charge weight of the primer.

Then plot both total weight vs velocity and charge weight vs velocity and see if there is any meaningful correlation. If there is one, I would expect it to be stronger for the charge weight than the total weight, but who knows. You can calculate a correlation with xcell (0 is no correlation 1 is perfect correlation), and plot a trend line as well that will tell you how much impact the primer weight has.

 

[Joe Salt]

CW take the cup off the Forester and catch them in your hand, a little more work but I have done this before!

Joe Salt

 

[Mulligan]

CW take the cup off the Forester and catch them in your hand, a little more work but I have done this before!

Joe Salt

Gotcha, thanks Joe
CW

 

[Joe Salt]

Oh make sure you mark the case that has the heavy or light primer! Just say.

Joe Salt

 

[Prose]

Just make it clear for me, are you suggesting one set of 5 pieces of brass will be used for one light primer test and one heavy test? With each primer test being 5 primers either light or heavy.
CW

2 pieces of brass total. Each piece is loaded 5 times for light and 5 for heavy. Total of twenty rounds fired as you originally suggested. Purpose is to reduce chances of component failure during testing, but maintains the ability of direct comparison within the overall data group. No matter what you do, it seems you have an eye for accurate data so I'm sure you'll do fine.

I heard 74% of statistics are made up anyways.

 

[CharlieNC]

The limited data you presented in the file sufficiently shows the primer weight made a statistically significant difference on the velocity average, meaning there is less than a 1% change that it was not real. You can easily determine this yourself using the ttest function in excel, google this test for more info.

In general there is no magic number of samples answer that applies to everything, so why do you always hear 20 or 30, etc? Simply put a statistical test is based on separating the signal ( in this case the difference in the average velocity ) from the noise ( the sd of the shots ). So how do we use these values to test the signal vs the noise? One of the simplest yet most powerful concepts in statistics is the standard deviation of the mean (the average); meaning what is the expected standard deviation of averages, given we know the sd of individual results.

In this case your sd is based on the deviations of the INDIVIDUAL shots. The SD OF THE MEANS is calculated as SD(Mean) = SD(Individuals) / Square Root (n). So as n increases the noise (variability of the averages) is reduced; eg if n=4 then the SD is cut in half (square root of 4 = 2). You will see that the impact of increasing n quickly diminishes for n>20. This is the basis for "20 or 30", but again not required for statistical significance.

In your case the difference of the average velocity vs the sd is sufficiently large to achieve significance. The number of samples needs to be larger if you wanted to detect smaller differences. Now that you "know" the sd, it is straight forward to calculate the n required to detect a smaller difference.

For the next test I would sort 5 primers into 5 categories: lightest, average, heaviest, and half-way between the extremes and the average. Then you can get a better correlation of velocity vs primer weight to decide that ranges to use for large scale sorting.

 

[Dusty Stevens]

Can you actually shoot the difference in primers- say shoot a group with a few brands, find one better and actually repeat that? Like your groups are better with gmm every single time? Until you can prove this on different days with unsorted primers theres no need to run this test

 

[Joe Salt]

Dusty let him do the test, I know at 1000 yards I can sure tell the difference in different brands of primers! So I for one would like to see what he comes up with.

joe Salt

 

[Mulligan]

Can you actually shoot the difference in primers- say shoot a group with a few brands, find one better and actually repeat that? Like your groups are better with gmm every single time? Until you can prove this on different days with unsorted primers theres no need to run this test

Dusty,
How does my ability to shoot have anything to do with the test?
the question is, “Does primer weight effect muzzle velocity”?
That is the only question.
Don’t get me wrong, I fully understand your question of practical application of the data. There are plenty of folks around who can shoot that good......... look at all the long range records that have been broken in recent years. 1/4” 600 group, 1000 yard group barely over an inch just to name two. The quality of components, materials, craftmanship, understanding of the trade, understanding of cause and effect, measuring everything, and most importantly the desire to prepare to win have all contributed to these new records being shot.

As I see it, this is simply one more little measurement to add to the list of things to test. At short range, 26 FPS could make no difference at all, I get that.
CW

 

[denobravo]

very interesting
I take it it is a given that necks are clean inside and annealed as in any test changing only one variable at a time to PROVE the test
I had never considered primer weight
I am finally getting single digit es
but only at certain charge weights
usually during the seating depth test
which may be a fluke
I am looking forward to hearing(reading) your results
thank you

 

[IdahoSharpshooter]

Shooter ability is the one variable you did not figure into all of the testing. I shot a friend's 6PPC LV rifle a couple weeks ago. No wind flags, not much wind. I shot a group with zero horizontal, and about 1 1/2 bullet holes of vertical. At 100yds, that gave me a .290" group. What does that tell me, other than the wind blows all the time here in Idaho? Was the rifle or the shooter at fault?

 

[alinwa]

View attachment 1065886 Here is a perfect example of what Dusty is referring to. You will see far more difference in type than weight

unless those groups are 400yds plus, they all look the same to me...... ???

 

[JFrank]

unless those groups are 400yds plus, they all look the same to me...... ???

Ha good one Al"
Those are 300 yard and very different

 

[CharlieNC]

Ever get an unexplained flyer? Would you turn down a reduction of 25fps just for doing a little no-brainer grunt work? Thats enough to kick me out of the xring on ftr at 500yd, which I need to maintain to get all the help I can for crappy wind calls. While this degree of improvement may not be the norm, the effect is enough to motivate me to evaluate for myself.