[STEVE M]

How many rounds do you crono when testing a new load?
I did an experiment yesterday, 2 different loads thru 2 different guns (4 combinations). I did 5 shots, and then did 10 shots. My standard deviations and extreme spreads where split down the middle 1/2 each were higher with 5 rounds, and 1/2 were higher with 10 rounds. The only thing that seemed consistant was the overall average velocity was LOWER in all of the 10 shot groups.
One of these loads was a (45ACP) 200Gr. CSWC and the other was a 230Gr. JHP. I really don't know why the velocites were lower with 10shots or why some 10 shot groups had a lower SD.
Any comments or ideas? How many do you test?

[popbang]

When you chronograph ammunition you working with statistics and all that comes with it. I believe in statistics a small sample is anything of 30 and under, or somewhere about there, and a small sample does not have as good reliability as a large sample. the more ammunition you shoot across the chronograph the more accurate numbers you will receive. I try and shoot at least 24 shots to get my numbers for a lot of ammunition, and I try and do it at least on two occasions then enter all the numbers on a spread sheet. This gives a pretty good idea.

A five shot string tells you very little especially if there is a large extreme spread. A ten shot string tells you more, and larger strings give you a better idea. Remember you are dealing with statistics and numbers will stand still for anything.

[antsi]

Amplifying on the statistics ideas...

The greater the SD, the larger sample size you need to measure it. If you made perfect ammo that was 100% consistent, you would only need to chrono one shot. If you ammo is very random and inconsistent, you might need 1000's of rounds for an adquate sample size.

With the trend of lower velocities on 10 shot strings, there are two possibilities:
1) a confounding variable - if for instance, you were shooting the 10 shot strings later in the session and there was some difference with a hot gun or a dirty gun or whatever - that's just an example; really any outside factor that's different between the short strings and the long strings could account for something like that
- or -
2) the "lower velocity in 10 shot strings effect" is just random chance -- if this is so, then you definitely need a larger sample size to try and "average out" all the random factors.

If you really wanted to know whether it's 1) or 2), you could calculate a confidence interval based on your sample size and the SD of your ammo, and see if the variation in velocity is attributable to randomness or if it seems to be a real effect. But gosh that'd be a lot of work and you'd need someone better with math than me to show you how to do it.

[STEVE M]

That's ok Antsi, it sounds like you are already better in math than I am.

Statistics is not something that I have really studied but I do understand

some basic ideas. I just thought it was curious that some of the longer

shot strings would have a lower SD (and extreme spread) but the only

constant was a lowering of average velocity.

As far as outside factors, I don't think so. I shot a 5 shot first then a ten

with the same gun/ammo. Then switched to 5/10 shots with second gun

and the same ammo as the first. I went and did some group shooting with

same, then came back and repeated the procedure with the other ammo.

Weather conditions were the same through out and less than an hour past

from first group to last.

Maybe more interesting than useful I suppose.

[popbang]

Hey try combining your 5 shot strings and ten shot strings the result should have a smaller SD even if the velocity lowers.

[STEVE M]

OK, I give up, why would my SD go down by combining the 5 and 10 shot

groups?

[popbang]

It should have a smaller SD based on the way you figure small sample SD's, The other reason is as your sample grows you numbers will form a bell curve with the majority clustered within 1 SD above or below the mean. Oh heck, just get a bigger sample.

[Cherokee]

I have always used 10 shot samples. Not perfect size but most loads I stay with have very low SD so I feel the results are indicative and that is what I am after. After all, we really shoot under different conditions most of the time.

[saands]

Steve,

If all your 10 shot groups were shot following the five shot groups, then you WERE adding a variable ... the barrel was warmer for the 10 shot group than it was for the 5 shot group in all cases ... that increase in temperature would make the barrel grow which might actually reduce the friction in the barrel, reducing the pressure and yielding lower velocities. I haven't ever done that experiment, but it sounds like an easy one to check on ... the real interesting test would have been to shoot some 10 rd strings followed by 5 rd strings and see if it is just the latter string that was always slower, or if the 10 rd strings were always slower. Within your 10 rd strings, do your last 5 shots have a lower average than the first 5?

Saands

[Dave R]

My stats perfesser is college was a very practical guy. He told us, when you don't know how much to sample, then keep sampling until the data stops changing.

IOW, keep shooting strings until your average and SD stop changing with new groups. Shouldn't be too many. And yes, the more consistent your ammo is, the fewer rounds it will take until the average and SD and extreme spread stop changing.