When I chrono a load I normally shoot 10 rounds across it (if it's a new load I normally shoot 10 over the chrono and then 10 more for patterning since I can remove my Magnetospeed in between so as to not affect the pattern).  Out of those 10 chrono rounds I will normally have one that's significantly high and one that is significantly low, both just enough to affect the standard deviation quite noticeably.  Other times a couple excessively high or low.  In any case, enough to affect the averages.  I'm sure I'm not the only one with this issue so when you guys run into something like that do you include all shots in your records or just the ones that you feel are "Typical" of the load?

There are pros and cons for both, I realize that.  The biggest Pro being that it is typical of how they will load most of the time, but at the same time I don't feel I'm getting a true indication of what the actual load is capable of.  To date I've kept everything in spreadsheets (15+ years of them) but will occasionally go back and clear a few of the shots just to see what happens to the numbers without the outliers, and then undo the changes while making a note somewhere of how the numbers look without them.  Sometimes it's significant, other times it's insignificant.  So I'm just curious what others do or recommend.

Try this next time..when you setup your target add a small target that is out of the way but easy to move the barrel onto with out rebuilding your set...Now before shooting for group, first round is a dope shot, then 2nd shot on move to your "target".. on the last shoot, I am going to guess you are worrying over getting that perfect group and end up pushing the shoot off...practice..concentrate on being identical in every action from shot to shot..Although if its a thin enough barrel it could be heat warp...

Define "significantly". Are we talking a couple hundred FPS or 40-50 FPS?

I never toss any data unless it's really clear there's some sort equipment (chrono) error. It happens with optical chronos, I've never used a magnetospeed but my impression is they're pretty reliable if set up properly. Is your loading process precise and consistent enough to place the doubt on the chrono?

Keep going, guys.  I get involved in this discussion later on.

Throwing out the high and low data points is a common practice to deal with outliers in statistical analysis. Done this for many years in business and product development.

Just be consistent and not cherry pick your data. If you throw out high and low data points, always to it for a particular study.

You could always increase you N to 12 rounds if you like working with the number 10 in your analytics.

No I don't throw out outliers with only 10 shots unless the are way off.
It generally turns out that they aren't outliers. They repeat the "outlier" again and again, loads that are genuinely tight repeat and don't have outliers even at 30 shots.

Throwing out a data point is only fooling yourself. The correct thing to do is to repeat the string or continue with the string.

On a precision load if one shot is 50fps high from the next highest or 50fps low from the next lowest then I will ignore it but watch for it again.
It has to be pretty far off from the next closest to toss it.

The greater the numbers the worse it will get.

10 shots vs 30 shots, conclusion, it's a 9/35 Sd/Es load not a 3/15.
Attached File
Attached File


With this load the low shot number  pretty accurately represented the load consistency as shown by the higher shot number.
Attached File
Attached File

Magnum handguns are notorious for giving bad data. The shock wave (muzzle blast) moves faster than the bullet. Use a blast screen, (a piece of plywood with a hole in it) placed between the chronograph and the firing point.

Wide data points shouldn't ignored. Are all the primers seated fully? Did you weigh charges to insure they are identical? Mixed headstamp brass?

If you think your instrument is giving you errors so large as to create "outliers" (higher, lower or both), how do you know what to do with the data?  If it's doing that, why do you trust the instrument to give you a decent mean value and standard deviation in a small sample test?

Consider a shot whose value is almost spot on the average.  How do you know it isn't a serendipitous error?  It might be erroneous but because it matches what you want to see you decide to keep it, when in fact it was 50-100 fps lower (or higher) and its displayed value is actually in error?  How do you know?

You can do as you please but keep an open mind on data quality.

I don't think my instrument is giving errors, where did I say the instrument is giving errors. If there is an outlier it is likely the round itself.

I have used both the Magnetospeed and Labradar side by side, I have seen they agree on the readings for each individual shot over and over again, enough to have confidence in them.

I have confirmed the velocities by shooting well into transonic ranges, drop predictions coincide with actual drop.

I cannot help you with your mistrust of chronographs, I see verification of the Magnetospeed and Labradar numbers over and over and over again shooting long range.

If they were not accurate two different devices would not agree on the same shot velocities.

If they were not accurate certain loads would not always have low SDs and ESs and another load would not always have higher SDs and ESs.

If they were not accurate drop predictions wouldn't match actual drop time after time.

If they were not accurate higher velocity extreme spreads wouldn't show up as large vertical spreads on the target.

.
.
.
I'm not saying you won't see an erroneous number once in a great while with a Magnetospeed or Labradar but they are very rare and usually obvious like 14,000fps or something way off, not just a 25 high and a 25 low giving a poor 50ES on a 10 shot string. You shoot the same load again and again and you see it is giving that same poor ES again and again.

Look midway down on my first post.HERE  I was using all three chronographs in a row measuring the same shots, look how close the the Labradar and Magnetospeed readings are, I have repeated this test with these two multiple times with the same agreement.

Sorry for the delay in responding I've been playing RSO at the rifle range all afternoon.  As always I'm grateful for the replies and advice.

I don't shoot a precision bolt gun, it's an AR-15, and while I do take time with my reloads and do the best I can with them I'm also sure that I don't go to the effort that many of you do so single-digit SD's are few and far between.  Here's an example of what I typically see:

1 - 2,726
 2 - 2,722
 3 - 2,722
 4 - 2,732
 5 - 2,699
 6 - 2,753
 7 - 2,741
 8 - 2,720
 9 - 2,720
10 - 2,789

AV: 2,732
ES: 90
SD: 23.2

If I omit the two in red (2,699 and 2,789) the ES drops to 33 fps and the SD drops to 11.2, and the average to 2,730.  Perhaps I should ignore those two values, perhaps not.  That's the basis of my question because they typically follow a similar pattern.  Not exact, of course, but similar.  I'm OK leaving them in and I'm OK omitting them, I'm just not sure which method would give the most realistic view of the results.  Since the chrono readings are normally taken when I'm working up a load in powder charge increments I'm not really interested in loading 50 rounds or so of each increment just to get a more granular sample to work with, I'm just trying to find the most accurate snapshot of the incremental velocities so I know what I have to work with.

That's easily done.  I normally shoot "Scope Adjustment Targets" that have 10 individual targets on each sheet with a 1" grid so I can easily use a separate target.  I typically do use one of them for a "Fouler" target to foul the barrel and warm it up a bit.

Typically I see an extreme spread of 50 to 75 FPS in each 10-shot block of data, and no my loading process is probably not accurate enough to place the blame on the chrono.

That was my understanding and the reason for my question.

That's what I'm unsure of.  This pattern typically repeats itself in that I'll have a couple of shots in the group that are noticeably higher and/or lower than everything else as shown above so I'm not sure whether they are truly "Outliers" or whether they are just representative of my methods and loads.

Just shooting .223 Remington, not even into the 5.56 pressure range.  My chrono is a Magnetospeed and clamps to the barrel of my rifle.  It isn't optical.

Primers are fully seated, all brass is Norma and I keep track of the number of times it's been fired so that each case in a batch is the same.  I don't weigh them or check neck tension so it is quite possible there are some variances.  Powder charges are weighed using a Hornady AutoCharge.  It will indicate a load more than 0.1 grain over however I watch the progress and I'll toss anything that indicates 0.1 grain over or under.

Again, thanks to all for the help and advice.  It is much appreciated.

Looks to me like those numbers are likely accurate and you say it’s a regular occurrence, so IMO, they need to stay. Same reason we don’t toss “fliers” on a target if we’re being honest with ourselves.

Removing them barely shifts the average, anyway. What would be the point other than lowering (artificially, IMO) SD and ES numbers?

I'm OK with that.  I really was asking a real question of whether I should omit them or not since I couldn't convince myself one way or the other for much the same reasons.  The primary reason for omitting them is that since I use the velocities in load determination the closer they are the better.  On the other hand, as you said, they did happen, they are real, and ignoring them is akin to pretending history didn't happen.  I can work with the data either way, I just can't convince myself one way or the other which is the best way to go about it.

And, yes, I am aware that in some respects I flip-flop in my answers because I really am undecided at this point ;)

Those aren't too far out. I'd include them. SD of 10-30 is typical.

Thanks.  I didn't really know what a typical SD was since everything I read says to shoot for single-digit SD's but I've only managed to do that once.  I'm sure a semi-auto rifle is going to have more variance than a bolt action and I'm also sure that my reloading practices aren't as precise as they could be.  My SD values typically run in the mid 20's, sometimes lower sometimes higher, so that makes me feel better.

Hmmmmm. Hope you don't mind me piggybacking. Today's results are unusual of the few times I chrono. Norm is between 15-25fps SD.

35*F and consistent overcast. I wonder if the screens weren't bright enough. ETA: I just learned the screens aren't necessary in overcast.
Only thing I did differently was shoot unsupported prone VS. slung up or on bags, which I expected to open the numbers up some. Bore untouched since the match last week; no more than 200 rounds since cleaned.

Load 1:
2648, 2636, 2647, 2681, 2618, 2560, 2565, 2631, 2631.
Avg: 2624 / ES: 121 / SD: 39

Load 2:
2936, 2985, 2893, 2886, 2891, 2890, 2904, 2856
Avg: 2905 / ES: 129 / SD: 39

Not only are the ES and SD higher than usual, they're almost identical.

No sir, I don't mind at all.  It's all about knowledge ;)

I will point out something I read recently, and I can't remember where or when, but it was a discussion of how recoil management affected muzzle velocity.  I may have saved a link to it at work, I'll look tomorrow.  The author's point was that if the rifle is allowed to recoil freely as opposed to being supported at the butt that the muzzle velocity is different (and I don't remember whether it was faster or slower, only that it was different).  Perhaps unsupported prone vs. using a sling played into that arena.

From the US Rifle Teams forum:

"Data from a few who tested with the 308 Winchester from a bench .

Fixed, non-recoiling barreled action: highest average (AVG), lowest extreme spread (ES) and standard deviation (SD). Not good for accuracy tests 'cause the barrel vibrates different during bullet's barrel time compared to being stocked.  Best test method for ammo interior ballistic uniformity.

Free recoiling (untouched except by two fingers pinching trigger towards guard), rifle resting on bags: lowest AVG by 40 to 80 fps compared to fixed, often 10 to 15 percent more ES and SD than free recoil.  Friction between stock and bags will influence numbers. Excellent for loads' accuracy test if friction is minimized and repeatable for all shots. Preferred by most benchrest winners and record setters.

Shouldered rifle resting on bags: between the free recoil and fixed AVG numbers. Larger ES and SD, sometimes a 3X to 4X spread in SD and ES compared to that of fixed and free recoil.

Would be interesting to see data from 223 ammo in AR platforms.

My own tests in the late 1980's were with a 12 pound NRA match rifle chambered for 308 Win.  Got about 40 fps lower AVG shouldering it lightly  compared to hard as I could as it rested on bags atop a bench.  ES and SD numbers holding hard were about 2/3rds that when held lightly.

Slung up in prone on the 1000 yard line at 6 AM, all velocity numbers with that scoped rifle and ammo were about the same as medium hard shouldering at the bench. Most interesting; some shots with about 10 fps lower velocity than average went about 1/2 MOA above call, some with higher velocity went the same amount below call.  Typical tiny spread in bullet BC coupled with position and hold variables we all have, in my opinion, even if all shots called inside a 3/4 MOA area.

Mid Tompkins mentioned that his benched, shouldered 308 Win AVG numbers with his rifle were 100 fps slower than his friend got with it."

Popnfresh, my post was directed at the OP who stated he plans to drop data from his calculations.  I only quoted you because I agree with what you said in that sentence.

Okay, after checking my data from a couple years ago, the 2nd load seems to have an inherently high SD - 3175/114/37fps.

While not the exact article I was thinking of that is the same content and results.  It could well be the reason I'm seeing some differences since I can't say for certain that I hold the same amount of pressure on my rifle for every shot.  I try to be consistent between shots but can't be sure that I do.

Thanks for posting it.

Huh never knew that. Would explain some 200-300fps deviations I have seen while chronoing my .458 SOCOM.

I have always thrown those types of deviations out, as with the .458, you notice a huge difference in recoil on a 1300fps round vs 1600fps round. I won't throw out 50 or 75 fps stuff, especially with .223 or 6.8 SPC.

If the measurement system is consistent (and yours should be)...  If you fire ten rounds, then you have ten data points.  Not ten minus two that are inconvenient.

If you want to pretend that your load is better than it is, you should shoot fewer rounds per string.  Like 1, 2 or maybe even 3 shot groups.

But seriously... if you decide to throw out data points, a valid scientific approach would require some sort of technical basis for doing so.  It's a lot easier to be casual with data with an optical chronograph which can be very flaky under lots of different conditions.