Weight sorting cast bullets: - all that and a bag of chips?

[Bjornb]

-Sorry, I couldn't resist using a Goodsteel-ism in the thread headline.

For some time, and especially since I got involved in the XCB rifle and bullet project, I have religiously weight sorted all my bullets cast for High Velocity. The bullet I shoot the most is the NOE 31-165 (XCB), and I spend long hours building bell curves, sorting and bagging the bullets I cast. Then they are loaded in 10-shot strings with every bullet in each string weighing exactly the same down to 1/10 grain (the limit of my scale).

So I got to thinking: Is there a way to quantify the importance of this practice of weight sorting, in a way that's statistically valid? You be the judge, but here's what I came up with:

From a casting session that yielded a gross quantity of 474 bullets (linotype), I culled 29 for physical defects, for a net of 445. With the 445 bullets sitting in a plastic container, I asked my daughter to pick out 30, and to put them in a plastic bag. They were put aside and no further weighing or inspection was done with these bullets.

This left a balance of 415 bullets, which were weight sorted into this bell curve:


Not a spectacular bell curve, but not totally cr@ppy either.

I then selected 30 bullets from the largest group in the curve: these weighed 154.4 grains naked.

All 60 bullets were then gas checked (Hornady), lubed (Lars 2700+), sized (.310 Buckshot push thru) and loaded in cases that were from the same batch of originally formed W-W cases. CCI 200 primers and a thrown charge of 52 grains WC867 was used. From experience this load gives me right around 2360 fps on the average, so I saw no need to hook up the MagnetoSpeed.

Shooting protocol: 2 foulers (same lube and powder) from a clean barrel, then 30 shots with 2 minutes between each shot. Every 7-8 shots the pause would be slightly longer when Cease Fire was called at the range (we shoot 15 minute volleys).
Then the barrel was cleaned (1 wet patch Ed's Red, 2 dry patches), then repeat as above.

All said the shooting took 3 1/2 hours, and I have to say the results were not what I expected:


I made a double take and re-checked my ammo boxes, but the picture correctly shows what happened. I was surprised.

[sdcitizen]

What are the odds that linotype is less susceptible to forming internal voids? Your results boggle the mind!

[Hannibal]

Perhaps, to some extent, this speaks to the skills you have developed as a caster/sorter. But the question also begs to be asked, following visual inspection, at what point do weight variances become important to down range performance?

VERY interesting.

Thank you for your continued efforts. Never a dull post !

[Pb2au]

I was wondering how long before someone did this exact test when the weight sorting topic was running hot.
It would be interesting to repeat the same test again, although 30 per batch is a nice sample size.
In industry when we run CPK testing for client machine acceptance, clients typically measure at least 50, if not 100 (thanks GE,,,,,) samples off of our machines to measure and understand accuracy and repeatability. ( I work for a manufacturer of machine tools)

thank you very much for sharing this, it is very interesting.

[Bjornb]

I was wondering how long before someone did this exact test when the weight sorting topic was running hot.
It would be interesting to repeat the same test again, although 30 per batch is a nice sample size.
In industry when we run CPK testing for client machine acceptance, clients typically measure at least 50, if not 100 (thanks GE,,,,,) samples off of our machines to measure and understand accuracy and repeatability. ( I work for a manufacturer of machine tools)

thank you very much for sharing this, it is very interesting.

I will most definitely run the exact same test again. Mainly because I didn't believe my own eyes!

[MBTcustom]

It's my personal opinion that weight sorting does very little once you achieve a certain proficiency with your casting. I think that if you were to dial in your casting till that bell curve was tighter, you would see both groups shrink, simply because you have a more controlled process. Using the weight of bullets to plot bell curves is different than weight sorting. I support using bell curves to learn to be a more consistent caster, but once you have the prescription dialed in for each mold, it really comes down to controlling things that you cannot possibly measure by observing the trends of the things you can measure.

[RogerDat]

There is just one more variable to remove that would make it double blind. You need to remove knowledge of which batch of 30 is which to remove any subtle bias that might impact your shooting.

[pworley1]

Both groups speak well of your casting and shooting skills. The main question I have is "did this experiment bring you enjoyment or frustration?". If enjoyment then it was very worthwhile, and I suggest further testing. If frustration, then I suggest you move on. Life is too short to frustrate yourself on purpose.

[Bjornb]

"There is just one more variable to remove that would make it double blind. You need to remove knowledge of which batch of 30 is which to remove any subtle bias that might impact your shooting."

True, but since I'm a one-man operation I have no way of "blind drawing" one batch or the other, and then know for sure which is which after the shoot.

[Bjornb]

Both groups speak well of your casting and shooting skills. The main question I have is "did this experiment bring you enjoyment or frustration?". If enjoyment then it was very worthwhile, and I suggest further testing. If frustration, then I suggest you move on. Life is too short to frustrate yourself on purpose.

That's a valid question. Answer: I thoroughly enjoy doing these "tests". I usually confer with Goodsteel, Larry Gibson and Sgt. Mike about what I'm planning, and they always give good advice that I take note of, and sometimes even follow!

[MUSTANG]

Scratch head over results.
Consider variables, stand back and look again.

The one way out possibility I see is the weighing and sorting was done before lube/sizing. What is the possibility that inconsistencies in the sizing/lubing operation resulted in the outcome being 180 degrees from what most of us would have forecast if asked before your results were published?

[texassako]

I tried a similar test in the fall with my favorite rifle and its favorite mold, the NOE 311331. I got similar results and just visually sort now, or occasionally put them on a digital scale to cull any outliers if I am doing something special with them. I boiled it down to experience with the mold and how it casts making the weight curve small enough that other factors play a larger roll in accuracy at the relatively short 100 yard range I shoot at. I made it double blind by having my wife number my 2 load sheets, holding them for me, and just putting the number in the box.

[scottfire1957]

One grain is one seven thousandth of a pound. That is a very very small number.

A small, small number. .00014 lbs. Teeny tiny. Lilliputian.

Add that to the fact that each and every shot travels at a different velocity, as small as that difference may be.

AND what does the one or 6 granules/flakes of powder difference dispensed for the same load make? We're talking about thousandths of a grain here. Thousandths of a thousandth of a pound.



A lot to worry about.

[Sekatoa]

Another observation...I'm not feeling doing the math tonight, but just looking at bell curve, I'd say it he odds a high percent of randomly pulled bullets are in fact the same weight as your measured group, or from the next closest on either side, are pretty good. So, your are not really comparing your control group to the full +/- spectrum of this casting session. But I understand that wasn't exactly your goal here either.

I would say in addition to the blindly pulled bullets, I would consider a batch be loaded from each end of the curve. Comparing those two groups to your other two could further illustrate the total possible spread for the casters ability.

Then, it also occurred to me that these results not only reflect casters ability, but likely the alloy, mold, rifle, etc.; meaning that the results of sorting the cast bullets may worth it more for some alloys, et al, than others?

[Cowboy_Dan]

Personally, I would have weighed the random group. Might have even rejected it if it too closely resembled the all one weight group.

[mongoose33]

I have a background in statistics; I must say I'm amazed and delighted to see these kinds of analyses being done. What's especially delightful is the search for alternative explanations of the result.

A few comments coming late to the parade:

1. If the one group was shot prior to the other group, what might have changed is atmospheric conditions, humidity, temperature, wind. Over a 3 1/2 hour time frame, that's possible. Especially wind. Should you be inclined to repeat this, a way to reduce that possibility is to alternate shots on two targets side by side.

2. If you take away the top, far right, and bottom right rounds, and compare groups, they are essentially identical. It is those three holes that enlarge the group. These could be flyers, gusts of wind, etc. I took the liberty of downloading the pic and erasing those three holes:





The groups, save for those three instances, are the same to my eye. So there's not as much difference here as might appear. The vast majority of boolits from both groups performed similarly, which tells me the weight differences probably aren't meaningful.

One other thing I note is that if you look at the rounds outside the ragged hole, on the left they were random-appearing in their distribution around the hole; on the right, the errors tend toward the right. That partly makes me wonder if there were wind gusts from the left on that trial.

3. Were they shot from a sled? I'm guessing you just used a sandbag rest or similar, because undoing from a rest like a Lead Sled would have taken longer to clean and reset. That introduces shooter error--I'm not being picky here because those are pretty darned good groups, and the patience needed to do this for 3+ hours is amazing. Just noting that as, with observation number 2 just above, all it takes is three flyers to make the groups look different.

[Pb2au]

It's my personal opinion that weight sorting does very little once you achieve a certain proficiency with your casting. I think that if you were to dial in your casting till that bell curve was tighter, you would see both groups shrink, simply because you have a more controlled process. Using the weight of bullets to plot bell curves is different than weight sorting. I support using bell curves to learn to be a more consistent caster, but once you have the prescription dialed in for each mold, it really comes down to controlling things that you cannot possibly measure by observing the trends of the things you can measure.

I got that from your thread on subject. Another way to put it is that your bell curves act as the canary in mine. When the curve flattens, it indicates an shift in the repeat ability/consistency in the methodology.
Bjornb's testing now starts to shed light on a bigger picture. It would serve people quite well to subscribe and pay attention to this thread. It is quite revealing.
When we pareto information like boolit weight (IE your bell curve) it should, and in most cases does reveal that out of say a matrix 30 boolit weights against 400 boolits, we will want to see a small amount of weights accounting for the bulk of the interrogation.

Again, thanks Bjornb for posting this test. I will eagerly await the next round.

[Love Life]

Which group did you shoot first?

[JonB_in_Glencoe]

A while back, I recall someone [who's opinion I respect] saying that these small weight differences probably won't show up on the target til 200 yards or 300 yards or beyond.

[captaint]

Mongoose has a point indeed. Very interesting test just the same. Thanks for the effort.