Satterlee 10 Round Load Development

[Klem]

Here is the next load workup comparing Satterlee with groups on a target. Unfortunately, velocities fail to predict the best long range group, again.

223, 77RDF using four round groups of 8208. The ladder starts 10% below SAAMI climbing in 1% increments, shooting at a 100yd target while velocities are recorded on a chronograph.

Raw data


Four 10-shot Satterlee's in a row, and then the average of all four.


Velocities and 100yd groups graphed together to see whether any relationship exists...There is none.


The data has been extrapolated to 1,000yds using JBM ballistics and the combination of vertical spreads and 100yd group sizes used to determine the overall size of the predicted group. Velocities have been given the best chance of success by assuming the fastest round is at 12 O'Clock while the slowest round is at the 6 O'Clock of the cone of fire. Even doing this the smallest groups at 100yds are the best predictor of smallest groups at 1,000yds. Note, the two best groups at 1,000 (highlighted in yellow on the raw data) do not have the lowest MV standard deviations. One of the best (lowest) velocity standard deviations produces a long range group almost twice the size of the best actual groups.

The four 10-shot Satterlee's show little consistency between each other, and a poor relationship to actual groups on the target.

This is the second work-up comparing [predicted] best group at 1,000yds from muzzle velocities and actual 100yd groups. Both work-ups have group sizes at 100yds the better predictor. The theory of low velocity SD is fine as long as the barrel is pointing exactly the same direction each time, and it never does. You can take the shooter out of the equation, the wind, clamp the gun and the barrel will still be whipping around releasing the bullet in different directions, regardless of velocity nodes.

In both tests, group size at 100yds is the better predictor of results at 1,000. In this test muzzle velocity as a predictor is haphazard.

Disclosure:
Using a bench-rested bolt gun with 23" Krieger barrel, March scope and Jackson trigger. No wind.

Here are the targets. On Target used to calculate group sizes. No shots discounted and barrel cooled three times.

[imaguy3]

Honestly I think you're still missing the point of satterlee's method and how it achieves what he wants in a load.

I bet he would take your load at 21.3 over your load at 21.8. I know I would based on the SD alone. That node b/t 21.1-21.5 is not bad at all for consistent velocity, which is what he's after. The smallest group, in the middle of the node is negligible compared to the others for his purpose, as he's not a 1k yard bullseye shooter. He wants a consistent load that allows for variations and also is accurate enough for his target.

If it were me, I would load on that 21.3 for the consistent velocities, play with seating depth a bit, and go kill many deer (or ring a bunch of steel) and be happy that I didn't waste a ton of components or time.

[Klem]

I beg to differ Sir, you are missing the point.

Satterlee and followers believe a ladder of 10 shots over a chronograph is a predictor of consistency. My critical view is that a 10-shot ladder is a consistent predictor of little more than a volume/velocity relationship, which when trended should give a high correlation value (R-squared) approaching 1, unless compressed. But it is certainly not an efficient predictor of results on a target, let alone a consistent one.

Let me run another experiment. I will load and shoot multiple groups at the best 'velocity node' (21.3) and multiple groups at the best 100yd group on the ladder (22.9), and compare whether the velocity node is the better predictor of group size consistency.

[Klem]

Here are the results of the comparison between imaguy's suggested load; 21.3gns, based on Satterlee emphasis on consistency, and the 22.9gns tightest groups at 100.

Same gun same bullet same powder same conditions. I alternated between Satterlee and tightest 100 group to keep the same barrel conditions.

6 x 4rd groups at 100yds using 6208, 77RDF in 223. bench rest bolt gun. 3 rds did not register on the chronograph so I dropped one velocity from one column so both columns have the same number, to be fair.



Results
The 'Best Gp' has the better consistency of velocities. This is the opposite of what the earlier work-up predicted.
The Satterlee groups are fractionally tighter when averaged. If you discount the flyer in target Four they would be the same.
The "Best Gp' groups have better consistency of precision.

If you believe in consistency then the 'Best Group at 100yds' wins on both standard deviation comparisons (Satterlee).
If you are going for best results on target then the Satterlee wins by a hairs whisker. In a competition, the Satterlee would have won. In hunting, you wouldn't notice the difference.

For me I cannot resolve how none of this seems to be repeatable and therefore predictable. The less consistent string here was more accurate whereas the earlier work-up the results were the opposite.

To conclude
If you shot only 10 rounds in a ladder using the Satterlee technique you would have honed in on an unrepeatable load that is little more than a guess.
If you shoot a few more rounds at each increment trying to average out the randomness you still end up in a situation where future predictability is shaky at best.
Choosing the best group at 100yds from a few rounds however is no better a predictor of future results it seems.

The only thing you can conclude with certaintly is that velocity as a predictor of accuracy, or even consistency is not proven this case.



Targets

[Lemonaid]

Klem, congratulations on that .071 group! Also, it's awesome that you actually did the work, testing and evaluation of the data to find out if what the Satterlee theory claims actually delivers. I don't know of anyone else who has done testing of this nature but there should be more of it!

I read in a Zediker book something like "Three round groups won't prove that a certain load is good, but it can prove that it isn't". Could the Satterlee method be used in negative way, weeding out unproductive loads?

What is your take on the relationship between group size and low SD, or low SD and group size. Should one be more important than the other?

[Klem]

Lemon,

This is my understanding to your questions, however I may be wrong.

3rd groups won't prove a certain load is good...(Zediker). That sounds like the tenet of Logical Positivism applied to reloading; That you can only prove with certainty that which you can disprove. Your groups can all be good but you can still not say with certainty that all future groups will be just as good because one day, one group might not be. But as soon as this poor result comes along you can now say something with certainty, that this particular load is not a consistently good one.

SD is a metric which describes the shape of a Normal Curve. The SD for a number of groups describes the distribution of those groups. A low SD describes a number of groups that are all about the same MOA's. Those MOA's can be large or small on average but the SD cannot describe that aspect. One is not more important than the other because they both describe some aspect of the groups. In a sense you are asking me to choose between Satterlee and the best group size. You can have very consistent groups but they are all bad. Or you can have a few good groups, a few bad groups, but no consistency across all the groups. One is not necessarily better than the other. What we are trying to predict is the load that gives consistently good groups. i.e the best of both.

Be aware the distribution of a number of groups might not be Normal however ('skewed'), and in that case a single number like a SD cannot properly describe the spread of gp sizes.

The tightest group in the above test is not the average of the group so it represents an outlier. Singling it out for praise is less realistic than singling out the average group size for praise. Notice it was the first group of the series, hardly a coincidence. After about 10 fouling shots the first group was the best. Pretty quickly there was mirage above the barrel.The reticle is clear but the target was getting blurry. It was dead calm so the boil above the barrel was making it difficult.

[Popeye212]

Klem, congratulations on that .071 group! Also, it's awesome that you actually did the work, testing and evaluation of the data to find out if what the Satterlee theory claims actually delivers. I don't know of anyone else who has done testing of this nature but there should be more of it!

I read in a Zediker book something like "Three round groups won't prove that a certain load is good, but it can prove that it isn't". Could the Satterlee method be used in negative way, weeding out unproductive loads?

What is your take on the relationship between group size and low SD, or low SD and group size. Should one be more important than the other?

That is exactly what I use it for and I have to disagree with Klem that the Saterlee is not useful. I have seen just the opposite. Excellent tool for weeding out bullet powder combinations.

[Lemonaid]

Klem, I did not phrase my questions well. Would you say that a load that tends to produces small groups does so because it has low SD velocities?
And that a loads that have horrible SD velocities tend to have large group sizes?
I've had loads that had excellent SD's but the groups were large so other factors are in play (harmonics likely) but how often do you find a great group without pretty good SD's? Is there some sort of link between the two that could be a math function?

[Sinclair]

Klem, I have a question, what procedure would you use if time, resources, and opportunity to shoot are all in short supply? I am a hybrid shooter, the most consistent, long range, 1 MOA or better accuracy is my goal, not counting X's or bringing down Antelope on the far horizon.

[Klem]

Klem, I did not phrase my questions well. Would you say that a load that tends to produces small groups does so because it has low SD velocities?
And that a loads that have horrible SD velocities tend to have large group sizes?
I've had loads that had excellent SD's but the groups were large so other factors are in play (harmonics likely) but how often do you find a great group without pretty good SD's? Is there some sort of link between the two that could be a math function?

Lemon,

We can come about a list of reasons why bullets go where they go and then control by keeping constant as many as possible, but then varying one in particular. This is what we do when working up a load.

If all other variables are controlled then no doubt we will see groups with vertical spreads that correlate with their velocity SD's. So yes I agree, there has to be a correlation between velocity SD and group size.

But, as in these tests all other variables are not under control. The barrel can be pointing in a different direction each time yet the bullets are coming out at the same velocity (low SD). In this case the groups will be large while the velocity SD will be low. The correlation between the two variables is suddenly not as clear. Not that the shooter is deliberately pointing the barrel in different directions but the barrel is whipping around each time it fires and apart from maybe using a barrel tuner or screwing something on the end like a suppressor there is little you can do about it. In theory, velocity SD's should be a predictor of group size but in reality there are other factors defeating the robustness of that predictor. Try as we might, we don't have absolute control over all the variables that cause bullets to go where they go.

You acknowlege this in the last couple of sentences so we're on the same song sheet. Your last couple of questions are the killers. Surely there's still a correlation. Intuitively I agree with you, even though these other variables are polluting what should be a clear correlation over time and multiple firings it should still reveal itself. That's if the other variables that we cannot fully control are at least random. But, if they are not random, like if the barrel whips badly at a particular load every time then we will never be able to claim a positive velocity SD/Group size correlation at that load.

My thoughts are that any focus on velocity SD is going to be a clumsy at best, predictor of future group sizes.

[Klem]

Klem, I have a question, what procedure would you use if time, resources, and opportunity to shoot are all in short supply? I am a hybrid shooter, the most consistent, long range, 1 MOA or better accuracy is my goal, not counting X's or bringing down Antelope on the far horizon.

Sinclair,

Mate, that is the $64,000 question which I don't have a definitive answer to.

I will say that I am a believer in paying attention to results on a target because that is the end goal. Whereas velocity is a mediating variable, and only one variable to achieving the end goal. I am also a believer in statistical distribution; that results can be plotted on a distribution graph to realise a Normal curve. This implies small sample sizes are less of a predictor of future performance than large sample sizes. But, how many shots are enough to be a reliable predictor of future performance?

I think the 10-step 10% workup is still sound because it is a safety check, plus also reveals the load/velocity relationship. I still prefer seeing results on a target because that is the end goal. Velocity is only one, and therefore a clumsy predictor of future results on a target. Results on a target is another future predictor and while no guarantee either, at least represents a combination of all the variables (including velocity) that cause bullets to go where they go.

If you want to save bullets and time you could fire only one round at the lower levels of the work-up but when you get closer to max fire a few more at a target to see how they group. We all prefer the best groups at the highest velocity so concentrating closer to max makes sense. Say 20 rds, single shots for the first five increments and three round groups for the last five increments. Then depending on what you see on the target take it from there. If all groups are tight it's an easy decision. If not, then you can load up some more to check or settle on the best at the time and hope it's repeatable.

[Lemonaid]

Klem, your help thinking through these concepts is outstanding, thanks! I really like this part:

"If you want to save bullets and time you could fire only one round at the lower levels of the work-up but when you get closer to max fire a few more at a target to see how they group. We all prefer the best groups at the highest velocity so concentrating closer to max makes sense. Say 20 rds, single shots for the first five increments and three round groups for the last five increments. Then depending on what you see on the target take it from there. If all groups are tight it's an easy decision. If not, then you can load up some more to check or settle on the best at the time and hope it's repeatable. "

I was starting on my own to do exactly what you have stated.

Now something sort of related I would like some sorting out of.
Gun magazines often use as a testing protocol an average of five, five shot groups as an accuracy standard for their evaluation of a gun's accuracy with a brand of ammo. So if I had a load for my rifle that gave sub 1 moa with a load at 100 yard after repeated tests and say it still gets 1 moa at 300 yards will it stay one moa at 1000 yards? (no wind in this thought experiment).
Now for the tricky part, is the velocity SD baked in the cake so to speak? Good SD, bad SD, in between plus harmonics all combine and don't matter. Is a good SD 1 moa load just as accurate at long range as bad SD 1 moa load?

[centerfire]

Klem, I have a question, what procedure would you use if time, resources, and opportunity to shoot are all in short supply? I am a hybrid shooter, the most consistent, long range, 1 MOA or better accuracy is my goal, not counting X's or bringing down Antelope on the far horizon.

A thrower and press that you can take to the range.

Just to add to the thread, a good group with high SD with be a poor group at 600 yards and beyond. A bad group at 100 with a good SD generally may only need a seating depth ladder to find the optimal depth to get it to group well. AR15's may not allow a productive seating depth ladder so you might have to chase down 10 different bullet and powder combinations until you get both tight groups and a low SD.

[centerfire]

Is a good SD 1 moa load just as accurate at long range as bad SD 1 moa load?

High SD means you have a wider dispersion of velocities (over simplifying). Take out your ballistics calculator, add or subtract 50fps, and look at the drop at various ranges. The difference in drop between the highest and lower velocity will be the minimum mechanical accuracy of your ammunition. A high SD will always group worse than a low SD at longer ranges, everything else being equal.

The real question should be whether or not you should sacrifice low SD's for low load development time and cost.

[Klem]

Gun magazines often use as a testing protocol an average of five, five shot groups as an accuracy standard for their evaluation of a gun's accuracy with a brand of ammo. So if I had a load for my rifle that gave sub 1 moa with a load at 100 yard after repeated tests and say it still gets 1 moa at 300 yards will it stay one moa at 1000 yards? (no wind in this thought experiment).

In your hypothetical where there is literally no wind [an underground 1K tunnel?] the stability of the bullet still deteriorates as velocity drops. The Miller stability calculator on JBM [online] demonstrates this. I imagine less stable bullets translate into wider groups. In reality there is always wind and in my experience of 1K F-Class slow bullets can blow out the group dramatically. With 223 as it approached Mach 1 about 900yds the groups would blow out like a cross section of a trumpet. But the other extreme can also blow groups out. Consider an overstabilised bullet resisting pointing down to the target coming out of apogee. As it flies through the air it will yaw.


Now for the tricky part, is the velocity SD baked in the cake so to speak? Good SD, bad SD, in between plus harmonics all combine and don't matter. Is a good SD 1 moa load just as accurate at long range as bad SD 1 moa load?

SD's are not 'fixed', they are simply a single digit reflection of a group of numbers. Change any of the numbers and the SD changes.

Your question regarding the relationship between short and long range is considered because long range is where the velocity of a bullet could have more of an influence on extreme spread (group size) than group size at short range. I had not considered this until it was brought up on this forum. I understand this is also what Satterlee advocates are concerned about and use it as justification for focussing on velocities.

Consider the group as cone of fire that proportionally increases as range does. You allude to this with your 1MOA hypothetical. All things being equal and controlled the best case scenario is going to be 1MOA at 100 is the same all the way to 1MOA at 1,000. We know stability drops with velocity, which has consequences but let's stick to the hypothetical for now.

The group at 100 is made up of rounds with different velocities. Using a ballistic program we can calculate bullet drops for each of those velocities and work out the maximum possible spread by imagining the fastest bullet is at the top of the cone and the slowest at the bottom. If the muzzle velocities are so different that the vertical spread on the target exceeds the 1MOA cone of fire then this becomes a situation where velocities are more influential than group size at 100yd in predicting long range results. In that case velocity SD's can be used as a predictor of max spread at 1K, but are not as definitive as actually working it out.

If we look at those 44 muzzle velocities from the most recent test we get the following for both loads;


The group size at 100 is extrapolated to 1,000yds (using 1MOA = 1.047inches at 100yds).
The extreme spread of velocities at 100 is also extrapolated to 1,000 using JBM ballistics.

In both loads the group size at 100 has more influence over group size at long range than the velocities. For 21.3gns, the original Satterlee predictor of best load the group at 100yds will translate to a 3.654" group at 1K. The velocity differences of those bullets will translate into a group no bigger than 3.2" at 1K. In this case velocity is a clumsier predictor of 1K group size than actual 100yd group size. This, because the cone of fire will never be tighter down range and as we've discussed will likely be even larger. groups physically cannot contract down range. They always get bigger. But, the velocity prediction of the group at 1K is based on the best possible scenario; the fastest bullet at 12 O'Clock and the slowest at 6 O'Clock. It is more likely then that the velocity groups at 1K will be smaller than predicted.

Using the most recent test the group size at 100 is a more accurate predictor of group size at 1,000, and even more so in reality. This is another nail in the coffin of using velocities to predict results on a target.