Joined
·
9,941 Posts
What’s the trouble with patterning?
A week ago a member posted a question about how to go about patterning - distance, number of shots, etc., the usual reasonable things people who “want to get into patterning” want to find out. In the old days I would have typed it all out in detail once again, but these days I just can’t make myself do it. I’ve corresponded with the poster and know he’s a serious shooter with a lot of drive to get better. But what are the chances, I ask myself, that after I lay out the whole process he will actually do it? Probably no chance at all. After all, realistically speaking, no one really patterns with the accuracy and care and effort which seem to be needed to actually learn anything, though certainly many walk back to their cars with some shot-up newspaper pages under their arm thinking “OK, finally I know all about my gun’s pattern!”
The problem – the fact that useful patterning requires more work than almost anyone is willing to put out, more dedication than almost anyone has – is apparent in the recent “Patterning is a waste of time” thread. As you read it you can tell that almost no one posting there has ever done it and even fewer intend to. There are a number of imaginative ways to say it but “too much work” is the prevailing undercurrent with “the grapes are probably sour anyway” detectable in almost every post as well.
Is there a shortcut?
That’s why when Lon Lauber in Washington state emailed me, wondering what I thought of Ron Jones’ “Pattern Analysis for Busy Sportsmen” in the March, 2013, issue of Shotgun Sports, I perked up my ears. Ron Jones says that you can shortcut the demanding standard requirement of 10 patterns and just shoot 3 and “almost always produce a composite average within 2-3 percentage points of the value arrived at using all 10 patterns.” This is a welcome much-reduced challenge. Shooting just 3 patterns is a far more realistic goal than 10 and maybe there’s enough in the modest “effort budget” so many shooters seem to live by to get even a few to do it.
What does the article tell us?
I borrowed the issue from the Buffalo Gun Club and took a look. I was immediately put on my guard when I saw the front page featured the pattern from page 364 of Greener’s “The Gun and its Development,” an example sometimes cited as being “too good to be true” to the extent that John Brindle wondered “(Did) the draughtsman spread the pellets toward the periphery, to spread them more evenly?”
Jones uses as his standards for choke performance the Oberfell and Thompson charts from 1960 and so is basing his assessments on shells with soft shot and no plastic wads. He calls a full choke one that puts 70% of its charge in a 30-inch circle at 40 yards which may still be the standard but shells and guns do a lot better than that now, as a rule. I’d say 75%+ is more typical these days and the difference is important because for long-yardage handicap you need all the pellets you can get you shouldn’t think that 70% is acceptable “full choke” performance these days, it’s not. Almost everyone does better, many a lot better
He presents a table of choke difference, all 5% low but good enough, some tables showing the effect of various distances, and a clear and concise plan for getting your three patterns. And (pending verification) the 3-pattern average plan, the real point of the article, looks like a promising way to make patterning “accessible.” Within two or at most three percent of the “real” average? That’s good enough, isn’t it?
How much corroboration does Jones provide?
But why do I caution “pending verification?” Well, Jones never presents any evidence that just thee patterns do what he says they will. Maybe in the original manuscript he did, writing “2) the three-shot patterns analysis did a remarkable job of predicting the pattern percentage for any given load.” But I don’t see anything about that now. But we don’t need proof to at least try it, to see if the benefit Jones promises is really there: “It will give you a good idea just what your gun and loads are doing . . .”
What data will we use and why?
I’ve always liked the approach of the NRA technical staff to patterning statistics, in that they base their recommendations on the data they have collected over the years rather than relying solely on academic statistics which, though certainly more mathematically rigorous, may lack the insight into what’s likely to make a difference that a workaday practitioner might be able to offer.
We will use two examples, patterns produced by 1-ounce and 1 1/8-ounce shot charges shot from the same gun. And what we want to know is whether these loads both give us the 75% -or-better patterns we think are reasonable to expect. Said another way, we want to know if Jones’ test will answer our question “Are these loads good enough or should we do more work?”
How do we choose our examples?
We know the ten-shot averages for two loads, 1ounce and 1 1/8 ounce, and we will set up some example three-shot mini-tests. I’m sure 20 such examples will be enough; we will just use a random number table, scanning successively and creating triads. If we see a 7, a 4, and an 8, we will pick those three shot numbers, add them up, divide by 3, and have our 3-shot average. If we see 7,4,7, we will move one digit over for the last one pattern, since we wouldn’t, on the field, shoot the same shot twice. When the number given is 0, we will use shot 10. When we have used three digits in the random number table, we will move down the next three.
Here is the source of our pattern data and why choose them.
![Text Line Slope Plot Diagram]()
We’ve selected those data-sets because, though they are both “typical,” they vary in how “consistent” they are. The 1-ounce patterns only span a range of about 8 percentages points, the 1 1/8 cover about 12. Doing them both will tell us about the generality of Ron Jones’ program.
The results based on unusually consistent data.
Generating 20 “synthetic” 3-shot patterns gave these results for the 1-ounce shells.
![Text Line Font Parallel Design]()
The article’s assertion that 3-shot averages would fall within three percentage-points of the 10-shot average was supported in this test.
Will that hold true for the more variable 1 1/8 ounce shells?
The results from more typical, more variable data.
![Text Line Font Number Parallel]()
Again, Ron Jones’ prediction is supported. The 3-pattern averages are within three percentage points of the 10-shot average.
Does this mean we can start right away?
And save all that work we’ve been putting in on those apparently unnecessary additional seven patterns? We’d better go slow here and back out carefully. Before we go, shouldn’t we just make a quick check to see what we’ve done? What does the averaging of three patterns do that is not done by just looking at only one? Let's make sure that 3 shot give us better answers to the question we were asking. You remember the question: “What we want to know is whether these loads are good enough or should we do more work?”
Compare our raw results with the averaged ones.
We have twenty synthetic 3-shot tests but the underlying data comprises just ten, so let’s spilt the artificial ones into two runs and look and see how this averaging works. We’ll do the 1 1/8-ounce data:
![Text Line Plot Slope Parallel]()
Now we see how it works.
It’s pretty easy to understand it when you compare the individual and the averaged data. Why it works that way should be equally clear. The most deviant values, two in this case, the 67.1% and 79.2% patterns have been averaged with more typical results nearer the center of the distribution. The data appear (but really aren’t, of course) more consistent without those two inconvenient outliers.
But in 4/5’s of the cases, nothing much has changed. There are both averaged and singular results which are about the same. The extra two patterns have bought you a little, but not much; maybe only a little peace of mind. That “bad” pattern, way down there at 67.1%, might be shouting to warn you that there’s a problem, but if you knew more about patterns you would know that you get them now and then even when everything is fine.
What do patterns 2 and 3 get us?
About all those extra two patterns can do is inform you about patterns in general – how they vary, what they look like – rather than about the specific gun/shell combination you are testing. That’s an advantage, for a while at least; knowing more about patterns can’t hurt. But should you do your testing like this from now on? I don’t think so. Really, all it represents is organized “data-improvement,” a bit of intentional self-delusion that your gun “shoots very consistent patterns” when it almost certainly shoots about equally consistent patterns as every other gun at the club. You are still limited to qualitative (What are my patterns like?”) rather than quantitative judgments and as I will discuss later, there are easier ways to get qualitative results if that’s all you want.
Aren’t those 3-shot averages enough “better” to be useful?
But “Wait!” you are thinking. “Look at those data. Pretty smooth, pretty nice really; can’t I just use them to compare choke-brands or wads or any of that other stuff that interests me?”
I have to apologize that my way of presenting data, sorted and graphed, makes things look a lot more consistent, more organized, “better,” than you will ever run into at the pattern board or with the calculator. Looking at the data as they really did come in makes the problem more clear.
![Text Line Plot Diagram Slope]()
It’s still not good enough to use for comparisons.
Even if you had some choke-tube data which was reliably three or four percent different, it would be easy to make your decision exactly wrong by unknowingly comparing a high result from one with a low results of another. It happens all the time and you never know it because you don’t have the rest of the 10 patterns to tell you what’s really going on.
If three patterns aren’t enough, how many do I need?
The technical staff at the National Rifleman breaks the bad news in the April 1980 issue in the article “Patterning your Shotgun” by Robert N. Sears.
“It is much less appreciated how many shots must be fired to establish the long run average with acceptable precision.
The number of patterns required depends on how closely the average needs to be determined and with what level of confidence. The table below is based on a statistical evaluation of actual fired patterns.
![Text Font Line Number Parallel]()
The important thing shown is that fewer then 10 patterns is generally not adequate for valid results. If you are not disposed to fire and count at least this many patterns with each barrel and load to be tested, it is best to forget about a quantitative pattern evaluation.”
“A practical alternative is to use a steel patterning plate. (Description of how to use one.) This method of pattern evaluation if recommended for shotgunners not disposed or equipped to conduct tests which yield quantitative results.”
Is there an alternative that works?
I think that the plan offered in Section 3 of my POI and Patterning booklet
http://www.mn-trap.org/tech_corner/n_winston/nw_poi_pattern.html
is much better, but not everyone will have the motivation for even that, I suppose. Though it only gives qualitative results (as does Ron Jones’ 3-shot program,) it only needs 18-inch paper instead of 36 or 48 inches wide and you can take a lot of shots and take them home to study any time you want. Just remember that you have to follow the plan in every detail. A measured 13 yards, light factory 7 ½ premium factory shells, paper not cardboard. And take a lot of shots - you have spent the time to set it up, why not get a lot more out of it with just a little more work?
Conclusion.
I hope you have enjoyed is review of an article many of you probably read. Has this changed or reinforced your original impression? Are you going to try 3-shot pattern averaging? Tell us what you think.
Yours in Sport,
Neil
A week ago a member posted a question about how to go about patterning - distance, number of shots, etc., the usual reasonable things people who “want to get into patterning” want to find out. In the old days I would have typed it all out in detail once again, but these days I just can’t make myself do it. I’ve corresponded with the poster and know he’s a serious shooter with a lot of drive to get better. But what are the chances, I ask myself, that after I lay out the whole process he will actually do it? Probably no chance at all. After all, realistically speaking, no one really patterns with the accuracy and care and effort which seem to be needed to actually learn anything, though certainly many walk back to their cars with some shot-up newspaper pages under their arm thinking “OK, finally I know all about my gun’s pattern!”
The problem – the fact that useful patterning requires more work than almost anyone is willing to put out, more dedication than almost anyone has – is apparent in the recent “Patterning is a waste of time” thread. As you read it you can tell that almost no one posting there has ever done it and even fewer intend to. There are a number of imaginative ways to say it but “too much work” is the prevailing undercurrent with “the grapes are probably sour anyway” detectable in almost every post as well.
Is there a shortcut?
That’s why when Lon Lauber in Washington state emailed me, wondering what I thought of Ron Jones’ “Pattern Analysis for Busy Sportsmen” in the March, 2013, issue of Shotgun Sports, I perked up my ears. Ron Jones says that you can shortcut the demanding standard requirement of 10 patterns and just shoot 3 and “almost always produce a composite average within 2-3 percentage points of the value arrived at using all 10 patterns.” This is a welcome much-reduced challenge. Shooting just 3 patterns is a far more realistic goal than 10 and maybe there’s enough in the modest “effort budget” so many shooters seem to live by to get even a few to do it.
What does the article tell us?
I borrowed the issue from the Buffalo Gun Club and took a look. I was immediately put on my guard when I saw the front page featured the pattern from page 364 of Greener’s “The Gun and its Development,” an example sometimes cited as being “too good to be true” to the extent that John Brindle wondered “(Did) the draughtsman spread the pellets toward the periphery, to spread them more evenly?”
Jones uses as his standards for choke performance the Oberfell and Thompson charts from 1960 and so is basing his assessments on shells with soft shot and no plastic wads. He calls a full choke one that puts 70% of its charge in a 30-inch circle at 40 yards which may still be the standard but shells and guns do a lot better than that now, as a rule. I’d say 75%+ is more typical these days and the difference is important because for long-yardage handicap you need all the pellets you can get you shouldn’t think that 70% is acceptable “full choke” performance these days, it’s not. Almost everyone does better, many a lot better
He presents a table of choke difference, all 5% low but good enough, some tables showing the effect of various distances, and a clear and concise plan for getting your three patterns. And (pending verification) the 3-pattern average plan, the real point of the article, looks like a promising way to make patterning “accessible.” Within two or at most three percent of the “real” average? That’s good enough, isn’t it?
How much corroboration does Jones provide?
But why do I caution “pending verification?” Well, Jones never presents any evidence that just thee patterns do what he says they will. Maybe in the original manuscript he did, writing “2) the three-shot patterns analysis did a remarkable job of predicting the pattern percentage for any given load.” But I don’t see anything about that now. But we don’t need proof to at least try it, to see if the benefit Jones promises is really there: “It will give you a good idea just what your gun and loads are doing . . .”
What data will we use and why?
I’ve always liked the approach of the NRA technical staff to patterning statistics, in that they base their recommendations on the data they have collected over the years rather than relying solely on academic statistics which, though certainly more mathematically rigorous, may lack the insight into what’s likely to make a difference that a workaday practitioner might be able to offer.
We will use two examples, patterns produced by 1-ounce and 1 1/8-ounce shot charges shot from the same gun. And what we want to know is whether these loads both give us the 75% -or-better patterns we think are reasonable to expect. Said another way, we want to know if Jones’ test will answer our question “Are these loads good enough or should we do more work?”
How do we choose our examples?
We know the ten-shot averages for two loads, 1ounce and 1 1/8 ounce, and we will set up some example three-shot mini-tests. I’m sure 20 such examples will be enough; we will just use a random number table, scanning successively and creating triads. If we see a 7, a 4, and an 8, we will pick those three shot numbers, add them up, divide by 3, and have our 3-shot average. If we see 7,4,7, we will move one digit over for the last one pattern, since we wouldn’t, on the field, shoot the same shot twice. When the number given is 0, we will use shot 10. When we have used three digits in the random number table, we will move down the next three.
Here is the source of our pattern data and why choose them.
We’ve selected those data-sets because, though they are both “typical,” they vary in how “consistent” they are. The 1-ounce patterns only span a range of about 8 percentages points, the 1 1/8 cover about 12. Doing them both will tell us about the generality of Ron Jones’ program.
The results based on unusually consistent data.
Generating 20 “synthetic” 3-shot patterns gave these results for the 1-ounce shells.
The article’s assertion that 3-shot averages would fall within three percentage-points of the 10-shot average was supported in this test.
Will that hold true for the more variable 1 1/8 ounce shells?
The results from more typical, more variable data.
Again, Ron Jones’ prediction is supported. The 3-pattern averages are within three percentage points of the 10-shot average.
Does this mean we can start right away?
And save all that work we’ve been putting in on those apparently unnecessary additional seven patterns? We’d better go slow here and back out carefully. Before we go, shouldn’t we just make a quick check to see what we’ve done? What does the averaging of three patterns do that is not done by just looking at only one? Let's make sure that 3 shot give us better answers to the question we were asking. You remember the question: “What we want to know is whether these loads are good enough or should we do more work?”
Compare our raw results with the averaged ones.
We have twenty synthetic 3-shot tests but the underlying data comprises just ten, so let’s spilt the artificial ones into two runs and look and see how this averaging works. We’ll do the 1 1/8-ounce data:
Now we see how it works.
It’s pretty easy to understand it when you compare the individual and the averaged data. Why it works that way should be equally clear. The most deviant values, two in this case, the 67.1% and 79.2% patterns have been averaged with more typical results nearer the center of the distribution. The data appear (but really aren’t, of course) more consistent without those two inconvenient outliers.
But in 4/5’s of the cases, nothing much has changed. There are both averaged and singular results which are about the same. The extra two patterns have bought you a little, but not much; maybe only a little peace of mind. That “bad” pattern, way down there at 67.1%, might be shouting to warn you that there’s a problem, but if you knew more about patterns you would know that you get them now and then even when everything is fine.
What do patterns 2 and 3 get us?
About all those extra two patterns can do is inform you about patterns in general – how they vary, what they look like – rather than about the specific gun/shell combination you are testing. That’s an advantage, for a while at least; knowing more about patterns can’t hurt. But should you do your testing like this from now on? I don’t think so. Really, all it represents is organized “data-improvement,” a bit of intentional self-delusion that your gun “shoots very consistent patterns” when it almost certainly shoots about equally consistent patterns as every other gun at the club. You are still limited to qualitative (What are my patterns like?”) rather than quantitative judgments and as I will discuss later, there are easier ways to get qualitative results if that’s all you want.
Aren’t those 3-shot averages enough “better” to be useful?
But “Wait!” you are thinking. “Look at those data. Pretty smooth, pretty nice really; can’t I just use them to compare choke-brands or wads or any of that other stuff that interests me?”
I have to apologize that my way of presenting data, sorted and graphed, makes things look a lot more consistent, more organized, “better,” than you will ever run into at the pattern board or with the calculator. Looking at the data as they really did come in makes the problem more clear.
It’s still not good enough to use for comparisons.
Even if you had some choke-tube data which was reliably three or four percent different, it would be easy to make your decision exactly wrong by unknowingly comparing a high result from one with a low results of another. It happens all the time and you never know it because you don’t have the rest of the 10 patterns to tell you what’s really going on.
If three patterns aren’t enough, how many do I need?
The technical staff at the National Rifleman breaks the bad news in the April 1980 issue in the article “Patterning your Shotgun” by Robert N. Sears.
“It is much less appreciated how many shots must be fired to establish the long run average with acceptable precision.
The number of patterns required depends on how closely the average needs to be determined and with what level of confidence. The table below is based on a statistical evaluation of actual fired patterns.
The important thing shown is that fewer then 10 patterns is generally not adequate for valid results. If you are not disposed to fire and count at least this many patterns with each barrel and load to be tested, it is best to forget about a quantitative pattern evaluation.”
“A practical alternative is to use a steel patterning plate. (Description of how to use one.) This method of pattern evaluation if recommended for shotgunners not disposed or equipped to conduct tests which yield quantitative results.”
Is there an alternative that works?
I think that the plan offered in Section 3 of my POI and Patterning booklet
http://www.mn-trap.org/tech_corner/n_winston/nw_poi_pattern.html
is much better, but not everyone will have the motivation for even that, I suppose. Though it only gives qualitative results (as does Ron Jones’ 3-shot program,) it only needs 18-inch paper instead of 36 or 48 inches wide and you can take a lot of shots and take them home to study any time you want. Just remember that you have to follow the plan in every detail. A measured 13 yards, light factory 7 ½ premium factory shells, paper not cardboard. And take a lot of shots - you have spent the time to set it up, why not get a lot more out of it with just a little more work?
Conclusion.
I hope you have enjoyed is review of an article many of you probably read. Has this changed or reinforced your original impression? Are you going to try 3-shot pattern averaging? Tell us what you think.
Yours in Sport,
Neil
