Home | 

BSA Supersport .25 cal.: Part 1

by B.B. Pelletier

The BSA Supersport breakbarrel spring rifle is not an old model. It probably dates from the late 1990s, but the Blue Book of Airguns makes no mention of this model. Also, other references exclude it entirely, which leads me to think it’s a recent model.

The BSA Supersport is an attractive, lightweight breakbarrel.

There are two versions of this model. This one is the plain Supersport. The Supersport XL has the fancier stock, but otherwise is the same rifle. The gun I’m testing for you is in .25 caliber, so those who starve for a mention of the quarter-inch bore will find this blog interesting. The rifle I’m testing (Mac is doing the shooting and velocity testing while he’s here in Texas with me) is serial number AAR-840865-10.

Very light
You pick this rifle up and it surprises you with how light it feels. It’s barely over six and one-half pounds, yet it looks large and robust. So, it’s a Beeman R7-sized rifle with R9 power.

Cocking is a real bear! Now some of that is due to the newness of the rifle, and I know that this model wears in over time. I would estimate the cocking effort in the 40-lb. region right now. I’ll test it during the velocity report.

It’s a .25!
Of course, the biggest news is the caliber we’re testing for you. And what’s even better is that Mac has access to a broken-in .22 caliber Supersport that’s still stock, so I’ve asked him for a second report on that rifle when he gets back home. Now, you really will have something to compare to. I’ve owned Supersports in the past and can tell you that the design of the rifle hasn’t changed much over the years. It’s a bread-and-butter breakbarrel that has more than enough quality to make most owners proud.

The basic Supersport that we’re testing has an attractive beech stock with impressed checkering on the forearm and pistol grip that’s slick and does nothing to grab the hand. The wood is finished satin while the metal has a semi-shine and dark black. The wood is very figured, which is not common for beech, though I do note that the one on the Pyramyd website is also figured. The post-and-bead sights are not fiberoptic, thank goodness, and they’re as sharp as any 1950s Winchester .22. I really like them. And, the rear sight is fully adjustable. Both sights are plastic, but that’s the extent of non-wood and metal on the gun.

The cocking link is articulated, making the cocking slot in the stock shorter that if there was only a single link. The result is very little vibration, even though this is a powerful, lightweight spring air rifle.

This rifle fits me very well. The Monte Carlo comb rises up enough to put me right on the sights. The slightly raised cheekpiece is only found on the left side of the butt, but other than that and the placement of the manual safety on the right rear of the receiver, this rifle could be considered ambidextrous. I don’t think left-handers will have a problem with it.

I’ll be shocked if we see 700 f.p.s. in the velocity test. But in the past, Mac has seen as much as 625 in .25 and 675 in .22. That makes the rifle capable of all the Pyramyd Air claims. The interesting thing about this rifle is that it can generate so much power in .25. We’ll see.

63 thoughts on “BSA Supersport .25 cal.: Part 1”

  1. To all,

    Sorry if I come across as a whack job about statics. I have a character flaw that I’d rather be right than be happy. 😉

    The overlying message is simply that small sample statistics stink. I am trying to help, and I trying to tone down my frustration. I’ve generally played the expert from afar who has folks wanting his advice. I’ve also been the boss. So I find it frustrating try to be building basic creditability.

    As been noted before we can only measure the overall group size once it is on paper. To a first approximation it would be reasonable o assume that:

    [Equation 1] GS(Total)^2 = GS(rifle)^2 + GS(Shooter)^2 + GS(pellet)^2

    where GS is group size and the values in parenthesis indicate particular error sources. The point is that if:

    [Equation 2 ] GS(pellet)^2 >> GS(rifle)^2 + GS(Shooter)^2

    then GS(pellet)^2 controls the overall error. ( >> much much greater than…)

    The third interesting scenario would be if the shooter had a good rifle and good pellets. Then the overall error is mostly due to shooter problems.

    [Equation 3 ] GS(shooter)^2 >> GS(rifle)^2 + GS(pellet)^2

    So give me a Olympic quality gun and such quality pellets, then I can preform a shooting exposition so bad that the quality of the rifle and the quality of the pellet are insignificant.

    Obviously if you had a good gun and a good pellet, you could confirm that range conditions were not to bad. for a second gun, or a different kind of pellet.

    It is all about what do you want to test, what factors do you want to test for importance.


    I’d be curious as to what kind of drill or tests were used to evaluate your shooting. The better the shooter, the better the overall system. Also smaller error in pellets and other equipment become more visible when the error due to the shooter gets smaller.


    • Herb, I believe you had said some time ago you were a chemist? If true, with your love for statistics, why didn’t you ever come over to the dark side, my industry which is insurance and become an actuary? Just “funnin” with you, Herb.

      Fred PRoNJ

      • Edith,

        The statement is some people would rather be Right than Happy. Here is an illustration. Suppose you knew your boss has made several serious errors in his report to his boss. His errors offset each other so no harm will come in the end.

        But you have to be right, so you go to your boss and he tells you to “mind your own business” and submits the report to his boss as is. You then go to his boss and tell him all you know. He subsequently chews your boss out royally!

        Two months later your boss fires you on some trumpeted up charge!

        So wanting to be right becomes a character flaw when it gets to the point you just MUST be right even if the consequences are bad for you or others!

        It is never a character flaw to do the right thing because it is the right thing to do. But it becomes a flaw if your motive is simply to be right and prove some one else wrong over trifling things.

        • pcp4me,
          Another scenario is where you point out to your bosses that their new approach to management is undermining employees abilities to do their job, destroying moral, and creating a non-team environment, and they respond by ignoring you. I’ve learned that in this situation, continuing to point out what they are doing to the team will only get me in trouble. They’ve been told, and concrete examples have been provided. Furthermore, as is often the case, others are afraid to speak up to management themselves (as they do amongst each other). It’s a no-win situation.

    • Herb,

      There were two basic drills that we did; dry firing, and calling shots.

      Lots of people dry fire in their home with mirrors, to offset the smaller space that they have at home. Dry firing takes away a bit of the stress associated with shooting for score. This allows you to better see your wobble area, position errors, and follow-through. Because of this, you can better refine your position. There are typically 3 kinds of physical elements that contribute to your motion; lack of “natural point of aim”, balance, and subtle needed adjustments between parts of your body (including sling, rifle, and ground). You MUST find your natural point of aim!

      Calling your shots adds the element of shooting live rounds, but also takes away the negative influence of the spotting scope, as someone else has it. I was taught to ignore all previous shots, because “they were already down range, and so there was nothing that I could do about them” (sort like crying over spilled milk), AND to “never think about future shots, as they didn’t exist”. So, “the only thing that matters is the shot that you’re working on”. A consequence of this is that we were never allowed to look at our scores until AFTER a tournament. Our coaches looked at our scores and our targets while we were shooting. IF they questioned a score, then we were told to “challenge” the score, by putting up a buck and seeing the target “plugged” for the first time. One point to be made here is that, if shooting indoors (where there’s no wind), once you’re sighted in (using the sighter bull), there’s really no reason for rubber necking to your spotting scope. In fact, there’s no reason for using your spotting scope at all. Position matches always start with the prone position, so we pretty much know that we’re sighted in at the beginning.

      My best years were my last two years competing. During that time, there were lots of tournaments (league, state, and invitationals), so our tests consisted of these tournaments. At the end of the year, a set of tournaments were used to determine state ranking. Air-rifle state championships allowed me to test my offhand shooting.

      During that period, and especially just prior to that period, I did a lot of experimentation with different positions, especially with the kneeling and offhand positions. My prone shooting was pretty solid, except for needing to learn how to read wind better. The thing about this “experimentation” was that it took me away from some of the “basics” that I had been taught by my coaches, so they felt that I was kind of on my own, since they didn’t agree with some of what I was doing.

      What I really like about the book, “The Way of The Rifle”, which I only read a year ago, is the fact that they don’t tell you that there is “a best” way to shoot any particular position. Instead, they show you what many of the best shooters do. THIS is what I had to learn on my own, and in the end, the proof was in the pudding, as they say.

      These days, your question can now best be answered with SCATT, which didn’t exist 30+ years ago, when I competed. SCATT takes the equipment out of the picture, to a point, and leaves you with the shooters contribution to errors.


      • Victor,

        Thanks for sharing so much of your experience. As I said, I used shotguns as a kid. I tried out for the rifle team in high school and probably had 2 or 3 afternoons shooting .22 rifles. I didn’t make the team. (As a side note, it wasn’t till about 15 years later when taking a driver’s test for California that I found out my vision was 35/20 and I needed glasses!)

        I also started playing golf again a few years back. Some of the same basics apply. You have to develop a routine. You need to understand the basics about the “standard” swing, then you have to practice until you develop your own technique.

        The SCATT system looks interesting. Such an electronic system would of course work well with an electronic trigger on the actual rifle. Just need a jack to grab the signal.

        The other thing in common between golf and shooting is that electronic systems exist to provide much more feedback than the typical participant will ever get. The systems are just too expensive for the typical amateur.

        I really wonder how some of the drills you did could be adapted to shooting with a scope at 10 yards. I don’t need a spotting scope, I can generally see where at least the first few shots hit on the target. The one drill that has helped is to close my eyes, slowly could to 5, then check my scope position again. I’ve learned to develop a position where the sight picture stays relatively constant.

        Thanks again,

        • Herb,

          The late, great, multi-Olympic Gold Medalist and UK shooter, Malcolm Cooper, use to have someone turn off the lights before a shot. It was said that his natural point of aim was so good that he could clean the kneeling position this way. My pistol coach, Stan Hulstrom, also put a great deal of emphasis on perfecting your natural point of aim. Stan held numerous national and world records, but liked to tell us about an eastern European pistol shooter who could clean slow-fire targets, blind folded. His natural point was that good! This is essentially what you are doing when you close your eyes, open them, and check that you’re still on target. According to Stan, the secret was exercise (at least for pistol shooting). In rifle shooting, the secret is in applying just the right amount of muscle stretch. The non-stretched muscle has more motion, as does the hyper-stretched muscle. In any case, muscle tone is critical to great shooting.

          I rarely used a scope in competition, except for prone matches, which are split between “iron sights” and “any sights”. The interesting thing for me was that I shot my best scores with iron sights. Scopes are not used in international competition. The one benefit of using a scope in prone matches is that you can see the mirage, and things like weeds blowing at the target. Aside from that, a scope won’t make you more still so other benefits (if any) are purely psychological, I think.

          I’m a huge fan of aperture sights (front and rear), and think that springer’s should be offered with them, as an option to scopes. They won’t break, like many scopes do, and they are less of a distraction for target shooting, and accuracy tests. Case in point. – Not too long ago, we saw excellent test results that Mac provided, and that everyone thought could only come from using a scope.


  2. Now that’s a fine looking airgun. Everything you need and nothing extra. I especially like the machined dovetails in the BSA Supersport vs. the raised scope rail on the BSA Polaris. Glad Mac is testing this one. 40 lbs. of cocking effort isn’t a plinker in my book. I can only slap a barrel open for so long.

    Smashburger is also in Colorado. They’re very good but I have them hold their “secret sauce”. Wish we had a Fuzzy’s Taco Stand here too. Great soft taco’s, terrific habanero hot sauce and very cold beer. Hmmmm.


  3. BB: You have made my day with this blog on the BSA Supersport .25 cal.! I have serial #AAR -840866-10. I will be very interested in seeing your crony numbers. I haven’t yet put mine through the Pact, but I will say this though . The 24.4 gr JSB Exact Kings I have shot in mine will totally pancake at 25 yards on the paddles of my Devin Bialalon RF trap and they knock down the paddles with decent authority. I also researched this gun on the forum searches and found that many posters felt that the barrels of the supersports in .25 were very tight, and difficult to even load. They also felt that velocity suffered due to the undersize bores in BSA rifle in this caliber. Most of these comments were of the 2000-2005 or so vintage. On my gun I have, found just opposite, as the pellets I have fit quite loosely. I fact ,I would like to get some thin skirted , 20-22 grain pellets that were a bit bigger in dia to try. We may have the only two new supersports in .25 in the USA right now as Pyramd only re-listed them last week and I bought one the first day I saw it. The next day I see that they are now out of stock! In the book “Total Airguns by Pete Wadeson(ISBN 1 904057 38 1, 2005) he mentioned the supersports as a bread and butter springer in the UK, and what many folks there cut their airgunning teeth on . Thank-you for writing this one up and recommending it ,Robert.

    • Robert from Arcade,

      Very encouraging endorsement of the BSA Supersport.

      Seems that you’re describing the ideal pellet as a JSB RS Exact in .25 caliber. I heard that they’re making them but not imported yet.

      Maybe a 20.06 gr Beeman FTS or H&N Field Target Trophy in the meantime?


      • Kevin: I did order some H&N field target trophy’s with the supersport. I only shot about twenty of them so far ,so I have no real experience with them yet. They shoot lower by about 2” than the JSB’s as they are only 20 grs. The H&H’s also ,are a loose fit in the breech of the supersport. I only have these two types of pellets right now, and am just only about 2/3’s of a way through my first tin of 350 JSB’s. I plan on using up this first tin to break in the gun a bit, and then putting the gun over the crony. Then I will shoot up the rest of my pellets to break it in further. That will be 1100 pellets from now, and probably take some time as I’m into the busy season now. I will watch for those lighter JSB’s. I wonder if Pyramd would also consider stocking the BSA Plyarms pellet in .25.( Please ,Pyramd consider stocking some other springer wt pellets in .25! )I also am going to experiment with the alloy pellets in .25 ,as I think that of all of the common airgun cals, the non-lead pellets show the most promise IMO in .25. Regards ,Robert.

  4. BB:
    My brother used to base his purchase choices on what the ‘Processionals’ used.
    His car(a Nissan Bluebird saloon) was popular amongst mini cab drivers.
    His bike(a Kawasaki GT550) was popular amongst dispatch riders.
    And his air rifle(a .22 BSA Supersport) was used by London Underground(Subway)line workers to dispatch rats in the tunnels.
    A great rifle which he had to reluctantly sell when his twin boys were born and they moved house.

    • DaveUK,

      I was at the rifle range here in Texas today and what did I see but a British taxi. It was black, with suicide doors, a redone pearl gray interior and a 327 Chevy V8 with $50,000 worth of modifications to the drive train to make it all work. It only looked like a British taxi, but it was quick off the line.


    • Hmm……….he should have kept the rifle. It does not eat anything in storage and the day will come when he will want it. Hope the twins are well.

      “When you really need a gun, it’s often an airgun you want.”


  5. B.B., where are you? Did the airgun show come and go? Will we get a report?

    Herb, I don’t think anyone is questioning your knowledge of statistics. The questions have to do (and not in an unfriendly way) with applying statistics in a way that is relevant to plausible shooting scenarios. Nobody is going to shoot a 30 shot group much less ten 30 shot groups, so to say that all statistical results outside of this standard are a waste of time is…not really relevant to shooting. If small sample statistics stink, then all statistical discussion of normal shooting scenarios stinks. But is that really true? Consider the case of a three shot clover leaf group at 100 yards with an M1 Garand. Yes, from a laboratory standard that figure has uncertainties, but to say that it is a totally useless data point is not really right either from my perspective. For one thing, it is what we have. Secondly, given the roughly normal distribution of shots, I suspect that it is much likelier that this one sample is representative of the shooter and gun than that it is totally anomalous. With Mike being an experienced shooter with an accurized rifle and lacking any obvious called fliers, the spectrum of evidence of which statistics is just one element seems to me to be weighted more toward a useful result than otherwise.

    Victor, as a matter of fact, Brian strikes a chord with me in some respects relative to abstract mathematics. I can use the function notation well enough, so to read abstract discussions that reformulate this in terms of injective, surjective, and bijective functions with left and right inverses seemed baffling and pointless. But now that I have given it a chance, there is definitely something to this abstract terminology. Call it power and a way of seeing things from a fresh perspective. I believe that the physicist Richard Feynman said that progress in science was often based on finding a new terminology. So, the shooting situation is definitely onto (except where you miss the target…) but not one to one. Interesting about the USAMU seminar where they said that experienced shooters lose their fundamentals. Exactly my experience. I had supposed that with enough shooting this would become automatic, but maybe not so. I was watching a show about Jerry Miculek, the revolver champion, as a he was preparing for a contest, and he said his emphasis would be on keeping his feet moving (for action shooting) and staying hungry for each shot. Interesting advice from a grandmaster, but I’ve noticed that just when I get complacent about my technique is when it fails.

    Thanks to all for the information on the Mosin Nagant. Duskwight, good memory about the Mauser. So what is the story on the recoil of the Mosin? Some people have said it kicks hard, but others say it is slight. Perhaps the difference is between the short-barreled M44 carbine which has a big flash and maybe a big kick and the longer barreled versions. Here are some accuracy statistics. I understand that Russian Mosin sniper rifles were required to put 10 shots into 1.38 inches at 100 meters with other sizes graduated for distance up to 600 yards where 10 shots went into about 2 MOA. So, we’re not the first to use 10 shot groups. For 100 yards, when you convert to 5 shots and yards instead of meters, I come up with a figure that is very close to 1 MOA. Wow. I don’t believe that the Russians made use of match ammo, so this would be with issue ammo. As a comparison, I read of a test of the Springfield 1903A4 sniper rifle which shot a .58 inch group at 100 yards with match ammo but 1.5 inches with standard issue. Mosin triumphant! The big question for me was how they did that with what sounds like an atrocious trigger pull. One guy said that his rifle had a 10 pound pull and with some complicated shimming technique, he reduced it to 8 pounds. So, does anyone know if the Mosin trigger can be modified to be decent?

    My imagination has been filled with lovely Russian women snipers and their Mosin-Nagant rifles in environments of unimaginable harshness and savagery. Feminists would have a field day with what is purported to be the reasoning of the Russian leadership in choosing women snipers: better shots, more psychological stable than men, more physically resilient and resistant to cold (that’s good!), but also more patient and cunning than men. He he.


    • Matt61,
      It’s not at all uncommon for competitors, of any kind, to lose their fundamentals.

      I think what Jerry Miculek is describing when he says, “.. keeping his feet moving (for action shooting) and staying hungry for each shot”, is HIS way of finding and keeping his rhythm. Very interesting, indeed! But Jerry Miculek practices a great amount (which I would expect), and he routinely walks the course to re-enforce a solid mental picture of what he wants to do.

      A good example that combines both of these elements is what we see with baseball hitters who have their longest hitting streak, or who set personal best batting averages over the course of a month or two. Ron Cey, of the Los Angeles Dodgers, once had a period in which he was batting close to 400 for a couple of months. The timing of this small miracle got the Dodgers to the World Series. This was not a one day, or two week, fluke, but a stretch that lasted for over a month! What happened? Why was he never able to duplicate this kind of performance again? On the other hand, you had guys like Pete Rose, and Rod Carew who could consistently perform at a very high level in these categories. These two professionals new how to find their rhythm, and perform “on demand”, as Dave Kimes puts it. It’s all mental.

      The thing about mathematical rigor is that it helps where intuition fails.


      • It does indeed. And if you calculate the odds of a 56 game hitting streak over 20th century baseball, and the average batting average, it appears that one Joe DiMaggio was just about right.

    • I have owned and shot both the 91/30 and the M 44 Carbine. I would describe the recoil as “average” or about the same as other WWII rifles. Of course the muzzel blast is greater with the carbine. As to accuracy, I doubt that you can get 1 MOA with the sniper version and service ammo on a regular basis. The issue scope is low powered with a heavy reticle so that makes it even harder. While a good rifle, the standard version’s bolt handle is too far forward for easy manipulation and too short for good leverage. The bolt handle is longer and turned down on the sniper version which is an improvement.


  6. Matt,

    On Mosin rifle in general – it really kicks unless you hold it tight to your shoulder. When held properly nothing serious, just like any other .30 caliber bolt rifle shooting full-sized rifle ammo.
    On accuracy – during WWII it’s very unlikely that any sniper could obtain any other rounds than standard issue. However, there was a difference. There were 3 types of bullets – “L” (light) and “D” the latter having a slightly tapered tail and a bit heavier bullet with better ballistics – for use in machineguns. This one was preferred round for snipers. Closer to 40’s they were all substituted by cheap uniformed “LPS” (light, mild steel core) bullet. There were also match 7,62×54 rounds, made (and still made) for sport, and they had far superior grouping compared to mass-production round.
    In late 50’s a special sniper round was developed – 7N1. It had strict manufacturing tolerances, boat tail bullet with mild steel core moved a bit forward, to insure its right seating inside bullet’s jacket and deny its influence on a process of cutting grooves into bullet. It is marked “Снайперские” (“sniper’s”). http://weaponland.ru/images/stat_ammo/ammo_of_sniper-1.jpg
    Its improved penetration version is marked 7N14 with a hardened steel core. Unofficial name is “puncher” due to the fact that it hardly notices body armor.


    • Call me a schizophrenic for commenting on myself,but…
      Pull and type of pull are set by a file, a hammer and steady hands. It’s possible to make them somewhat 3.3 pounds.
      I could translate this part for you, but I say – I do not recommend it, unless you are qualified enough, and anyway it’s your own fear and risk, no claims in case you mess something up 🙂

  7. Small number statistics have their place, but they also need to be used with great caution when you are making decisions (such as which gun to buy). Let’s take your 3-shot Garand cloverleaf. Given all you know about Garands, is this typical of the breed, or much better (I assume the shooter is perfect)? Assume that it’s roughly typical. Then the cloverleaf tends to confirm that you’ve got a typical gun (some will recognize this as using Bayesian statistics in a rather cavalier manner). Assume that most Garands are much better. Then from the single datum you might be tempted to conclude to choose another gun. But I wouldn’t have much confidence that was correct.

    But let’s take a different case. Reach into the box and pull out a random Weihrauch gun. Fire a three shot group and measure c-t-c of 6mm at 10 meters, not entirely implausible. Reach into the box again and pull out a random RWS-Diana gun of comparable quality and reputation. Fire a three shot group, and measure a c-t-c of 5.5 mm at 10 meters. Which of these two is a more accurate gun?

    The answer is you have no idea, and the test did essentially nothing to help you make a choice. If you really want to know, you’re going to have to do a lot of shooting and measure the rms dispersion of the impacts from each gun. Oh, if you had to say something based on 3 shots, I suppose you would suck it up and go with the Diana, but I stress again that the test was wholly inconclusive.

    A minimum number of shots to draw any serious conclusions would be 30-50, and then you would only have maybe 90% confidence if both guns were comparable. And it could all be out the window if you changed pellets and reran the experiment.

    I think I would make essentially the same comments if somebody fired 3 or even 5 shot groups to choose the best among several pellet types — but I recognize from experience that some pellets are terrible performers in some guns, even if other guns love them. I’ve sworn off H&N Match; I can’t get consistent grouping from them using either my rifle or pistol. RWS Meisterkugeln do significantly better, and R-10 maybe a hair better. No, really better. The number of loose-fitting R-10s in a tin is much smaller than in the MeisterK tins. So the R-10 probably gives more uniform performance. But I stress I don’t have test targets to prove it, while I did have targets to tell about RWS vs H&N.


    • Pete, it’s not typical for the breed unless it is a tuned match rifle and a good one at that. The one I have just happens to shoot really well with the right load. The average is more like 1.5 inches at 100 yds but on some days I can do much better. It has been said that a rifle that shoots is priceless!


      • So, I’ve learned something about the Garand. It is not as accurate as I thought, but it’s still pretty good.

        Now we try another thought experiment. You reach into a black box full of Garands, maybe 25 or 100 in the box. Pull out one rifle. Shoot three rounds, and find out they group within a circle 2.0″ in diameter. And let’s agree you know only one thing about Garands in general: a typical rifle in decent shape (no obvious dings to the barrel or crown, etc) will shoot within a 1.5″ circle. Now the question:

        Is this rifle from the black box better than typical, typical, or worse? I argue that from only 3 shots you cannot tell a thing. Oh, if asked to guess you might say a tad worse, not much. But you could fire off another 27 rounds to make a 30 round pattern and find that one of your first 3 was an outlier, and the real group was < 1.0". Or you might find that the 3 together was a fluke and that your group was 3" across. Both with about equal probability.


    • pete,
      5.5mm vs 6mm? You didn’t mention anything about 10m competition in your scenario, so I would say that your statistics will prove both rifles to be excellent shooters and near impossible to improve significantly in our non-professional hands to be useful in the coming Great Accuracy Test. Your calculations negated the “or” condition proving both rifle are worth owning. I would conclude that I would choose the least expensive one and be proud of it. Now, go do your scenario with a Ruger Air Hawk and a Gamo CFR shooting 1.2″ – 2″ groups and you’ll see there is room for us to work with and improvements to be made (hopefully) and your statistics will then highlight the wiggle room.

      • You missed the point, Chuck, but I wasn’t entirely clear. The numbers are wholly imaginary. Make it 0.5″ and 0.6″ and it’s the same thing. What I am trying to make clear is that three shots from a gun that is in reasonable shape prove absolutely nothing useful if what you want to know is “which is better?” My point has absolutely nothing, nichts, nada and zilch to do with whether a group can be improved by your being a better shooter.

        The point is that 3 shots (or 5) tells you not very much about the inherent dispersion of the gun, or the pellets, or the gun-pellet system.

        Also, I wasn’t saying anything about the great accuracy test, whatever that finally turns out to be.

        • If I forced you to make a decision between the two rifles I used in my thought experiment, you might well be best off just buying the cheaper one. But you might want to invest a buck in lead and shoot 7 more rounds through each to get a little more statistically valid feeling for the likely true dispersion. I would.

          I said nothing about whether or not both were “worth owning.” Based on the givens, of course they both are. I might buy the prettier or the one that fitted my shoulder better.

        • Ok, I’ll accept that I missed the point. But all this talk about accuracy, to me, is being generated because of the Great Accuracy Test (GAT) so I assumed that ‘s where all this discussion is headed. Also, I got off track because the GAT is not about comparing rifles but about comparing the results of implementing changes to the one rifle we, each individually or collectively, choose to use.

          I still think three shots would be enough for me to tell if I wanted to consider a rifle or not. After all, if three shots were 5″ apart that’d tell you something as much as if three shots were touching. However, if I couldn’t be the one shooting them, the three 5″ shots would have to be by someone I trust is an acceptable shooter, by my standards. I would trust anyone’s 5mm group. I mean, how could you cheat to do that?

            • Maybe you could use a very bad 5-shot group to decide to reject a gun, particularly if the gun were misbehaving all day long. I wouldn’t be comfortable buying on that kind of test. But Edith is right too. If the weapon has a good reputation (the Germans use Waffe for air guns), and you know folks who shoot it (or its brand-mates) with results that leave your mouth watering, one bad group shouldn’t lead you to reject the choice.


              • pete,
                You said “shouldn’t” and I agree with you when the word shouldn’t is used, but I contend, in a real situation, you would reject it. If there were two “Waffes” in the rack each with a picture of its 5 shot group under it and you only had enough money for one….goes without saying what you’d do.

            • Exactly! But my position is there are more air guns than fish in the sea. I can afford to be choosey and willing to miss a good gun or two. Eventually, it’ll come around again and try to bite me and maybe give a better showing. But I’d just as soon move on the the next gun. On the other side of the coin, I still think you can’t fake a 1/4″ group at 30yds or be that lucky that one time so that would be the one to get my attention.

          • Re: “the GAT is not about comparing rifles”

            It certainly could be….

            Let’s have 100 shooters each shoot one 10 shot group with their 0.177 RWS 34 and Hobby pellets. Find the average, calculate a 95% confidence interval based on Taylor and Grubb’s charts and throw out any groups outside the confidence interval as being abnormal. Recalculate the average. Now this performance becomes the baseline.

            Geeze – JT’s group size is only 10% of the average and was discarded as “too good”. Was JT lucky, or is he that good? Have JT shoot 3 more 10 shot groups. Does his performance degrade towards average or is he still doing better? (Here of course the tables by Taylor and Grubbs and don’t give us a good way to calculate the expectation for the average of three 10 shot groups. So we’re on shaky ground as to how much better than average JT would have to be to be considered the superman of our little derby. We could shift gears and note that it is 50/50 to shoot better than average. So it JT shoots 4 groups, first plus next three, better than average then he would do this only 1 time in 16 simply by chance. )

            Ok, now we split a tin of pellets between JT and Larry who was an average shooter. Each shoots three 10 shot groups and reports. (Does JT just have a great tin of hobby pellets stashed from 2004?)

            Ok, JT lives near Sam. Sam’s results were discarded because he had a much larger than expected group size. We have JT and Sam meet and swap rifles for three 10 shot groups. (Does JT have a great gun? Does Sam have a poor gun?)

            I’m just trying to make the point that by design each of the subsequent experiments is trying to test a different hypothesis. Given the hypothesis that we want to test, then we design the experiment to get the data that we need to test that hypothesis.

            This isn’t magic. You simply drive a stake in the ground somewhere and then measure relative to that stake.

            • I think in this example you’re testing the skills of the shooters and not the mechanics of the guns. What I want to see done is not test JT against Sam but let JT shoot with whatever he has in his hand, then give him another pellet brand and do it again, THEN use your statistics to see if that made a difference and how much. Sam would then be doing the same thing with his gun.

              The result would be, “See how changing pellets can affect the accuracy of a gun?” Now, one gun could have gone worse and the other gone better but you proved that changing pellets, indeed, changed the results.

              Now, if later you wanted to have them swap guns and do it over hoping to identify the BEST gun/pellet, that’s another blog, perhaps a sequel to the GAT.

    • Pete,

      I totally agree with the points you are trying to make.

      I’d make one other point. In order to use statistics correctly you design the experiment, then you test to collect the data to analyze. It is poor form to just take test data, then decide how to use statistics.

      In designing a good experiment you have to be sure that the design takes into account all the factors that you already know or suspect are important. You also try to allow somewhat for the surprise element, but you can’t really design a good experiment to elucidate a surprise. A surprise generally leads to a different experiment and more testing.

      The necessity to actually <> the great [choke] accuracy test is the larger point I’ve been trying to make.


      It is hard to stipulate what one 3 shot cloverleaf patten means with writing 20 paragraphs of stipulations. Suffice it to say that with my shaky hands a 3 shot cloverleaf would be so much better than my expectations that I would attribute such a result to pure luck.

      Let’s take this in a somewhat different direction and talk golf for a moment. Using nonparametric statistics a shot is either it is a hole in one or it isn’t. So Tiger Woods and I go to a par 3 and we each hit 100 balls. In reality it would be highly unlikely that either of us would make a hole in one. Does this “prove” that I am as good a golfer as he is? How would we design a reasonable experiment to tell who is the better golfer?

      If we just counted holes in one, Tiger and I would have to hit balls for a week to get enough data to establish who is the better golfer. Hitting 10 shots and measuring the average distance to the pin would provide the same conclusion in 10 minutes.

      So I have not said that since all shooting uses relatively small sample statistics that we should just throw up our hands and flip a coin. I’m also certainly not advocating that massive amounts of data are an absolutely necessity. It isn’t the absolute amount of data, it is all about the quality of the data which depends on the design of the experiment.

      To get back to shooting, think of the problem this way. I have a great shooter and a really lousy one. I have a great rifle and a really lousy one. I have a tin of great pellets and a tin of really lousy pellets. How do I identify the great shooter, the great rifle and the great tin of pellets? One three shot group isn’t likely to do the job.


      • Grrrr….

        The necessity to actually DESIGN the great [choke] accuracy test is the larger point I’ve been trying to make.

        Had deign in double “less than” and “greater than” signs. So much for being cute…

      • Herb,
        You may be contributing your clover leaf to luck but at least you have demonstrated that the gun can indeed shoot a clover leaf if treated right. That is the gun to own. Making it up to you to take the necessary measures to increase your luck. As far as the great shooter getting a bad gun and/or bad pellets, well, that’s resolved buy the shooters swapping equipment.

        Ok, I think the point you’re trying to make is one can’t tell from looking at a picture of a 5″ three shot group if the gun is really bad. Agreed! But one CAN tell by looking at a picture of .08 three shot group that the gun is good. As I said in my previous comment, I don’t believe that a shooter can fake that (not with witnesses around) or be that lucky.

        As for your golf analogy…forget it…bad example! Golf is done by magic. No normal human being can do what those pros do. Someone needs to have them investigated to see if they are from this planet.

        • Chuck,

          RE: “You may be contributing your clover leaf to luck but at least you have demonstrated that the gun can indeed shoot a clover leaf if treated right.”

          I disagree. As I said my hands are so shaky that a cloverleaf group with an M1 would be pure luck. I would have absolutely no chance of repeating such a group no matter how many times I tried. A single group, and particularly the best group, is not a good statistical estimator. In order to get a decent perspective on performance you would need to take an average over some number of groups.

          Please look at figure four in the following reference. It has a nice histogram of the distribution of measurements that would be expected from one 3 shot group when the known true group size averages 1 inch. It is like you shot a million 3-shot groups and plotted a histogram of the individual measurements for each group. The variation between individual groups is due purely to chance.

          Also note that the distribution is skewed. You can’t get a group size less than 0, but the group size can be “infinitely” large.

          Group Therapy
          © 2007 Denton Bramwell

          Now let’s assume that we take the average of five 3-shot groups and plot that number. The distribution of the average (of the 5 3-shot groups) will be more symmetrical and narrower than the plot of individual 3-shot groups. The average of twenty 3-shot groups would be almost Gaussian, and narrower yet. So the more 3-shot groups that we measure, the better our estimate of the performance of the rifle.

          Now in order to make this analysis I have used “hidden” information. I just “know” that I have absolutely no reason to expect that my performance as a shooter is anywhere near good enough to get a cloverleaf at 100 yards with an M1. There is no statistical data to support this assumption. So in order to gather such evidence, I shoot ten 3-shot groups. So my (Herb’s) average group size for the ten groups is now 6 inches. Now we have a baseline of sorts.

          The problem is that the baseline also includes errors due to the ammo, and errors due to the gun.

          Let’s say that my imaginary friend Harry is with me at the range. Harry takes the same rifle and the same lot of ammo. He shoots ten 3-shot groups and gets an average group size of 0.75 inches. We don’t know how much of the error is due to Harry, how much to the ammo and how much to rifle. But we know:

          [Equation 1] GS(Total)^2 = GS(rifle)^2 + GS(Shooter)^2 + GS(pellet)^2

          For Herb:
          6^2 = GS(rifle)^2 + GS(Herb)^2 + GS(ammo)^2

          For Harry:
          0.75^2 = GS(rifle)^2 + GS(Harry)^2 + GS(ammo)^2

          Let subtract the equation for Harry’s performance from the equation for Herb’s performance. we get:

          6^2 – 0.75^2 = GS(Herb)^2 – GS(Harry)^2

          or GS(Herb)^2 = GS(Harry)^2 + 5.95^2

          Now at worse GS(Harry)^2 is 0.75^2, and at best GS(Harry)^2 is 0. So the best estimate of Herb’s performance is:

          5.90^2 <= GS(Herb)^2 <= 5.95^2

          This isn't a very sophisticated analysis, but I think it makes the point. Herb's poor groups are due to Herb, and the rifle+ammo errors contribute very little error to Herb's overall group sizes. So If Herb gets a "better" M1, or get better ammo, his performance won't improve much.

          • Herb,
            Thanks for the help. I wish I had the time and resources to do an analysis like you describe when shopping for an air rifle. Instead, I have to go with the gut feel I get from BB’s and Mack’s tests plus the rest of the accuracy advice offered on this blog. Maybe when we get into the GAT we will find a way to feed a statistical analysis. I did read the link you sent. Interesting. I’m too skeptical to trust a computer simulation. I know many scientific, business, and government organizations use them but I’ve worked with computers, programming and programmers for too many years.

        • Chuck,

          I truly do understand your point. If I grabbed one rifle that shot 5 inch groups, and you grabbed a different rifle and shot one 0.08 inch group, then I’d tend to believe that you were a good shooter, had a good rifle, and that you had good ammo for the gun. However that is a “gut feel” not statistical evidence. The whole point of a proper statistical test is to AVOID the bias created by our gut and look at the data alone.

          To understand the point, think of a lone measurement of 40 something another. Is the process on average better or worse that 50 something another? Think of mean and standard deviation. The 40 gives us an estimator for a mean but we don’t have any degrees of freedom left which we can use to measure the spread of the 40 measurement. Now if we have at least two measurements we can estimate the mean and standard deviation. If we have more than two measurements, then the system is overdetermined and we could start to test for outliers. The more measurements the better the statistical estimates for the mean and standard deviation.

          If we assumed that the measurement were Gaussian, then a sample size of about 30 gives us a pretty good estimate of the mean and the standard deviation. Good in the sense that we could also have a good test to throw out a few outliers. From a different perspective let’s say that we have a million measurements. We establish a 95% confidence interval and throw out the 2.5% high measurements and 2.5% low measurements. If we recalculate the mean and standard deviation we will get results very close to the original ones with all the data.

          Note that there is something else here. If we just measure group size then we just have one measurement. But we have three shots. So we take the average X-Y position of each shot and calculate an average X-Y position for the POI. Now we calculate the RMS distance from the average POI to each shot. We get three measurements. We lose one degree of freedom because we calculated the average POI. We lose another degree of freedom to calculate the average distance to the average POI. But we still have one degree of freedom left to estimate the variability of the distance to the average POI. So we can calculate the average distance to the POI and the standard deviation for that measurement.

          So by using a poor test, the group size, we got less information out of the three shots than was there. This is why your gut is telling you one thing when a single measurement of group size is less meaningful. It is all about how you setup the tests…

          Hope this helps,

Leave a Comment