To expand on what is mentioned in the article; mathematics (statistics to be exact) tells us that a 3-shot group will only give us 87.5% confidence and a 5-shot group will give us 96.9%. A 7-shot group on the other hand will provide 99.2% confidence; so obviously a larger sample size is better. Of course, this is all on paper (as in drafting paper - not target paper ;)). However, as you know, shooting a full mag for grouping opens us up to additional factors that we may not even see at 7 shots (as heat enters the picture where it might not have played a part during the smaller sample groups) - and these are results that would be of even greater interest to those of us who tend to shoot quite a few rounds at a time (for a sniper group size can often be irrelevant as it is the accuracy of the CBS - ie the location of the 1-shot group - that is of the greatest interest).
Also, in the world of shooting and ballistics we can often fool ourselves by testing the wrong things (or rather by performing the wrong tests). An example would be to shoot a .338 Lapua for groups at 50 or 100 yards - where it often will be outperformed by less whiz-bangy cartridges - when we really should be testing at 600 meters or more.
Not only within different cartridges. Also with different barrel length for same cartridge. Friend was testing Sako TRG-22 (660mm barrel) and Tikka T3 Tactical (506mm barrel) out of special testing fixture with all gizmos around (Doppler radar, trajectory analyser and what not). With same sniper grade ammo (testing was for military) showed that T3 is more accurate than TRG-22 up to 300m, then TRG-22 start to shoot tighter than T3 over this distance.
BTW I know of “serious” gun writer who use to shoot 3 rounds group (“becouse more is good only for Internet, real sniper or hunter does not need more”) until he finds one that supports his point. This guy has always sub-MOA groups from crappy rifles, with Chinese scope (with his own designed reticle!) and Russian steel cased .223 55gr Barnaul.
How about firing 10 shots, saving the target and shooting the same target 2-3 times a month for 3 months. I personally value consistency over the ability to shoot that one “1/4 inch group” out of 10 groups on a mild windless day in May. If that 30 round group RSilvers shot was from one of the AAC uppers that is not bad at all. If you could do that 10 times in a row on 10 different outings that is great! For me the other part of the equation is the optic/mount, toss your rifle in the truck and keep bringing it back to the range and shooting that same target, take it hunting, carry it and keep bringing it back to shoot the same target, I like to be confident in the equipments consitency. I am just a hunter, but I like stuff to work, those things that don’t go down the road.
Is it a correct interpretation of that group that the sight needs to be adjusted about 1-1.5" to the right?
If I understand correctly, the total 30 rounds represents 100% of where the shots can be expected to land, and the central 15-20 of them is where 75-80% of them will land and thus where the zero should be adjusted to fit. Thus it’s futile to chase one’s scope around the paper based on 3 or even 5 shot groups because inevitably the 4th or 6th shot will make you think you’re wrong and try to adjust according to it.
Excellent work, Mr. Silvers, and I hope others will follow suit and do more of the same. This is much, much more useful than anything I’ve seen before from others and definitely gives me a lot of perspective on my goals with my rifles. I’m going to apply this to my practice and evaluations from now on.
The basic problem in discussing shot groups is that it requires a somewhat peculiar mathematical language if a person is to adequately express the facts. The simplest way to describe the phenomena, as you are likely aware, would be to count the number of shots that land at bucketed distances from the “X”. All other factors being kept equal, those numbers would describe a normal distribution. Since a rifle target has both height and width, a more complete view would be a 3D plot where the vertical dimension represents the number of shots in that location. Picture a haystack. The more pointed the haystack, the more accurate the rifle and its load; the flatter the haystack, the more random the spread of shot. While 3 or 5 shots thru the same hole certainly indicates a propensity for accuracy, it does not tell the entire tale. These two comparisons of 5.56 ammunition show 30-shot overlays that can be examined more completely than the usual target picture:http://www.box.com/shared/static/hgdpmkhkw0.jpg , http://www.box.com/shared/static/61n5als9sm.jpg
Is statistically calculated mean radius (as on presented pictures) good way to present this difference? Of course to have it mean something, statistically significant probe must be used. Like 30 shots for example.
I do actually use mean radius, as OnTarget can calculate it nicely for me :agree:
Yes - that is what I do for internal testing - mean radius (good and easier to understand) or radial standard deviation (best but slightly more abstract).
My dream is where no one can make an accuracy claim without something like how the EPA does fuel economy. Govt tests it, and it would be average mean radius - 30 shot group at 100 yards, 10 guns tested.
Something I think about when referring to group size is, well, the group size.
With each shot, your point of aim on the target gets more and more distorted.
After three shots, you no longer have a crisp 1 inch dot to focus on. So you may be aiming dead center of a 1" dot for the first couple rounds, but then you’ll be aiming dead center of half of that dot, or maybe just the ragged hole you have left.
I make up for this by adjusting the elevation on my optic so my rounds impact 2" low. This way, even after ten shots on paper, I still have a crisp spot on the target to aim at.
Robert and Todd–Thanks for speaking the TRUTH! 30 rd group is a good start, as is averaging 5 x 10 rd groups. An adequate sample size is the only way to begin to see what the true dispersion pattern will be. Using 3 rd and even 5 rd groups is bogus, as are any “guarantees” of rifle accuracy using such “groups”.
