(Technique given in bold half way down)

For those in the Garand Collectors Association (which I highly recommend), there is an excellent article in this quarters magazine on group size measurement. The classical approach that almost everyone uses is to pull out a ruler and measure the distance between the two farthest holes. They described a better method - a much better method.

Years ago, I've seen a better method that some professional level folks on here do, using what appears to be professional software packages. They plot each hole on a computer target, and then the computer performs a statistically analysis on it. This is more meaningful, but who is going to have this software? Sounds complicated, time consuming, and expensive. You could do it by hand, by mapping it out on a grid, performing a moment analysis to find the center point, and then measure mean distance from this theoretical center point, but again - wow, who's going to do that. That's a lot of work. I thought about it pretty hard actually, but it's tough to pull off in the field. It's much much better statistically though, as this method should give consistent data regardless of round count - meaning a 10 shot group should have the same mean radius as a 3 round group. And it dampens out the effects of only 1 or 2 outliers. Point being, it'll help you identify good loads better than the common extreme spread method.

So for years, I've been doing it the old way of measuring extreme spread - a flawed technique that does not normalize for larger group sizes, and is completely screwed by one or two outliers in the group. As an engineer, this has always bothered me (what can I say, engineers are cursed), but rather than fix it, I just accepted it (hence the "lazy" in the name).

Anyway, back to the article, Hallelujah, they explained an easy method to achieve the goal. I'm actually kind of embarrassed I didn't think of this myself. The technique is thus:

Take your target with let's say a 10 shot group on it. Draw a horizontal line on it such that half the rounds are above it, and half below it. Use best guess, but placement isn't critical. Then do the same thing with a vertical line. Where those two lines bisect, that's your center point. From that point, measure distance to all 10 rounds fired. Average is your mean radius.

Wallah, that simple. If your lines are a little off, it won't have a significant impact on the mean radius. I.e. if you drew the line a little higher than the true mathematical moment analysis requires, it doesn't really matter, because the upper ones will be closer, and the lower ones will be further, and the difference will wash out in the average. So long as it's even ballpark, the mean radius will be correct. Seriously, how did I miss this??

My biggest delima now is that all my performance data for the last 20 years is based on the simple (and flawed) extreme spread measurement method. So new measurements are going to be harder to compare. The solution, obviously, is to measure both!

This is what being an engineer is- on a Saturday night, this is what gets me excited.

Google on target software. It is free, and calculates all that stuff. You need a scanner or a digital camera. I like it a lot. Though I suppose for those that don't use computers your method would be easier. Then again they would not be on this site.

Tag for later.

+1 for On Target

It's free and it works.

That makes sense!

Thank you for the reply.

Sample size is indeed crucial. Multiple, 10 shot groups give you increasing confidence levels for the (potentially) open-ended population of a particular load. It also requires extremely well thought-out rifle positioning for repeatability. If you do everything possible to eliminate the shooter's impacts on the accuracy of each shot, you can virtually eliminate those as factors. Ideally, you would have the rifle set up on a rest as solidly as possible, with the stock also solidly supported and placed in exactly the same spot for each shot. You could also include a cable release for the trigger if you don't think you can keep your trigger technique consistent.

You cannot simply use a test barrel in a universal receiver to determine how a load behaves in YOUR rifle, though. Barrels are unique, and the dynamics of one are going to differ, even if only slightly, from one to the next. Otherwise, I'd have a big old chunk of I beam with a barreled receiver fastened to it for accuracy testing and never have to worry about any of the positioning and trigger technique stuff.

(wups, wrong spot)