A little about the bullet: It is a dual core (lead & polymer), swaged bullet with an open tip and flat base. I've got the core proportions figured out to give me a 27gr finished bullet (+- 1%) for all three versions (I make them from jacket nominal lengths of: .600", .705", and .740"). Here's a shot of the components, a couple finished bullets and one that I sawed in half.
As you can see, the plastic core fills the bottom two thirds or so allowing me to push the lead forward to improve stability (or at least maintain it given the different densities of the cores). Figuring out this center of gravity was the most recent breakthrough that allowed me to get the bullet I wanted. There were a hundred others, which is why it took forever.
My last trip to the range allowed me to capture the results, which I summarize below. For those unfamiliar with stats, I posted a shooting analysis primer on my blog here: http://www.dhxbullets.com/Shooting-Anal ... r_b_4.html (I know, shameless plug). Feel free to refer to it if this analysis doesn't make any sense. I've read you guys for a couple years now and this is a pretty smart forum so no sense dumbing it down.
Ok, so the test is between the DHX195 Reloads and FNH's SS195LF factory ammunition, with the goal to see whether the two lots had the same or different levels of precision. I loaded my bullets with 5.2gr of TB and seated them to an overall length of 1.585". Both lots were shot from my PS90 tri-rail with a 4x32 scope at an indoor 50' range. I used a Stoney Point steel bench rest modified to fit the PS90. Targets were 1 3/4" diameter, with 12 sequentially numbered targets to a sheet. My wife mixed up three round groups across two magazines, wrote down the sequences and kept them until I got home. This way I did not know which groups were mine and which were FN's while shooting. Here are what the two lots looked like after decomposing them with my wife's list and entering them into Minitab:
Aside from having to zero that scope about a 3/8" down & a 1/4" right, it's hard to see anything obvious here. Maybe I was a little more wild on a couple of the outliers in the upper right. The scale here is in 1/8" increments or about 2 3/16" tall by 1 9/16" wide (scale modified for best fit). But to make any sense of this it's got to be plotted by shot group. Here's what that looks like for DHX and for FN, respectively:
So now the question becomes which of those lots is better in terms of group precision? Well, the residuals (distance between each shot and its group average) tell the story. Here are what the two residual distributions look like:
Now we've got some actual stats to compare. My bullets had an average residual of .2585" while FN's came in slightly higher at .2663". That's only a difference of .0078", which is practically insignificant. FN's standard deviation was smaller, but again the difference was negligible (.0047"). Here are some of the other stats:
Descriptive Statistics: Residual (in)
Variable Lot N N* Mean SE Mean StDev Minimum Q1 Median Residual (in)
DHX 41 0 0.2585 0.0215 0.1377 0.0466 0.1801 0.2438
FNH 29 0 0.2663 0.0247 0.1330 0.0659 0.1627 0.2610
Variable Lot Q3 Maximum Residual (in)
DHX 0.3177 0.6156
FNH 0.3473 0.5867
I ran a few statistical tests to confirm that the means, medians and variances are all statistically equivalent and I'm proud to say that they are. Practically speaking that means I've finally made a bullet that shoots as good as FNs. I've got a couple other things I'm going to try to see if I can improve on this, but for now I'm pretty pleased.
I'd appreciate any feedback and am happy to reply to questions about the bullet or anything else. Also wanted to offer them to anyone who wants to try them out. Just look up DHXBullets.com to order or give me a call and I'll make you up a lot of a hundred. They are excellent bullets for the 5.7 cartridge. And, as I mentioned earlier, I keep a blog there that I post about new insights to the world of swaging, shooting and the associated stats.
Thanks for reading.
Dave
Test results:
Two-Sample T-Test and CI: Residual (in), Lot
Two-sample T for Residual (in)
Lot N Mean StDev SE Mean
DHX 41 0.259 0.138 0.022
FNH 29 0.266 0.133 0.025
Difference = mu (DHX) - mu (FNH)
Estimate for difference: -0.0078
95% CI for difference: (-0.0735, 0.0580)
T-Test of difference = 0 (vs not =): T-Value = -0.24 P-Value = 0.814 DF = 68
Both use Pooled StDev = 0.1358
Kruskal-Wallis Test: Residual (in) versus Lot
Kruskal-Wallis Test on Residual (in)
Lot N Median Ave Rank Z
DHX 41 0.2438 34.9 -0.32
FNH 29 0.2610 36.4 0.32
Overall 70 35.5
H = 0.10 DF = 1 P = 0.752
H = 0.10 DF = 1 P = 0.752 (adjusted for ties)
Test and CI for Two Variances: Residual (in) vs Lot
Method
Null hypothesis Variance(DHX) / Variance(FNH) = 1
Alternative hypothesis Variance(DHX) / Variance(FNH) not = 1
Significance level Alpha = 0.05
Statistics
Lot N StDev Variance
DHX 41 0.138 0.019
FNH 29 0.133 0.018
Ratio of standard deviations = 1.036
Ratio of variances = 1.072
95% Confidence Intervals
CI for
Distribution CI for StDev Variance
of Data Ratio Ratio
Normal (0.724, 1.450) (0.524, 2.104)
Continuous (0.653, 1.565) (0.427, 2.449)
Tests
Test
Method DF1 DF2 Statistic P-Value
F Test (normal) 40 28 1.07 0.859
Levene's Test (any continuous) 1 68 0.02 0.884
