Test between old and new bottles of HP-38

Nick_C_S

New member
Hello All,

Over the last few months I ran a couple performance tests between old and new propellants. I ran tests between old bottles and new bottles AND new lot numbers of both Power Pistol and Unique. I said I would run a similar test between old and new bottles of HP-38; but of the same lot.

For some background: Over the years, I have passively observed that newly opened bottles of propellant tend to weigh more (higher density) than their old counterparts. As I was only making casual observations of this, I decided to finally run some actual tests.

I was almost out of a 1# bottle of HP-38, and the next one in line happened to be of the same lot. I bought these (4#'s total) on line at Cabela's on January 13th 2014. The lot number implies they were manufactured on 12/12/13. Yes, this was quite a find, considering it was during the peak of the shortage.

The old bottle was opened on 6/18/16; so it had been opened - and used many times - for 26 months - plenty of time to "air out."

The load used was a 200gn LSWC for 45 ACP; at 5.0 grains charge weight. This is my oldest continuously used load recipe. First minted in October of 1984 and I have loaded many thousands since then. The load recipe is so popular with me that I have a dedicated seat die for it - it loads nothing else. Primers are CCI 300's. Brass is once-fired (by me) WWB. COAL=1.240".
Test gun is a full-size Kimber 1911. 18-1/2Lb Wilson Combat recoil spring - if anybody needs to know that. 20 round samples, both old and new.

I set my Uniflow to throw at exactly 5.0 grains with the old bottle of propellant. I loaded 88 rounds - and weighing every charge as the hopper ran low. When it started throwing wonky, I stopped and dispensed the remainder into my scale pan and tossed it onto the lawn (probably 30 grains or so).

I then poured in the new bottle of HP-38; did a couple test throws, then weighted the next. Wadda know, it weight in at exactly 5.0 grains. That's right, there was no difference in density between old and new bottles. This really surprised me; as my previous observations left me certain that the outcome would be different. I double, triple, and quadruple checked. The scale doesn't lie. (My 10.10 scale was zeroed with a 5 grain check weight, and not touched between old and new bottle weight tests, btw.).

In spite of my surprise, and like any good scientist, I followed through with the test. So off to the range I go - with chronograph in tow. :p

Results: (20 round samples)

Old bottle: 860.9 f/s; 10.13 SD
New bottle: 846.5 f/s; 8.53 SD

Is the difference in velocity significant? I don't know. Could somebody with a firm knowledge of statistics please tell us? I believe it is - with my meager knowledge of the subject. I however, don't believe the difference in Standard Deviation is significant. But the velocity, yes.

Same lot, same weight, same volume - different velocities. Interesting. I don't have an explanation. But those are my results.

And I chronographed 10 rounds of the old; then 10 of the new; repeat. Further, I "pre-warmed" the gun with 14 rounds (two mags) before chronographing.

I don't care for the expression: But it is what it is.

Opinions welcome.
-Nick.
 
I've been following your powder tests and want to thank you for this effort, it is interesting info! Good work!

My recall of statistics is deficient, but I would think that there is enough overlap of the data so that you couldn't call the difference significant. I'll defer to those smarter in statistics than me, but I would guess that a larger sample size is required to detect any true trends.
 
Good test! I would conclude that there is no significant difference.

I did similar tests with HS-6 in 9MM 115 grain FMJ loads a few years ago.

The difference in age of the powders was 12 years

Like you, my scale was set with Lyman check weights.

Difference in average velocity between old and new powder was less than 50 FPS for three 25 round test batches. 1375FPS average for new powder. 1340 FPS average for old powder.

You've proven something to yourself that lots of folks refuse to believe.:)
 
Just a layman here, I took the minimum of probs and stats classes and that was during the dark ages of the slipstick. That considered if I could get single digit SD's with a 20 round test I would be ecstatic

I have no idea how much ES's and SD's affect pistol ballistics but to me those are some damn fine numbers
 
Standard Deviation

I have no idea how much ES's and SD's affect pistol ballistics but to me those are some damn fine numbers

In the pistol world, with pure target ammo (like these), Standard Deviations are generally in the low-10's, or high single digits. This is no place for mixed brass, of course. I've got some 38 wadcutters that are in the 5's. Hot ammo generally opens up. As does lighter bullets; and or slower propellants.
 
Do you have the ES on each of the 20rd test . Asking because one really bad ( either high or low ) velocity and it can throw off the avg but the SD will still be OK
 
In the pistol world, with pure target ammo (like these), Standard Deviations are generally in the low-10's, or high single digits. This is no place for mixed brass, of course. I've got some 38 wadcutters that are in the 5's. Hot ammo generally opens up. As does lighter bullets; and or slower propellants.

Thanks I'll keep that in mind. My club is planning some bullseye matches and I may try my hand at it
 
The SD on both sets of data are good, so your loading techniques are solid and you're obviously doing something right. However, your question was whether the difference in average velocity is significant. The exact percentage of difference depends on which you take as the baseline, but the difference as a percentage is either 1.7% (using the new powder data as the baseline) or 1.7% (using the old powder as the baseline). Neither is exactly 1.7%, but they both round to 1.7.

That's not a huge difference, but it is a difference. Given that each sample set consisted of 20 shots, I think it is "statistically significant." What it leads me to wonder is what the numbers were in each data set, and how much (if at all) the high numbers for the new powder overlapped the low numbers for the old powder. If the two data ranges have significant overlap, then perhaps we could conclude that there is no statistical significance to the variation.
 
What’s important to note is that the new powder is slower for the same volume. The average of 846.5 fps is outside the range of your first data set...the low side being 850.77 fps. In application, it means you need to work up a new load if trying to achieve 860.9 fps with HP-38 and your 200gr LSWC bullet. For plinking, no big deal. I don’t have a chrono or shoot competition so it wouldn’t be a big deal to me.

The other interesting thing is that in Ben Dover’s HS-6 trial, the new powder was faster, while your new powder was slower. We don’t know if the lots were the same between his powders.

Great study...very scientific! Now I’m going to have to get a chronograph to start my own research studies.
 
markr6754 said:
The average of 846.5 fps is outside the range of your first data set...the low side being 850.77 fps.
How did you conclude that 846.5 fps is outside the data range for the old powder? Did I miss something? All he gave us was the average of each set, and the standard deviation.
 
The other interesting thing is that in Ben Dover’s HS-6 trial, the new powder was faster, while your new powder was slower. We don’t know if the lots were the same between his powders.


Different lots.

I only had ~1/2 lb of the old powder and a brand new 8lb jug of the new, so I didn't have the opportunity to expand the tests to lot numbers.

Having different lot numbers of both powders could very well have shown very interesting results.
 
The interesting or maybe less interesting thing in all this is it really does not mean anything to us as a whole , unless we are using the same powders in the same lots , in the same environments . I did a test using old and new lots of IMR-4895 in 308 loadings and did record a difference but it was not much . I don't remember the specifics but think there was an average of 25fps difference . Again that only mattered to me and the lots I was testing with the specific components I was using . I'll add mine were all weighed charges and not thrown at all .
 
Quote I'll add mine were all weighed charges and not thrown at all .Quote

Absolutely. all test charges are weighed. I don't weigh charges for normal loading, but for testing they are all weighed.
 
How did you conclude that 846.5 fps is outside the data range for the old powder? Did I miss something? All he gave us was the average of each set, and the standard deviation.
Simple definition of standard deviation. The bulk of shots from his old bottle fell between 850.77 fps and 871.01 fps, average at 860.90 fps.

The bulk of shots from his new bottle fell between 837.97 fps and 855.03 fps, average at 846.5 fps. Thus...some bullets landed within the low end of his extreme distribution, but the bulk did not. The average velocity of 846.5 fps is lower than the 1st sd of 850.77 fps, though they did fit within the 2nd sd of 840.64 fps.

If one were to plot this out you would see two distinct sets of data that overlapped at the low distribution and the high distribution respectively.

I'm exaggerating the effect, but he has a bactrian camel, not a dromedary camel...ie. two humps, not one.
 
That's not how SD works and or is defined . SD is the likely difference you should expect from shot to shot . My SD's are never 1/2 of my ES . In fact if I recall they tend to be around 1/3 of the ES . Meaning if I have a ES of 60 the SD will be around 20 not 30 .

Another point would be that just because you have an avg of 200 does not mean most of your shots were close to that number . Half could have been 300 and the other half could have been 100 leaving you an avg of 200

You know , it's late and I've had a heck of a day maybe I'm all backwards on this . Please correct me if I'm wrong .I'm so burnt out right now I probably shouldn't be thinking or writing anything anyone else could read right now :o
 
Last edited:
That's not how SD works and or is defined . SD is the likely difference from you should expect from shot to shot .

You know , it's late and I've had a heck of a day maybe I'm all backwards on this . Please correct me if I'm wrong .I'm so burnt out right now I probably shouldn't be thinking or writing anything anyone else could read right now :o

He's comparing the 2 different data sets. The average FPS are far enough away from each other that you can determine with a fair amount of certainty that the old and new powder are not providing the same results.
 
reddog81 said:
He's comparing the 2 different data sets. The average FPS are far enough away from each other that you can determine with a fair amount of certainty that the old and new powder are not providing the same results.
But I asked the question how he [markr6754] could determine that the average from one data set is outside the range of the other data set without knowing anything other than average and standard deviation.

I agree with Metal God -- I don't think markr6754's conclusion is valid.

That said -- it has been almost 60 years since my last class in probability and statistics.
 
That's not how SD works and or is defined . SD is the likely difference from you should expect from shot to shot . My SD's are never 1/2 of my ES . In fact if I recall they tend to be around 1/3 of the ES . Meaning if I have a ES of 60 the SD will be around 20 not 30 .

Another point would be that just because you have an avg of 200 does not mean most of your shots were close to that number . Half could have been 300 and the other half could have been 100 leaving you an avg of 200

You know , it's late and I've had a heck of a day maybe I'm all backwards on this . Please correct me if I'm wrong .I'm so burnt out right now I probably shouldn't be thinking or writing anything anyone else could read right now :o
No....that is exactly what standard deviation means...and what it says...it is a measure of spread from the mean...if the data is normally distributed. Your example is meaningless as 50% 100s and 50% 300s is not normally distributed (bell curve) data. Shots from a firearm with a given powder and projectile weight will be normally distributed. The 1st sd covers 68% of the data.
 
But I asked the question how he [markr6754] could determine that the average from one data set is outside the range of the other data set without knowing anything other than average and standard deviation.

I agree with Metal God -- I don't think markr6754's conclusion is valid.

That said -- it has been almost 60 years since my last class in probability and statistics.
I stand by my remarks. Results are statistically different. You had all the data you needed. The remainder of the details are inferred...Same firearm, same distance to chrono, same temperature, and so on.
 
Last edited:
Back
Top