There is a lot of folklore out there about black powder, and I must confess, I enjoy dispelling a few of those myths from time to time. One of them is that black powder is very hygroscopic and can absorb substantial proportions of its weight in moisture directly from the air. To test if this is indeed a problem, Pete Mink in Iowa, and Ed Stutz in Ohio ran a simple but fairly conclusive test using unopened tins of 2fg Elephant powder of different vintages. Pete used the fabled 25/99 batch, while Ed used a 1997 powder.

The test was simple. Basically, they started early in the morning using Redding powder measures set initially to throw about 90 grains of powder. Every \BD hour, changing to every hour or every other hour as the day wore on, they took four samples and weighed them as carefully as possible while also recording the time, temperature, humidity and barometric powder at the time of sampling. In Pete's case, he also weighed a sample of this powder and then dried it in an oven for a considerable amount of time at 150 degrees and reweighed it. From this we know that his powder started at 1.23% moisture.

I plotted and regressed the time series data in simple linear regressions to look for some sign of an effect of these environmental variables. This is a very simple test, but I think it suffices adequately to say, there ain't a whole heck of a lot going on.

In Pete's data, there is no regression that explains more than 12.7% of the variation in Average Weight. That is to say, of the variation in the average from one reading to the next, there is really very little of it that can be explained by changes in the environment. That said, temperature and barometric pressure did better than Humidity and Time by a lot. Further, as Temp increased, powder charge decreased. As it did for Baro. Humidity was a pretty useless variable when used in this way, and while powder charge did increase with humidity, it only explained about 6% of the variance. And the slope was VERY shallow. Over a very wide range of humidities from 50-90%, the regression predicts only 0.1 gr of increase in the powder charge. That is to say, over the widest range of humidities that you are likely to encounter, Pete's powder charge might fluctuate one tenth of one percent due to humidity. That's damn little, and quite a bit less than the range of variation in the four samples that Pete took at each time interval. In other words, there is high probability that these trends that I've mentioned are really produced by simple measurement variation, and not by environmental conditions.

Now, this is hardly a sophisticated analysis, and one might rather use a multivariate approach that uses all or at least more than one variable at a time. There is little reason to expect to do much better, but it's possible. Second, it might be better to use some other indication of water availability since relative humidity is a relative thing - that is to say, it's not the amount of water per cubic foot of air so anything like that. But the bottom line seems to be that there is very little going on here.

I hope that this makes sense to everyone. If anything I said is unclear, just ask. I do not find this easy to explain without assuming a certain amount of background in statistical analysis, but I've tried my best.

Ed's TEST#3

Was to try to duplicate what I had observed on the range during match shooting from time to time and what Pete was trying to sort out. I set up a powder measure [Redding 3br] and set to 90 grs with Elephant 2f lot 099 34/1997, Iused this powder through out these tests. Started at 7am weighed four charges after setting scale [RCBS 10/10 beam scale, for those who need that info] then weighed at 11am and again at 3pm noting average weight, temperature, humidity, and barometric pressure. this was a new gauge and was calibrated to instructions except for the barometer as I couldn't get a local reading, doesn't matter as we are only interested in the amount of change if any.

TEST#4

This was to duplicate the test that Pete did so I set up another powder measure again at 90 grains of the same powder mentioned above, took four weights and averaged as above and recorded these along with temp, humid., and baro press. and also noted the weather conditions.

Here are the graphs from Ed's data. I have plotted Exp #3 in pink squares and used a dotted regression line. For Exp #4 the data are dark blue diamonds and a solid regression line. The equations and R2s on the graphs are for only Exp #4. I added a copy of the table from the book that Pete has been telling everyone about. Pass these along to whomever you wish.

I've plotted both Experiments #3 and #4. The results by themselves are interesting. First, over any range of variables that average charge wt is plotted against (time, temp, humidity, baro), the powder in exp #3 changes in the opposite direction from Exp #4 even though they are largely the same test. Ed, that tells me that something else is going on and I sure don't know what it is.

x` BUT, taken in all, the results are a lot like Pete's. Not much change, nor much correlation with any of the potential explanatory variables. There are some differences however. First, Ed's powder responds to changes in these variables more than Pete's. Note that the value in front of the "x" in each graph's regression equation. This is the slope of the line. Bigger numbers mean steeper lines. Smaller numbers mean flatter lines. The sign of this slope (+ or -) tells you whether the powder charge increases or decreases with increases in the variable plotted on the x axis. In comparable graphs, Ed's slope values are sometimes about 2 times Pete