Denton, thanks for posting, but I believe you have made an unacceptable assumption in your work. Two, actually.

Specifically, you assume that the variation is linear. I don't think that is correct.

Second, you assume that the discrepancy between crushers and strain gauges is random error. I do not accept that assumption. The two measure two different things. A strain gauge (or piezo, for that matter) is capable to measure peak PSI much closer to it's true value than a crusher. Simply because of the mechanical delays involved in the crusher system.

The two issues are related, of course.

I would suggest to try both running a non-linear regression, and include the bore expansion ratio of each cartridge as a variable. I would expect at least half of the "error" to disappear.

Oh, and try without the 45/70. There's something fishy about that data point. JMO, Dutch.

Try kernel regression. Gaussian or Epanechnikov kernel.

Most of my articles are posted at the RSI site,

http://www.shootingsoftware.com/tech.htm

I hope you enjoy them.

Probably my next investigation will be into the effects of magnum primers in 30 cal rifles. Casual investigation hasn't shown any effects at all, but I want to give them a try at really cold temperatures. The manager of Federal Cartridges 22 rimfire operation also says that he doesn't think there are any... but that's why we do the tests.