LOL

Simon,

I wrote a little spreadsheet to calculate the partial pressures of the various constituent gases in the mix, using the Clausius-Clapeyron relationship to find the PPs of each gas, and then calculated the total pressure, and gas used as the cyclinder is used.

I used two approaches, actually, on using the C-C realtion, and one using quoted pressures.  The problem with the C-C method is that it gives the true partial pressure, so ignores the surrounding atmosphere, and gives a more promising result.  The approach used to draw the curves that Mole linked takes account of atmospheric pressure when calculating which gas is used, even though it's a less elegant calculation.

If you're interested, drop me a PM with an email, and I'll send the spreadsheets.