From what I can gather when reading about the early Mark I's from Springfield, it all came down to if it was sunny or cloudy on the day when receivers were being treated. Apparently the 'old hands' they hired to do the treating only went by the glowing color of the metal when determining when the receiver was at the right temperature. They did not compensate for how the metal glowed on sunny days as opposed to cloudy days. If a trained eye did not see the proper glow, it went back into the treatment furnace, which lead to overtreating and crystalizing of some of the components in the steel, which in turn lead to receivers exploding when fired.
A few GI's lost eyes when the bolts were blown backwards, others were more critically injured and I think there was one death.
By 800,000 the problem was solved and receiver failure was down to much less than 1 : 100,000. Now the problem is: of all Mark I's that survived and have numbers under 800,000, which ones were treated on sunny days. I spotted a Mark I with a number under 800,000 on one the Gun Auction sites the other day, so I know some survived.