This whole thing is quite relevant today, since the usgus has got everybody convinced that it is shallow faults and injection volume that is driving this whole thing. Thus, Kansas and OK are reducing volumes. The next big earthquakes will blow that out of the 'water'.
So, we look in detail at my 'straw dog' principles, or working hypothesis. I've almost promoted all this to a theory, but that's just me. :)
1. All the earthquakes are reduced to quartz and water.
That's right, two quartz points and water. When a Precambrian (granitic) fault slides, it hangs up on the quartz points, no matter what the other minerals. This gives us an 'initial state'. Further motion (critical displacement) shears these points. If you baked a sample dry, and put it in outer space, you could not create an earthquake, since further shearing would cause further adhesions. Unfortunately, this is what most people do when they are studying rock in a testing machine.
In the real world we have water. Even pieces of rock on your desk have a thin film of water. On the small scale, these are little ball-bearings all over the quartz. When you shove two quartz points together with enough force, the ball-bearings move away, but this takes some time. Think of a kid sinking into the ball pit.
All of this causes the drop to dynamic friction. Earthquakes are caused when the quartz points are under shear stress, move a little bit (dc) to break the adhesion, and can't reform bonds quick enough. This is called stick-slip. The exact same mechanism gives you a bump on the head when you slip on an old-fashioned enamel bathtub. Once second you have firm footing, and the next .... whoops!
With this model set up, we can have fun with changing the water properties, for the last bit on stress corrosion.
-to be continued.