I built this project to see if crystal cells could actually be used in a real world application. I have built/seen other peoples prototypes of crystal cells and found them to be something of a novelty and lacking; they didn't really power useful devices except multi-year motors that had no load, and multi-year led night lights that couldn't light up a room or be useful as a flashlight. I also seen people use the same electrolyte in lead batteries to extend them and make them indestructible at low voltages (can be shorted for days on end and they bounce back). I thought why not combine them both and make something that you can actually use and walk around with...albeit carefully. Its been 5 days so far, and the light hasn't extinguished. The light is as bright as 3 days ago when the voltage dropped to about 3.4-3.5v and hovered in between. The amperage has been steady at .5 amps. Time will tell if this battery lasts a long time but so far the curve has disappeared.
Have you thought about datalogging current and voltage through the discharge cycle so you can quantify watt-hours? That seems like the most relevant measurement for battery performance IMHO.