I started a runtime test last night, but I do not have a light meter. I'm not sure what my criteria for "cut-off" will be.
I do not have another red light, so I was thinking I'd get a red Photon and take beamshots at various time milestones vs a new Photon. Would that be easier, or should I take beamshots at known (fixed) exposure values throughout the test?
Since no better means may exist at your disposal, you can try this (though correlating this method's results to specs or other's measurement means will not be possible)...
Using something like Microsoft Word, type out a line of text in a very large font size.
copy and past it in a number of times (10-20 times) with a couple of blank lines between each line of text.
make the font successively smaller for each line.
See how far down you can read at 10 or 20 feet away when the batts are fresh. Take hourly "readings" (pun intended) and keep notes of when you are no longer able to read lines of smaller text that you could read before.
Like i said, it's NOT a great method (it's not even a good method), to say the least, but perhaps it's better than nothing.