If I understand the lighting situation, i.e. dark as all get out, the light is waaaay below the threshold of sensitivity of any meter I know of.
Calculation goes out the window at that point. Tests and comparable situations become the way to establish exposure times etc. I'd just open the lens up all the way and try a series of exposures - 1, 2, 4, 8 hours. Then process and see if any show any exposure at all.
I hate to say this, but what about testing with digital? Most DSLRs will go up to ISO 1600, even if it is noisy as hell. They don't suffer reciprocity and you get really quick feedback. Why not use one of them to test for the "metered" exposure and then do the reciprocity calculations from that? You're probably looking at a measured exposure that is in minutes, so bumping the ISO to 1600 should let you start with a simple series of 1, 5, 10, 20 minute exposures as test cases. Much quicker than doing a 6 hour test, developing it, doing another with a different f-stop.
I regularly use a DSLR to test exposure before I burn large format film. My 20d is a very expensive exposure meter. I just have to remember that when the highlight warning flashes my film will probably still have detail.
I was actually thinking of trying this. I asked the original question to tap into the vast pool of knowledge here. Great idea!
One of the problems with digital and really long exposures is the battery can die... My film SLR or my rangefinder uses little or no batteries on bulb.
I have a tested reciprocity chart that shows a metered reading for Tri-X of 10 minutes requires an exposure of about 9.5 hours. Tri-X ain't gonna work in this situation.
Srangely, that is very close to what my equation predicts when I use Kodak's value of 1 second for the amount to be added to a calculated exposure of 1 second. The equation gives 600 + (1*600^1.6) = 27864 + 600 seconds = 7.9 hours. Working backwards to find the amount of correction at 1 second that would predict 30000 seconds correction at 600 seconds I find that value to be 1.07 seconds.