Its desirable IMO with calibration tests to replicate actual shooting conditions to eliminate variables. With bellows extensions, filters, incandescent light, etc., there are some variables that are getting introduced that could effect results.

After testing for awhile, I found that I was really only using zone I, V, and VIII readings to determine the spread that my dev time was giving. So my tests evolved to a method based on Hutching's described in the back of his pyro book. Using a board with a strong light at one end that graduates through a full 10 zones, you can mark the points where my meter reads I, II, and VIII, The advantage is that I'm using my regular spot meter and a normal lens with a reliable shutter speed.