You are right, there is always a bit of variance but my point was that the same light, if properly regulated will draw the same regardless of cell used. I wasn't saying that all lights of a particular model or build will read the same as it's siblings. A single regulated light will be the same brightness on a TF Flame as it will on a Panasonic PD or a VTC4. Each LED of a particular model bin can vary as much as 10%. It's why one guys sample light could read 10% higher than another guy who has the same make and model of light.
I guess I simply shouldn't worry if my reading is a bit low for my TK35UE. The same tested by Fenix may have been brighter than normal and my sample might be weaker than normal. It's really only a 5% difference. :D