Originally Posted by PeterB
You need to think of testing the film and testing the camera as two separate things. Take a look at the flare test I uploaded. What I did is to make identical exposures using the black box described by Davis. The only thing that changed were the boards I had surrounding the opening. One was black, one was gray, one was white, and one was a mixture of all three. After shooting them, I then make a sensitometric exposure (that's where the curve is from). Notice how even though the exposure never changed, there's a 0.39 log-H difference between the black surround and white surround.
Now ask yourself, how can you get a reliable film speed from this method? Your test showed a great deal of flare. How much should it be and how can you confirm the amount?
Contacting is simply the way it's done. Ralph's correct in that it's unrealistic to assume there isn't flare in the system, but that doesn't mean it has to be included during the exposure part of the film testing only that it needs to be factored in. This can be done at the interpretation / evaluation stage. The ISO standard for B&W film speed specifies contacting the film, but it's the methodology of the film speed determination that has a stop of flare factored into it.
I recommend reading Photographic Materials and Processes by Stroebel et al or Sensitometry for Photographers by Jack Eggleston.