For what it's worth, as I am re-reading BTZS, I have just decided to plot my own curves (using R) from the flawed test data, which I enclosed in the first post on this thread. Thanks to your comments, especially Stephen's patient demystifying of the concepts, I feel more able to do it without having to ask Bill for his services. This is the resulting set of curves, I have smoothed them using LOESS before plotting, and I am glad to say this looks almost identical to Bill's hand-made plot:
Flawed Film Test Curves.png
Reading page 91 of BTZS, "Troubleshooting the Film Test" section, Figure 7-6d, shows a similar situation, where the curves' toes stack up on top of each other this way. Mr Davis helpfully comments: "You'll get curves like this if you test films in the camera (by photographing the step tablet on a light box, for example). These curves are not usable.". I suppose I know from you, gentlemen, that the key reason for the curve stacking is the flare. I wish I read BTZS before the test, but at least I have learned something useful my usual way: try and err.
I am looking forward to retesting, most likely using the flash exposure technique. I hope I get a repeatable exposures, but just in case, I will shoot 25, or so sheets, so I can average the readings from each of the five development batches. I need to do that, as I usually develop 5 at a time in my HP CombiPlan. I wonder, however, if I should agitate a little less than usual, perhaps down to 3–4 inversions only every 30s.