Michael, all sorts of subjects were photographed and printed on multiple grades. Of course they emphasis “normal” conditions. BTW these tests also determined what are considered normal conditions. I have to refresh my memory as to the details, so I’ll have to come back to this point. In Minimum Useful Gradient as a Criteria of Photographic Speed, Jones writes, “The one factor of fundamental importance which determines the quality of a picture made by the photographic process, whether it be a print on paper or an image projected on a screen, is the relation between differences as they exist in the original object and the corresponding luminance differences in the reproduction. Therefore, from the standpoint of tone reproduction and this is the most vital consideration in judging the quality of photographic results, the gradient characteristics of the negative and the positive materials are of primary importance.”
It occurs to me, the print evaluation studies originally used in deriving 0.3G involved exposures of average scenes. So, to me, it would mean in the context of the tests, developing the films to lower gradients (for example) would have constituted “errors” to be corrected with an increase in paper grade – rather than deliberate attempts to accommodate a higher than normal subject luminance range. This could explain why I was initially startled by the six graphs on p.325. Before going into this, I had actually expected at least some of the print judgment speeds and 0.3G speeds to be lower than the fixed density speeds (eg: the long toe, low gradient curves). But this is the opposite of what was concluded in the print evaluation studies. While the math made sense to me, intuitively the graphs just didn’t look right – unless one simply increased the paper grade to effectively normalize say the bottom two curves in Fig.2. This is my take on why the fractional gradient method is not more concerned about a precise contrast. A constant relationship of 0.3G seems to work in print judgment tests, apparently regardless of how low the absolute toe contrast is, because a correction via increase in paper grade would bring both the toe G and average G up. I think this is sort of a different way of saying the same thing as the possible explanation you outline in your paper for why fractional gradient is not more concerned with a target contrast, and why the Delta-X film speed stays relatively constant with deviations in delta D.