.. and the real question is: what mistake did I make?
None; I made one. The datasheet I pulled up turned out to be the one for graded Fomabrom, which shows more linear curves...
Koraks: I would be interested how did you come into those conclusions based on my plotting - it would be nice to learn this (too).
Basically it's a matter of looking at the curves and then specifically at a few properties:
* The toe, which in paper contains the (very) light values/low densities. As you can see particularly in your first image (the second is a little less obvious in this respect), the grade 0 curve for Fomabrom becomes noticeably less steep towards the lowest densities (i.e. it exhibits a long toe). This means that the very high light values become less distinguished, as an equal change in exposure in these high values results in less change in density compared to the middle of the curve (which is steeper).
* The shoulder, which is similar to the toe but then for the high densities i.e. deep shadows. Again looking at the Fomabrom curve but this time the one for grade 3, you can see that the shoulder becomes less steep at around 1.50logD. This means that given the same increase in exposure leads to less increase in density compared to the steeper lower end of the same curve.
I have plotted it in your second image by means of illustration:
The red set of tangential lines is for the Fomabrom grade 0 curve, clearly demonstrating the difference in steepness (i.e. contrast) between the toe and the middle part of the curve. The blue set of tangential lines for the Fomabrom 3 curve demonstrates the difference in steepness (i.e. contrast) between the shoulder of that curve and its straight middle part. This illustration also shows that the blue shoulder tangent for grade 3 is similar in steepness (a tad steeper, but not all that much) to the red tangent for the middle part of the grade 0 curve.
As you can tell, the placement of the shoulder and toe tangents is somewhat arbitrary, as the contrast changes continuously in those parts of the curve; the effect is more or less pronounced depending on what point in the toe or shoulder you are examining.
Concerning the shoulder of the grade 3 curve, this is the part that as a printer would worry me most, for the reason that the human eye has a bit more trouble discerning density differences in low light values (i.e. shadows), so a lack of contrast in shadows is usually perceived as more pronounced as the same lack of contrast in high values (highlights). We are pretty good at determining the difference between pure white and just off-white, but we as humans do a pretty bad job at discerning pure black and something that approaches pure black. When examining negatives, this becomes even more problematic as the shadows in the print are the thin parts of the negatives, so when looking at the negative, we can easily see quite well the shadow gradation in our negative (i.e. the low-density areas), while when we print this, this contrast will be less pronounced to our eyes. Add to this my personal preference for good distinction in shadow values and the Fomabrom grade 3 curve becomes a bit problematic at lest for me personally.
Perhaps this also helps to explain my preference for overexposing negatives (in films such as Foma400, which we discussed recently) in order to push the shadow areas a bit upwards in the film curve so I get good contrast in the shadows, as they are particularly challenging to translate well to paper for the reasons outlined above.