Laurent,
I just checked your graphs and must say: nice work. As you may know I'm doing a lot of reviews for various magazines and my testing guidlines are also very strict.
Looking at your graphs I could see the following:
ACS: total weight difference is 300 grains, speed loss/gain is 45 fps = appr. 6,7 grains for 1 fps.
Fedora: total weight difference is 300 grains, speed loss/gain is 41 fps = appr. 7,3 grains for 1 fps.
Black Widow: total weight difference is 300 grains, speed loss/gain is 49 fps = appr. 6,1 grains for 1 fps.
RER: total weight difference is 300 grains, speed loss/gain is 42 fps = appr. 7,2 grains for 1 fps.
I'm testing my bows with 6 to 10 gpp, so my findings are different, especially with foam core limbs. I used part of your charts and looked at the lighter arrow weights. If I take the part between 550 grains and 400 grains I see the following:
ACS: total weight difference is 150 grains, speed loss/gain is 26 fps = appr. 5,7 grains for 1 fps.
Fedora: total weight difference is 150 grains, speed loss/gain is 22 fps = appr. 6,8 grains for 1 fps.
BW: total weight difference is 150 grains, speed loss/gain is 25 fps = appr. 6 grains for 1 fps.
RER: total weight difference is 150 grains, speed loss/gain is 22 fps = appr. 6,8 grains for 1 fps.
If you go towards 6 gpp you'll get close to 3 - 5 grains for a speed difference of 1 fps, especially with ILF limbs. This thread shouldn't end as a pi$$ing contest, but you can see that my numbers even match your charts. I never tested bows over 11 gpp, so you have the edge in this department. When I do quick calculations, I use 4 to 5 grains for 1 fps, which ever is easier to calculate.
Blacky