I wasn't saying to use the 50 fps per grain as an absolute value, and I understand it's not completely linear through the spectrum of the entire charge, but you did get me curious. So in two examples of ammunition manufacturing, the 5 fps per grain.10 grain, or 50 fps per grain isn't far off.
In the Hornady manual, the use of H4350 in 6.5 CM (which is what the OP conversation was about), the charge of 41.1 for 2600 MV and the charge of 42.8 for 2700 MV. That's about 58.8 fps per grain.
In the Nosler manual, the charge of 37.0 for 2485 MV and the charge of 41.0 for 2699. That's about 53.5 fps per grain.
Again, I never said this should be used in load calculation or load development. It was in reference to show an approximation for how much velocity was being possibly left on the bench by using a lower charge.