The compressibility correction is different
This is a new one for me, also. My take -
(a) the early instrument makers were not able to model the full equations into an analogue device (and, yet, Huber did that in his computer in the late 40s - go figure ?)
(b) the outcome was that they had to take a shortcut. My understanding has always been that this was to restrict the ASI to standard SL conditions.
(c) my speculation is that the folk on opposite sides of the pond developed slightly different analogue models which would account for differences in calibration as suggested in Bennett's document ?
Perhaps, sometime when I have a (very) slow day, I might play with the equations and see if I can come up with the numeric deltas in the document.
Once again it never ceases to amaze me at the incredibly interesting stuff which comes out of the woodwork in PPRuNe. And, yet again, my good colleague, Megan, demonstrates his quite incredible ability to find obscure stuff in the said woodwork ... well done, once again, good sir.