Thanks.
One nit: Baxandall has a side-note "...while an f.e.t. will normally have a lower gM than a junction transistor at, say, lmA, the more gradual fall-off in gm with working current for the f.e.t. would, if continued, give it a much larger gm than a junction transistor when operated at a low enough current. In view of the very basic quantities involved in equation (15), I felt this result was probably too good to be true!"
I believe that in fact a modern FET gives more gM than a BJT when you get down below about 10uA. And I think this fact is critical in medium-speed low-power logic design. I may be wrong about this... I may have picked-up something from someone who had not really tried it.
In audio, it is moot. Below 10uA, we'd be drowning in stray capacitance and clobber gain at the high end.
It is also irrelevant to Baxandall's studies, because he runs his test FET at 1.5mA or 1500uA, and his own curve shows his FET following the square-root law very well down to 30uA, far outside the swing you would expect from a 1.5mA idle current.
