I wondered about that measured radius vs rollout (revs per mile)
specification.

Let us assume we have a tank or some other tracked vehicle (D9 Cat or such).
One rev per mile is one length of the track.  Once around.  Now this
parameter is independent of how the tread/track is routed.  In a tank the
tread describes a rough oval with 'trapezoidal' ends, in a top drive Cat it
is more like a pyramid.  Regardless, one rev is one track length.

The question is, if the tire is grossly underinflated so that it is
'squashed' does one revolution in 'squashed' mode, equal in rollout, one
revolution at maximum pressure?

If not then why not?  All that tread still has to go around once and the
tread although positioned closer to the axle is still the same length.

If so then revolutions per mile is not dependent on measured axle to ground
distance although overall gearing is slightly affected as is braking and
acceleration.

So if the tire is underinflated the effect is that the last factor in the
thrust value, the radius arm of the torque applied at the axle is less
producing a slightly lower overall final drive ratio.  The reverse is true
for fully inflated tires (larger axle to ground distance).

So for speedo accuracy, rollout is important.  For gearing selection, axle
to ground distance is important.  IFF the arguments above are valid and not
just the usual ravings.

Pensioner

"If you profess knowledge but cannot express that knowledge in numbers, then
that knowledge is of a meagre and insufficient kind" -- Lord Kelvin