Jim M0BMU,

Some further comments on selected points:

I am confident that there was a slight increase in FS, and that it was
near 0.9dB

A decrease in current from 2.7 to 2.2A would result in FS
decreasing by 1.8dB if the antenna remained the same. So overall,
there was an improvement of 2.7dB from the radiating point of
view, although offset by increased losses, apparently.

That is a good presentation, to summarise:
   impact of the top loading coil on new current distribution  +2.7 dB
   impact of the higher losses in that loading coil                   -1.8
dB
   net result for that setup
+0.9 dB

What ZL2CA says about increasing TX power to make up for
reduced antenna efficiency is only true up to a point, as some of
us have found already. There comes a point where the voltage on
an antenna becomes so high that breakdown occurs. Since for a
given antenna, the voltage is directly proportional to the current,
this places an upper limit on the ERP for a given antenna.

What ZL2CA would reply to that point is to find better insulators.  They too
are far cheaper than managing the overheads of an elevated loading coil.

By the way, I find that the loss resistance of the antennas I have
used tends to increase by a factor of 2 or 3 when frequency is
reduced from 136k to 73k - in indirect support of Rik's idea about
ground loss being reduced at increased frequency.

I too found that trend.  Results of impedance measurements of an earlier
vertical at my QTH appear in Table 8 of the LowFer section of the 18th
edition of The ARRL Antenna Book, which is:

           Frequency       Resistance     Capacitance
           100 kHz           14 ohms         790 pF
           165                  11                  800
           190                  10                  805
           250                    9                  810
           300                    8                  815

The capacitance increase is expected, from the transmission line effect as
antenna becomes an increasing fraction of a wavelength with increasing
frequency (a hyperbolic tangent function if I recall correctly).  However,
the increase in series resistance is an enigma.  My current thinking is that
it is better to express the impedance in parallel components rather than
series i.e. admittance rather than impedance.  The antenna environment of a
top loaded T does physically involve equivalent parallel RC in the "foot
print".   I'm soon departing to do some shifts on a field day contest so I
can not develop comment on parallel rather than series analysis in this
posting.

Cheers for now,

Bob ZL2CA