Frank, as you say, the height of the radials in the NEC2 model is only 6
inches above ground.

The radials are shallow-buried in the RADIALS2 model. It can't do elevated
radials.

The ground loss resistance as height decreases, as seen by the antenna,
increases very fast percentage-wise as the radials get within a few inches
of the ground. It is due to very close magnetic and electric coupling to
ground. Radials are transmission lines, insulated from but running very
close to a resistive slab of soil.

This would account for the computed higher input resistance of the radials
( 3.5 - j*3.3 ohms ) ( for 29, 25-feet long radials. Rg=77, K=13 ) in
program RADIALS2.

The calculated antenna input impedance in RADIALS2 is that of the antenna
alone. For feedpoint impedance add the input impedance of the radial system.
Presumably, NEC2 does not compute the input reactance of radials.

Efficiency is calculated in the usual way from the sum of antenna input
resistance and radials' input resistance.

If you contrive to change the radials input reactance without changing
frequency or the antenna, you will notice the loading coil tunes it out
along with antenna reactance.

Incidentally, when elevated radials are near the ground their velocity
factor decreases fast which makes a mess of the usual recommendation to
prune them to 1/4-wave free-space length.

When radials are actually lying on the ground surface the velocity factor
decreases to roughly 0.5 of the velocity of light. When buried the
underground VF can fall to as small as 0.15 depending on soil permittivity.
(or moisture content.)
----
Reg, G4FGQ