Keith Dysart wrote:
Consider a quarter wave-length of open circuited line connected
to a 100 W source. After 1/2 cycle it is fully charged. At 4 MHz
it stores 0.125E-6 J of energy.

Only if the 100 W source is turned off after the first
1/2 cycle. If it is a constant power source and is not
turned off, the forward and reflected power in that stub
would increase without bounds except for I^2*R losses
and dielectric losses. Only when the power output of
the source equals the losses in the stub will steady-
state be reached and that could be when the forward
power is 1000 watts or more.

I can not see a mechanism
where a 1/2 wavelength antenna would store more than this

As I said above, if you have a constant power source,
the energy in that stub would increase without bounds.
The forward power is certainly not limited to the source
power. The source is only having to supply the losses in
the stub. The stub could be storing magnitudes more energy
than your calculation.

(SWR on the line would be inifinity, much worse than 20:1),
and it will be radiating 100 J/s. 0.125E-6 J is not much in a
system that is moving 100 J/s.

As pointed out above, your logic is flawed. I estimate
that with 100 watts being fed into the dipole and 100
watts being radiated from the dipole, the forward power
is about 500 watts and the reflected power is about
400 watts at the feedpoint of the 1/2WL dipole.

There is also another flaw in your logic. You are equating
the length of time it takes to charge the stub to one
second so you are off by almost 10 magnitudes.
--
73, Cecil http://www.w5dxp.com