Topband: Variable Flag Termination?

[Dan Zimmerman N3OX]

I took a pass at optimizing my flag termination a few days ago and got
a pretty respectable pattern on an AM broadcast station, WWWT AM 1500
about 10 miles from my house during daylight hours:

http://n3ox.net/files/flagpattern_lg.JPG

I ended up with something around 1300 ohms.  I'd been using a 940 ohm
termination before that and hadn't really checked the pattern, but it
still seemed to have some useful F/B

However, last night when I got on 160m I felt like I'd actually made
the antenna worse F/B-wise.  I **felt** like I was hearing U.S.
stations a lot more clearly off the back than usual.  EZNEC doesn't
give much of a clue as to what might be going on; a good termination
in the model for 1.5MHz is good at all higher frequencies that I'm
interested in. It might have been that I had a better elevation null
in the rear with the old configuration, but I'm not sure that's it.

The first step, of course, will be to actually measure the pattern on
160.  I don't have a good local source, so I need to rebuild my TTL
oscillator chirper and tiny magloop and toss it out in the park again
... maybe the degradation is all in my imagination... (alternatively,
if someone around here is within groundwave-dominant distance of me
and wants to hit me with a carrier later tonight, let me know ;-) )

But, I was just wondering if those of you who are using flags (or
pennants or deltas or other ground-independent terminated loops) have
found variable terminations to be useful across the bands?    I know
some K9AY users find them useful, and I know K9AY doesn't favor
Vactrols but I do have some light dependent resistors and I also think
I have a hank of old fiber optic cable somewhere and I'm thinking I
can hack something together that won't require any extra wires near
the flag...

I run low power and doubt I'll damage the LDR with transmitted energy...

So what does everyone think?  Put together a variable termination or
just get a good steady signal source on 160 and re-adjust if
necessary?  Has anyone else observed much frequency variation in their
pattern? That sure doesn't show up in a simple EZNEC model, but maybe
it's some sort of soil frequency dependence issue...

73,
Dan