This line
GW 37 3 0 y 0 0 -y 0 0.0404
is an imaginary wire used to specify the feed point with the "EX" line below it.
the last field in the line/card (0.0404) is radius in inches of 12 GA wire.
(same as all the other wires)
But models should be able to pass an AGT test (f-7 pick "far field pattern then look for the check box)
So to pass the AGT test this model needs the radius of that imaginary wire to be "0.077" At least in the TV UHF frequency range.
This also reveals another 0.6 dBi of gain.
So be careful of models you find posted (or create) as this could be used to cheat even if inadvertently done.
At the same forum dave pointed to there is a modeling thread here
http://www.digitalhome.ca/forum/show...ighlight=4nec2 it's long and wanders a tad but the search function can weed out some of the less relevant stuff.
As far as this fractal 2 bay whisker model I think the elements are to long (Because the SWR decays at higher frequency >ch.44). Most whisker antennas are 9-10 inches long (for uhf) these are 12"
So the sections (not segments they are different) might be better at ~2.5"
for a total wire length 10" per whisker.
Overall applying fractals to an antenna does not increase gain/performance
(at least that Ive ever seen). It's a mater of "how little do I lose if" game.
The payoff is in smaller easier to stamp out antennas without giving up to much performance
So fractals are generally compared to the parent design as a best you can do max.