...it shows 50 ohms at 14.185 but has an SWR of 3 to 1...
How are you measuring the antenna? A 50 ohm antenna can't have an SWR of 3 : 1 unless
there is significant reactance. If that is the case, then adjusting the wire length should
bring the SWR to a reasonable value. (Or inserting a coil or capacitor in series with the
wire at the feedpoint.) But this may not be the case if you are measuring the antenna
through any significant length of coax.
Loops can have a wide range of input impedances depending on the shape. If you make
it rectangular and about twice as tall as it is wide, then feeding it in the middle of the
bottom wire gives very close to 50 ohms. If you spread the sides out so it is wider than
it is tall then the impedance will be closer to 200 ohms (with slightly lower gain and wider
SWR bandwidth than the 50 ohm version.) The same approaches apply to delta or diamond
shapes.
...the antenna was 4 feet off the ground...
Antenna impedance varies with height above ground. I would guess that the resistance will
increase and the resonant frequency will shift up the band by perhaps 100kHz when the
antenna is raised into the air. Getting the bottom wire 5m off the ground will give you a
better sense of what it will be at the final height. (This also depends on whether you are
using a single loop or a quad, as the parasitic elements of the latter will increase the
variation.)
I suspect that the elongated rectangle will end up giving you more gain than a circular
loop and a more convenient feedpoint impedance (besides being easier to construct.)
Here are some references:
http://www.cebik.com/content/a10/quad/qloop.htmlhttp://www.cebik.com/content/a10/vhf/cs.html(Yes, they require a free registration, but this is one of the best antenna resources
on the internet.) The circular loop can give about 0.3dB improvement over a square,
while the elongated loop is about 1dB better.
Generally I find that I can get close enough to match any full-wave loop using either
direct feed (50 or 75 ohms), a quarter wave of 75 ohm transmission line, or a 4 : 1 balun,
depending on the shape of the loop.