I did not see this in L.B.'s article ( in a quick read through) and it
would
seem to go against the basic premise that takeoff angle is solely a
function of
height above ground. I did a loop model in AO and did not see any changes
in
takeoff angle as I changed loop circumference. Can you point me to the
section.

Look at the first three figures with elevation plots for an 80m loop at 3.5,
7 and 14 MHz and see how the elevation angle falls with frequency (as well
as with height). This performance is scalable, so consider these 3 figures
as a loop of size one, two and four wavelengths. Then 80m/3.5 is the same as
a 160m loop at 1.8 MHz, and 80m/7 MHz is the same as a 2*160m loop at 1.8
MHz and so on. From this follows the results that a loop is a cloud warmer
(NVIS) at 1 wavelength and becomes a better and better DX antenna as
frequency increases. But only up to a point, as a 160m loop at 28 MHz does
not fully develop its main lobes and loses gain compared to a 80m loop at 28
MHz, the same with a 2*160m loop at 14 MHz and so on.


Sverre
LA3ZA