Thanks all.

Roy, I was not implying that I believe one can assume that the power
is from a point source and one can consider the power density passing
through a sphere to determine RF safety. I was looking for some
guidance as to how to determine a "safe" distance from a small tuned
loop assuming a particular frequency and power.

It appears that the simple sphere approach works reasonably well
beyond a wavelength or so, and may be an acceptable first-order
approximation at 1/2 wavelength [from a small loop].

-JJ