Roy Lewallen wrote:
So, does the current go clockwise or counterclockwise? How much goes
that way compared to the radial component? Where can I find a
quantitative or explicit statement of your interpretation?

I think, based on the excerpt Richard has provided, it does both.
Imagine a leaky hose with water diffusing into the surroundings. So, if
you were to integrate over the entire width,or over any region which is
symmetric over the wire, the *net* is entirely radial, but if you look
at a small region, directly adjacent to the wire, there will be current
diverging from the wire as you move outward (assuming current flow is
outward... obviously, on the opposite half cycle, it converges toward
the wire, as it moves generally inward)...

I suspect one could also analyze it as a wave propagating away from teh
wire in the lossy surrounding medium, where the medium has a lower
propagation velocity than in the wire. (e.g. imagine a waveguide made
with the walls being soil)

Another sort of "hydraulic" model would be if you represented the
radials as below grade drainage ditches which have a lot more pitch than
the surrounding soil, so the water tends to flow diagonally down the
ditch walls.

The interesting question would be whether this is important at all..

One might go through lots and lots of analysis, worrying about the small
incremental effects of non-radial current, and find that the inherent
variations in soil properties are orders of magnitude larger.

Sounds like a good exercise for a graduate level E&M or calculus class..
you could cast it as a similar exercise in heat flow.. both temperature
and electrostatic fields satisfy Laplace's equation.