Walter Maxwell wrote:

Roy, it seems to me everyone has missed an important point concerning a circumferential component of the
current.

We know that the current flowing on the radial wires is radial in direction. What seems to be missed is the
current that returns to earth between the wire radials. That current is going to flow in the direction of the
lowest resistance. As such it's not going to flow radially alongside the currents flowing on the wire, because
the radial resistance of earth between the radial wires is much greater than the resistance of the wires.
Consequently, currents reaching earth between the wires will find a lower resistance by traveling toward the
nearest radial wire instead of continuing in a perfectly radial direction. This new direction of current flow
will not necessarily perfectly circumferential, but will certainly be somewhere between radial and
circumferential.

Walt,

I hadn't missed that phenomenon, but didn't mention it because it
doesn't produce a circumferential current. If you look at the current
flowing from the earth to each radial wire, you'll see that the sum of
these currents will be purely radial, assuming that the system is
symmetrical, i.e., radials are equally spaced and equal length, the
ground is homogeneous, and the radiator is vertical. Consider a bit of
current returning between two radials, which is a little closer to the
radial on the right. It'll detour to the right, giving it a rightward
component as well as an inward radial component. But for every such bit
of current, there's another one the same distance from the radial to the
left which will have leftward and inward radial components. The radial
components are in the same direction (inward) so will add but the
circumferential ones (leftward and rightward) cancel, leaving a net
radial current flow. You can say that the returning currents bend to the
right or left as they propagate toward the antenna base, but not that
there's a systematic circumferential current flow -- no current crosses
from radial to radial in a clockwise or counterclockwise circular
pattern like Richard implied.

I recall reading a paper which showed that connecting radials with
circumferential wires actually degrades a ground system's effectiveness,
but I wasn't able to lay my hand on it when I looked.

Roy Lewallen, W7EL