In the case of a circle, you should still have a radiated signal that arrives before the conducted signal. But it would be far weaker, further compounding the problems with the "light bulb illuminating" way of illustrating that a signal has arrived. In the original experimental design, it was already reckless at best to leave the viewer with the impression that the radiated signal was capable of powering the light bulb rather than merely serving as a trigger for a self-powered light. With a circular setup, I doubt that equipment sensitive enough to detect the arriving radiated signal (and reliably distinguish it from other electromagnetic noise) could even fit into the light bulb's black box.
