One point of clarification is that my suggestion of adding long buried radials to improve the performance of a ground mounted vertical, only pertains to a ground plane antenna. Not the GM.
Marconi, your EZNEC results are shocking! At first glance your ground mounted GM did exactly what I thought it would do. The pattern is horrible with a peak gain at 45 degrees. Too sharp of an angle for DX and useless for line of sight.
I was stunned to see EZNEC report a significant change at only 6 inches above ground. The angle of radiation here at 17 degrees may be ideal for much DX work! From what I see here, the idea looks plausible. The one thing that concerns me is that it only takes 6 inches to make this difference in the model. It is such a small change that I suspect it will vary depending on soil conditions.
I agree SW.
I was surprised with the patterns too. The 6" raised model suggest to me that Penn's idea for DX might work just as he notes, even though the gain looks to suffer badly. With that said, I've made some very good long distance contacts with very low watts out, and when conditions were working for me. DX is just unpredictable and depends a lot on how the antenna propagates...with waves combining somehow or not.
If the ground mounted model is even close to accurate, it would appear like the bottom lobe is out of phase with the top, but far worse than a regular 5/8 wave ground plane. IMO this might account for the very low gain we see reported by Eznec. SW, wouldn't that too be sorta' predictable also?