I have done a comparison of vertical vs horizontal loop but I could not do it using the full wave 40m loop because of size constraints, so I used 20m loops.
Both loops were triangular in shape. The horizontal loop was 7.5m at the highest point. One of the corner poles was 6m high, the other 4.5m high. It was fed at the highest with 450ohm "window line" down to the smartuner at the base of the pole and then with coax back to the shack.
The vertical loop was 7.5m high at the apex, 5m high at one end and 6m at the other. and connected directly to the smartuner in the middle of the lowest side. An obvious problem with this configuration was that the low side was only 1m above an iron roof for most of its lenght.
The results were as follows:
40m: No obvious difference in signal strenght in either receive or transmit but the vertical loop was noticably noisier.
30m: As for 40.
20m: Signals 1-2 S points stronger on vertical loop but it was noisier, so no real advantage.
17m: Not enough signals heard for comparison.
15m: As for 20m.
12m: Vertical loop would not tune on this band. Weird - must be some sort of coupling effect with the other antenna, or maybe the roof.
10m: Signals 2-3 S points higher with vertical loop but it was noisier.
6m: Both loops worked equally well.
So in summary, although there was a slight increase in signal strenght with the vertical loop the increased noise negated the advantage. The results might have been different if the vertical loop had been fed at a corner as suggested by Felix VK4FUQ in his article on loops in AR magazine but that was not possible at this QTH. The other possible confounding factor was the difference in radiation pattern between the two loops.
I suggest that a horizontal loop, if possible fed by 450ohm line and auto-tuner, is a good antenna for space and height challenged situations.
73
Wayne VK4WDM