thats a question ive been keen on finding out the answer to as well.

thing is, with current heatpipe technology, i.e. something from Artic Cooling for example... i have yet to see different temperature scores from both vertical or horizontal orientations. The scores to vary, but by 1-2 degrees celcius only, just enough to fall within the acceptable margin of error.

probably because the heatpipes are used are too small, and too short. Maybe with something thicker and longer, we will see gravity taking effect?