Regarding some comments earlier about GPS being inaccurate on slopes, we had some discussion of this on another forum and after some dredging up of rusty maths skills found that you would have to be going up a very steep slope for the the error to be significant.
Good point as although I'd previously commented that a straight and level road would produce the greatest accuracy, I'm aware that calibrating GPS units against vehicle speedometers doesn't produce a signifcant difference, in practice, when going up or down hills. Without doing any maths, I'd been thinking that the change in subtended angle observed by an averaged-out, mulitple satellite fix is probably very small when compared with that obtained when the vehicle is travelling on a level road.