This paper concerns the use of the l(p)-norm for estimating travel dis
tances in a transportation network. In previous empirical studies, val
ues of the parameter p > 2 have been employed to characterize road net
works. After a detailed study of the directional bias of the l(p)-norm
, it is shown here that such values need never be considered, since th
e same degree of accuracy is obtained with a value of p in the interva
l (1, 2) after rotating the axes 45-degrees.