r.slope.aspect

[martijn at meteoor.frw.uva.nl]

Recently I wanted to find out what the relationship between archaeological
sites and aspect was in an area of the southern Netherlands. So I ran a
report on site occurrence. I noticed that r.slope.aspect seems to prefer
the main aspects (E, NE, N, etc) to those aspect cats lying in between:
in fact they are just about twice as popular. In my area I have 33% of 
cells with no aspect, 34% with a main aspect and 33% with a minor aspect.
Note that there are 8 major aspects and 16 minor aspects.

So, if there ain't something wrong with my area, what's wrong with the aspect
calculation? Did anyone else have comparable results?
Solving this one would be a great help in my research, because I can't run
a correlation test involving aspect as long as that map layer can't be 
trusted.

Martijn van Leusen