vector vs raster statistics

I would appreciate if some one can address why I am getting a huge
difference in statistics between a raster and a vector map.

The vector map is a streams map and I ran through v.report and found
24 Million meters of length. Then I converted the vector to raster
at 100x100m grid and ran the r.report to find how many cells are there
in that stream, and found 0.12 Million meters and then if I multiply
0.12 with 100 meter side of the grid I get only 12 Million meter length
stream, even if I multiply by the sqrt(2)*100*0.12, I get only 17 Million meters.

I looked at the v.report code and found the program is using

hypot - Euclidean distance routine

#include <math.h>

     double hypot(x, y)
     double x, y;

DESCRIPTION
     hypot() returns

          sqrt(x*x + y*y)

to find the distance between two points.

Can any one have similar experience? I streams are not much complex, these
are pernnieal streams.

Thanks in advance.

Srinivasan
srin@brcsun0.tamu.edu