[Proj] Datum shift for Dutch maps 1850-2000

[Jan Hartmann]

  This is not a question about the program PROJ, but perhaps there are 
people on this list that can help me. I am georeferencing the first 
topographical map of the Netherlands from about 1850. That map is in a 
Bonne projection centered at  51.5N / 4.8838828E,  and uses  its own 
ellipsoid, computed by the Dutch engineer Ackermans (semi-major axis 
6376950.4m, inverse flattening 309.65m). Its PROJ-string is:

+proj=bonne +lat_1=51.5 +lon_0=4.8838828  +a=6376950.4 +rf=309.65 
+units=m +no_defs

The present Dutch topograpical maps use a stereographic projection on 
the Bessel ellipsoid, defined (without the WGS84 datum shift) as:

+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 
+k=0.999908 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +no_defs


We have the complete lists of triangulation points used for both  maps, 
1000  in 1850 and  5000  in 2006.  Quite a few points are identical, 
mostly church towers, so for these we have coordinates in both 
projections. The  lists  also contain  the lat-lon values for these 
points.  When I use  cs2cs to  reproject  the old points  into the 
present system, I get a difference of about 650 meter, almost 
exclusively in the Y-direction: the reprojected 1850 map lies about 650 
meter to the north of the present map. This different is exactly the 
same over the whole country, give or take a few meters that look random 
to me (could be measurement errors, plus I am not sure they used the 
same spot on the church towers in 1850 and the present).

This means I can get a reasonably accurate historical map by just doing 
a linear shift of -10/-650 meters, but I am curious what happens here. 
Of course the center of projection in 1850 could be actually 650m away 
from its recorded position, but I am wondering if there are other 
possibilities for such a shift. Am I doing something slightly wrong? Any 
help would be appreciated.

Jan

Dr. J. Hartmann
Department of Geography
University of Amsterdam