[Proj] Accuracy Cassini Solder -> lat lon 200m?!

[Sebastian Schubert]

Dear,

I want to convert from Solder system used in the city of Berlin 
(EPSG:3068) to the standard lat lon system.

Here is what I did:

schubert at linux-8ubn:~> cs2cs -v +init=epsg:3068
# ---- From Coordinate System ----
#Cassini
#       Cyl, Sph&Ell
# +init=epsg:3068 +proj=cass +lat_0=52.41864827777778
# +lon_0=13.62720366666667 +x_0=40000 +y_0=10000 +ellps=bessel
# +datum=potsdam +units=m +no_defs +towgs84=606.0,23.0,413.0
#--- following specified but NOT used
# +ellps=bessel
# ---- To Coordinate System ----
#Lat/long (Geodetic alias)
#
# +proj=latlong +datum=potsdam +ellps=bessel +towgs84=606.0,23.0,413.0

20281.021 19760.743
13d20'12.334"E  52d30'21.666"N 0.000

20302.749 19797.597
13d20'13.478"E  52d30'22.861"N 0.000

(I entered some newlines for a better overview). These are virtually the 
same results I get when I use the conversion formula given on the epsg 
website with the parameter listed there.

The problem is that the result is about 200m away from the place it 
represents! I had a look in google earth and on a couple of different 
other sites:
http://stable.toolserver.org/geohack/geohack.php?params=52_30_21_N_13_20_12_E
The data represents the Gedaechtniskirche in Berlin,  52°30'17.33"N, 
13°20'6.49"E using google earth. It is also depicted in this screen 
shot:
http://img443.imageshack.us/img443/6214/74483355.jpg

Do you know where the problem is? In the conversion, the data or the 
parameter?

Cheers
Sebastian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url : http://lists.maptools.org/pipermail/proj/attachments/20090703/09118991/attachment.pgp