Hello, Tim, Dan and I have just had a chat and the obvious thing I would say is to stick with 2.1.3 if you still have it. We'll hopefully get to 2.1.5 pretty damn soon. There is a flaw in the update code in 2.1.3 which means that for source code (not pre-compiled) updates it is not setting the symbolic link for the Bayes code in 2.1.4. That can be fixed *after* getting updated to 2.1.4 but only by doing the "make links" mentioned in the previous email. An alternative is to fix the update code in 2.1.3. The relevant file is ccpnmr/ccpnmr2.1/python/ccpnmr/update/UpdateAgent.py. The line number 758 says: for directory in ('memops/c', 'ccp/c', 'ccpnmr/c'): and should say: for directory in ('memops/c', 'ccp/c', 'ccpnmr/c', 'cambridge/c'): I've attached that file as well. I've put this fix on the 2.1.4 update server and if 2.1.4 source code people do not have this patch then they in turn will have a problem when going to 2.1.5. Wayne