I am trying to do modelling using neutrons with this example.  I made a small
modification to ExN03DetectorConstruction.cc to add extra molecule types.  The
problem is this:  neutron paths don't seem to be correctly modeled.  For any
kind of absorber/gap material types, neutrons seem to go straight through.  As
far as I can tell, other particles work properly, e.g. alpha particles, protons,
muons, electrons, and positrons.
This occurred when I entered the commands in interactive mode:
/gun/particle neutron
/gun/energy 14 MeV
/run/beamOn 25
This gives straight lines in all cases where the gun particle that was used is
a neutron.  It really doesn't matter what absorber/gap material is used, even
for molecules specified by the original ExN03DetectorConstruction.cc.  I noticed
an interesting and obviously wrong behavior by setting the run & tracking output
to verbose.  The neutron's energy seems to never get initialized (i.e. it is at
zero before and after all the runs are executed).  Unfortunately, I don't know
enough about Geant's architecture yet to understand where the problem might be.
I can provide additional information, e.g. output from exampleN03 binary,
screenshots of output, etc.