I found that the HOSE codes generated are wrong in some cases. I would expect it affects quite a few actually. The problem occurs if there is more than one atom of an element on a sphere, the atoms attached to these atoms are randomly ordered in the next sphere, not (as it should be) per atom in lower sphere. In some cases of course it could be right be accident. A patch with a test is below. I know what's going wrong in the code, if somebody wants to fix it I can give some hints.

patch.txt