Andrea Petric writes: HGEOM (and its fellows GEOM, LGEOM, PGEOM) all had a curious aspect to their I/O that caused them to read 1 too many rows per plane when the rows were long enough (512 was too short to cause the error) and a subimage was being read. It will be fixed as asoon as aips is unfrozen enough to do that. Thanks for reporting this Eric