There is an inconsistency between the examples and the struct_sheet_hbond definitions. Look at Example 1, the simple beta barrel. 11 N is bonded to 30 O and 19 O is bonded to 40 N in the diagram. However in the mmCIF version of the example, 11 N is identified as range_1_beg and 30 O as range_2_beg, which seems fine, until you read the text of the definitions, which takes the range_1_beg items as "one partner of the first hydrogen bond" and the range_2_beg items as "one partner of the last hydrogen bond" I would suggest changing the definitions rather than the examples, as follows: For the range_1_beg items: ... one partner of the first hydrogen bond ... For the range_1_end items: ... one partner of the last hydrogen bond ... For the range_2_beg items: ... one partner of the first hydrogen bond ... For the range_2_end items: ... one partner of the last hydrogen bond ... And add a general comment on the meaning of "first" and "last", which is tricky but important. -- H. J. Bernstein P.S. Life is simpler if you follow the PDB approach of just giving one representative hydrogen bond for registration, in which case worrying about "first" and "last" is a non-problem. This simplification could be achieved by making the range_*_end items implicit rather than mandatory, and allowing "first or representative" for the range_*_beg items.