The discovery of intervening sequences (introns) in DNA led Gilbert and Tonegawa to suggest that a new protein could have been produced by bringing together certain segments of pre-existing ones. However, Blake argued that if DNA was so organized that coding sequences (exons) correspond to structural as well as functional units of proteins, then combinations would be much more likely to yield a stable globular conformation through being 'sums of parts'. In immunoglobulin heavy chain, four separate exons encode four different units, all with distinct functions and three of which have clear domain structures. However, in haemoglobin, which has no obvious domain structure, no clear conformational characteristics have so far been recognized for the segments encoded by exons. From a close inspection of their conformations by drawing various stereodiagrams and the Calpha-Calpha distance map, I now propose a conformational characterization of the segments as structural units.