If the collagen triple helis is so built as to have one set of NH . . . 0 hydrogen bonds of the type N3H3(A) . . . O?(B), then it is possible to have a linkage between NIHI(B) and O,(A) through the intermediary of a water molecule with an oxygen 07 leading to the formation of the hydrogen bonds Nl(B) . . .O; and 0; . + .01(A). I n the same configuration, another water molecule with an oxygen 0; can link two carbonyl oxygens of chains A and B forming the hydrogen bonds 0; . . . 0 1 (A) and 0; -* . 00 (B). The two water oxygens also become receptors a t the same time for CH . . . 0 hydrogen bonds. Thus, the neighboring chains in the triple helix are held toogether by secondary valence bond linkages occurring regularly at intervals of about 3 A along the length of the protofibril. The additional water molecules occur on the periphery of the protofibril and will contribute their full share towards stabilizing the structure in the solid state. I n solution, they will be disturbed by the medium unless they are protected by long side groups. It appears that this type of two-bonded struct,ure, in which one NH . . . 0 bond is to it water molecule, can explain several observations on the stability and hydrogen exchange properties of collagen itself and related synthetic polypeptides. The nature of the water bonds and their strength me found to be better in the onebonded structure proposed from hladras than in the one having the coordinates of Rich and Crick.