In experiments designed to determine the position of the —S—S— bonds in insulin, cystine peptides were separated from a partial acid hydrolysate, and their structures determined after oxidation to the corresponding cysteic acid peptides. It was found that many more cystine peptides were present than could be accounted for by any one unique structure for the insulin molecule, and it seemed likely that a transfer reaction of the type: had taken place during the hydrolytic treatment, leading to a random rearrangement of the —S—S— bonds. That such a reaction can take place was illustrated by the following model experiment, in which mono-2 : 4-dinitrophenyl-cystine is formed from a mixture of cystine and bis-2 : 4-dinitrophenyl-cystine.