It has not often been possible to consider the relation between the composition of proteins and their behavior. This has been due, in part, to the complexity of this class of substances; a complexity often as bewildering as the organizations and symmetries of the biological systems which proteins constitute, and always more intricate than the stereochemistry of the simpler organic compounds of which proteins are, in turn, composed. The large molecular weights of the proteins, and the large number of amino acids that combine with each other to produce these vast molecules give abundant evidence of their structural complexity. The failure to correlate the behavior of proteins with their composition has probably depended less upon this complexity, however, than upon incomplete knowledge of the amino acids they contain.Recent advances in the methods of isolating and of estimating amino acids have greatly increased our knowledge of the composition of certain proteins. Thus the composition of casein is now 94.2 per cent 1 known (1). As a result of Dakin's new butyl alcohol method 1 The convention has been followed of referring to the ratio of the weight of the amino acid recovered, to the weight of the protein analyzed as the percentage composition of the protein. The water absorbed by the amino acids during hydrolysis should, of course, be added to the denominator, in order to yield the true percentage composition.