Heart-muscle preps. were obtained, exhibiting strong cytochrome and succinic dehydrogenase activities and showing not only the usual [alpha] and [beta] bands of components a, b and c, but also the three Soret or [gamma] bands, lying at 448, 432 and 415 m[mu]. The prep. also showed a new component a3, the bands of which were fused with those of a. While the main portion of the y band (605 m[mu]) belonged to component a, most of the y band (448 m[mu]) belonged to component a3. The existence of a3 was not previously recognized because in the oxidized state its absorption bands were invisible and in the reduced state they coincided with the corresponding bands of component a. The absorption bands of the a component of cytochrome were modified by certain respiratory inhibitors. However, the absorption bands act and ay did not belong to component a but to a mixture of a and a3. One of the principal properties of a3 was its marked autoxidizability and in this respect it differed from other components of cytochrome. a3 was thermolabile, very sensitive to organic solvents, alkalies and acids, and in the reduced state combined with CO, forming a compound which showed absorption bands, a, 590 m[mu]. and [gamma], 432 m[mu]. In the oxidized and reduced states, it formed 2 compounds with KCN of which the latter was very easily autoxidizable but the former was not easily reduced. In the oxidized state, a3 combined reversibly with H2S, NaN3, NH20H and possibly with NaF and C2H50H. These properties supported the view identifying this component with cytochrome oxidase. The catalytic activity of this enzyme depended entirely on cooperation with components a, b and c of cytochrome. Component a3 was identified with the O2-transporting enzyme of Warburg and his associates. However, there still remained a few weak points in the theory identifying a3 with cytochrome oxidase. Outstanding among them was the failure to demonstrate a direct or indirect reduction of the oxidized component a3, by addition of reduced c under strictly anaerobic conditions and in complete absence of other reducing substances. The co-existence of a and a3, their proportionality, the identity of their haem nuclei, the great similarity in some properties of their protein suggest that these two components were intimately connected if not interconvertible.