From Fig. 1 it may be seen that the effect of elevated temperature during the pyrexial period upon 1/K and therefore on the dissociation curve of oxyhemoglobin was, on the average, greater than would have been expected from experiments on normal blood in vitro, and greater than would be expected in view of the alkalosis occurring See PDF for Structure during fever. Temperature rise, and excess hydroxyl ion acting in vitro in the opposite directions, seemed to indicate a more stable state of affairs than was found. Apparently other factors have come into play, as, for example, alterations in the proportions and concentrations of the various electrolytes. In pneumonia, for instance, there is a retention of chloride during the febrile period with excessive loss of phosphates. The variations were not due to variations in the hemoglobin molecule itself since from the work of Adair, Barcroft, and Bock (18) hemoglobin must apparently be reckoned as having identical properties in normal individuals of the same species. If Barcroft's (19) hypothesis be right, namely that the CH within the corpuscle is higher than that of the plasma, the observed variations of 1/K may not be so surprising. In view of the fact that the hemoglobin inside the corpuscle is enclosed within a semipermeable membrane, the possibility arises of the setting up of membrane equilibria which will protect the respiratory pigment from excessive changes of reaction that may occur in the plasma, and thus the optimum conditions for the carriage of oxygen to the tissues may be maintained. Krogh and Leitch (10) in 1919 also drew attention to the protected situation of hemoglobin inside the corpuscle. In Case 6 it seems as if the alkalosis consequent on the febrile state had gained the upper hand and had extinguished the normal temperature reaction. This is rather confirmed by the fact that clinically the case showed one of the earlier signs of an alkalosis; namely, twitching of the facial muscles. Case 10, who had been on salicylate, also showed an analogous effect, when 6 days after the first observation the temperature shift was practically nil. The relationship between pH, 1/K, and the febrile temperature still awaits investigation. The extent of the shift of the dissociation curve was not by any means uniform; in neither Fig. 1 nor Fig. 2 is the highest value of 1/K at the highest temperature recorded. Fig. 2 shows the effect of temperature rise upon 1/K after cessation of the pyrexia; the effect is not so marked. Some cases, however, showed a variation in excess of the normal as if there was not yet complete return to normal. See PDF for Structure Biologically these changes are of importance in that this shifting of the dissociation curve to the right in fever means that there is more oxygen available for the tissues than normally, more especially at higher pressures. The tension of unloading is raised. This, in addition to the accelerated circulation and the probable increased velocity o[ reaction, means that even in a localized area of inflammation, if there is increased temperature, the tissues are placed in a better position for resisting infection as a result of their better oxygenation. That there is increased metabolism during fever has been conclusively shown by Du Bois (20) and others using large bed calorimeters. Du Bois has shown that the increase in metabolism obeys van't Hoff's law, increasing 13 per cent for each 1°C. rise. This shifting of the curve then falls into line with these observations as an adaptive response to the febrile condition, and the febrile temperature, if not too great, would seem to be a purposive attempt to aid the combating of infection. This shifting of the curve probably explains Uyeno's (21) observation on the effect of increased temperature on the circulation in the cat; namely, increased coefficient of utilization, and increased fall in the saturation of the mixed venous blood. Turning now to Table II, we find that, if the CO2 dissociation curve of Haldane (12) is accepted as normal, the bicarbonate reserve of five of the cases was above normal, of one normal, and of the rest below normal. Gastric secretion was not the cause of the varying curves, since the time of drawing the blood in all cases was during intestinal digestion. No observations having been made on the blood pH or the alveolar CO2, we cannot be absolutely certain as to the actual reactions, more especially in the last cases; Case 6, however, was probably one showing a partially compensated CO2 deficit in view of the absolute lowering of the total bicarbonate and evidence clinically of a tendency to alkalosis. Case 3, which had a lowered reserve, was probably similar. Koehler (22) gives a series of blood pH determinations in acute fevers in which ten out of twelve cases showed an uncompensated alkalosis when the temperature was 103°F. (39.4°C.) or over. Pemberton and Crouter (23) in a study on the response to the therapeutic application of external heat also observed a tendency for the reaction to shift to the alkaline side as shown by the alteration in the pH of the sweat. Hill and Flack (24) and Bazett and Haldane (25) observed that in thermal fever there was an excessive loss of CO2 comparable to the effect of hyperpnea. These facts are of importance in view of the above results regarding the bicarbonate reserve. That there was a definite alkalosis in some of the cases is at least shown by the value of 1/K at 37.0°C. during the pyrexial period in Cases 1, 2, and 7. The upper limits of 1/K at 37.0°C. both during pyrexia and after were similar. Of other factors that might be considered, reference may be made to the work of Barbour and his associates (26–29), who showed that in hyperthermia and fever there is an alteration in the concentration of the blood. But the changes were hardly of such magnitude as to cause the variations above detailed.