In a previous report (Rahn et al., 1946) some effects of hyperventilation on performance test scores were described. In these experiments oxygen was breathed at a simulated altitude of 30,000 feet. The results could be related to changes in alveolar carbon dioxide tension alone, since the alveolar oxygen tension never fell below the normal ground level range. The experiments reported below constitute an attempt to relate performance to alveolar gas composition when both the oxygen and carbon dioxide tensions are varied simultaneously and when one or the other is varied singly. METHODS. All experiments were conducted in our high altitude chamber. Alveolar gas tensions were varied by 1, hyperventilation with the aid of the G.E.X. pneumolator; 2, by the addition of dead space (rubber hose) between mask and demand valve, and 3, by change of pressure in the high altitude chamber. Performance was measured bv the Hecht contrast discrimination test and by a modification of the hand steadiness test described in a previous report (Fenn et al., 1946). In the hand steadiness test, as originally described, the score w ‘as taken as the total number of contacts made by the subject during a 30 second test period. Preliminary experiments showed that this method of scoring may involve a serious error when the subject is very anoxic. Such a subject tends to allow the stylus to remain in contact with the plate for relatively long intervals of time during the test period.2 The number of hits, which tends to be relatively few in such a case, is a poor measure of the subject’s steadiness, because it is much easier for a subject to rest the stylus against the side of the hole than to hold it in the hole without touching. This problem has been solved by taking as the score, not the number of hits, but the total time during which the stylus makes contact in a period of 30 seconds. This time was measured in units of & second by an A.C. impulse counter included in a circuit that was closed whenever the stylus made contact with the plate. The general plan for an experiment was as follows. The subject entered the