236 



SCIENCE 



[N. S. Vol. XXXIII. No. 841 



Table I ' 



The large discrepancies existing between Ever- 

 sheim's determinations in the helium, cadmium 

 and mercury spectra* and the earlier results of 

 Eayleigh, Michelson, and Fabry and Perot were 

 mentioned. 



As conditions of apparent significance in con- 

 nection with conclusions 2 and 3 above, the two 

 following facts were emphasized: 



1. The line of division between the group of 

 wave-lengths considered in (2) and that consid- 

 ered in (3) is sensibly coincident with the green 

 cadmium reference wave-length. 



2. Fabry and Buisson's and Pfund's correction 

 curves for " dispersion of phase " cross each other 

 at about this same wave-length. 



As a tentative hypothesis to account for con- 

 clusion 4 above, the speaker suggested an insuffi- 

 cient approximation in computing. In support of 

 this hypothesis, he stated that he had recomputed 

 from the published data, the results under II., 

 p. 836, Vol. 30, Ann. der Physik (1909), and 

 obtained values systematically lower than the 

 ones there published. These recomputed results 



° Data from same sources as mentioned in foot- 

 note 1. 



*Zs. filr wiss. photog., 8, 148, March, 1910. 



were obtained on a ten-place computing machine 

 and so involved no approximation in computation. 

 Computation by seven-place logarithms gave re- 

 sults systematically high, while computation by 

 eight-place logs checlied the machine results. It 

 happens that the errors of the seven-place table 

 are additive instead of compensating in this case, 

 so that the error in the final result may amount 

 to + 0.002A. Tliere is a possibility of the ap- 

 proximate logarithmic computation introducing a 

 systematic error owing to the fact that all wave- 

 lengths are referred ultimately either to the 

 green or red cadmium wave-lengths as standards; 

 and to the fact that the values of K (see p. 835, 

 Vol. 30, Ann. der Phys.) nearly enough equal to 

 fall at the same point in the log table may be 

 expected to frequently occur. As to the impor- 

 tance of this latter condition, nothing can be said 

 without consulting the original data. It seems 

 possible, however, that this condition if it occurs 

 often enough, in connection with the error due to 

 approximation in log X, may cause a discrepancy 

 about large enough to account for the observed 

 discrepancy between the results of Eversheim and 

 the other investigators. 

 Ocean Currents and Barometric Highs and Lows: 



Dr. W. J. Humphreys, of the United States 



Weather Bureau. 



In the first part of the paper the speaker dealt 

 with the five barometric highs on the oceans 

 which remain substantially fixed in position 

 throughout the year though varying in intensity, 

 three of which are in the southern, and two in 

 the northern hemisphere. In the second part of 

 the paper the speaker discussed the Aleutian and 

 the Icelandic regions of low barometric pressure. 



A brief review was given of the explanations 

 advanced by past investigators to account for the 

 existence and character of these regions of high 

 and low barometric pressure, none of which ap- 

 peared adequate to account for all of the observed 

 facts. 



Lantern slides were exhibited showing the iso- 

 bars, isotherms and ocean currents over the ocean 

 areas, and the relation of these to the existing 

 high- and low-pressure areas was discussed at 

 some length, the purpose of the paper being to 

 show the physical reasons for the existence of 

 these highs and lows and to explain why they are 

 where they are actually observed to be. (This 

 paper will appear in full in an early number of 

 the Bulletin of the U. S. Weather Bureau.) 



R. L. Fabis, 

 Secretary 



