452 



SCIENCE 



[Vol. LVr, No. 1451 





weiglits, M, at tJie same hour, on successive 

 days. As the graphs are often quite divergent, 

 the interpolations will lose in accuracy ; but the 

 general relations of the results will neverthe- 

 less appear much more clearly. These static 

 deflections, Aj/, are given in the lapse of time 

 in the figure. For 1922 the graphs are drawn 

 for l'"", 3'"', 8h P.M. of the successive days,- and 

 are distinguished by circles or crosses. For 

 1921 the night observations (at about 8 P.M. 

 on the average) only are given, as the other 

 lines would lie too close and confuse the dia- 

 gram. In fact the variations in 1921 are of a 

 smaller order and must be given on a scale ten 

 times larger to be adequately shown. 



The diagram brings out the striking differ- 

 ence of the results very well. For 1921 the 

 observations lie practically on a straight line, 

 A^/ = 13.42, for which the normal period of the 

 needle in vacuo would be 752 see. In the re- 

 sults for 1922 the :time of the successive exhaus- 

 tions (Ex) is indicated approximately. It will 

 be seen that the cooling or other influence of 

 such an exhaustion (though carried from 

 1 mm. to .001 mm. only) is still effective in 

 exaggerating the radiant forces, for at least 

 six hours or more (c/. July 24, 30) after the 

 exhaustion has been completed. Consequently 



" S denotes sunshine, G cloudy, C partly cloudy, 

 B rpjn. Vac. shows the vncuum in mm. of mer- 



the graphs for V^ and 3^ should probably be 

 joined by the dotted lines as indicated. 



In all eases the extraneous radiant disturb- 

 ance which is strong in July, 1922, gradually 

 recedes more and more, as the observations 

 enter the days in August. On July 24 at 

 8 P.M. the combined gravitation and radiant 

 effect of the attracting mass M. was repulsive 

 (At/ negative), the radiamt repulsion being 

 about twice the gravitational pull. Positive 

 values are not reached until after July 26. 

 From July 28 on, the 8 P.M. increase is deter- 

 mined, though it has not quite reached the 

 values of Aj/ even at the end of the diagram 

 (August 14). In the afternoon observations 

 (1922) the rain effect (or the absence of sun 

 effect) is brought out very clearly by tthe 

 marked depressions on August 2, 8, 11, 12. At 

 night this effect may be reversed. When the 

 day's radiation is scantily received, the needle 

 fails to radiate at night. 



In case of the observations of 1921, the 

 small fluctuations of the Aj/ curves throughout 

 a month showed instances of resemblance to 

 the run of atmospheric temperature. But in 

 the large variations recorded in 1922 (as a 

 consequence perhaps) I was unable to detect 

 such resemblances in the night observations, 

 which are here alone of interest. The same is 

 true of the change of temperature per day, 

 etc. Nevertheless it is possible that relatively 



