( 244 ) 



A temperature gradient in \ertical direction of 1 centigrade per 

 meter causes, rougiily calculated, a variation in rate of ()^25. 



tliermometers (upper — lower) of one cenligrade corresponds willi a relardalion of 

 0\30 a day. 



This difference of lemporalure, apart from variations in connection with mete- 

 orological conditions, clearly shows a yearly periodicity ; in winter it is small, in 

 summer on some days it increases to above 0^.50 G. It can hardly be doubted 

 that, at least to begin with December 1898, the yearly periodicity must be ascribed 

 to the temperature gradient; for it is since that dale that the standard clock has 

 been placed in the vestibule in a niche cut out from the pier of the 10-inches 

 refractor. Since about March IS'JO the place on the Eastern pier in the transit 

 room where the standard clock hung during the former periods ISüï!— '74 and 

 1877— '98, treated by E. F. van de Sande BakhuvzeiN, has been occupied by the 

 clock Hohvvii 46 with a Rienier-pendulum. 



in order to obtain accurate data ai)out the temperature gradient also for this 

 place, two thermometers, which had been compared with each other and were 

 graduated to tenth parts of a centigrade were suspended in the case of Holnvii 46 

 at a vertical distance of (55 cms. on i*'ebruary 27, 1903. It appeared that here the 

 differences in temperature were in general greater than in the pier-niche, and in 

 .hily 1903 it was derived from the observations that a difference of temperature 

 between the thermometers of 1 centigrade corresponded with a variation of 0^.40 

 in the daily rate of this clock. 



A yearly periodicity in the temperature gradient appeared also distinctly in 

 Holiwii 46, as will be seen from the fo'lowing monthly means, which are given 

 by the side of those of Holiwil 17. 



lloliwii 17. Hohwii 46. 



1903 March 4-0^14 +0\25 



„ April + 0.09 + 0.16 



„ May + 0.31 + 0.42 



„ June + 0.28 + 0.37 



, July + 0.29 + 0.32 



„ August + 0.18 + 0.22 



„ September +0.16 + 0.26 



„ October + 0.06 +0.12 



„ November + 0.02 + 0.10 



„ December 4 0.01 + 0.03 



1904 January — 0.02 + 0.08 

 „ February + 0.05 +0.12 



Moreover in the temperature gradient in the clockcase in the transit room a 

 daily inequality was observed, which was hardly perceptible in the niche. 



In the case of Hohwii 46 the mean values of the temperature gradient from 

 0'' to 12'' mean time, are regularly greater than the mean values from 12*>to24'> 

 moan time. From the differences between these mean values in connection with the 

 differences in the relative rates of the two clocks for the corresponding half days 

 I could derive the influence of a variation in the temperature gradient on tlie 

 rate of Hohwii 46 for a shorter period. 



The investigation is not yet huished, but in connection with the theoretical 



