OF CLOCK RATES UPON LONGITUDE WORK. 129 



variations of latitude. But any good clock can be relied upon during 

 a period of one day to test a diurnal variation. 



If there is a physical cause for a diurnal variation in our observed 

 results, the best clocks will give the best defined variations. 



Since latitude observations with the zenith telescope have been 

 confined to the night hours, we can not expect much contribution to 

 the solution of a diurnal term from that delicate differential instru- 

 ment. If the maxima of longitude variation occur at sunset and 

 sunrise, the maxima of latitude variation should come at noon and 

 midnight. 



The current observations at the international zenith telescope sta- 

 tions are made in two groups, at nearly equal intervals each side of 

 midnight. 



The closing error of the groups, which is about o".2 distributed 

 among twelve periods, might be due to a variation with a daily maxi- 

 mum that does not fall exactly half way between the two daily 

 groups. 



At your neighboring institution, the Flower Observatory of the 

 University of Pennsylvania, a distinct difference in latitude results 

 was derived by Prof. C. L. Doolittle between early and late hours of 

 the night. This diff'erence could be attributed to an error in the 

 adopted constant of aberration, and a correction (o".o8) was com- 

 puted by that most thorough and capable observer. It will be recalled 

 that the zenith telescope observations have pretty uniformly given 

 larger values for the constant of aberration than those derived from 

 other sources. With the value of the constant, 20". 47, only one 

 quarter of the difference derived by Prof. Doolittle would be ac- 

 counted for. 



Observations with the prime vertical instrument by M. Jean Boc- 

 cardi, at the Turin Observatory, in 1920, were designed to show a 

 differential effect in latitude results during the night hours. 



By comparing the observed differences between stars separated 

 about three hours in interval, through a cycle of nearly one year, he 

 derived a cosine term with an average coefficient of o".07. 



The extensive series of prime vertical observations of a Lyrce at 

 the U. S. Naval Observatory, seventeen hundred observations in 



