128 TUCKER— EFFECT OF DIURNAL VARIATION 



The weights have been assigned according to the number of nights per 

 year, and the number of stars per night. 



Small undetected progressive changes in the position of the in- 

 strument would be represented in the computed clock corrections and 

 rates. The changes in instrumental corrections were usually meas- 

 ured over a period of at least four hours. 



It is hardly credible that uniform progressive changes would per- 

 sist undetected in this long term of years if they were of sufficient 

 size to account for the hourly rates as observed. 



To explain the sunset and sunrise results, systematic differences 

 would need to be of a decided character. Instrumental corrections 

 and the indications given by the mire readings have had careful 

 scrutiny in this connection. 



Physical or mental fatigue might be presumed to affect the per- 

 sonal equation of the observer, and thus influence the computed clock 

 rate. The effect would more probably produce erratic results, with 

 larger accidental errors of observation. 



The reaction times at sunset and sunrise would necessarily be of 

 quite different character, also, from those at night to make plausible 

 this explanation. Our current series of observations will give a test 

 of such a possible effect, as we shall have mean hourly excesses during 

 periods of six consecutive hours, in each of which the systematic 

 errors of right ascension will have been eliminated. 



A diurnal term in longitude, similar in character to the fourteen- 

 month variation, would produce a diurnal periodic variation in clock 

 corrections and rates, as observed. If the maxima occur at sunset 

 and sunrise, the most rapid changes would occur at noon and mid- 

 night. 



Clocks do not run over long periods of time with the uniformity 

 requisite to test the fourteen-month term, and we derive the long- 

 term variations in the longitude from the corresponding observed 



