OF CLOCK RATES UPON LONGITUDE WORK. 127 



Changes of temperature in our well-protected clock cases have 

 had no sensible effect upon daily clock rates for short periods, such 

 as one day, with which we are here concerned. 



Ordinarily the range is less than one degree, and nearly always it 

 is progressive, so that the temperature at night falls between those of 

 successive days. 



The variation of atmospheric pressure, as recorded by the barom- 

 eter, affects the rate of a pendulum clock not hermetically sealed. 

 For our Riefler clock, installed in 1907, a change of one inch in 

 atmospheric pressure changes the daily rate 0^.46. 



In our fine summer weather the average barometer reading at 

 midnight is 0.04 inch below that at noon, and the lowest reading com- 

 monly occurs in the early morning, following midnight. In fine 

 weather in winter the reading at midnight is 0.0 1 inch below that at 

 noon, and the lowest readings occur in the afternoon. 



The summation of the hourly excess at night has been divided 

 into two periods, corresponding to the use of the Dent clock, unsealed 

 up to 1907, and the use of the Riefler, following that date. 



The first period gives a mean hourly excess of o^oo7, and the 

 second period gives 0^.004. The difference between the two results 

 and the mean results are too large to be accounted for b}' the variation 

 of the atmospheric pressure. 



The observation of clock corrections and rates during the night 

 hours should be uninfluenced by any possible deviation of the meridian 

 plane, due to barometric or thermal gradients in the atmosphere, such 

 as might be suspected at sunset and sunrise. 



The effect of such gradients has been found to be very small, even 

 at these epochs of greatest disturbance. 



Excess of Hourly Rates at Night. 



