hazy, with but little wind. On Aug. 16th, the set was slightly better and no observa- 

 tions were marked bad ; temperature 40 F., slight wind at times. On Aug. 17th, the 

 whole set is marked very bad ; temperature 35 F., sky very hazy, very marked 

 temperature changes alternately depositing crystals on the lenses and removing them 

 again, marked refraction effects in the direction of the fixed mark. Owing to the haze 

 over the sky, the stars were too faint for good work. 



On Aug. 18th, ^=40 F., sights marked good, little wind, clear sky. 



In the above Table it will be noted that a number of star observations (in brackets) 

 have been neglected, and to decide which ones were to be thrown out Chauvenet's 

 Criterion was applied. It seems doubtful, however, if this method is quite allowable in 

 observations such as those of Aug. 15th. Here, by successive applications of the method, 

 three stars are thrown out, leaving only six sights from which a value of the rate is 

 to be calculated. 



The actual rate taken for this series is not the rate shown by these observations, 

 but one arrived at in the following way. The observations of the 18th, appearing to 

 be the best, are taken as correct ; and the mean of the rates deduced from it and from 

 the observations of the 15th and of the 16th is applied to the whole period. 



(1) Aug. 15th Di = 57-13sec. 



(2) 16th 56-37 



(3) 17th (neglected) 



(4) 18th 55-96 sec. 



From (1) and (4), the daily rate of " S.C. " is 0-39 sec. gaining, and from (2) and 

 (4) 0-20 sec. gaining. The mean 0-30 sec. gaining is therefore applied over the whole 

 period. (A) 



The prime reason for assuming a constant rate for " S.C." lies in the comparisons 

 with mean chronometer " E." 



Thus " S.C." gained on " E " from the 15th to the 16th. . 237-03 sec. 1 



16th ,, 17th.. 237-01 1 mean 237 -00 sec. 



17th 18th.. 236-96 



This assumption seems a fair one, as it is exceedingly unlikely that the two chrono- 

 meters would both be erratic to the same extent. 



During August, " E " was going very well, observations by Commander Evans 

 giving from 



July 27th to Aug. 2nd " E " 0-071 sec. losing daily. 



Aug. 2nd to Aug. 15th 0-144 



Aug. 15th to Aug. 29th 0-144 



42 



