10 



GOULD — REDUCTION OF D'AGELET S OBSERVATIONS. 



Data for computing refractions. 



Inasmuch as the thermometer was generally read to even degrees only, never more 

 closely than to half degrees, and the barometer was seldom read more closely than to half 

 lines, errors of half a degree and half a line may easily be apprehended, corresponding to an 

 uncertainty of about 0.001G5 in log TB. The aggregates of the nicer corrections, b, I', &c, 

 are under no circumstances likely to attain so large a value as 0.0023 p, and are so much less 

 than those due to the errors of reading the meteorological instruments that they have been 

 disregarded. 



From the data already given, and a study of the tables of meteorological means for Paris, 

 deduced from observations in more recent years, the following table has been empirically 

 constructed and used for computing all the refractions. It presents, to three decimal places, 

 those values fur the meteorological factor which seem most probable, after taking into account 

 the estimated diurnal changes of temperature and probable fluctuations of the barometer. 

 (10) 



