the Barometric Formulce, <^c. 633 



and highest temperatures occurring during the day-time. It also appears by 

 inspection, that the very greatest error of calculated height takes place at one 

 hour p. M. of the month of July ; this error amounts to 31"14 metres, as may 

 be ascertained by an easy calculation with the data of the Table. The greatest 

 error, in defect, of the entire year takes place in December at five hours after 

 midnight, and amounts to 26'88 metres. The error in defect of the month of 

 July, which takes place from two to three hours after midnight, amounts to 

 16'7 metres, little more than one-half of the error in excess during the greatest 

 heat of the day of the same month. 



This Table also tells us that the horary corrections vary more between the 

 hours of any given month than between the corresponding hours of two con- 

 secutive months. It also shows us that the errors in excess are individually 

 greater than errors in defect, although the sum of errors in excess is less than sum 

 of errors in defect, which is evident from the fact that the height of the Con- 

 vent above the Observatory, as calculated from the mean often years' observations, 

 by night as by day, is less than true height by nearly 4 metres (vide Table I. 

 of this Paper, page 630). It also appears by inspection, that the errors are 

 greatest near 1 o'clock p. m. of summer months, consequently, such moments 

 are the most unfavourable for observations wherewith to calculate a constant ; 

 yet these are the moments chosen by Laplace in the calculation of his con- 

 stant. Doubtless, by careful inspection, other useful facts may be elicited from 

 this Table : one, however, of peculiar importance remains to be noticed, and 

 the attention of the Academy is particularly invited to it. 



Wliereas we find calculated heights to be in excess during day-time, and 

 in defect during night-time, there must be moments when the errors of such 

 calculated heights are reduced to zero. Such are, obviously, the moments 

 most favourable for barometric observations, and they are indicated by changes 

 of sign from plus to minus, or from minus to plus. 



By inspection of the large Table, I have ascertained the moments when the 

 horary corrections vanish, which are given in the small Table which I here 

 subjoin. Before I produce this small Table, I have to remark that the months 

 of January and November have but one such moment; that December has no 

 moment whatever when the horary correction = 0, all the calculated heights 

 for this month being in defect, as indicated by the fact that all the horary 



4 2 



