28 Meteorological Observations on ParisnatJi Hill. [No. 1. 



I will now assume the value found for each of these quantities 

 to be nearly correct and will consider the magnitude of the error 

 which would be caused by smaller deviations from the truth in each. 



Iu doing this, an error in the height cannot be considered sepa- 

 rately as it would alter at the same time the values of t and b. A 

 variation in the height would change the value of the real mean tem- 

 perature in the same direction, but in a greater proportion. 



The influence of this error upon the value of H , or the 



a -j- t 



volume of the air, reduced to 0° C, would in the present case be 



very slight. The alteration in A for + 10 feet would be Hh 0.1. 



The value of the mean pressure cannot be obtained nearer than 



within about ± 0.005 inches. It would generally, the pressures 



remaining unaltered, for a greater height become a little smaller and 



for a smaller height greater and thus partly compensate the error in 



the temperature, the difference only of both appearing in the final 



result. In the present case, the effect of both, for a difference of + 



10 feet would cause a variation in the value of S, amounting to 



about ± 0.002 inches or less. An error in the barometrical readings 



at either station would tend to produce similar irregularities in the 



computed hourly variations of the weight, as have been found in the 



observed barometrical differences. 



If my assumptions regarding the correctness of the values which 

 have been found for H., t and b are right, the foregoing considera- 

 tions would lead us to expect, that the computed barometrical dif- 

 ference would not differ from those observed by more than a few 

 thousandths of an inch. 



This result will however be modified by the presence of moisture 

 in the air, which I have hitherto neglected, and which would alter 

 the specific gravity of the air and therefore the value of p. 



The computed values of the barometrical differences are given in 

 Table VII. Column III. Corresponding as they do to the weight of 

 a column of dry air, they are much higher than the observed weights 

 in Column II. The deviations between the computed and observed 

 values of S show very little variation, ranging between 0.028 and 

 0.031 inches. In every other respect the computed values show 

 exactly the same course as those observed. "We shall see how far 

 their deviation may be attributed to the moisture in the air. 



