22 Meteorological Observations on Parisnath Hill. [No. 1. 



perature solely depends on the quantity Log w, all other terms of the 



equation being constant quantities for all hours. An error in the 

 temperature would therefore indicate an error in one of the baro- 

 metrical readings, or in both. Having corrected the temperature, we 

 shall be enabled to find the magnitude of the error in the barometri- 

 cal reading at the one station, provided that of the other can be 

 relied on. 



The error to which barometrical observations are liable,is composed 

 of two parts — the unavoidable error, which is attached to every read- 

 ing and depends on the imperfection of the instruments and of the 

 eye of the observer, and the incidental error, caused by unfavourable 

 external circumstances attending the observation, as for instance 

 shaking of the instrument, a bad light, unsteadiness of the eye from 

 muscular exertion, &c. In converting the readings of one instru- 

 ment into those of another with which it had been compared before- 

 hand, the unavoidable error is included in the small allowance which 

 must be made in adding the correction. In the present case this 

 allowance amounts to + 0.003 inches. The incidental error cannot 

 well be estimated. It must be more or less irregular and may be 

 greater or smaller than the unavoidable error ; generally greater. 



In cases like the present, wishing to correct the errors as far as 

 possible, we must be satisfied to assume the readings of one of the 

 stations as correct and alter those of the other, being guided in our 

 selection by the circumstances of the case. Such a proceeding can 

 not lead wrong so long as the corrections do not exceed the proba- 

 ble limits of the errors of observation. Should they appear to do 

 so, this would prove that in the first instance the correction of the 

 temperatures must have been erroneous. So we have in these 

 barometrical corrections a means of controlling to some extent the 

 alterations made in the curve of real mean temperatures. In our 

 case, the readings of the Barometer at Calcutta, the instrument 

 being a standard, fixed in its position and never removed from its 

 place, necessarily had been less exposed to incidental errors than 

 those of the mountain barometer, which was frequently changed 

 from one place to another and could not be suspended so as to avoid 

 oscillation. 



