168 M, Ramo)id*s Instructiom for the Application of [Sept. 



instrument for an exactness of which the observation itself is not 

 susceptible. The uncertainty is in the thing itself, not in the indi- 

 cation of it. You have perhaps just observed the thermometer, 

 and noted its indication : observe it again ; it has varied : 

 observe it again, it rises ; it falls ; and however small the range 

 of these variations may be, that which was at first considered 

 certain has become in a great measure problematical : you know 

 not what to think for certain of the temperature of the air. 

 There are some cases, however, where the choice is pointed out, 

 whether by the nature of the place where we observe, or by 

 that of the circumstances which manifestly act upon the ther- 

 mometer. If it rise during short intervals of sunshine, we may 

 lav this to the charge of reverberation. If it sink during gusts 

 of mist, it is the temperature of the meteor which produces the 

 change ; but frequently also the change is owing to causes of a 

 more general nature ; as to the encounter of two currents of air 

 of different temperature. To attain the greatest exactness we 

 can, we should prolong the observation for some minutes in order 

 to judge of the changes to which the instrument may be 

 exposed ; to inquire into the cause of its apparently capricious 

 movements ; and in a case of uncertainty, to take the mean 

 between its extreme variations. 



Si/stem of Observations. 



To find the real mean barometric pressure is a matter of more 

 diflBculty than is commonly imagined. The mercury has in fact 

 two species of oscillations, essentially different, although 

 frequently confounded, from the effects of the agitations of the 

 atmosphere. One sort is periodical and regular, the other, 

 accidental and irregular. The idea of a barometrical mean 

 necessarily imports that of a complete compensation between the 

 one and the other. 



If there were none but periodic variations, the mode of pro- 

 ceeding would be equally expeditious and simple. We should 

 in this case soon settle the length of the periods, and the epochs 

 of their recurrence. We should merely have to observe the 

 barometer for some days at the particular hour of these periods, 

 and to take a mean between the different heights of the mer- 

 cury. This is actually the case within the tropics. There the 

 accidental variations are almost reduced to nothing, and the 

 periodical are very evident and regular. The barometer is at its 

 maximum at 0, a. m. and 11, p. m. : at its minimum at 4, p. m. ; 

 and at 4^, a. m. this variation is constant : it is uniformly 

 repeated every day : the succession of the seasons produces no 

 alteration: the elements on which the mean is founded are, 

 therefore, simple, distinct, and free from all mixture of interfer- 

 ing causes ; and M. de Humboldt, who has founded the system 

 of his observations on the phsenomenon of the daily variation, 

 has give^i us a barometric mean completely unequivocal. 



