735 



The extent of the horary oscillations at Lima 

 (lat. 12° 2' south); appeared to us a little less 

 (1.7 to 2.3), than near the equator (2.6 to 3.3), 

 in the forests of Atabapo and of Rio Negro. 



bruary, 1822, at 0.76603m (th. 25° cent.) ; February 28th, 

 at 0.76510m (th. 26.5°); March 1st, at 0.76465m (th.26.5*); 

 and March 5th, at 0.76325m (th. 26.6°). If the relative cor- 

 rection of the horary variations be not neglected, there re- 

 mains only, as a source of error in the measurement of 

 mountains, without a corresponding observation, in the tro- 

 pics, the difference of absolute barometric heights $ and in 

 distinguishing between the difference of extreme dispersion 

 and the oscillations of that dispersion around the mean baro- 

 metric height, we may conceive that the probable limit of 

 the error arising from the cause we discuss, will rarely be 

 above fifteen or twenty metres. This estimate is important 

 for those who, in the barometric levellings projected for ex- 

 amining provisionally the isthmusses of Huasaeualco, Da- 

 rien, and Panama, may employ only one barometer. In order 

 to know exactly the number of locks which a canal requires, 

 we must, even between the tropics, where every circum- 

 stance is so favorable to the use of the barometer for the 

 levelling of the soil, employ two instruments : the one 

 should remain on the seashore, or, which is preferable, 

 should follow the second barometer from station to station, 

 as in the levelling operations executed by MM. de Parrot 

 and Engelhardt, between the North and Caspian Seas. If, 

 on the contrary, we seek only to know approximately (at 

 about twenty metres) the height of the ridge of partition 

 which presents a favorable chance for cutting an isthmus, one 

 barometer will suffice, which must be observed in going and 

 returning, as ought always to be done in the chronometric 

 measurement of distances. 



