OLIMATJi. 69 



scending (but with occasional starts) till August or Septem- 

 ber, when it again rises gradually, till the cold weather sets 

 in. But here (as is found to be the case elsewhere within 

 the trQpics), I have not been able to satisfy myself that any 

 accurate prognostication of the state of the weather is to be 

 deduced from the fluctuations of the mercurial column. I 

 have seen it rise suddenly before or during heavy showers 

 of rain, and sink, equally inexplicably, before a course of fine 

 4ry weather. The only agent which appears uniformly to 

 act in the same way upon it is wind, the mercury always 

 rising before or during the prevalence of high wind. I have 

 also occasionally been able to predict wet weather, from ob- 

 serving the top of the column to be flattened, or concave, 

 but not with any degree of certainty. 



The daily range of the barometer is very trifling, probably 

 never exceeding 'O-iO or '060 of an inch, and seldom great- 

 er than '035 ; but on this head, as on that of its horary 

 oscillations, I am unable to speak confidently, from want of 

 leisure to make the necessary observations ; the horary 

 oscillations occur, as far as I have observed, exactly at the 

 same hours, and in the same succession, as elsewhere all 

 over the globe ; but according to Dr. Dalmahoy, only to 

 half the extent observed at Madras, and they are not inter- 

 rupted during the monsoon, as conjectured by Baron Hum- 

 boldt. 



The mean annual height of the barometer appears to vary 

 considerably, and to have diminished annually for the last 

 three years : this may have depended on the situation of the 

 instrument. The mean of ten months : 

 in 1831, was 22-933. 



Six do. in 1832, „ 23-067. 



Eight do. in 1833, „ 23-054. giving as an annual mean 



