OCT. — mae. 1858-59.] Report of the Magnetic Survey. 351 



The haze, during the experiments at Umballa, though generally 

 accompanying the hot winds, modified the heat, particularly for the 

 boiling thermometer; also, the small clouds, though very thin, and 

 not before the sun at the moment of the reading, had caused a sen- 

 sible depression in the black thermometers on the wool; the other 

 black bulb, freely suspended, being much less sensible for the rays 

 of the sun, since the objects against which it radiates change their 

 temperature but very little {see " Surface of Ground,") went on 

 steadily rising. 



The great heat of the air lasting till sun-set is very characteristic 

 for days with hot winds in general ; also the surface of the ground 

 keeps remarkably warm in the first evening hours, though it loses 

 25 degrees C. during the night. 



Notwithstanding this great variation, no particle of dew is depo- 

 sited in these regions during nearly three months, the wet bulb 

 thermometer sinking still 10 degrees lower during the night than 

 the dry one. 



The power of the sun's rays was also determined at Benares and 

 Cawnpore. 



At Benares the boiling point thermometer, in an arrangement 

 like the one just described, was exposed on the 3rd of April. 



30 M 78° 15' Centigrade. 



5 „ .... 82° 60' 



10 „ 84° 23' 



15 „ 84° OS' 



The maximum sees to fall decidedly after 12 (the sun's power 

 decreasing less rapidly than the loss of radiation is diminished by 

 the general increase of the temperature of the air continuing till 2 

 or 3 p. m.) The maximum on normal days, without clouds, and 

 with very light wind, seems to be reached pretty regularly at Ih. 

 10m. to \h. 20m. p. m. (See Allyghur.) 



At Ih. 29m. the wind at Benares became more violent, about 6*5 

 metres per second, and was felt very hot. To my surprise the ther- 

 mometer immediately went down, the thermometer with white bulb 



It stood at 12 H. 



Note.— The read- 1 „ 



ings are correct- j 



ed for index er- " 



rors. 1 ,, 



