On Solar Radiation. 191 



on the watch, but on the end of the rising column of 

 liquid. At the 20th beat read off, and register the reading 

 (12° 0), as in column 3 A, of the annexed form. Then 

 wait, watching the column of air above the liquid, to see 

 that no bulbs of liquid are in it, or at the opening of the 

 upper bulb (which will cause the movement of the ascending 

 column to be performed by starts), till the minute is nearly 

 elapsed. At the 50th second begin to watch the liquid 

 rising, at 55s. begin to count 0,1,2, up to 20 beats as before, 

 attentively watching the rise of the liquid, and at the 20th 

 beat, or complete minute (2h. 16m. 0s.) read off, and instantly 

 shade the instrument, or withdraw it just out of the sun and 

 penumbra. Then register the reading off (43° 3) in column 

 3 B, and prepare for the shade observation. All this may 

 be done without hurry in 20 seconds, with time also to 

 withdraw the screw if the end of the column be incon- 

 veniently high in the scale, which is often required. At the 

 20th second prepare to observe ; at the 25th begin to count 

 beats, 0, 1, 2, 20 ; and at the 20th beat, i. e. at 2h. 16m. 

 30s. read off, and enter the reading in column 3 A, as 

 the initial shade reading (45°.2). Then wait as before 

 till nearly a minute has elapsed, and at 2h. 17m. 20s. again 

 prepare. At 17m. 25s. begin to count beats ; at 17m. 30s. 

 read off, and enter this terminal shade reading (42°.8) in 

 column 3 B, and if needed withdraw the zero. 



" Again wait 20s, in which interval there is time for the 

 entry, &c. At 17m. 50s. remove the screen, or expose the 

 instrument in the sun; at 55s. begin to count beats; and 

 at the complete minute, 18m. 0s. read off (14°.8,) and so 

 on for several alternations, taking care to begin and end 

 each series with a sun observation. If the instrument be 

 held in the hand, care should be taken not to change 

 the inclination of its axis to the horizon between the 

 readings, or the compressibility of the liquid by its own 

 weight will produce a very appreciable amount of error. 



