to the Antarctic Regions. 



211 



where n is the number of sun observations, the time corresponding 

 being the middle of the middle shade observation. 



A complete actinometer observation cannot consist of less than 

 three sun and two shade observations intermediate ; but the more 

 there are taken the better, and in a very clear sunny day it is highly 

 desirable to continue the alternate observations for a long time, even 

 from sunrise to sunset, so as to deduce by a graphical projection the 

 law of diurnal increase and diminution of the solar radiation, which 

 will thus readily become apparent, provided the perfect clearness of 

 the sky continue, — an indispensable condition in these observations, 

 the slightest cloud or haze over the sun being at once marked by a 

 diminution of resulting radiation. 



To detect such haze or cirrus, a brown glass applied before the 

 eye is useful, and by the help of such a glass it may here be noticed 

 that solar halos are very frequently to be seen when the glare of 

 light is such as to allow nothing of the sort to be perceived by the 

 unguarded eye. 



It is, as observed, essential that the instrument be exposed a few 

 minutes to the sun, to raise its temperature in some slight degree. 

 If this be not done, owing to some cause not very obvious, the first 

 triplet of observations (sun, shade, sun) will give a radiation percep- 

 tibly in defect of the truth, as will become distinctly apparent on 

 continuing the series. But it may be as well for a beginner to com- 

 mence at once reading as soon as the instrument is exposed, and re- 

 ject the first two triplets, by which he will see whether he has all 

 his apparatus conveniently arranged, and get settled at his post. 



When a series is long continued in a good sun, the instrument 

 grows very hot, and the rise of the liquid in the sun observation de- 

 creases, while the fall in the shade increases ; nay, towards sunset it 

 will fall even in the sun. This phenomenon (which is at first 

 startling, and seeming to impeach the fidelity of the instrument) is, 

 in fact, perfectly in order, and produces absolutely no irregularity in 

 the resulting march of the radiation. Only it is necessary in casting 

 up the result (in col. 5.) to attend carefully to the algebraic signs 

 of the differences in column 4, as in the following example (which, 

 as well as that above given, is one of actual occurrence). 



P2 



