pritchard's wedge photometer. 307 



Here the probable error of the mean is T \ of one division, or rather less than ^\ 

 of one per cent of the whole. The probable error of a single observation is about 

 T 8 3- of one division, or rather less than £ of one per cent of the whole. Even in 

 this case, however, the probable error is compounded of that due to the instrument 

 and that due to the changing temperature of the source. 



If, however, we measure the solar heat, the result is less favorable^ for, as I have 

 elsewhere remarked, there is always found an incessant fluctuation of the heat trans- 

 mission from minute to minute under an optically clear sky, or, in other words, the 

 bolometer constantly perceives haze and mist which the eye does not. 



The same day on which the above Leslie cube readings were taken, presented 

 to the eye a fair, blue sky, with some cumulous clouds. Except for the passage 

 of a cloud between the ninth and tenth reading, in the following series of twenty 



readings there was no interruption. The sky was watched critically, and even 

 to a practised eye it appeared as clear at one time as at another. The conditions 

 are the same, except that a shunt was introduced in the solar observations, so that 

 the image should not be sent off the scale by the greater heat. 



deflection. 



(Source of Heat, Sim, in optically clear Skt.) 



Time 11 h. 



434 



440 

 455 



468 

 464 

 468 

 479 

 475 

 476 



Cloud . . . 



421 

 407 

 471 

 474 



447 

 456 



468 

 468 

 475 

 441 

 449 



Mean 456.8 ± 3.1 



