70 



MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 



MAY 2. 



This table sliows the effect of using a coii.staiit value of c between the hour.s of 7.00 a. ni 

 and .5.00 p. ui. Had the value c = 1.78 been employed, the computed values would have been 

 diminished by about 0o.5, and the mean of the residuals would have been nearly 0. There still 

 would have remained, as was expected, plus residuals for the morning and evening observations, 

 and minus residuals for those taken near the middle of the day. The magnitude of the residuals, 

 while larger than was hoped for, is not sufficient to prevent the use of the method, but it renders 

 advisable the computation of the solar intensities by equation (1) as well as equation (2). In the 

 latter equation it was decided to use the value c = 1.8. 



The following tiible contains the observations selected for the computation and the solar 



intensities obtained by equations (1) and (2). Those observations only were used in which 



the sun had been clear for some time before the observation. The mean of the corresponding 



readings with the two pairs of instruments was taken and the resulting values converted into 



centigrade degrees : 



Oomputation of solar intensities. 



[Conjugate thermometers.] 



APRIL 28. 



