SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 69 



two days chosen, one of greater haziness and one of greater clearness 

 than September 17, as follows: 



A mean between the values of sky brightness for these two days 

 was adopted as the sky brightness for September 17, 1918. It is 

 certain that on the first of these days the sky was brighter than on 

 September 17, and on the second it was less bright than on Sep- 

 tember 17. If we were to adopt the values of either of these two 

 days, the resulting values of total sky and sun brightness of Sep- 

 tember 17 would not be altered by so much as one per cent from those 

 given in the table. 



Comparisons of the pyranometer with Pyrheliometer IV on the 

 sun alone were made before this work, 1 and again after the work on 

 October 8 and October 9. A mean of 10 values on these last two 

 days gave a constant 2.\ per cent higher than the earlier comparisons. 

 Taking a mean of all comparisons, the constant of the pyranometer 

 was regarded as 24.1 instead of 23.8 as used for the eclipse reduc- 

 tions. The computed values originally obtained from measurements 

 of dimensions, electrical resistances and assumed absorbing power 

 of the pyranometer strips is 25.9. That recent observed values are 

 so much lower is doubtless due to rough usage of the blackened sur- 

 face necessary in fastening in new thermo-couples. 



It is to be noted that a very considerable irregular galvanometer 

 drift was present throughout the cloud observations. This seemed 

 due mainly to the changing air currents as the balloon basket swung 

 in the wind. Table I shows that in general the higher the wind 

 velocity, the greater the range of zero drift. Inadequate protection 

 of the galvanometer from vibrations caused by passing trains and 

 auto trucks also contributed to the drift. However, since each indi- 

 vidual determination required but five seconds and in each group 



1 See Report of Eclipse Expedition, Smith. Misc. Coll., Vol. 69, No. 9, p. 6. 



