SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8/ 



few preliminary ones of June 24, at which time the silver-disk pyr- 

 heliometer measurements were affected by a timing error. From these 

 comparisons the constant of pyrheliometer S.I.5bis results as 0.3625. 

 Many comparisons were made at Washington in Avigust, 1931, 

 and in September, 1932, between silver-disk pyrheliometers S.I.5bis 

 and A.P.O.Sbis bis- The latter instrument has been used for many 

 years to standardize silver-disk instruments sent abroad. The former 

 was furnished with 1 2-1/2 inch vestibule and the other changes made 

 upon it in August, 1931. The comparisons just referred to resulted 

 as follows : 



Date 



Aug. 8, 1931. 

 Aug. 17, 193 1 

 Sept., 1932. . 



Ratio 



S-I-Sbh 



-'^•PO-8bis bis 



I . 0206 

 I .0177 

 I .0170 

 I .0182 



According to these results and to the observations made with Stand- 

 ard No. 5, the constant of A.P.O.Sbis bis could now be taken as 0.3691. 

 We have hitherto adopted 0.3786. A change of — 2.5 per cent is 

 indicated. 



Is this difference to be regarded as due to error in the experiments 

 with our water-flow and water-stir pyrheliometers No. 3 and No. 4 

 with which we established the Smithsonian scale of 1913? First of 

 all we recall that in the use of these instruments at Washington and 

 Mount Wilson their sky exposure, reaching an angular diameter of 

 about 16°, is to be contrasted with the sky exposure to only 10° 38' 

 in the original silver-disk instruments. More than twice as much sky 

 area was observed by the standard instruments as by the silver-disk pyr- 

 heliometers in 191 3. This may have made several tenths of a per cent 

 too high a scale value in the work of 191 3. 



In the second place, the calibration of the platinum resistance 

 thermometers used in standard pyrheliometers No. 3 and No. 4 in- 

 volved the whole technique of exact mercury thermometry. Those 

 who have had occasion to use mercury thermometers for exact work 

 will know what was involved in the determinations of temperatures, 

 as given in Tables 10 and 19 of Volume 3 of the Annals, and that 

 appreciable inaccuracy there may have been possible. Indeed, the 

 irregularity of run of the numbers in the latter part of Table 10 seems 

 to throw some doubt on the sufficient accuracy of the coefficients of 

 temperature change determined. 



