2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 87 



water streams outflowing from the two chambers we employed eight 

 thermoelectric junctions of nickel-platinum connected in series with 

 their elements passing alternately from one stream to the other. .Short 

 glass outflow tubes pierced to admit these thermoelectric elements were 

 skilfully made by L. B. Clark of this Institution. These delicate parts 

 of the apparatus were enclosed completely by an air-tight box made 

 of cellophane and wax. Thus air currents could not directly affect the 

 thermoelectric system. Plate i, Figures i, 2, 3, show the instrument 

 in various stages of completion as thus constructed and used. The 

 electric heating coils are wound on the rear surfaces of the receiving 

 cones at the extreme rear of the absorption chambers. In water-flow 

 pyrheliometer Xo. 3 we wound heating coils not only in that position, 

 but also on the front edges of the cone. We found in our use of that 

 instrument that coils in these two positions give equal results ; hence 

 we felt justified in omitting the edge coils in water-flow pyrheliometer 

 No. 5. 



With the cooperation of Doctor Siiring, Messrs. Abbot and Martens 

 compared at Potsdam in October, 1931, silver-disk pyrheliometers 

 S.I.5bis and S.I. 12. The latter instrument has been for about 20 years 

 in the possession of the Meteorological Observatory at Potsdam. The 

 constant of S.I.5bis, 0.3715, as used last October, was determined by 

 Messrs. Abbot and Hoover in August, 193 1, by 24 comparisons with 

 A.P.O.Sbis bis- We have used the latter instrument since 1912 solely 

 for standardizations at Washington. Using this constant, 0.3715, read- 

 ings of pyrheliometer S.I. 12 were found to satisfy within 0.3 per 

 cent the identical constant found for it in 1912. In order to carry 

 through a direct comparison with S.I. 12 in use at Potsdam, we 

 employed at Mount Wilson in June, 1932, silver-disk pyrheliometer 

 S.I.5bis as the comparison instrument, together with our improved 

 water-flow pyrheliometer No. 5. For convenience of observing, the 

 latter was supported in a fork carried by an equatorial telescope 

 mounting, and pyrheliometer S.I. 5b is was so attached that its tube 

 for admission of solar rays was exactly parallel to the two tubes of 

 No. 5. In this way the observer of S.I.5bis kept No. 5 as well as 

 S.I.5'bis continually pointing at the sun during comparisons. Times of 

 reading of the silver disk pyrheliometer were governed by a sounder 

 beating seconds. Observers C. G. Abbot and L. B. Aldrich inter- 

 changed duties in reading the pyrheliometers. No difference in result 

 was noted. 



Water currents of approximately 45 cubic centimeters per minute 

 in each branch of pyrheliometer No. 5 were found to give good re- 

 sults, but different rates of flow were tried at various times without 



