40 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



the corrected readings of the balloon pyrheliometer were, at the first, 

 about 16 per cent higher when the current of air was in operation 

 than when read in still air. 



We reduced this source of error very greatly, by attaching to the 

 instrument, a flat plate of blackened tin (r, fig. 5), level with the 

 copper ring diaphragms which admit the light to the aluminum disk, 

 and extending out from the copper disk to about 25 centimeters in 

 diameter. This tin plate deflected the current of air in such a manner 

 that the magnitude of the error we had found became reduced to 

 4 per cent. It seemed to us that the error must be proportional to 

 the number of molecules carried down by the current of air, and that 

 it would therefore decrease directly in proportion to the pressure of 

 air in which the instrument found itself. Accordingly we believe 

 that at the altitude reached by the instrument, namely, 24 kilometers, 

 where the pressure of the air is only one twenty-fifth of that which 

 prevails at sea level, the effect of this source of error will be to 

 increase the reading of the pyrheliometer by only about 0.2 per cent. 



2. VARIATIONS IN SKY EXPOSURE 



As indicated in Mr. Abbot's letter, there was expected a difference 

 in the radiation exchanged by the instrument with the sky, depending 

 upon whether the shutter is opened or closed. This difference grows 

 less and less as the instrument goes to higher and higher altitudes, 

 but there could readily be a source of error here if the instrument 

 were compared on the ground with another instrument exposing the 

 disk very differently. 



To avoid this source of error, one of our older pyrheliometers, 



No. Y, was reconstructed, so that it might be exposed to the sun and 



sky in exactly the same manner as the balloon pyrheliometer. In 



fact, one of the balloon pyrheliometers was taken to pieces, and the 



copper diaphragms and the shutter were transferred to pyrheliometer 



No. V, so that, in respect to its exposure, pyrheliometer No. V became 



identically similar to the balloon pyrheliometer No. 3. The two 



instruments were then compared, and the result of the 16 determina- 



No. 3 

 tions gave us the ratio of their readings: kT y = 1 .882 ± 0.024. 



We then returned pyrheliometer No. V to its original condition, 

 except that we retained the same copper diaphragms, so as to prevent 

 any error from the measurement of the size of the aperture; and we 

 compared it with silver disk pyrheliometer No. 9. By 14 compari- 

 sons we determined the constant of pyrheliometer No. V in these 



