NO. 4 SOLAR RADIATION ABBOT, FOWLE, AND ALDRICH 4I 



circumstances to be 0.849 ±0.003, t0 reduce its readings to calories 

 per cm. 2 per minute. From this we find the constant of balloon 

 pyrheliometer No. 3 to be 0.451 ±0.006. 



In this way, it appears to us, the source of error above mentioned 

 was avoided. A few comparisons were also made at Omaha directly 

 between balloon pyrheliometer No. 3 and silver-disk pyrheliometer 

 No. 9. These show the magnitude of this error, for assuming that 

 no such error as above considered exists, the results of these com- 

 parisons yield for the pyrheliometer No. 3 the constant 0.414, which 

 differs by 8 per cent from the value obtained by the preferred process. 



3. ROTATION OF THE INSTRUMENT 



Another source of error. which was not inconsiderable depended 

 upon the rotation of the balloon during its flight, for the instrument 

 not only rotated, but swung around a small cone, so that the average 

 angle made by the sun rays with the surface of the aluminum disk 

 was not given immediately by a knowledge of the latitude of Omaha 

 and the declination and hour angle of the sun at the time of exposure. 

 Fortunately the record of the flight gave means of determining this 

 small correction. The record of the degrees marked upon the ther- 

 mometer stem, instead of being a series of parallel fine lines as they 

 are shown in figure 9, became broadened out as the instrument 

 rotated. By measuring the distance apart of the edges o'f the broad- 

 ened lines, as compared with results found in check experiments made 

 by moving the instrument through known angles, the half angle of 

 the cone during the highest part of the flight was determined and 

 found to be about 9 degrees. It was then computed that a correction 

 of about 1 .2 per cent should be added to the readings over and above 

 that of about 8 per cent which was due to the zenith distance of 

 the sun. 



4. RATE OF THE CLOCKWORK 



At Omaha, on July 2, 1914, during calibrations, the mean period 

 occupied by a complete rotation of the shutter was found 8 m 17 s ; 

 at Washington, on December 26, 1914, during calibration, 8 m 18 s . 

 Other records give similar indications of substantial constancy of 

 rate of the clockwork. However, on February 4, 191 5, at +19 C, 

 the mean rate of the drum was .02154 mm. per sec, while at — 46 ° C, 

 the mean rate found was .0217 mm. per sec. This indicates a 

 change of 1 per cent for the range of temperature +34 to — 37°> 

 which occurred on July 11, 1914. This error would tend to diminish 

 the results by 1 per cent. 



