42 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



5. HORIZONTAL THERMOMETER STEM AND CALIBRATION 



A difficulty was encountered in the experiments of 1913, for, owing 

 to the horizontal position of the thermometer stem, the mercury 

 thread sometimes separated, and failed to return after a rise of 

 temperature. This was overcome by drilling a hole into the upper 

 bulb, just before the flight, so that air pressure came upon the 

 mercury column. In 1913, this worked perfectly satisfactorily, but 

 in 1 9 14 the mercury column became foul in every case but that of 

 No. 5 pyrheliometer, owing probably to the creep of the lubricant used 

 in drilling the glass. This prevented the use of pyrheliometers Nos. 

 1 and 2, and required several washings with benzol and alcohol before 

 the bore of Nos. 3 and 4 was clean enough to be used. Even then 

 the upper temperatures were unavailable, so that no use could be 

 made of records at low altitudes in the flights of July 9 and July 

 11, 1914. 



The reader will perhaps wonder why there was not left a small 

 gas pressure above the mercury column in the original construction. 

 This was not done, for we were required to calibrate the thermometer 

 stems because their bores were not uniform. We could most readily 

 do so by breaking the mercury thread and moving a short column 

 from place to place in the bore, observing its length-changes. This 

 we did for. all the thermometers, and have corrected our results 

 accordingly. In view of our experience we should now prefer to 

 introduce gas pressure in the original construction, and calibrate the 

 thermometers in baths of known temperatures. 



6. OTHER CORRECTIONS 



The aluminum disk, during the highest flight, differed slightly in 

 its mean temperature from that which it had during calibration. 

 Owing to change in the specific heat of aluminum with change of 

 temperature, a correction of 0.5 per cent should be deducted for this. 



The suspending wires in their rotation shaded the disk. A correc- 

 tion of 0.2 per cent should be added for this. 



Variations in the absorption of the disk by deterioration of the 

 blackened surface between July and December are thought to require 

 a correction of somewhat less than 1 per cent to be deducted. 



Variations in reflecting power of the copper diaphragms used in 

 the calibrations are thought to require an additive correction of 0.25 

 per cent. 



