38 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



cylinder, rests upon the photographic paper on the drum, e, between 

 the thermometer, d, and the drum. A little longitudinal slot is cut 

 in the aluminum arm, o, at the point where it passes under the ther- 

 mometer, so that, as the drum revolves, the sun prints through the 

 thermometer stem and the slot, and makes a trace of the position of 

 the arm, 0, appearing as a dark narrow streak between two light 

 streaks. 



No temperature record was obtained in the pyrheliometer flights 

 of 1914. Certain corrections to the barometric readings depending 

 on the temperature were worked up by a consideration of the tem- 

 peratures found in other flights, as will appear in its place. It would 

 have been better if the mounting of the barometric element had been 

 wholly of invar, so as to reduce these corrections, but no essential 

 harm seems to have resulted. 



The size of the apparatus was made as small as seemed practicable, 

 and its entire weight, including about one-half pound of water but 

 exclusive of silk, feathers, and cotton used for wrapping, was only 

 three pounds for the water jacketed instruments. The electrically 

 heated instruments, with their battery 1 and devices for operating it, 

 weighed about four pounds. 



METHOD OF READING PYRHELIOMETER RECORDS . 



The records indicate the rate of rise of temperature of the alumi- 

 num disk during exposure of it to the sun, and the rate of fall of 

 temperature of it during shading. One desires to know the rate of 

 rise during exposure as it would be if there were no cooling due to 

 the surroundings. In reading a record, it was fastened upon a large 

 sheet of cross-section paper, with the degree marks of the balloon 

 pyrheliometer record lying parallel to the section lines, in abscissae. 

 A fine wire was then stretched parallel to a branch of the zigzag 

 trace, and the tangent of its inclination to the degree marks was read 

 upon the cross-section paper. Each such tangent was determined by 

 several readings. The tangent representing each solar heating was 

 then corrected by adding to it the mean value derived from the 

 coolings preceding and following it. Thus we obtained, in arbitrary 

 units, values proportional to the solar heatings. The same method 

 of reading was applied to the records obtained while calibrating the 

 balloon pyrheliometer, at Omaha, and at Washington, before and 



1 A special form of Roberts cell was developed, comprising tin, nitric acid, and 

 carbon. Each cell was of 20 grams weight, 1.3 volts potential, and furnished 

 an average of 0.4 ampere for 2 hours. 



