NO. II USE OF THE PYRANOMETER ABBOT AND ALDRTCH . 3 



second swing occurs. Of two galvanometers tried by us both gave 

 about 2 seconds' time of single swing on open circuit, but when closed 

 on the pyranometer alone, one gave no second swing at all, the other 

 a second swing of about 1/20 the magnitude of the first. By insert- 

 ing 75 ohms or more resistance in series with the first galvanometer, 

 it also gave a second swing, and the time of single swing of each 

 galvanometer, when just giving a second swing, was about 4 seconds. 

 Even with 75 ohms in series, the first-tried galvanometer gave about 

 twice as much deflection as the other, but for several reasons we at 

 length preferred the second one. This was a galvanometer by Rose, 

 of Upsala, Sweden, made for use with the Angstrom pyrheliometer. 

 Its resistance is about 15 ohms. The resistance of the pyranometer 

 is 30 ohms. When used with the pyranometer, whether with or with- 

 out additional resistance of 200 ohms or less in series, the Rose 

 galvanometer completes its first swing in about 4 seconds. 



We next made tests with the heating current and with radiation 

 at night to see if the first swing is proportional to the final deflec- 

 tion. We found this to be the case. We also found that both with 

 the heating current and with nocturnal radiation, not only the first 

 swing but the deflections, attained after 10, 15, 20, 25, and 30 

 seconds, maintained certain definite proportions to the final deflection, 

 no matter what the strength of the current, or the intensity of the 

 observed radiation. We also found that when proper allowance was 

 made for the non-uniformity of sensitiveness of the galvanometer 

 for large and small deflections, the deflections observed due to 

 heating currents were exactly proportional to the S(|uare of the 

 heating current employed. 



These facts ascertained, the way seemed clear to avoid the source 

 of error mentioned above, and at the same time to greatly increase 

 the rapidity of reading the instrument, and also to avoid drifting of 

 zero, so apt to occur in long exposures. In short for all daylight 

 observations we adopted the plan of reading first swings, and of 

 omitting exact adjustments of the energy of the heating currents to 

 equal that of the observed radiation. Our present procedure in day 

 work is as follows : 



(i) On exposure to radiation, read the first swing of the galva- 

 nometer, and immediately close the shutter. 



(2) After 30 seconds or more throw on a heating current sufficient 

 to cause a deflection approximately equal to that from the radiation, 

 and again note the first swing, and the exact strength of the heating 

 current. 



