2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 92 



a null reading of the sensitive galvanometer connected into the thermo- 

 electric circuit whose two sensitive junctions are in the outflowing cur- 

 rents of water issuing from the two chamhers. In preparing for a 

 series of readings, hoth chambers were opened simultaneously to the 

 sun rays, and the adoptetl zero of the galvanometer was that which 

 subsisted during this preliminary joint exposure. After the conclusion 

 of a series, the zero was again determined in the same way, and some- 

 times was found to have changed a little. Nevertheless, error from 

 such a drift of zero is ap])roximately eliminated by the device of 

 alternately exposing the two chambers to solar and electrical heat. 

 For if owing to an unknown drift of the real zero the electrical energy 

 is determined too large within one chamber, it will be found for the 

 same reason too small in the other immediately afterward by about 

 an equal amount. To secure complete elimination, we took one more 

 electrical measurement on one side than on the other and made our 



West plus West , t- ^ tt f +i 



comparisons as between and East. Users of the 



Angstrom compensation pyrheliometer will have employed and will 

 appreciate this corrective principle. In fact we found no appreciable 

 difference in the results of the various series as between occasions 

 when the drift of the galvanometer zero was relatively large and rela- 

 tively small, res])ectively. 



The total deflection for uncompensated solar heating ranged from 



3 to 10 centimeters on the scale of the galvanometer, depending on the 

 rate of flow of the water current used on different days. The zero 

 observation could be made to a tenth of a millimeter or better, cor- 

 responding to from - — to of the total deflection. No systematic 



'■ ^ 300 1000 



difference in results depending on the rate of flow of the water current 

 could be detected. 



During the entire campaign of comparisons, the silver-disk pyrheli- 

 ometer was read by C. G. Abbot, with timing by the eye and ear 

 method, listening to beats electrically sounded from an accurate sec- 

 onds pendulum. The galvanometer and current measurements were 

 made by L. B. Aldrich. 



To determine the electrical energy of compensation, an electric cur- 

 rent from storage batteries was passed through slide wire resistances, 

 thence to a milliammeter, and thence to the manganin heating coils 

 within the pyrheliometer chambers. The resistances of these heating 

 coils were repeatedly measured on a standard Wheatstone's bridge. 



They were found to be identical to within less than ■ with the 



