XO. 15 IMPROVED PYRHELIOMETER ABBOT AND ALDRICH 3 



affecting the results of the comparisons of pyrheHometers. There was 

 a deflection of about 9 centimeters on the scale of the moving coil 

 galvanometer when sun rays were allowed to enter one chamber of 

 pyrheliometer No. 5 without electric compensation. No wiggle as great 

 at o.or centimeter was ever detected during observations. Accidental 

 fluctuations were therefore less than i/iooo part of the deflections. 

 Slow drift of a millimeter or two during a run of half an hour some- 

 times occurred, but was eliminated by alternating the chambers ex- 

 posed to solar and electric heating. In short, water-flow pyrheliometer 

 Xo. 5 behaved far better than we had expected and indeed so perfectly 

 that we can not conceive of its improvement. 



In the use of it, we at first regarded as the zero of the galvanometer 

 that position on the scale which was assumed when no outside energy 

 of sun or electricity was entering either chamber. There was then 

 considerable inequality in the results from the two chambers. Later 

 we perceived the advantage of using as galvanometer zero^ the position 

 assumed when solar rays were shining fully into both chambers simul- 

 taneously. Using this zero, the results from the two chambers were 

 then very nearly equal, usually within i per cent. But no difference in 

 the mean result of the comparisons with S.I.5bis was discerned, which- 

 ever zero was employed. 



In using the water-flow pyrheliometer No. 5, we found that an ex- 

 posure of two minutes before interchanging chambers was ample. It 

 is clear that a certain time must be allowed to elapse before interchang- 

 ing, for the electric heating is applied where the water current is 

 about to issue from the chambers. A small part of the solar heating, 

 on the other hand, is absorbed near the front orifices of the chambers. 

 From that location the current of water must flow several meters 

 through its spiral channel within the hollow walls of the chambers 

 before emergence. Hence the full effect of electric heating reaches 

 the thermoelectric junctions quicker than can the full effect of solar 

 heating. It is therefore necessary to wait until both sources of heating 

 can exert their full effects before making a balance. We assured our- 

 selves that two minutes was ample time for this by trying much 

 longer exposures and finding thus no change of the results of com- 

 parisons of pyrheliometers. 



The accuracy of the results with water-flow pyrheliometer No. 5 

 does not depend on any measurements of the rate of flow or of the 

 temperature of the water .^ It depends on the measurement of the 



^ The question of temperature concerns us only in reducing the electric energy 

 in joules to its equivalent in calories. We use the factor 4.185 to reduce to 15° 

 calories. 



