80 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 



Mean relative intensities observed at the same Jionr on two or more (l((ys by the three methods. 



The accompanyiDg chart represents graphically the values given in the above table. The 

 full curve corresponds to the intensities given by the conjugate thermometers, the dotted curve 

 to those given by V^iolle's bulbs, the broken curve to those giveu by the ordinary black and 

 bright thermometers. The original values by which the curves wei'e drawn are represented by 

 0, •, and +, respectively. 



5. SUMMARY OF RESULTS. 



The discussion given in the preceding pages has given us values of the solar intensities on 

 the days of observation by three different methods. It has thus furnished incidentally a means of 

 comparisou of the methods, in so far as the observations themselves allow this to be done. The 

 uncertainties arising from the state of the sky have already been mentioned ; their effect is shown 

 in the discordances which the results at different hours on the same day exhibit. By including 

 favorable times only, the tables showing the relative intensities by the different methods rei^resent 

 fairly the results obtained under the best conditions during the series of observations. The 

 following summary gives the conclusions derived from an examination of the results : 



1. The method of the black and bright bulb thermometers freely exposed in the air gives 

 only an ajiproximate determination of the solar intensity. This was expected, and is due to 

 the constantly varying conditions of exposure, caused by the effect of the winds on convection 

 currents. 



2. Violle's, bulbs are affected by convection, but the effect is shown less than in freely 

 exposed thermometers, on account of the i)osition of the thermometer.s within. The observation.s 

 on the afternoon of May 2, when the air was almost perfectly still, show higher intensities than 

 tlie corresponding times on other days, but the observations are not sufficient to indicate how 

 much the results are intluenced by this cause. 



3. The intensities by the conjugate thermometers seem also to be affected bj' the varying 

 influence of convection, but in this case (and in the preceding also) direct experiments would 

 give more information as to this effect than any examination of these observations. 



4. The intensities by Violle's bulbs (see the curves) are smaller in the morning and greater 

 in the afternoon than tho.se by the conjugate thermometers. There is a marked difference 



