308 SCIENTIFIC RECORD FOR 1884. 



of September, the zenithal absorption being 29 per cent, on the latter 

 day and 15 per cent, on the former, and this is the more remarkable 

 because the air was more decidedly hazy on the 12th of September. 



By observing the temperature of a point in the same vertical plane 

 with the sun but on the opposite side of the zenith, one should theoret- 

 ically have the means of determining the influence of the atmosphere, 

 but his efforts to realize this method have shown that it is not success- 

 ful in the latitude of Berlin, partly because the observations are not so 

 accurate, and partly because the atmospheric irregularities have a 

 larger influence. 



In general, in the day as well as by night, the lowest temperature is 

 found in the zenith and the warmest towards the horizon. This, which 

 is due to the sky temperature, is apparently exaggerated by the sun's 

 heat without a material change in the relative temi)eratures. 



The diminution of solar heat with the length of path through the at- 

 mosphere is such that the former is a linear function rather than a log- 

 arithmic function of the latter. 



The observations on the Faulhorn show that the atmospheric con- 

 stant for that elevation is sensibly smaller than in Berlin. 



The most important result of the work is the proof that the solar heat 

 is n(jt constant, but subject to very considerable variations, as shown 

 by the following values of the " solar constant " : 



For June 29 574±18 



For July 5G4±15 



For August 14 ... 607±18 



For September 12 573±51 



For October 15 555± 19 



The probable errors here given are really three times the probable 

 error as ordinarily understood, and represent the limits within which 

 the probabilities are 22 to 1 that the true value must fall. In general 

 the change of solar heat is such that it increases by about 6 per cent, 

 from the 1st of July to the middle of August, and then diminishes by 

 about 8 per cent, by the middle of October. This change is entirely 

 parallel to the observed change in the development of sun-spots, as re- 

 corded by the observers at Potsdam, who state that at the beginning of 

 July and in September and October the development was quite rapid, 

 but in August much less, v, hence we may hope that this method of ob- 

 serving the solar heat may give a method of inferring the variations 

 in the sun-spots. 



The importance of this variation in solar heat may be estimated by 

 the following considerations : The mean temperature of the ground at 

 any place on the earth's surface is the result of two factors, namely, the 

 mean influence of the heat from the sky and that of the heat from the sun. 

 The first of these, according to the Konigsberg observations, amounts 

 to —82° for that place, and the latter influence amounts to -f 89° 0. 

 We may therefore assume that in temperate latitudes a change of solar 



