I 449 ) 



Williams makes the shadow of II larger than that of T whereas 

 in the estimation of Mr. Comas they were e(|ual. It seems hardly 

 doiihtful but tlie English observer must be right. 



(5) In 1901 See repeatedly measured the diameters of the satellites 

 of Jupiter at the 26 inch telescope of Washington. He made use of 

 the filar micrometer but took a special care to eliminate the syste- 

 matic errors peculiar to this instrument (Vid. iVstron. Naclir. N". 3764, 

 21 Jan. 1902. The communication of See is dated 19 Oct. 1901). 



During the months May — August (both inclusive) of the year 1901 

 he measured the diameters in the night. He was then much troubled 

 by the undulation of the limbs caused l)y the unsteadiness of the 

 air. Afterwards in the months of September and October of the same 

 year he observed a little before and a little after sunset. Artificial 

 illumination was then not needed; and the satellites appeared as 

 quiet discs. Moreover the held and the satellites were coloured greenish 

 yellow by a screen tilled with protochloride of copper and picric acid. 

 The results for the diameters turned out to be smaller in every case 

 than those formerly found. The difference w^as attributed to irradiation. 



The results, reduced to the mean distance of Jupiter to the sun 

 (5,2028), are as follows. 



Satellite 



At nisht 



In daytime 



Ditïerence, attributed 

 to irradiation 



It is remarkable that the brightest satellite, III, shows also the 

 strongest irradiation. If howe\'er we consider the difference insufTi- 

 ciently established, and if therefore we combine the several results 

 obtained for the irradiation, duly taking into account the weights 

 corresponding to the probable errors, we get 



Irradiation = 0",264 ± 0",012. 



This is the irradiation for the whole diameter and we thus get 

 0",132 for each of the limbs. This number however holds only for 

 the telescope at Washington for which, owing to its great aperture, 

 the diffraction must be exceedinglv small. 



