(453) 



Average mean error of a single observation. 



Disappearance Reappearance Mean | Dolambre *) Introd. p. LIV 



I 

 II 



III 

 IV 



± 258 

 45 

 37 

 80 



± 20s 

 29 

 (iG 

 00 



± 22s 5 

 37 



51,5 

 70 



1785 



1 88, 5 



'72,5 rejecting the observations 

 deviating more than 3 mi- 

 nutes). 



According to these nnmbers the comphiints abont the i'lcreased 

 inaccnraey of the tables of Damoiseau seem rather exaggerated, at least 

 for the (irst and secoiul satellites. 



Taking into account the mean errors contained in the last column 

 we get tlie most probable correction at the epoch 1894 — 98 



for I -f S^0 with a mean error of ± 2^6 



similarly „II — 3 ,8 „ „ „ „ „ ± 5 ,4. 



Both corrections can hardly be vouched for. 



For III the case stands otherwise. It is 'true, the subtractive cor- 

 rection at the reappearances as well as the additive one at the dis- 

 appearances may be attributable to the use of more powerful tele^ 

 scopes; still there seems to be a progression in the numbers of the 

 last column but one, which calls for a more exhaustive investigation. 



In regard to IV, we found great corrections foi' the year 1895, 

 After some years in which this satellite had not been eclipsed, owing 

 to the fact that at the opposition it passed to the north of the shadow 

 cone of Jupiter, there began a new period of eclipses in this year. 

 In such a case the satellite travels high above the plane of the 

 orbit of Jupiter, and describes only a small chord in the shadow. 

 The consequence is that any small error in the latitude appears 

 strongly magnified 'm the duration of the eclipse. The observations 

 of Mr, Winkler at Jena and of the observer at the observatory at 

 Uccle near Brussels, of 8 March 1895 are very suggestive in this 

 regard. The corrections were found to be: 



Jena. Brussels. Mean 



at disappearance +19'"48« -|-21'"58^ + 20'^53s 

 at reappearance — 19 36 — 18 33 — 19 4,5 



which shows that it is not the mean longitude of this satellite which 

 is mainly in error. 



*) Delambre gives mean differences ; we have multiplied his numbers by IV4 

 in order to set mean errors. 



