128 



were determined for the two periods 1892 — 94, and 1903 — 09, 

 while 1890 and 1910, which were a(h]ed later, were connected to 

 the other years as well as possible. 



Like my predecessors I liave arranged the A«, after applying 

 all the above mentioned corrections according to the days of true 

 age of the moon Dt, in order to investigate whether there may exist 

 a dependence on this age. These results, as well as those found b}- 

 Prof. Bakhuyzen, 1 give here, viz. the mean deviation for each day 

 diminished by the total mean for the limb. In the first table 1 have 

 collected the results for limb I, in the second those for limb II ; 

 the numbers in brackets give the weights. 



We must remember that the systematic errors of the observers are 

 ojdy correctly represented by these figures if the theoretical correc- 

 tions depending on D are quite correct. 



This condition is certainly not fulfilled by the old set of corrections. 

 Too much importance must therefore not be attached to the variation 

 of these ligures and by making the corrections too complicate we 

 run the risk of introducing periodic terms and spoiling our results. 



Newcomb, however, already indicated the possibility of one cause 

 of divergence, viz. the tendency of observers to estimate the moon''8 

 diameter smaller by day-light, that is at very small and very great 



