301 
for a value of [A] deduced from 100 observations the m.e. is thus 
for K + 0°020, for H + 0°018 and for a value of [Aly + |Alr, 
if in each position of the instrument 100 observations are used, it 
is for Meee 0028 vand for H == 05026, 
Notwithstanding the pretty considerable m.e. the following facts 
stand out clearly. For Kam in the first period, i.e. in 1864, the sum 
W 4E was relatively small — 0.031, whereas in the following 
years it remained very constant at about — 05.097. For HENNEKELER 
no distinct change can be seen; the mean value for him is 08.078 
and therefore possibly a little smaller than K.’s value in 1865 —68. 
Moreover for both observers the value may be somewhat larger for 
the threads V—VII than for the threads !—III. 
However that may be, the chief result is undoubtedly the constant 
amount for all the threads. The cause of this had to be inquired 
into, in order thereby to deduce the influence on the determination 
of the thread-intervals.. The value being as large as it is, it was 
not allowable without further investigation to take for the correction 
of the thread-intervals the mean of the results in the two positions. 
Before proceeding, however, the meaning of the results obtained 
may first be established, They depend upon the corrections which 
the intervals previously assumed require, and therefore also upon 
those former values themselves. The question might therefore arise, 
whether the anomalies found above, i.e. the values of the sums as 
differing from zero, may not have their origin in the results of the 
original investigation and therefore in the nature of the material used 
at the time. But it will be seen at once that this cannot be the case, 
e.g. thus: whereas before the reduction was made with one definite set 
of distances, errors in it cannot give rise to the differences between 
the corrections now found in the positions Cl. B. and Cl. W. 
At the same time, although no conclusion may be drawn from 
the fact that af present stars were used which were also observed 
‘in declination and formerly stars, observed in R. A. alone, still it 
could not but appear at once probable, that the anomaly found 
would be connected with the biseetion by the horizontal thread (the 
material originally used by the second of us also consisted, for the 
great majority, of stars observed in deel.) and it was then natural to 
look for the cause in a peculiar personal error of the kind as was 
previously shown to exist in the observations of the first of us, 
namely that after bisecting the star, probably owing to the dimi- 
nished brightness, the times of transit were observed later. The 
retardation in his case amounted to about 0.08. 
If we now assume, that the bisection, which with a very few 
