171
and a2 the co-ordinates of D in the grid-system were computed by
the approximate formulas
Pi
Pi
x, micr.
x0 micr.
In these formulas and a2 were expressed
in cgr. It will be clear that for this purpose
pjg 3 the normal readings, for which the unity
of the verniers is 1 cgr, give a sufficient
accuracy since for p 1 micr, a 2,5 cgr is.
From the 20 co-ordinates of D the mean error in the single ad
justment could be computed. The results are given in table I.
Table I.
JL
It could be expected that the mean value of the 20 determinations
of px and p2 in each case would be nearly zero. This seems not to be
confirmed by the data in the first two columns of table I. To judge how
far these values could be influenced by the accidental errors, the
mean errors in the average values of px and p2 were computed. When
■p
the ratio exceeds the limit of ca 3, a systematic error can be ex
pected. In both cases wherein this phenomenon appears, a reasonable
explanation could be found.
In the first case 1) there was a small amount of back lash in the X
movement that was not systematically taken into account. The serial
was repeated with a better result.
in the second case 2) the distance between the line AB and the
rotation-centre D was rather great (ca 0,15 mm) so that the setting
of M became difficult. When in this case the grid-lines are not exactly
parallel to the X direction, a systematic error can be easily expected.
5 mm
t
5mm
K
2
4
Pa~
At
-oc,
3
°C2
5000
6366
5000
6366
1
2
Instrument
Mean value
in microns
Mean error in
single deter
mination
Mean error
in mean
value
Mp
/t
P>2
mp i
mp2
MPl
Mpi
Zeiss
Left
9
5
I
3
12,6
9.3
8,3
5.6
2,8
2,1
1,8
1,3
3,2
2,4
0,6'
2,3
Right
2
4- i
5.6
4.3
1.3
1,0
',5
1,0
De Koningh
11 r 3985
L
-1- 2
- 13
9.3
17,6
2,1
3.9
0,9
3-32
R 3
5
8.7
8,7
1,9 I 1,9
i,9
2,6
1), 2) See reference in text.