( 222 ) 
as in the former observations. The second column contains the 
capacities corresponding to the respective positions and obtained by 
the above graphic interpolation. 
Column (3) contains the differences of the capacities in column 
(2) which should be proportional to one another. From the various 
consecutive quotients the geometric mean is obtained and the 
numbers calculated which are actually proportional to one another 
and which also agree as much as possible with the numbers in column 
(3). From these “calculated differences” column (4) is obtained by 
addition. The differences again between columns (4) and (2) are 
found in column (5) which we may safely call errors of observation. 
The greatest is 0.019 of the capacity of the calibrating condenser 
and is equivalent to a displacement of 0.13 mm. of the glass plate 
in the adjustable condenser. We can take this as the highest limit 
of attainable accuracy, being that with which a single capacity is 
itself determined. 
Tables III and IV were obtained in the same manner. The former 
relates to the part of the glass plate which was not used in Table II, 
while the latter represents the results of an experiment in which 
the former method was carried out with a considerably greater 
capacity. 
In will be seen that these tables show “observation errors” of 
the same magnitude as Table II, and the remarks made above 
concerning the accuracy of the determinations hold also with these 
values. On the magnitude of the errors we may notice the following. 
The accuracy of a single adjustment was at a maximum 0.1 mm. 
as shown by a number of observations, in which the readings 
TA BLE HL 
| 
| 
Penge | 
Kositan ‚Capacity | | Capacity 
of the F Dif. | A 
observed. | calculated. 
glass plate. b | . 
i 
52.25 | 7.000 | ees 
1.445 | 
43.15 | 5.555 | | 5.545 | 0.010 
| 1.395 | 
34.55 4.160 | | 4.164 0.004 
1.310 | 
26.70 2.850 | | 2.854 0.004 
| 1.255 
183900 e595) 1.611 0.016 
| Delf 
11.25 0.425 0.431 | 0.006 
