ti errors in principal distance right and left, no trouble will arise when measuring 
iud at one height; however, with different heights appreciable and constant errors 
de will arise in the distance between successive grid projections. On account of 
sat this fact too, measurements at only one height must be considered insufficient. 
jon 3) For a thorough test of an instrument it is not sufficient to measure only with 
1 a wide-angle setting. Precision survey with normal angle objectives with a basis 
led proportion 1:3 should also be taken into account. 
nd 4) The number of grid points to be measured as indicated by Mr. Pennington 1s 
the also insufficient. The study of the deformation diagrams which might arise 
M on account of the residual distortions, shows that at least the observation of 
rid all grid points indicated by Mr. Pennington in his figure, is necessary. The 
ed measuring of more points, e.g. on both main verticals and on the main hori- 
zontal, is likewise required. Similarly, the measuring of points in the middle 
All of the diagonals 1-4, 3-4, 8-9 and 8-11 could be recommended. If the middles 
his of the diagonals 4-5, 4-7, 5-8 and 7-8 in Mr. Pennington's figure are also 
Stc measured, we obtain exactly the 23 points measured by manufacturers for the 
ed control of the instruments. 
165 5) When carrying out a standard test a continuous triangulation of 2 grid plates 
nd should be prescribed. This kind of triangulation provides a proper insight into 
the systematic errors which the instrument produces, even with a perfectly 
nt constant base length. An extension might be possible by executing this trian- 
its gulation at a second height, i.e. with a different base length. This second value 
he shows to what extent the base length influences the systematic errors in the 
d, triangulation. This is important for the question whether the user is obliged 
fi= to triangulate with a constant base length. 
ed 6) We do not agree to the extension proposals mentioned by Mr. Pennington 
he under 1). When judging the instruments it is important that the model be con- 
he trolled up to the corners. 
7) With regard to repetition of each observation, we would like to state the fol- 
of lowing: like Galileo, we may repeat each setting five times and introduce the 
mean of these readings as observation. It is, however, preferable to repeat 
SO each setting and its corresponding reading only once; the whole grid measure- 
al ment as well as the grid triangulation should be executed twice at three dif- 
ts ferent heights. Thus each base length would be used twice. Only such a number 
of measurements as would result in two different figures for each value of the 
ze grid coordinates, and provided sufficient conformity exists, would furnish an 
g impression of the accuracy of the instrument. By the two settings executed for 
h each point, the actual accuracy of measurement may be judged. The difference 
|- between coordinates of the same grid points — resulting from both series of 
LS observations — when compared with the accuracy of measurement, provides 
an impression of the sensitiveness and stability of the instrument. À comparison 
2, of the observed and actual values of the grid point coordinates finally shows 
iS the accuracy obtainable with the instrument. 
e 8) If, after application of these repetitions and comparing the observed and 
d actual values of the coordinates, the final result is desired in mean square 
e errors, the method of this calculation must also be indicated. 
r 
| Besides the aforementioned considerations regarding the accuracy of the 
5