i6 4 
METHODS OF PETROGRAPHIC-MICROSCOPIC RESEARCH. 
The difference is not great, but it is noticeable, and is sufficient to make 
it advisable to use unmounted plates wherever possible, in optic axial meas- 
urements, if results of the highest accuracy are desired; but ordinarily 
this precaution is unnecessary, since such accuracy is not required. 
A rotation of the crossed nicols through 90 also generally produces a very 
slight shift of the axial bars from mounted plates, as indicated by Fig. 94, 
which is a direct record to scale of the observed phenomena. In each case 
the points along the central line of the axial bar were plotted. The posi- 
tion of this central line for an angle of rotation of 15 of the crossed nicols is 
indicated by the curve /, Fig. 94; its position for an angle of rotation of 105 
is shown by curve II. These two curves do not coincide and, although 
such measurements can not be made very accurately, they show that a 
rotation of the crossed nicols through 90 may, under certain conditions, 
FIG. 960. 
cause a slight though perceptible shift of the axial bars of the interference 
figure of a mounted crystal plate. The amount of shifting rarely exceeds 
several degrees and is usually less, but it is often sufficient to be perceptible 
and shows the importance of referring the data, when plotting, to the cor- 
rect position of the extinguishing plane of the upper nicol. It is, therefore, 
not immaterial which one of the principal nicol sections be chosen. If the 
observations themselves were of a higher order of accuracy, this fact would 
be a serious objection to Professor Becke's method. 
ANHYDRITE. 
A series of observations on a cleavage plate of anhydrite (unmounted, 
Fig. 960 ; and mounted, Fig. 966) corroborates the conclusions stated in the 
last paragraph. The degree of accuracy of the two methods in question is 
about the same here as in muscovite. A rotation of the crossed nicols 
