316 MR. R. T. GLAZEBROOK ON PLANE WAVES IN A BIAXAL CRYSTAL. 
of n 1 are greater than those given in Table III., and agree better with the observations 
in the neighbourhood of the other optic axis. 
Another set of measurements in the same neighbourhood gave as values of p 2 
1-68124 1-68131 
1-68123 1-68125 
Hence, on the whole, it seems likely that the results of the observations given in 
Table III., lines 12, 13, 14, are too small, and that an error has been made in the 
measurement at that point, owing to the indistinctness of the images which nearly 
overlapped, so that we may fairly replace Table II., lines 12, 13, 14, by Table XXVII., 
lines 12, 13, 14, and instead of the results of Table XXV. referring to these we shall 
have 
/O Theory. 
yu x Experiment. 
Difference. 
1-68114 
1-68130 
•00016 
1-68104 
1-68116 
•00012 
1-68096 
1-68127 
•00031 
This modification renders the results similar in the neighbourhood of the two optic 
axes. 
Let us now consider the results given in the latter part of the table for the wave E. 
The two sets of observations there recorded were made under entirely different 
circumstances. The telescope and collimator of the goniometer were removed in the 
interval between them, and both reset and refocussed ; and yet in but one case is the 
difference between the results of the two observations as great as '0001, while the 
average difference, irrespective of sign, is about "00005. 
This comparison forms a strong test of the accuracy of the experiments, and the close 
agreement of the results seems to show that the error in them is at any rate not greater 
than "00005. 
It would appear that more trustworthy results might have been obtained by com¬ 
bining the series of results registered in this last table with those discussed throughout. 
But experiments covering the whole ground embraced in the series we have chosen 
had not been made at the same time previously. 
Moreover, we must recollect that each value of <f) and D + i, given in Tables I., IV., 
&c., is the mean of three for the first prism and of two for the second. 
And, again, this comparison assures us that the difference between the results of 
experiment and theory amounting, as it does in some cases, to "0005 in the value of p, 
is far too great to have arisen from experimental errors or errors of observation. 
It seems, then, that assuming the position of the plane of the prism, and the 
values of the principal axes to be accurately determined, we may assert that in a 
central section of the wave surface inclined at a small angle to the plane of the optic 
