4 B, E. Moore 
Intensities are recorded for the components of the first order 
lines, whether the lines were measured in first or second order. 
The lowest intensity I is just observable. It is measurable only 
when very favorably located in a group of components, and then 
with no special accuracy. A line of intensity 2 is capable of fair 
measurement in groups of several components, but is not satis- 
factory in a doublet of broad separation. Intensity 3 is the lowest 
that is satisfactory under the latter conditions. Diffuse lines of 
higher intensities may also be unsatisfactory. A pair of com- 
ponents which are diffuse outward or inward indicates the pres- 
ence of outer or inner weak components, which may be brought 
out when self-induction, capacity, slit-width, and exposure of 
plates have been properly balanced. Broad single lines suggest 
a similar possible resolution. Hence a record of these facts may 
be helpful even if the resolution has not been effected. The ac- 
curacy of the readings also depends upon the sharpness of the 
lines. Where weakness, diffusion, or presence of overlapping 
“components has made a reading less accurate, the result is tabu- 
“lated in brackets. The bracketed values are still close enough to 
give a reasonable idea of the magnitude of the separation. Over- 
lapping components can usually be circumvented by varying the 
field strength. However, this did not always suffice, and it was 
necessary to omit an occasional component in a line, e. g. the outer 
violet component of zircon 3573.3. There is large variation in 
definition among the lines of a substance. This is particularly 
noticeable in barium. The lines of osmium are uniformly excep- 
tionally sharp, so that the readings in the latter possess nearly 
three times the accuracy of the former. Yttrium and zirconium 
occupy an intermediate position. The separations are deterinined 
from five readings for each component. These were all repeated, 
and in most cases the repeated readings were taken upon other 
sets of plates. If these readings showed an unwarranted deviation 
from the old result, the line was subjected to further analysis upon 
both plates. In zircon there is certainly an error of .oo5 mm. 
possible. Cases of double the magnitude may have escaped my 
attention, but such cases are few. This value gives a minimum 
error in AA/A? in the second order at 4500 of .o25, and a maximum 
