The Magnetic Separation of Lines 5 
error in the first order at 3250 of .og. Hence it was desirable as 
far as possible to limit the readings in the shorter wave-lengths 
to the second order photographs. 
The identification of the lines in the violet and ultra violet 
spectra was made by Exner-Hascek’s spark spectrum tables, which 
generally proved very satisfactory. An occasional difficulty was 
met; for example, in zirconium a very well-defined line of eight 
components was found at about 4214 which failed to identify with 
impurities or with Zr. 4214.58. It does, however, identify with 
the arc light table line 4214.05. Such a difference in arc line wave- 
length and spark line wave-length is very exceptional, but in the 
absence of line 4214.58, it indicates that line in question is the 
same. Rough measurements frequently gave readings closer to 
the arc light lines than the spark lines, and many such might have 
occurred had very careful comparator measurements of distances 
from line to line been made. Sometimes such comparator meas-_ 
urements are necessary in identifications, as it may be ‘either a 
required line or an impurity. Difficulty was met with in three 
lines toward the red of 3392.20. These eight lines measured from 
3392.20 gave 3393.30 instead of 3393.30; 3394.96 instead of 
3394.79, or possibly this is another line; and 3396.87 instead of 
3390.71. A fourth line 3396.49 is certain!y in close enough agree- 
ment with 3396.51. Several other lines appeared which have nct 
been identified. 
lil. EXPLANATION oF TABLES 
The abbreviations in the tables have the following significance, 
viz.: A, wave-length; AA/A?, the change in vibration per cm.; 
s, vibrations perpendicular to the lines of force, and /, parallel to 
the same; H, principal series; G, greater wave-length: K, smaller 
wave-length; NV, subordinate series; S, satellite; h, principal line; 
R, observations according to Runge; M, observations according 
to Moore; i, intensity. Column A gives the approximate value 
of the components represented in terms of a small separation, 
called the “interval,” multiplied by small numbers called ‘“‘factors.” 
The factors represent the ratio of the distances of the sticcessive 
ul 
