362 
MR. E. G. BILHAM ON THE SPARK SPECTRUM OF 
being made on lines which reverse under pressure. The head of the micrometer screw 
is divided into 60 divisions, and whole turns are read off on the comb d. 
In measuring displacements the following procedure was adopted. The micrometer 
was first adjusted so that the wires a, b u b 2 were at right angles to the direction of 
travel of the microscope. The plate was then placed on the stage of the measuring 
machine so that the line of separation of the pressure spectrum and one of the normal 
comparison spectra bisected the field of view, and so that the spectrum lines were 
parallel to the cross-wires. Having brought the line to be measured into the middle 
of the field of view, the final settings were made by means of the eye-piece 
micrometer. The normal and displaced lines were made to bisect the space between 
the wires b l and b 2 alternately, six settings being made on each. In the case of 
reversed lines greater precision is obtained by using a single thread. Consequently, 
Fig. 1. 
in such cases, the wire a was brought into requisition, the normal line being 
measured as before. A constant difference is, of course, introduced corresponding 
to the distance in scale divisions between a and a line midway between b 1 and b 2 . 
Determinations were made using each comparison spectrum and measuring in both 
directions. Four sets of readings were thus obtained from each plate. The mean 
displacement determined in this way bad to be multiplied by a dispersion-factor to 
o . 
convert scale divisions into Angstrom units. The factor was obtained by measuring 
the separation of suitable lines in the iron arc spectrum on the same plate, using 
Dr. K. Burns’ values for the wave-lengths. One division of the micrometer head 
corresponded to about 0'0027 A.U., the factor varying slightly in different parts of 
the spectrum, being smaller towards the red. Headings were taken to tenths of a 
division, but the final results are only expressed to O'OOl A.U. In the case of good 
lines the error is probably not more than two units in the third place of decimals. 
