LYMAN. — TIIE SPECTRUM OF IIYDKOGEN. 141 
once suggests itself: Is this test sufficiently delicate for the present purpose? If very 
accurate results are required the question must be answered in the negative. A little 
consideration makes it obvious that the relative position of the images J and /', and 
hence the shift, changes with the focus more rapidly than can usually be detected 
the change in sharpness of the lines. In other words, if the shift were given, the proper 
focus could be accurately determined, but if the sharpness of focus must be relied on, 
then the true shift can only be approximately inferred. Or again, for practical pur- 
poses, the apparent shift varies slightly more rapidly with variation in adjustment 
than does the sharpness of the spectral lines. The foregoing is, of course, somewhat 
dependent on the manner in which the adjustment is made. In the apparatus in 
question the slits and the photographic plate are rigidly fixed on the arc of a circle. 
This arc is capable of being thrust in or out parallel to itself along a line connecting 
the centre of the grating and the centre of the photographic plate ; it is also capable 
of rotation about its middle point. By these two movements perfect adjustment can 
be attained, but the test of this adjustment is not absolutely adequate. 
The practical application of the method is as follows: The spectrum 
f 
selected for comparison work. The grating was so turned that known lines in the 
spectrum of aluminum fell upon one end of the plate when the right hand or direct 
was illuminated by light from a spark between terminals of the metal. The shift 
a 
of the principal spectrum with respect to the comparison spectrum was then deL. 
mined by comparing the positions of these lines in aluminum with known lines in the 
spectrum of iron. In order to insure accuracy this shift determination was recorded 
on the same plate as the spectrum of hydrogen whose lines were to be measured. 
This was conveniently brought about by admitting the light from the aluminum spark 
directly through the discharge tube, for which purpose the tube was fitted 
window of quartz at the end not attached to the face plate. Upon the spectrum to 
be measured was superposed the comparison spectrum of iron, and in this spectrum 
fiducial lines were selected. The relative value of these lines was then obtain. I by 
subtracting the shift from their real value, previously corrected ^ nflnn. These 
relative values were then used as points of departure to deter 
igths of 
the unknown gas spectrum. In practice 
ft was 1180 A 
point in the iron spectrum falling on say X 1400 of the gas spectrum has 
of 1400 + 1180 = 2580 Angstroms. 
Owing to the dimensions of the plate only a region of about 760 tenth mc 
be photographed at one time. Thus if the aluminum line 1935.29 falls upon 
me of the plate the 
d corresponds to wave-length U175. I" order to 
