42 HH. M. Parkhurst on the Photo-mapper. 
3. If the disk is small, the accuracy of comparison is im- 
paired. Hence points of light cannot be as accurately compared 
as disks. I have used a disk of an apparent magnitude of at 
least 1°, and have not investigated the amount of error with a 
smaller disk. 
e error from the variation of the artificial disk during 
observation. I have made many and careful experiments to 
render this error as small as possible, and to avoid ascribing to 
the irregularity of the sky diserepancies which might be owing — 
to variations of the artificial disk. The brightness of the arti- _ 
ficial disk may vary from four causes: et 
1. The illuminating quality of the gas may vary. Hence © 
observations on different evenings cannot be compared directly. 
But the change of the gas will be so slow, excepting probably — 
from the heating of the apparatus when it is first lighted, as not 
ito appreciably affect the results. : 
2. From variation of the pressure the flame may be made 
larger and brighter. I first partially corrected this by forming 
upon the screen an image of a circle of the flame ‘05 in. in di- 
ameter. I afterwards made a gas regulator, admitting the gas 
into an inverted receiver suspended at one end of a lever, the _ 
other end of which gradually shut off the gas as the receiver — 
rose; so that whatever the pressure of gas in the mains there — 
should be no variation in the flame. 
and reduced the flame and moved it further back, so as to use 
the whole flame. 
4. The light from the star is a sharp cone, wholly entering 
the eye even if withdrawn several inches from the me | 
; Sy : nl 
therefore the eye-piece renders the rays parallel, which it will 
not if the observer is either near-sighted or far-sighted, a differ- 
ence in the distance of the eye will greatly affect the results. 
I have therefore turned the plane glass in the eye-piece so as to 
ly corresponds with that from the star. It is necessa that the 
eye should be held so that both cones shall completely enter it 
at once, which can-be easily accomplished but requires care. 
__ The error from the variation of the artificial dise, without the 
