264 
SIR J. CONROY ON THE AMOUNT OF LIGHT REFLECTED 
attached to the goniometer was made as small as possible and blackened, and the 
goniometer itself was covered with a black cloth whilst the observations were being 
made. 
In order to measure the intensity of the light reflected by the glass both wdien 
incident normally and at various angles, that is, to compare the intensity of light 
which under certain circumstances would be partially polarised w^ith the intensity of 
unpolarised light, it was necessary that the photometric surface should be normal both 
to the incident light and to the line of sight; hence, the photometer wdiich had been 
used in the transmission experiments clearly could not be used, nor indeed could a 
Bunsen’s disk or any of its modifications. A new form of photometer was, therefore, 
devised. Two wooden screens were fixed to the sides of a block 10 cm. across, similar 
to the one which had carried the photometer in the first set of experiments; in these 
rectangular apertures were cut, 3 cm. by 2 cm., and “ parchment ” paper fastened 
over them ; the two right-angled glass prisms which had been used in the third 
method for determining the amount of the transmitted light (Plate 8, fig. 3a, p. 260) 
were placed between the screens, and in a line with the apertures. The two papers, 
each illuminated by the light of one of the lamps, were seen by reflection in the 
prisms, and by moving the block, to which an index was fixed, along the photometer 
board a position could be found in which the two images appeared equally bright. 
Glass only reflecting from 4 to 5 per cent, of the light incident normally upon its 
surface, and the photometer scale being only 2 metres long, it was impossible to 
compare the intensities of the dkect and reflected light when the two lamps had the 
same illuminating power. During the first set of experiments, those marked A, the 
necessary difference was obtained by keeping the flame of the comparison lamp turned 
down rather low. Subsequently, in the determinations marked B, the same result 
was obtained by different sized apertures in the metal chimneys ; that in the chimney 
of the goniometer lamp was 10 mm. by 18 mm., \vhilst the one in the chimney of the 
comparison or fixed lamp was only 10 mm. by 6 mm. ; the gas flames were so regu¬ 
lated that the apertures appeared completely filled with a uniformly bright flame. 
When the gas pressure changes slightly, the size of a flame, and not its intrinsic 
brilliancy, is mainly what alters ; and, therefore, by limiting the visible portion of the 
flame in this way a greater constancy in the ratio between the illumination produced 
by the two lamps was obtained ; but, as the tables show, the measurements made 
when the whole flame, and those in which the central portions only were used, agree 
satisfactorily. 
This method for determining the reflective power possesses the obvious defect that, 
owing to the necessary alteration in the course of the light, the direct and reflected 
light cannot be interchanged, and, therefore, a constant source of error may easily 
exist. It seems probable that, to a small extent, such was the case, and that the 
measured amounts ol* the reflected light were slightly too high. 
When the lamp attached to the goniometer Avas so placed that the light fell directly 
