IN QUAETZ OF LIGHT IN THE VISIBLE REGION OF THE SPECTRUM. 
285 
placed at one end or the other of the water-jacket, according as the quartz 
required to be tilted forward or backward. By means of the reflecting eye-piece, it 
would have been easy to set the quartz within a fraction of the thickness of a card 
but this was not required, as it was necessary to insert or remove some half-dozen 
cards before the readings were altered by more than 0°'01. The appearance of the 
reflected image was the best test of the optical contact of the cylinders ; if these came 
apart (as always happened if water was spilled into the jacket) a series of widely 
separated images was seen ; but if the optical contact held, the grinding was so perfect 
that it was barely possible to recognise the double character of the image reflected 
from the ends of the column of quartz, the error in parallelism being probably only a 
fraction of a minute of arc. 
4. The axis of the quartz, which in the earlier plates had been located in the ordinary 
way by means of the interference rings, was set much more carefully in the later 
' plates, using the very sensitive method described by J. Walker (‘Phil. Mag.,’ 1909, 
vol. 18, 195). The error in setting the axis normal to the surfaces was estimated not 
to exceed 5 minutes of arc. As the cylinders could be tilted more than this without 
increasing the readings, there was no reason to suppose that the setting was not 
sufficiently accurate, even when taking readings to 0°'0001 per mm. This conclusion 
is all the more reasonable in view of the fact that readings concordant within about 
0°'001 per mm. have been obtained by a number of different workers for the optical 
rotations of sodium light in thin plates of quartz, set to axis by the rougher methods 
generally in use. 
5. The temperature was maintained at 20° C. by means of a generous flow of water 
at constant temperature. The spiral gas-regulator used in the thermostat was of tlie 
pattern designed in 1905 (‘Trans. Chem. Soc.,’ 1905, vol. 87, 1030-1034) and was 
capable of maintaining the temperature of the well-stirred bath, over long periods, 
within 0°'01 of the desired temperature. The water was drawn out of the bath and 
through the jackets by means of an Albany pump of |-inch bore, the arrangements 
being in other respects very similar to those described and figured in an earlier paper 
(‘ Trans. Faraday Soc.,’ 1907, vol. 3, 119). As the pump works best with an ample flow 
of water, the water jackets were always arranged in parallel, thus avoiding the obvious 
drawbacks of a series arrangement. Thermometers were provided for reading the 
temperature of the return-flow, which usually differed from that of the bath by 
something of the order of 0°'01 for each degree of difference of temperature between 
the water bath and the atmosphere. All the jackets were double, the inner flow of 
water being shielded by an outer flow at practically the same temperature ; with the 
exception of about 2 feet of rubber tubing, all the cooling surface was on the return 
flow and it is reasonable to suppose that 90 per cent, of the drop of temperature took 
place there. Thus, although the readings of the polarimeter changed by about 0°'01 
for each 0°‘01 C.,the regulation of temperature was probably quite sufficiently accurate 
for the purpose. 
