ME. W. CEOOKES ON EEPULSION EESULTING EEOM EADIATION. 
535 
quartz window in front of the pith surface on which radiation is to act. The latter tube is 
furnished with card shutters, h, i, at each end, capable of easy movement up and down. 
The whole apparatus is then closely packed on all sides with a layer of cotton-wool, 
about 6 inches thick, and outside this is arranged a double row of Winchester quart 
bottles, j, j, filled with water and covered with brown paper, spaces being only left in 
front of the paper tubes. Jc and l represent the positions of the candle 140 and 280 
millims. distant from the pith. The whole arrangement has the appearance shown 
in fig. 8. 
Fig. 8. 
105. I will not discuss at present the phenomena presented when the apparatus is 
full of air, or when the vacuum is imperfect, but will proceed to the effects observed 
when the exhaustion has been pushed to the highest attainable degree. However much 
the results may vary when the vacuum is imperfect, or when the apparatus is full of 
air, I always find them agree amongst themselves when the residual gas is reduced to the 
minimum possible ; and I have also ascertained that it is of no consequence what this 
residual gas is. Thus I have started with the apparatus filled with various vapours and 
gases, such as air, carbonic acid, water, iodine, hydrogen, or ammonia ; and at the 
highest rarefaction I find no difference in the results which can be traced to the residual 
vapour, assuming any to be present. A hydrogen vacuum seems neither more nor less 
favourable to the phenomena than does a water or an iodine vacuum. If moisture be 
present to begin with, it is necessary to allow the vapour to be absorbed by the sul- 
phuric acid of the pump, and to continue the exhaustion with repeated warming of the 
apparatus until the aqueous vapour is removed ; then only do I get the best results. 
When pith surfaces are used at the extremities of the glass beam, they should be per- 
fectly dry ; and they are more sensitive if the apparatus has held a vacuum for some 
weeks, as the residual moisture in the pith will then have been absorbed by the sul- 
phuric acid in the pump. 
106. It was found that when a source of light and heat is suddenly allowed to shine 
on the pith surface and not removed, a deflection rapidly takes place, attaining its 
