356 
ME. W. CEOOKES ON EEPULSION EESITLTINGr EEOM EADIATION. 
and timed its oscillations. 10 half-vibrations were taken by a chronograph. The times 
recorded during three experiments were 
9 - 5 seconds, 9-6 seconds, 9 ‘6 seconds ; 
so that in vacuo the beam ought to take about two seconds for each complete oscillation. 
The pith surfaces at the ends of the beam are 13 millims. square ; they are very thin, 
and are lampblacked on the surface. The window in the centre (d) is of plate glass, 
the window at d' is of quartz. They are on opposite sides of the apparatus, as will be 
better seen by referring to fig. 13. 
187. When fitted up for spectrum observations the whole arrangement is shown in 
fig. 13. a b is the torsion-apparatus, shown in plan, the pith disks being represented by 
black lines at the ends of the central fine line representing the beam. The suspended 
Eig. 13. 
mirror is shown in the centre of the beam. The quartz window is shown at the end 5, 
and the other window opposite the central mirror, c, c, c are Winchester quart bottles 
full of water, and each encased in brown paper to prevent it accidentally acting 
as a lens, and condensing light on any part of the apparatus. Between the bottles, 
and surrounding the apparatus on all sides, as well as above and below, cotton-wool is 
well packed, spaces being left only for the rays of the spectrum to pass to the pith disk, 
and for the index ray of light to pass to and from the mirror. The cotton-wool acts as 
an excellent non-conductor of heat, and also prevents air-currents. The water in the 
bottles keeps the apparatus of a nearly uniform temperature, and entirely prevents sudden 
