122 
Proceedings of the Royal Society of Edinburgh. [Sess. 
previously calibrated, a suitable resistance, R, being placed in series. 
Owing to the fact that it was impossible either to have a point contact of 
the thermo-junction with the foil, or to ensure that the two points at which 
the junction wires left the foil should be on an equipotential line, it was 
always necessary to read the galvanometer deflection before and after 
reversing the direction of the heating current through the foil. The mean 
of these deflections (whose difference was never large) gave the true thermo- 
electric E.M.F. generated. 
The thermo-electric junction itself was calibrated against a reliable 
mercury thermometer up to about 300° C. and at 800° C. by immersing in 
molten sodium chloride at its freezing point. A more accurate calibration 
was unnecessary for our purpose. 
The platinum foil was heated by the current from two large secondary 
cells, b, a suitable rheostat, Rh, ammeter, A, and reversing switch, D, being 
placed in series. All these, as well as the instruments in the thermo-electric 
circuit, were carefully insulated from earth in order that the platinum foil 
might be raised tp any desired potential by small storage cells (W. G. Pye’s 
pattern). 
The copper electrode, Cu, was connected directly to one pair of quadrants 
of a Dolezalek electrometer of high sensibility, and, except in the intervals 
required to take readings, to earth through the key K. A standard 
Muirhead condenser, the capacity of which could be varied from 0001 to 
0*2 microfarads, was also placed in parallel. The second pair of quadrants 
of the electrometer and the other plates of the condenser were permanently 
earthed, whilst all wires leading to the electrometer, the condenser, key, 
etc., were enclosed in earthed metallic casings. 
The measurements of the photo-electric currents were made as follows : — 
The primary circuit of the induction coil was closed, starting the ultra- 
violet light ; about three seconds later the pair of quadrants of the electro- 
meter connected to the electrode Cu were insulated by breaking the earth 
connection at K electromagnetically, a stop watch being started simul- 
taneously. The light was switched off exactly ten seconds later, and the 
electrometer deflection read at leisure as soon as it had become steady. 
Accuracy to 2 per cent, was always attained. 
The platinum or sensitive electrode could be charged up to any potential 
up to 200 volts from small storage cells, b, and up to 435 volts by utilising 
the city mains. 
The leak apparatus was connected to a Topler pump, M‘Leod gauge, 
and suitable drying tubes. Plugs of gold and silver leaf were inserted 
between the pump and leak apparatus in order to exclude any mercury 
