UNDER THE INFLUENCE OF CHEMICAL ACTION. 
23 
frequency G -89 X 10 14 gave a large deflection shows that this frequency was considerably 
higher than the threshold value. If the small value of the current with X 5460, fre¬ 
quency 5 -49 X 10 14 could be relied on as genuine, it would establish v<s as very close to 
X 5460 and just on the low-frequency side of it. On the other hand, if this small deflec¬ 
tion is spurious or due to a trace of light of higher frequency mixed with the X 5460, 
the threshold value must be well on the low-frequency side of X 4350 on account of the 
large deflection given by X 4350. This experiment shows that v 0 was somewhere 
between 5-49 X 10 14 and 6 -89 X 10 14 at this stage. A plot of the photo-electric data 
got during series T. x shows that the photo-electric current with X 4350 met the voltage 
axis at — 0 -20 volt. This was immediately below the 
point at which the chemical current attained 0 -50 of its v \ Fig.II. 
saturation value. Taking the frequency as 6-89 X 10 14 
for X 4350 and 5-49 X 10 14 for X 5460, the value of 
(v — i/ 0 ) h/e is 0 -58 volt for v t> = 5-49 X 10 14 and zero for 
„ 0 = 6 -89 X 10 14 . Thus the true zero must lie between 
a point which is coincident with the — 0 -20 volt on the 
voltage scale and with the 50 per cent, point on the 
chemical characteristic and a point which is 0 -58 volt 
negative to these points. At this period the contact 
potential difference between the two electrodes then 
was between 0 *2 volt and 0-78 volt, and the true zero 
between the points where the chemical current had 
attained between 0 -50 and 0 *86 of its saturation value. 
Comparing with the composite curve in fig. 4, this makes 
the true zero lie between -f 0 T2 and +0-70 volt on that 
diagram. It must be definitely to the right of + 0 *70 
on account of the large deflections given by X 4355, but 
how far to the right these data do not determine. 
The importance of fixing this zero led me to make a 
renewed attack on this part of the problem recently. 
For success in this determination it is necessary to have 
a series of intense sources of monochromatic radiation 
not too far apart on the frequency-scale. It is also 
necessary to carry out the successive measurements with the utmost rapidity. 
By this time the apparatus shown in fig. 1 had come to grief and a new 
testing vessel of simpler construction, shown to scale in fig. 11, was employed. 
Apart from the smaller dimensions, the principal change consists in the substi¬ 
tution of a cylindrical copper electrode, 2 cm. in diameter, instead of a spherical 
one. The alloy used was NaK 2 and the COCl 2 was the Kahlbaum specimen already 
referred to. The monochromatic sources were the lines of the mercury arc 
spectrum projected on to the drops by a Hilger glass monochromatic illuminator. 
