UNDER THE INFLUENCE OF CHEMICAL ACTION. 39 
The full curve in fig. 19 is a replica of fig. 12, and the crosses denote the series of 
experimental points on which it is based. The auxiliary series of points in fig. 12 have 
been left out for clearness. The currents calculated as explained are shown by the 
points marked thus •, and when necessary the curve joining them is indicated by a 
broken line. This line, therefore, gives the characteristic as it would be if it were deter¬ 
mined entirely by the distribution of velocity of the electrons, and if this were a Maxwell 
distribution for the temperature T = 3300°K, which is the value given by fitting the 
points referred to above. It will be seen that the theoretical curve agrees with the experi¬ 
mental characteristic for all retarding voltages exceeding about —0-20 volts. There 
is a small deviation in the neighbourhood of zero volts which is the counterpart of the 
discrepancy between the theoretical curve and the experimentally derived points in 
fig. 18 in the same region. This process places the true zero voltage at —1 -84 volts in 
figs. 12 and 19 ; the limits assigned by the photo-electric measurements were —1-78 
and —1-88 volts, and the value here found lies almost midway between them. 
Fig. 20 deals similarly with the earlier data for COCl 2 given by the spherical electrode. 
The full curve is the composite curve of fig. 4 and the points © denote the best single 
set of data on which this curve is based. The crosses denote the calculated values 
on the assumption of Maxwell’s distribution, the fit at the two assigned points requiring 
a value of T = 3000 in this case. The broken lines show the theoretical characteristic 
as thus calculated. It will be seen that the agreement with the composite curve is satis¬ 
factory at the higher retarding voltages (shown on the left of the diagram), but that 
the deviation in the neighbourhood of zero volts is much more considerable than in 
fig. 19. However, this deviation is greatly minimised if the circular points are considered 
rather than the composite curve, and it may be that the rather considerable errors in 
the determination of this part of the curve have displaced it unduly to the right-hand 
