40 
PROF. O. W. RICHARDSON ON THE EMISSION OF ELECTRONS 
side. In any event the deviation is of such character and magnitude that further 
investigation may assign it to causes outside the energy distribution. The true position 
of zero volts given by this calculation is + 0 -40 volts on the scale and is well within 
the limits, marked by vertical lines at A and B on the diagram, set by the photo-electric 
observations. 
It will be seen that figs. 18, 19 and 20 lead to three separate estimates of the value 
of T for the case of COCL, viz., 3600, 3300 and 3000. Of these the first and second use 
the same experimental data but treat them in different ways. The third employs 
different data but treats them in the same way as the second. The mean of these 
determinations is T = 3300°Iv. It is probable that an agreement to within 10 per cent, 
of the value of T is as much as it is reasonable to expect at the present stage of develop¬ 
ment of the subject. 
Turning to fig. 21, the full curve is a copy of the chlorine curve in fig. 7 and the broken 
curve represents the theoretical characteristic assuming a Maxwell distribution, the 
calculated points being shown by the crosses. The value of T for this curve was found 
to be 4350. The general appearance of fig. 21 is the same as that presented by the 
C0C1 2 data in fig. 20, and, in general, the same remarks apply to both curves. The 
triangular area CDE, which measures the discrepancy between the calculated curve and 
the experimental characteristic, is somewhat larger in fig. 21 than in fig. 20, but the 
increase is only about in proportion to the greater horizontal extension of the chlorine 
diagram corresponding to the higher value of T. Whatever the difference between 
the full and broken curves is due to, it is probably caused by similar factors operating 
in the case of both gases. The value of the true zero given by the calculated curve in 
fig. 21 is — 3T1 volts, which is well to the right of the left-hand limit —2-17 volts 
