18 
PROF. O. W. RICHARDSON ON THE EMISSION OF ELECTRONS 
attributable to tlie rather high gas pressures which developed in these experiments. 
The fact that III. in which the pressure was kept down to 0 -002 by operating the 
pumps continuously, does not show this tendency, would seem rather to support such 
a contention. An examination in detail of the data, however, is against it. For example, 
in I. x and VI. • the currents at the high voltages were measured at the lower pressures 
and in II. © conversely. On the whole it seems likely that any such appearance is 
illusory and, except in the case of I. x , where a displacement caused by change in 
contact potential during the series seems probable, due to chance errors. 
It appears, then, that fig. 7 determines the shape of the chlorine characteristics. No 
determination of the contact potential with chlorine present has yet been made, so that 
the true zero on the volt scale in fig. 7 cannot yet be assigned with certainty. All that 
can be said at present is that an examination of the relative positions of the chemical 
and of the photo-electric curves which were taken at the same time shows that the true 
zero of fig. 7 lies to the right of — 2 -2 volts. 
§ 4.— Hydrochloric Acid. 
The effects given by this gas are of a much smaller magnitude than those given by 
carbonyl chloride and by chlorine. It was impossible to make accurate measurements 
of them with the apparatus used with the latter gases, and such fragmentary data as 
were obtained only enable qualitative statements to be made. The characteristics 
appear to be similar generally to those given by the other gases, and to drop from satura¬ 
tion to zero current within a range comparable with one volt. 
§ 5.— The Characteristic Curves for Water. 
These experiments were made in 1915 with an apparatus which differed in some details 
from that shown in fig. 1 and used in the experiments described up to this point. Instead 
of the glass tube ABC of fig. 1, the alloy was fed into the testing vessel through a copper 
tube fitted with a tapered silver nozzle. Tins had a fine hole bored vertically down 
the centre and was screwed on to the copper tube. The other electrode was 7 -5 cm. 
in diameter. The alloy used was also different, having the composition NaK instead 
of NaK 2 . The electrometer sensitiveness was 540 divisions per volt and the capacity of 
the electrometer and connections 0 -00032 mfd. There is an element of doubt about 
the correctness of the voltmeter set up which was used in some of these early experi¬ 
ments, and I am only prepared to state the applied potential differences as relatively 
correct. The units are probably volts, but they may be as small as 0 -6 volt. The 
data for the series of observations which were completed are given in fig. 8, and some 
of the details in Table III. and in the remarks which follow. 
