20 PROF. O. W. RICHARDSON ON THE EMISSION OF ELECTRONS 
much exceed 0 -002 mm. at any time. In V. © the water vapour was obtained by 
placing crystals of CaCl 2 6H 2 0 in the generating tube and immersing it in ice. In 
IV. □ the same crystals were used but the container was immersed in a mixture of 
ice and salt. In all other cases mixtures of sulphuric acid with different suitable propor¬ 
tions of water were used, and the mixture was kept at the temperature of the room- 
In VII. A equal parts of the two liquids were taken. The water vapour was given off 
quietly and the liquid did not boil. In I. to V. the drops fell at the rate of about one 
per minute, in VI. at the rate of seven per minute. After VI. a large structure with a 
wart-like appearance had grown on the end of the nozzle, and in VII. the globules came 
out of this structure. They were very large and looked dirty when attacked by the 
IT 2 0. In the previous experiments, in which much larger currents were obtained, the 
oxidised layer either looked quite white or was invisible. It was noticed that the 
oxidised metal showed no appreciable photo-electric effect with the quartz mercury 
lamp except just when the oxidised skin broke and the bright metal shone through. 
In series I. to V. the electrometer deflections were reduced by adding suitable capacity 
from an adjustable air condenser. The added capacities are not recorded in these 
cases. In VI. 0 -2 mfd. was added from a standard condenser and the maximum satura¬ 
tion current was about 50 divisions per minute. In VII. the capacity was that of the 
apparatus alone (0*00032 mfd.) and the saturation current was 50 divisions per minute 
in tliis case also. VI. and VII. probably correspond respectively to the largest and 
smallest effects dealt with. 
In fig. 8 the individual data I. to VII. are plotted as they were experimentally deter¬ 
mined. The points for I. x and V. © fall so close together that they appear to fall 
on a single curve. The same applies also to each of the pairs IV. 0, VII. A and III. 
VI. §j|. Thus the seven sets of data appear to fall on four distinct curves. However, 
these curves are all nearly parallel to one another, just as was the case with the curves 
for C0C1 2 and Cl 2 . They have therefore been subjected to the same treatment, namely, 
given the arbitrary displacements which are recorded in Table III. parallel to the volt 
axis. The result is shown in fig. 9. It will be seen that, with the exception of the points 
marked a , b and c, all the points then lie on a single smooth curve to within the degree 
required by the probable experimental errors. Of the excepted points, b is doubtful 
owing to a rapid variation of emission with time when it was taken, and a and c should 
not, except for the sake of having a complete record of the observations, have been 
included in the diagram, as they are determined by minute deflections comparable in 
magnitude with the variations in the natural leak of the electrometer system in this 
particular experiment. 
Fig. 9 shows that the form of the characteristic curve for H 2 0 is of the same general 
nature as those for Cl 2 and C0C1 2 . No determinations of the contact potential K with 
water present have been made, so that the position of the true zero of potential difference 
on fig. 9 is uncertain. 
