208 PROCEEDINGS OF SECTION B. 
salts, we obtain the specific velocities of the different negative 
ions, all referred to that of the K ion as unity ; and thus these 
negative ions can be indirectly compared with one another. In 
the same way various chlorides, for instance, may be dealt with ; 
and thus the specific velocities of the various positive ions may 
be indirectly compared. 
Il. RELATIVE VELOCITIES OF THE IONS OF THE ALKALI METALS. 
In the paper already quoted, I gave the relative velocities of 
K, Na, and Li as deduced from experiments with their chlorides 
and sulphates in different states of concentration. The values 
were shown to remain nearly constant and to agree well with 
those calculated from Kohlrausch’s determinations of the con- 
ductivities of aqueous solutions of salts of similar concentration. 
I have since made experiments with half-rormal solutions of 
rubidium and caesium chlorides, thus completing the alkali- 
metal group. Fresh observations were made at the same time 
with half-normal potassium chloride, so as to connect the two 
sets of experiments. The found value for the K velocity was 
1.07 (putting the Cl velocity as 1.00), which differs slightly 
from the older value of 1.02. This is probably due to the fact 
that the gelatine employed in the two cases was not identical, 
though very similar, and that neither sample was wholly free 
from electrolytic impurities. The earlier sample gave nearly 
.5 per cent. of ash, while the later one gave about 1 per 
cent.; so that the 1.02 value is probably the more correct. 
There were other slight differences in the conditions, but none 
of a kind to affect the numerical results to an appreciable extent. 
The jellies contained 10, instead of 12, per cent. of gelatine. 
The temperature was about 15 deg. C, instead of 18 deg., which 
would reduce the absolute velocities, but not, appreciably, their 
relative values. The tube used was 12 cm. long instead of 
15 em.; which materially reduced the time spent on an experi- 
ment, since this varies as the square of the length of the tube, 
but did not otherwise affect the results. The internal diameter 
of the tube was the same as before, namely, 2.2 mm.; but of 
course the ionic velocities are independent of this factor. 
In the following tables the details of four experiments are 
given in condensed form, omitting all readings of position and 
time except those corresponding to exact scale divisions for the 
yellow advance. The simultaneous galvanometer readings are 
also omitted as they do not enter into the calculation of the 
relative velocities of cation and anion, their purpose being to 
afford data for proving that (as is the case) the actual current 
density agrees with that calculated from the observed velocities 
and the known concentration of the salt. The ratio of the blue 
advance to the yellow advance at each stage is given in a separ- 
