122 
^IR. P>. D. STEELE OX THE MEASUREMEXT OF TOXIC VELOCITIES 
could be found are the Cl and ^ Cl. 
IM.o' Li 
l)et\reen this and the Nernst theory of liquid cells. To establish the relation, 
liowever, it is necessary that a very much larger numljer of experiments should be 
carried out. In the table only five of the experiments correspond to the minimum 
potential fall. This follows from the fact that the investigation had for its first aim 
the production of a margin, and not the determination of the limits of voltage within 
which this may he produced. 
Two of the boundaries previously referred to for which no satisfactory conditions 
From Table II. it is seen that for this pair 
of solutions no change in potential slope is brought about. The densities and 
refractive indices of the two solutions, however, differ very considerably. Perhaps it 
is on this account that it is possible to obtain an indistinct and undulating refraction 
margin whether the Li follows Mg or Mg Li; l^ut in neither case is the velocity 
constant. This is undoubtedly due to the fact that, since there exists no difference in 
potential fall at the two sides of the margin, there is no controlling force which shall 
prevent the faster ion difihsing into the .slower, or vice verscl. Diffusion therefore 
takes place, tiie resistances become .slightly altered, and so also tlie potential fall and 
(M 
velocities. A similar case is that of the Oh N 7— SO, maroin. Here, with a low 
Cu ^ ^ 
voltage, a mixed colour boundary, which cannot be located nearer than peihaps 
1-2 millims., results; Imt when the voltage is increa.sed, a sharp colour boundarv 
idtimately appears at about 6h volts, and from the table it is seen that even with 
7 volts the change in potential slope only amounts to 0'30 volt. That no refraction 
margin can be detected witii this pair of solutions at this concentration (such a margin 
is obtained in more concentrated solutions) is explained 1)}^ the circumstance that the 
refractive indices of the two solutions lie very close together. Neex.st, in describing 
the experiment previously quoted, refers to the necessity for a high voltage that will 
allow the experiment to be completed in a comparatively short time, as the margiir 
would otherwise be lost l3y diffusion. Tliis is another case in which there is no 
change in potential slope on ])assing from one solution to the other, tlie velocities of 
the NOo and MnOj^ ions being practically identical. 
The case of cadmium cldoride as indicator followino' 2 N solutions of the chlorides 
o 
furnishes examples of another difficulty. Here the boundaries obtained are the most 
distinctly to l)e seen of any that have l)een investigated, but the motion is always 
irregular and much too slow. These irregularities are probably due to a mixing In- 
convection currents set up as previon.sly described by the greater heating in the 
indicator solution ; they may, however, Ije in some way connected with the nature of 
the anions in the cadmium chloride solution, which according to Hittoef, are not 
simple Cl ions, Imt something more complex. 
For the systems Cl and Cl also an undidatinv refraction maro-in results iu 
both 
cases 
Cu Li 
Imt in the former case the copper lags l)ehind tlie margin, and the 
