152 Messrs. Holborn and Austin on Cathode Disintegration 
the observation errors. The loss of weight of the aluminium 
cathode was only 0°01 to 0:02 mg. (V=800 to 1800), while 
iron and steel at 1000 V. did not seem to disintegrate at all, 
and at 2000 V. lost 0°15 mg. 
In the above-mentioned observations on gold, it appears 
that the amount of the disintegration is too small in com- 
parison with the chemical equivalent, considering gold as a 
uni-valent metal. As gold in some other cases, as we shall 
see later, behaves regularly, we can suppose that possibly the 
irregularities are caused by the method of observation *, or 
we may suppose that a portion of the atoms under certain 
conditions assume a larger valency. 
Observations on platinum and bismuth make the last 
assumption probable, since a few cases occurred in which 
both metals, which we have assumed to be di-valent, must be 
treated as tri-valent if we are to make the formula agree 
with the observed values. For example, for V=800 and 900, 
bismuth gave the disintegration y=44 and 51 (the tri-valent 
metal would give 39 and 52 as the calculated values). In 
the case of platinum, for V=1900 the value y=140 was 
once found (the tri-valent metal would indicate 143). If it 
is possible for atoms to be thrown off having different 
valencies, the curves calculated for the smaller valencies can 
only be ‘considered an upper limit for the possible loss in 
weight of the cathedes. 
3. Observations of Granquist. 
We will now undertake to determine how far our hypo- 
thesis can be applied to the observations made by Granquist. 
These were made with comparatively strong currents; the 
Icsses of weight, however, are given only to tenths of 
milligrams. The difference of potential between the cathode 
and the tube-wall was not often more than 1006 V. Several 
observations were made at very low potential, which on account 
of the above mentioned grounds we do not consider With 
these exceptions, we make use of practically all of his obser- 
vations, and, indeed, not only those from his Tables IV. to VI. 
where the aren was constant, but also the values in Tables I. 
to III. and in Table VII. ae it was varied. The cathodes 
were in most cases thin metal sheets, 12 mm. long and 
4-8 mm. broad, which hung parallel to the axis of the tube, 
so that both aie were subject to disintegration. It was 
* A recent observation indicates that gold which has stood under 
vacuum for a long time in connexion with a mercury-pump becomes 
amalgamated. 
