532 
ME. W. CEOOKES ON EEPULSION EESTTLTINGr EEOM EADIATION. 
and breaking contact by hand is not sufficiently certain, and hesitation for a fraction of 
a second would seriously affect the ultimate amplitude of arc. I tried making and 
breaking by clockwork, also by a seconds’ pendulum, but there were difficulties in each 
plan. 
Owing to the mode of suspension, there was uncertainty as to the length of the pen- 
dulum. I tried to make it the right length to beat seconds in vacuo. Assuming that 
I had succeeded in this, the pendulum would have executed fewer vibrations in the 40 
seconds when oscillating in air, and consequently I should not have got the full benefit 
from the making and breaking contact, supposing these were accurately timed to 
seconds. 
The battery-power varied, being stronger at the commencement, and gradually 
declining towards the end of the experiment ; and even were the battery to remain con- 
stant, the spiral became much hotter, owing to the removal of the air from the appa- 
ratus, ranging from a bright red heat in air to a full white heat in vacuo. 
Owing to the height of the centre of suspension of the pendulum from the stand of 
the apparatus, the slightest deviation from the perpendicular made an appreciable dif- 
ference in the distance of the weight from the spiral, and thereby increased or diminished 
the effect of radiation. Thus the tread of a person across the floor of the laboratory, 
or the passage of a cart along the street, would cause the image of the edge of the mag- 
nesium weight apparently to move from the cross wires in the telescope. 
Many of these sources of error could have been removed ; but in the mean time having 
devised a form of apparatus which seemed capable of giving much more accurate results, 
I ceased experimenting with the pendulum. 
Before proceeding to describe the apparatus subsequently employed, I may men- 
tion that a candle-flame brought within a few inches of the magnesium weight, or its 
image focused on the weight and alternately obscured and exposed by a piece of card 
at intervals of one second, will soon set the pendulum in vibration when the vacuum is 
very good. A ray of sunlight allowed to fall once on the pendulum immediately sets it 
swinging. The pendulum-apparatus above described was exhibited, and experiments 
shown with it, at the Boyal Society, April 22nd, 1874, and also before the Physical 
Society*, June 20th, 1874. 
101. The difficulty which attended experiments with the balances and bulb-apparatus 
used at first was to bring the moving part accurately back to zero, and also to measure 
the deflection produced. I therefore tried several plans of giving a fixed zero-direction 
to the movable index. Thus a piece of magnetic oxide of iron was cemented to one end 
of the index, and a permanent magnet was brought near it. This answered pretty well, 
but was inconvenient, besides not being sufficiently accurate. A bifilar suspension from 
cocoon-fibres seemed likely to succeed better ; but the difficulty of suspending the rod 
in this manner, so as to get exactly the same tension on each fibre, was very great, and 
unless this was done there was more tendency to move in one direction than in the 
* Phil. Mag., August 1874. 
