514 
MESSES. ELEEMING JENKIN AND J. A. EWING ON EEICTION 
attention, and it is sometimes difficult to see whether the rate is too great or too small, 
since both errors have much the same effect in preventing the siphon from vibrating 
properly. The deposit of ink must be made in a succession of particles by the rapid 
vibration of the siphon in the manner described, and not as a continuous “ brush ” or 
“ glow.” 
In our experiments the length of the longer limb of the siphon was about 3| inches, 
and its period of vibration when under the influence of the electrification was about 
0-028 second; in other words the particles of ink were deposited on the paper at 
intervals of 0-028 second of time. The distance between successive spots was in some 
cases as great as § of an inch. The thickness and length of the siphon were both much 
greater than those of the siphons used in the siphon recorder, and the period of vibration 
was consequently much greater also. We at first attempted to use siphons of the same 
size as those used in the recorder, but owing to the much greater speed of the paper 
in our apparatus, as well as the necessity of keeping the siphon-point at a greater 
distance from the paper, so that any irregularity in the paper band might pass without 
scraping against the point, we found it to be necessary to use much longer and coarser 
tubes. 
When the pendulum is at rest, and the disk is caused to revolve, if the rod l be con- 
tinuously electrified the vibrations of the siphon deposit a series of particles of ink forming 
a dotted straight line in the centre of the paper strip. If when the disk is revolving 
the pendulum is made to oscillate, a curved line will be traced out by the drops of ink, 
crossing the central straight line at intervals depending on the velocity of revolution of 
the disk. The distances between the successive points of intersection of the curve with 
the central line are the distances described by the circumference of the disk in equal 
intervals of time, these intervals being equal to the semiperiod of oscillation of the 
pendulum. 
The experiments were conducted as follows : — A pair of bearings ( b b) were selected 
of the material whose friction against steel was to be measured. Rectangular notches 
were cut in them, and they Avere secured in their places on the uprights as shown 
in figs. 1 & 2, and exactly levelled. The disk with its spindle was then set on them 
and the end-plates ( c c) adjusted. A strip of strong paper (E E) 6 - 3 feet long and 2-5 
inches broad was stretched round the periphery and its ends secured by gum, one end 
overlapping the other for an inch or two. The paper band was continuous except for 
this junction. The pendulum was then caused to oscillate, and the electrical machine 
in connexion with l Avas put in action. When the siphon was vibrating rapidly and 
the ink Avas being freely deposited on the paper, the disk was set revolving in the 
direction of the arrow (fig. 1) by means of an impulse given by hand during only a short 
part of a revolution. It was then left to itself, and usually performed from three to 
eight complete revolutions before coming to rest. The motion of the point of the 
siphon relatively to the paper on the disk was registered in the form of an undulating 
curve of dots crossing the central position at intervals which gradually diminished as the 
