s C, C', completing the circuit of the
battery B and lamp L. When working properly the lamp L lights up regularly
once every second. This regulator is an excellent one to use for
experimental work, although it depends a great deal upon the skill of the
operator, but good adjustment should be obtained in about two minutes. It
is a good plan to insert a clutch of some description between the driving
motor and the machine, so that the regulator can be adjusted prior to the
act of receiving or transmitting, the machine being prevented from
revolving by means of a catch. The motor used should be powerful enough to
take up the work of driving the machine without any reduction in speed. The
clocks M can be regulated so that they only gain or lose a few seconds in
{70} twenty-four hours, which gives an accuracy in working sufficient for
all practical purposes.

Connection is made with the contact springs S, S', by means of the springs
T, T', which press against the spindles J, J'.

Another important point is the correct placing of the picture upon the
receiving drum. It is necessary that the two machines besides revolving in
perfect isochronism should synchronise as well, _i.e._ begin to transmit
and record at exactly the same position on the cylinders, viz. at the edge
of the lap, so that the component parts of the received image shall occupy
the same position on the paper or film as they do on the metal print. If
the receiving cylinder had, let us suppose, completed a quarter of a
revolution before it started to reproduce, the reproduction when removed
from the machine and opened out will be found to be incorrectly placed; the
bottom portion of the picture being joined to the top portion, or _vice
versa_, and this means that perhaps an important piece of the picture would
be rendered useless even if the whole is not spoilt. It is evident,
therefore, that some arrangement must be employed whereby synchronism, as
well as isochronism of the two instruments can be maintained.

There are several methods of synchronising that are in constant use in
high-speed telegraphy, in which the limit of error is reduced to a minimum,
{71} and some modification of these methods will perhaps solve the problem,
but it must be remembered that synchronism is far easier to obtain where
the two stations are connected by a length of line than where the two
stations are running independently.

In one system of ordinary photo-telegraphy synchronism i