reasoned that if the paper strip could be imprinted with elevations
and depressions representative of sound-waves, they might be caused to
actuate a diaphragm so as to reproduce the corresponding sounds.
The next step in the line of development was to form the necessary
undulations on the strip, and it was then reasoned that original sounds
themselves might be utilized to form a graphic record by actuating a
diaphragm and causing a cutting or indenting point carried thereby to
vibrate in contact with a moving surface, so as to cut or indent the
record therein. Strange as it may seem, therefore, and contrary to the
general belief, the phonograph was developed backward, the production of
the sounds being of prior development to the idea of actually recording
them.

Mr. Edison's own account of the invention of the phonograph is intensely
interesting. "I was experimenting," he says, "on an automatic method
of recording telegraph messages on a disk of paper laid on a revolving
platen, exactly the same as the disk talking-machine of to-day. The
platen had a spiral groove on its surface, like the disk. Over this was
placed a circular disk of paper; an electromagnet with the embossing
point connected to an arm travelled over the disk; and any signals given
through the magnets were embossed on the disk of paper. If this disk was
removed from the machine and put on a similar machine provided with
a contact point, the embossed record would cause the signals to be
repeated into another wire. The ordinary speed of telegraphic signals
is thirty-five to forty words a minute; but with this machine several
hundred words were possible.

"From my experiments on the telephone I knew of the power of a diaphragm
to take up sound vibrations, as I had made a little toy which, when
you recited loudly in the funnel, would work a pawl connected to the
diaphragm; and this engaging a ratchet-wheel served to give continuous
rotation to a pulley. This pulley was connected by a cord to a little
paper toy representing a man sawing wood. Hence, if one shouted: 'Mary
had a little lamb,' etc., the paper man would start sawing wood. I
reached the conclusion that if I could record the movements of the
diaphragm properly, I could cause such record to reproduce the original
movements imparted to the diaphragm by the voice, and thus succeed in
recording and reproducing the human voice.

"Instead of using a disk I designed a little machine using a cylinder