ver a
sprocket, hanging bags of sand on either side of the sprocket of a
weight approximately equal to the pull that would be exerted on the
chains when driving the propellers. By measuring the extra amount of
weight needed on one side to lift the weight on the other, we calculated
the loss in transmission. This indicated that the loss of power from
this source would be only 5 per cent., as we originally estimated. But
while we could see no serious error in this method of determining the
loss, we were very uneasy until we had a chance to run the propellers
with the motor to see whether we could get the estimated number of
turns.

The first run of the motor on the machine developed a flaw in one of the
propeller shafts which had not been discovered in the test at Dayton.
The shafts were sent at once to Dayton for repair, and were not received
again until November 20, having been gone two weeks. We immediately put
them in the machine and made another test. A new trouble developed. The
sprockets which were screwed on the shafts, and locked with nuts of
opposite thread, persisted in coming loose. After many futile attempts
to get them fast, we had to give it up for that day, and went to bed
much discouraged. However, after a night's rest, we got up the next
morning in better spirits and resolved to try again.

While in the bicycle business we had become well acquainted with the use
of hard tire cement for fastening tires on the rims. We had once used it
successfully in repairing a stop watch after several watchsmiths had
told us it could not be repaired. If tire cement was good for fastening
the hands on a stop watch, why should it not be good for fastening the
sprockets on the propeller shaft of a flying machine? We decided to try
it. We heated the shafts and sprockets, melted cement into the threads,
and screwed them together again. This trouble was over. The sprockets
stayed fast.

Just as the machine was ready for test bad weather set in. It had been
disagreeably cold for several weeks, so cold that we could scarcely work
on the machine for some days. But now we began to have rain and snow,
and a wind of 25 to 30 miles blew for several days from the north. While
we were being delayed by the weather we arranged a mechanism to measure
automatically the duration of a flight from the time the machine started
to move forward to the time it stopped, the distance traveled through
the air in that time, and the number of revo