apes past the high-pressure cylinder instead of
blowing off at the safety valve. In escaping through this valve, a fall
of about 200 lbs. pressure per square inch is made to do work on the
surrounding steam and drive it forward in the pipe, producing a pressure
on the low-pressure piston considerably higher than the back-pressure on
the high-pressure piston. In this way a portion of the work which would
otherwise be lost is utilised, and it is possible, with an unlimited
supply of steam, to cause the engines to develop an enormous amount of
power.'

With regard to boilers, Maxim writes,

'The first boiler which I made was constructed something on the
Herreshof principle, but instead of having one simple pipe in one very
long coil, I used a series of very small and light pipes, connected in
such a manner that there was a rapid circulation through the whole--the
tubes increasing in size and number as the steam was generated. I
intended that there should be a pressure of about 100 lbs. more on the
feed water end of the series than on the steam end, and I believed that
this difference in pressure would be sufficient to ensure direct and
positive circulation through every tube in the series. The first boiler
was exceedingly light, but the workmanship, as far as putting the tubes
together was concerned, was very bad, and it was found impossible to so
adjust the supply of water as to make dry steam without overheating and
destroying the tubes.

'Before making another boiler I obtained a quantity of copper tubes,
about 8 feet long, 3/8 inch external diameter, and 1/50 of an inch
thick. I subjected about 100 of these tubes to an internal pressure of
1 ton per square inch of cold kerosene oil, and as none of them leaked
I did not test any more, but commenced my experiments by placing some
of them in a white-hot petroleum fire. I found that I could evaporate
as much as 26 1/2 lbs. of water per square foot of heating surface per
hour, and that with a forced circulation, although the quantity of water
passing was very small but positive, there was no danger of overheating.
I conducted many experiments with a pressure of over 400 lbs. per square
inch, but none of the tubes failed. I then mounted a single tube in a
white-hot furnace, also with a water circulation, and found that it only
burst under steam at a pressure of 1,650 lbs. per square inch. A large
boiler, having about 800 square feet of heating surface, including the
feed-