purpose. But let us assume that we have a compressor which shows an
adiabatic pressure line. We now have the air in the clearance space
acting precisely as a spring, compressed at each stroke, retaining its
heat of compression, and giving it out against the air piston at the
point when the stroke is reversed. There is no loss of power in such a
case as this, but, on the contrary, the air spring is useful in
overcoming the inertia of the piston and moving parts. The best air
compressors give a result about midway between the isothermal and the
adiabatic, and the net loss of _power_ directly due to clearance is so
small as to be practically unworthy of consideration.

It must not be inferred from the preceding remarks that the designer of
an air compressor may neglect the question of clearance. On the
contrary, it is a very important consideration. If we assume a large
clearance space in the end of an air cylinder of a compressor which is
furnishing air at a high pressure, we may readily conceive that space to
be so large, and that pressure so high, that the entire volume of the
cylinder would be filled by the air from the clearance space alone, and
the compressor would take in no free air and would, of course, produce
no compressed air.

Loss in _capacity_ of air compressors by clearance is in direct
proportion to the pressure.

Owing to the loss of capacity by clearance space at high pressures, it
is important that compound air cylinders should be used for furnishing
air at high pressure. With compound air cylinders the air is compressed
to alternate stages of pressure in the different cylinders, and the
clearance loss is thus reduced because of the reduced density of the air
in the clearance spaces. In ordinary practice air compressors deliver