lustration: FIG. 17.]

[Illustration: FIG. 18.]

The latter is left perfectly free to expand; and yet, owing to its
particular shape, the pressure in the cylinder during the compression
and explosion stroke only tends to make the joint between the tube and
washer more secure. The action of this ignition device depends upon the
tube heater H, which is merely a small bunsen burner, the flame of which
impinges on the tube at one particular spot, raising it to a very high
temperature--almost white heat. Most of my readers will know the
formation of the bunsen flame. It in composed of two distinct zones. The
inner one, marked A in fig. 18, is a perfectly cold part of the flame,
and appears to be a pale-blue coloured cone.

It is the outer zone which is the hot portion of the flame, hence this
part _only_ must be allowed to play on the tube. The tip of the blue
cone A must be kept about 1/4 in. below the tube, in order to ensure the
hottest part of the flame impinging precisely where the heat is
required.

The total length of the whole flame is, to a certain extent, immaterial;
but, generally speaking, it should be adjusted so that the length of the
inner cone A is about 1 in. or 1-1/4 in. The same methods which we
described in the early part of this chapter can be employed in the
adjustment of this burner, but some care should be exercised to get the
correct flame length.

The result of allowing the cold part of the flame to impinge on the tube
is observable in fig. 18. The black spot indicated on the drawing
actually appears as a black or sooty spot when looking at the tube under
these conditions; but in reality no discoloration whatever takes place,
the spot disappearing immediately the cone A is made shorter, or the
burner H lowered in the chimney B, so that the tip of A is just below,
and does not touch the tube at all.

The adjustment of the length of cone A may be accomplished in two
ways--(1) by keeping the supply of gas constant, and varying the amount
of air admitted at aperture K, fig. 18; (2) by keeping the supply of air
constant, and varying the amount of gas admitted through nipple N. The
first method is to be preferred when it is necessary to make any slight
adjustment due to the variation of gas pressure during the day, and may
be accomplished by fitting a small sliding shield G, as shown in the
figs. 16 and 17, and moving it round so that it covers, more or less,
the aperture K. Thus the length of cone