on: Fig. 32.]

[Illustration: Fig. 33.]

We have found that the pressure in a short cylinder gets less if it
begins to develop a waist, and greater if it begins to bulge. Let us
therefore try and balance one with a bulge against another with a waist.
Immediately that I open the tap and let the air pass, the one with a
bulge blows air round to the one with a waist and they both become
straight. In Fig. 32 the direction of the movement of the air and of the
sides of the bubble is indicated by arrows. Let us next try the same
experiment with a pair of rather longer cylinders, say about twice as
long as they are wide. They are now ready, one with a bulge and one with
a waist. Directly I open the tap, and let the air pass from one to the
other, the one with a waist blows out the other still more (Fig. 33),
until at last it has shut itself up. It therefore behaves exactly in the
opposite way that the short cylinder did. If you try pairs of cylinders
of different lengths you will find that the change occurs when they are
just over one and a half times as long as they are wide. Now if you
imagine one of these tubes joined on to the end of the other, you will
see that a cylinder more than about three times as long as it is wide
cannot last more than a moment; because if one end were to contract ever
so little the pressure there would increase, and the narrow end would
blow air into the wider end (Fig. 34), until the sides of the narrow end
met one another. The exact length of the longest cylinder that is
stable, is a little more than three diameters. The cylinder just becomes
unstable when its length is equal to its circumference, and this is
3-1/7 diameters almost exactly.

[Illustration: Fig. 34.]

I will gradually separate these rings, keeping up a supply of air, and
you will see that when the tube gets nearly three times as long as it is
wide it is getting very difficult to manage, and then suddenly it grows
a waist nearer one end than the other, and breaks off forming a pair of
separate and unequal bubbles.

If now you have a cylinder of liquid of great length suddenly formed and
left to itself, it clearly cannot retain that form. It must break up
into a series of drops. Unfortunately the changes go on so quickly in a
falling stream of water that no one by merely looking at it could follow
the movements of the separate drops, but I hope to be able to show to
you in two or three ways exactly what is happening. You may r