We "started something" that time. It's going on all by itself. The grid is getting more positive, the plate current is getting bigger, and so the grid is getting more positive and the plate current still bigger. Is it ever going to stop? Yes. Look at the audion characteristic. There comes a time when making the grid a little more positive won't have any effect on the plate-circuit current. So the plate current stops increasing. There is nothing now to keep pulling electrons away from plate 1 and crowding them into waiting-room 2. Why shouldn't the electrons in this waiting-room go home to that of plate 1? There is now no reason and so they start off with a rush. Of course, some of them came from the grid and as fast as electrons get back to the grid it becomes less and less positive. As the grid becomes less and less positive it becomes less and less helpful to the plate. If the grid doesn't help, the plate alone can't keep up this stream of electrons. All the plate can do by itself is to maintain the current represented by the intersection of zero volts and the audion characteristic. The result is that the current in the plate circuit, that is, of course, the current in coil _ab_, becomes gradually less. About the time all the electrons, which had left the grid and plate 1 of the condenser, have got home the plate current is back to the value corresponding to _E_{C}_=_0_. The plate current first increases and then decreases, but it doesn't stop decreasing when it gets back to zero-grid value. And the reason is all due to the habit forming tendencies of electrons in coils. To see how this comes about, let's tell the whole story over again. In other words let's make a review and so get a sort of flying start. [Illustration: Fig 34] When we close the battery switch, _S_ in Fig. 34, we allow a current to flow in the plate circuit. This current induces a current in the coil _cd_ and charges the condenser which is across it, making plate 1 positive and plate 2 negative. A positive grid helps the plate so that the current in the plate circuit builds up to the greatest possible value as shown by the audion characteristic. That's the end of the increase in current. Now the condenser discharges, sending electrons through the coil _cd_ and making the grid less positive until finally it is at zero potential, that is neither positive nor negative. While the condenser is discharging the electrons in the coil _cd_ get a habit of flowing from _c_ toward _d_, that is from plate 2 to plate 1. If it wasn't for this habit the electron stream in _cd_ would stop as soon as the grid had reduced to zero voltage. Because of the habit, however, a lot of electrons that ought to stay on plate 2 get hurried along and land on plate 1. It is a little like the old game of "crack the whip." Some electrons get the habit and can't stop quickly enough so they go tumbling into waiting-room 1 and make it negative.