That means that the condenser not only discharges but starts to get charged in the other direction with plate 1 negative and plate 2 positive. The grid feels the effect of all this, because it gets extra electrons if plate 1 gets them. In fact the voltage effective between grid and filament is always the voltage between the plates of the condenser. The audion characteristic tells us what is the result. As the grid becomes negative it opposes the plate, shooing electrons back towards the filament and reducing the plate current still further. But you have already seen in my previous letter what happens when we reduce the current in coil _ab_. There is then induced in coil _cd_ an electron stream from _c_ to _d_. This induced current is in just the right direction to send more electrons into waiting-room 1 and so to make the grid still more negative. And the more negative the grid gets the smaller becomes the plate current until finally the plate current is reduced to zero. Look at the audion characteristic again and see that making the grid sufficiently negative entirely stops the plate current. When the plate current stops, the condenser in the grid circuit is charged, with plate 1 negative and 2 positive. It was the plate current which was the main cause of this change for it induced the charging current in coil _cd_. So, when the plate current becomes zero there is nothing to prevent the condenser from discharging. Its discharge makes the grid less and less negative until it is zero volts and there we are--back practically where we started. The plate current is increasing and the grid is getting positive, and we're off on another "cycle" as we say. During a cycle the plate current increases to a maximum, decreases to zero, and then increases again to its initial value. [Illustration: Fig 36] This letter has a longer continuous train of thought than I usually ask you to follow. But before I stop I want to give you some idea of what good this is in radio. What about the current which flows in coil _cd_? It's an alternating current, isn't it? First the electrons stream from _d_ towards _c_, and then back again from _c_ towards _d_. Suppose we set up another coil like _CD_ in Fig. 36. It would have an alternating current induced in it. If this coil was connected to an antenna there would be radio waves sent out. The switch _S_ could be used for a key and kept closed longer or shorter intervals depending upon whether dashes or dots were being set. I'll tell you more about this later, but in this diagram are the makings of a "C-W Transmitter," that is a "continuous wave transmitter" for radio-telegraphy. It would be worth while to go over this letter again using a pencil and tracing in the various circuits the electron streams which I have described. LETTER 12 INDUCTANCE AND CAPACITY DEAR SIR: