In order to observe the spectrum of coal gas at atmospheric pressure, I passed a stream of the gas through a tube open at both ends, and into the sides of which the platinum electrodes of an induction coil were inserted. I noticed that the spectrum gradually became narrower. On observing this, I stopped the current, and found that a filament of carbon had grown upwards from the lower electrode. On examining it carefully it appeared to have numerous short branches, all of which pointed in the direction of the electrode towards which the filament was growing. The time which it took to form between the electrodes was, I think, under two minutes.