Statistical analyses are applied to records obtained under conditions of spontaneous activity. These demonstrate that all the cells examined are similar in the respect that their signals are composed of an approximately Poissonian shower gated on and off at random instants. A mathematical model is developed on this basis which characterises the signal of any one neurone with a number of simple parameters. The manner in which these parameters change under various types of stimulation is studied. It is found that the average frequency inside the sections of Poisson shower remains unchanged in all cases examined. Implications as to the types of network involved are discussed.