The paper, which is in two parts, describes an investigation of the nature and origin of the 'whistling atmospherics' or 'whistlers' which are sometimes observed at frequencies below 15 kc/s. The first part describes an experimental study of their properties, in the course of which a considerable number of whistlers were recorded and analyzed, and the law of the variation of their frequency with time determined. Some whistlers are heard to follow impulsive atmospherics, and these are found to be produced in the normal way by lightning strokes taking place within a distance of about 2000 km. Other whistlers are unaccompanied by atmospherics; they differ from the former type in several further respects. The diurnal and annual variations of the properties of both types of whistler have also been studied. In the second part of the paper a theory of the origin of the whistling atmospherics, originally due to Barkhausen (1930) and Eckersley (1935), is developed in detail. The theory proposes that they are due to waves which originate in normal impulsive atmospherics and travel through the outer ionosphere, following the lines of force of the earth's magnetic field and crossing over the equator at a great height. During their journey they become dispersed so as to arrive as 'whistlers'. They may be reflected from the earth's surface back along the same path, one or more times, to produce whistlers with increased dispersions. The effects responsible for the guiding of the waves along the lines of the geomagnetic field provide sufficient focusing action to prevent the energy from being spread unduly. Measurements of the degree of dispersion of the whistlers have been interpreted to yield information about the density of electrons in the atmosphere at very great heights. The density required seems considerably larger than could reasonably have been expected. If the free electrons are produced by ionization of the terrestrial atmosphere its temperature in these regions must be at least 7200 degrees K. The results might alternatively be explained on the assumption that the electrons are falling in from outside, and if this were so it might account for the relationship between the occurrence of whistlers and magnetic activity.