An attempt is made to obtain a coherent picture of the extent and mode of operation of the ‘southern oscillation.’ This term is used here, following Sir Gilbert Walker, to describe a standing fluctuation of opposed pressure anomalies in both eastern and western hemispheres. The existence of this opposition has been verified, using more recent data, for stations in the Indian and Pacific Ocean regions; results show the oscillation was less marked in recent decades.

The representativeness and physical meaning of the index devised by Walker to characterize the state of the oscillation are considered. The geographical extent of the phenomenon is examined using correlation and regression charts of pressures with Walker's index. The temperature and rainfall anomalies associated with it may be derived qualitatively from the pressure anomalies.

Recent data are used to verify persistence and lag correlations between station pressures; while there has again been some decline, the lag correlations of elements with previous South America pressures still hold good. The decline in these various quantities is indicative of a minor secular change commencing in the 1920's, which is also evident in a decrease in the variability of pressure.

What ‘periodicities’ appear to exist in elements affected by the southern oscillation may well be an outcome of sampling fluctuations in (often persistent) random series. This is suggested by the variety of supposed ‘periods’ reported, and their evanescence in space and time. An example of this evanescence in time is provided from the Darwin pressure record.

A mechanism for the oscillation is proposed in terms of variations in a direct toroidal circulation between warmer eastern and cooler western hemispheres. These variations are attributed, (following a model by Palmer of the synoptic climatology of the tropical Pacific) to variations in the south-east trades in the South Pacific and the consequent variations in cyclonic vortex generation in the West Pacific. The persistence of anomalies is then due to the extent of ocean areas in the south-east Pacific where the sea temperature is lower than the air temperature. The lag correlations observed may be due to this persistence and to a transmission of anomalies along the trades through air-sea interaction.