For the fiftieth anniversary of Berger's first EEG publication, some of his early recordings obtained between 1924 and 1931 are discussed and illustrated. Examples of his protocols from the Freiburg Berger Archives are reproduced. Three types of Berger's early investigations are described: (1) String-galvanometer recordings obtained between 1924 and 1926, mainly from trephined patients with cerebral diseases, which usually showed brain waves slowed to 6--8 per second; (2) Direct recordings from the cortex and white matter proving the cortical origin of the EEG in 1930; (3) Typical unpublished EEG recordings of epileptics and of petit-mal attacks obtained in 1930 and 1931. Berger's first six papers published between 1929 and 1933 described nearly all the main EEG findings of cerebral diseases and the EEG alterations of normals during attention, sleep, and narcosis, but they did not report on convulsive potentials in the EEGs of epileptics. Berger had, however, obtained excellent records of epileptic EEG features, here depicted in Figs. 4 through 7. These remained unpublished until 1933 and 1938, because Berger suspected that they contained artifacts caused by blinks and facial movements which he had recorded in his controls (Fig. 4). Only in 1933, after other authors had described large amplitudes of convulsive potentials in the cortex of animals, did Berger publish parts of the EEGs of a petit-mal attack and of focal attacks in progressive paresis. In 1938, Berger presented the EEG of the beginning of a petit-mal attack with large 3/s spikes and waves recorded in 1931 which were similar to those described by Gibbs and coworkers in 1935. In 1933 and 1938, Berger interpreted the abnormal brain potentials of epileptics as signs of a preconvulsive state of the forebrain and suggested that the periods of 3/s waves were cortical correlates of an epileptic absence.