The first 'training analysis' consisted largely of reading Freud's books. Later this was supplemented by 'demonstration': analysis for a few weeks or months, coupled with discussion. The third period, that of analysis proper, was established after much resistance in professional circles. The fourth phase is described as that of the 'fully completed analysis', more than is usually needed for therapeutic purposes. Balint considers that the fifth phase, research, reached in the last few years, has been the consequence of some undesirable developments. A covert collusion appeared between analysts and their candidates whereby the idealized analyst was introjected. There resulted a series of 'schools of analysis', with confusion of tongues and 'power politics'. Analysts had failed to deal adequately with the primitive aggression of their candidates: too early and too consistent interpretations led to dilution of feeling on both sides. It is not merely the candidate's transference but also the analyst's technique that is of decisive importance.