The author questions the validity of reconstructing events in the first two years of life from knowledge gained from adults or psychotics. He postulates a theory of nuclear ego development which allows an ample representation of early ego development arising from expression of the various components of early instinctual life. In this way ontogenetic ego and superego elements can be represented with more freedom than with a rigid concept of an oral ego. He does not believe the total ego function should be considered during the first two years of life, a period he prefers to call embryonic. Great emphasis is laid to the danger of over-evaluating fantastic reconstructions for the infant.