Comment on ``Conduction states in oxide perovskites: Three manifestations of Ti^(3 +) Jahn-Teller polarons in barium titanate''

In this comment to [S. Lenjer, O. F. Schirmer, H. Hesse, and Th. W. Kool, Phys. Rev. B {\bf 66}, 165106 (2002)] we discuss the electronic structure of oxygen vacancies in perovskites. First principles computations are in favour of rather deep levels in these vacancies, and Lenjer et al suggest that the electrons' interaction energy is negative, but data on electroconductivity are against.