From a defective-lysogenic Proteus mirabilis strain we isolated several clones differing in the pattern of their growth on agar plates. Using electron microscopy we have shown some of the selected clones to be efficient in producing mirabilicin D-52 after UV induction, while other clones produced defective mirabilicin polysheaths and polycores. Clones producing polysheaths and polycores can be detected electron microscopically only, since these defective particles are biologically inactive.