The type of birefringence described by Mitchison, which extends some 0.5 µ in from the surface of the human red cell ghost in glycerol and which shows a maximum retardation of about 7 A, is only found in ghosts which are sufficiently well hemoglobinised to be seen with the ordinary microscope. Ghosts from which all hemoglobin has been lost are not visible with the ordinary microscope and are not birefringent, although they are clearly visible with phase contrast. About 90 per cent of the ghosts in glycerol preparations are of the latter type, the exact percentage being a function of time. Mitchison's measurements of birefringence, although reproducible, accordingly apply only to ghosts in which some hemoglobin still remains complexed with the lipoprotein layers of the red cell ultrastructure, and do not enable one to draw conclusions as to the thickness and orientation of the lipoprotein surface layers.