patches at 1:30 o'clock, one being more anterior than the other. More likely by far is the occurrence of several patches, all of which are found along the same approximate circumferential arc (usually at about the level of the equator). Individual patches of fibrovascular proliferation may coalesce along this arc, in which case their feeding arterioles and draining venules frequently form complex anastomoses and arborizations. Although fluorescein angiography in Stage II confirms direct arteriolar-venular flow of blood, there is, in Stage III, an apparent shunting of blood from the feeding arteriole into the capillaries of the proliferative patch before venular flow commences (Figs. 7 and 8).

When a neovascular patch remains relatively isolated from similar neighboring

the

patches and coalescence does not occur, characteristic sea-fan anomaly may be observed (Figs. 7, 8, 9). The sea-fans were found to have predilection for the superotemporal quadrants of the fundus (Fig. 10).

Only nine of the 47 eyes studied failed to show neovascular or fibrous proliferations. Most of the involved eyes fell into substage 1.

Stage IV: Vitreous hemorrhage-Vitre ous hemorrhage in PSR occurs sporadically, without demonstrable systemic or exogenous factors, or may infrequently occur following rather insignificant trauma such as a mild blow to the head. In most cases, a neovascular patch can be demonstrated to be the source of the hemorrhage (Fig. 11). The role of vitreous contraction in hemorrhages associ ated with PSR is currently unkown. Such

nemorrhages often seem to remain localized in the cortical vitreous which overlies the responsible patch of proliferative retinopathy. Thus, once again, a circumferential pattern is observed in the fundus abnormality, and, as in the lesions in Stages I, II, and III, the peripheral temporal quadrants are preferentially involved. Such localized vitreous hemorrhages are frequently asymptomatic, even in the presence of the usually intact peripheral visual field, and are observed only if routine ophthalmoscopy is performed at a fortuitous time. Occasionally, however, a vitreous hemorrhage is extensive enough to spread into the visual axis or into the premacular area, thereby causing visual symptoms, including reduction in visual acuity (Fig. 12). Indirect ophthalmoscopy is the best means for detecting and quantitating vitreous hemorrhage associated with PSR.

Evidence for vitreous hemorrhage was present in 14 of the 47 eyes studied. Eight of these fell into substage 1 or 2, and four fell into substage 4.

Stage V: Retinal detachment-Retinal detachment is the ultimate and most severe stage of PSR (Fig. 13). Because such detachments are frequently related to adjacent vitreous changes (which in turn are related to neovascular proliferations and hemorrhages), they assume a temporal, equatorial configuration similar to the configurations of the antecedent Stages I-IV. Retinal tears, which are both round and horseshoe in shape, often are found adjacent to neovascular proliferations.

Four eyes had retinal detachments when initially examined and classified. In all these eyes, two or more quadrants of the retina were detached. In one case, J.B., there was