[en] During the 1975 PEP Summer Study, Garren and Kadyk studied systems for rotating the normal beam polarization into longitudinal polarization at one or more interaction regions. They developed a solution using vertical bending magnets in the long drift between Q2 and Q3 and leaving the 20-m drift region around the interaction point unencumbered. Considerable difficulty was encountered in matching the beam optics, but the possibility of such a match was demonstrated. Meanwhile, further studies of depolarizing effects have resulted in cautious optimism for the possibility of obtaining a useful degree of polarization in PEP. Consequently, it seemed worthwhile to re-examine the question of such a spin rotation system with the objectives of: improving the beam-matching optics, investigating the design requirements for the magnets and quadrupoles and defining the additional transverse space which should be reserved for the SRS in the interaction region housing. In the present note some variations of the Garren-Kadyk scheme are presented, and it is shown that the beam optics can indeed be improved considerably, along with a reduction in the complexity of the system. It is proposed that a design of this type be used as a basis for specifying the required enlargement of the appropriate straight section tunnel. The magnetic requirements for the beam transport elements appear to be reasonable, although some interesting mechanical design problems may arise. 3 refs., 4 figs., 4 tabs