Experiments are described in which carbanions are generated at the bridgheads adjacent to phosphorus in phosphonium salt 2 and in phosphine oxide 1. These experiments were undertaken to find ways to make the intermediates in a proposed scheme (Scheme 111) for the synthesis of derivatives of cubane. When attempts are made to prepare the conjugate base of 2 using sodium hexamethyldisilylamide in tetrahydrofuran (THF), the ylide apparently rearranges rapidly to a syn-tricyclooctadienyldiphenylphosphine ( 5 ) , a novel example of the electrocyclic process summarized as eq 5 . The conversion conatitutes a preparation of the tricyclooctadienyl ring system. On oxidation with hydrogen peroxide, 5 gives its phosphine oxide (211, but a t -2 "C this equilibrates with an isomer (22). At 138 "C these isomers rearrange to give 7. Similarly at 74.5 "C, 5 isomerizes to cyclooctatetraenyldiphenylphosphine (6), but shows no evidence of giving an isomer analogous to 22. Unlike the carbanion 3 derived from 2, lithiated phosphine oxide 1 is stable in solution at ambiient temperature. This lithium derivative can be made from 1 and phenyllithium in THF at room temperature, while at -78 "C the same reagents give 8. D20, CH31, and ( C G H~S )~ react with the lithiated material to introduce substituents adjacent to phosphorus. The thiophenyl substituent can be oxidized to the sulfone (16), and this with sodium hexamethyldisilylamide gives 18. With phenyllithium 16 gives the analogue 20. These last rearrangements are novel, and since 1 does not undergo them, they reflect the leaving ability of sulfone anions The cubic hydrocarbon CsHs, known as cubane, was first synthesized by Eaton and Cole in 1964.' Shortly afterwards a related synthesis was reported by Barborek, Watts, and Pettit,2 and additional ways to arrive a t intermediate molecules on the original routes were found by Chin, Cuts, and Masamune3 and by Eaton and Cole.4 These syntheses, and all of those developed since for derivatives of ~u b a n e ,~ employ as a key step the Favorskii rearrangement6 of an a-bromohomocubanone (Scheme I), and despite difficulties experienced in some laboratories,5f,c although not in other^,^^^^ and despite the length of the synthesis, the original route of Eaton and Cole' remains the most effective. Alternative syntheses have not been reported and seem to have been sought only rarely.'The availability of phenylphosphahomocubane oxide from the cyclooctatetraenyl dianion (Scheme II)8-10 suggested that another route to cubane might be found if a phosphorus analogue of the Favorskii rearrangement could be discovered (Scheme 111), and the research described here was toward this goal. A sulfur analogue of such arrangements, the RambergBacklund rearrangement,Il has been applied effectively in many syntheses, but because it proceeds by way of an episulfone and fails when the intermediate three-membered ring is highly strained, path B in Scheme 111 is unlikely. Path A, Br OH ---1 0 ; Scheme I Scheme I1 2 L i C , H , P C I O P C 6 H 5 prm Scheme 111 NU c 1Bp< C...