The above supports the intermediacy of a symmetrical intermediate of the type V whose cleavage could then be affected by the substitution pattern but whose formation would be only slightly affected by inductive and solvent effects, since no charge is developed.
VilaVllb VIII where the positive oxygen was three atoms away from the CH2X group, the relative insensitivity of the reaction to varying inductive effects and the good linear correlation would not be unreasonable. However, it is unreasonable to expect that Vllb could possibly be the exclusive intermediate over the range of substituents used in this study and in light of the strongly supported intermediacy of the two possible acylcarbonyl oxides in Bailey's study.3 The two acylcarbonyl oxides, II and III, could only arise from two four-membered ring precursors which differ in addition orientation, as in Vila and Vllb. If varying proportions of Vila and Vllb were formed, depending on the substituent, it would be highly unlikely that a linear correlation would exist; instead, as one went to more electron-donating substituents, Vila would be favored, resulting in marked curvature up as one went to decreasing a* values, something that is not observed.The initial rate-determining formation of a symmetrical complex, or of a symmetrical three-membered ring of the type VIII, cannot be ruled out. However, the relative insensitivity of the rate of reaction to changing substituents and changing solvent polarity does tend to make it less likely that charge is developed in the transition state.