Synthetic routes to the tricyclo(5,3,1,1 2,6)dodecane and tricyclo(5,1,0,0 3,5)octane systems have been investigated. Routes to the former class have involved double Michael additions to 4,4'-dimethoxycyclohexa-2,5-dienone, and attempted preparation of a bis delta-enol-lactone from diethyl succinylsuccinate, but no tricyclic compounds were produced. Since approaches to the latter system, involving carbene additions to 3,3',6,6'-tetramethoxycyclohexa-1,4-diene gave a novel rearrangement to hydroquinone dimethyl ether, 1,2,4-trimethoxy benzene and a pentamethoxy biphenyl, the tricyclo(5,1,0,0,3,5)octan-2,6-dione has been synthesised by reaction of 4,4'-dimethoxycyclohexa-2,5-dienone with an ylid reagent. A reported violation of Bredt's rule involving monodecarboxylation of bicyclo(2,2,2)octan-2,5-dione-1,4-dicarboxylic acid to bicyclo(2,2,2)octan-2,5-dione-1-carboxylic acid has been proved genuine, but the other expected decarboxylation product, bicyclo(2,2,2)octan-2,5-dione has not been detected.