“…M a 's test follows the colon and begins with three quanti®ers: ilil H u. The nonmanipulability condition of Zeckhauser (1973), Gibbard (1977), and Feldman (1980) weakens that test, strengthening M a , by weakening the universal quanti®cation to an instance of it: for us manipulability requires that a certain relation hold for every pair of lotteries with supports CP and CP H , but for them the relation need hold only for one pair, socially chosen along with CP and CP H . An even stronger nonmanipulability condition weakens the``all'' of our test to``some,'' ilil H u to ll H u.…”