Nan, generally speaking (e.g., in maths) tuples can contain any number of elements. Haskell specifically disallows one-tuples because of the ambiguity in parsing them. Instead, (expr) is just evaluated as expr. On the other hand, it does allow (), which is a zero-tuple. Haskell's compilers also impose an upper bound on the practical size of a tuple (for modern GHC, that's 61, as Max mentioned) but that isn't a restriction of the Haskell language itself. If you want, you can consider Identity a to be a one-tuple, but it obviously doesn't use tuple syntax.