Infinite dimensional representations of finite dimensional hereditary algebras.

Enrico Gregorio claims that Theorem 3.9 is not correct, see his paper Every quasitilted algebra is a quasitilted ring, to appear in J. Algebra. (Theorem 3.9 asserts that for a wild hereditary algebra, the torsion pair with torsion class the direct sums of preinjective modules does not split.)

On May 30, 2007, Enrico Gregorio informed me that the warning can be deleted. His preprint contained a mistake.