Whose Knowledge?

Sir Rudolph Peierls, in a reply to John Bell's last critique of the state of our understanding of quantum mechanics, maintained that it is easy to give an acceptable account of the physical significance of the quantum theory. The key is to recognize that all the density matrix characterizing a physical system ever represents is knowledge about that system. In answer to Bell's implicit rejoinder "Whose knowledge?" Peierls offered two simple consistency conditions that must be satisfied by density matrices that convey the knowledge different people might have about one and the same physical system: their density matrices must commute and must have a non-zero product. I describe a simple counterexample to his first condition, but show that his second condition, which holds trivially if the first does, continues to be valid in its absence. It is an open question whether any other conditions must be imposed.