Proposals, including those in Britain and France, for fairly draconian bans on smoking in public places have caused much anger and protest. This is as it should be, since the issue is controversial. But the contrary arguments demand -critical scrutiny. One line of critique questions the use of statistical evidence for policymaking. Another invokes the importance of liberty to do what one likes in one's own life.

David Hockney, the distinguished artist, has argued that he has read"all their statistics" about the connection between smoking and disease,but he must observe that "fateplays part in life, that mysterious forces are at work on life": "Medical statisticians cannot grasp this, but almost everyone else does." What, then, should we make of such foundational doubts about the relevance of statistical reasoning?