Not long ago this column deplored the deadening consensus that was gripping the Bank of England's interest- setting committee.

Month after month the members all voted the same way - to do nothing. Any consensus has a nasty trick of being wrong - because people in the habit of agreeing with each other are slow to notice when something altogether novel and unexpected starts to happen.

Well, that is not the way it is any more. They may not be scrapping like ferrets in a sack - bankers and economists are generally a civil lot - but for two months running they have voted five to four, as near to being split down the middle as a nine-strong committee can be.

Better still, they voted in opposite directions. July's losers were August's winners. For the first time ever the Bank's governor was on the losing side - and his respected chief economist, Charlie Bean, voted against him.

Under the rules laid down by Gordon Brown eight years ago the governor has a casting vote. If the committee is split evenly he can vote twice. But that is no use unless one of the members goes sick. So far they have proved as unfailingly healthy as they are civil.

All this is welcome. The fact that at least four of the nine members are wrong may dent their reputation for omniscience. It may unsettle the markets because it is harder to guess what happens next. But it is better than the risk of all nine being wrong month after month.