I have never understood why French polls are reported as if there was no undecided -- they simply set undecideds aside and report the responses those who did express a preference. Which vastly reduces the actual sample size, thereby increasing the margin of error. So when they claim a MOE of +_3.5 its actually not the case. (And in any event, there is almost never an MOE reported, so that a change of 1 percent, well within a given poll's MOE, is reported as a change of opinion.)

That said, there's a clear trend showing Bayrou gaining at Royal's expense and Sarkozy holding a lead, even if he's also losing some support to Bayrou.

I think a lot of the undecided vote is going to go to Sarko, in part because as afew points out, he seems to be getting very favorable coverage (what happened to the story about his townhouse in Neuilly?) -- and because his presence in tv, in print and on the radio is simply better organized than Royal. (See tomorrow's Libe, online now, which polled people about who was running the best campaign, and 50% said Royal is running a "bad" campaign, more than 3x the % which said that about Bayrou or Sarkozy.)

I'm not disputing this is due in part to an established media narrative of campaign "turbulence," but its something she's got to break out if she's going to have any chance.