OPINION polls showing the Yes and No campaigns neck and neck could be wildly inaccurate, a leading pollster has admitted.

Martin Boon, director of polling company ICM, said the independence referendum could prove a "polling Waterloo" and become the industry's biggest embarrassment since 1992, when surveys predicted a narrow victory for Neil Kinnock in the days before his eventual defeat to John Major.

He told the BBC all the companies gauging public opinion in the run-up to Thursday's vote were hoping to "dodge a bullet".

The latest poll of polls, the average of the last six surveys, gives the No campaign a narrow 51 per cent to 49 per cent lead. It includes two polls which put Yes ahead.

Mr Boon's comments echo the private feelings of campaign insiders, who believe the vote could produce a much clearer margin of victory for either side than the polls suggest.

He told BBC Radio Four the unprecedented nature of the referendum and the reliance by most companies on relatively small online panels made the result difficult to predict.

He said: "We are dependent on a pot of people which is defined, but we don't know how big it is and in my view it won't be big enough. In that lies a real danger for the accuracy of the polls in advance of this referendum."

Asked if he would be surprised if the result did not reflect the poll findings, he said: "No. This referendum has the potential to be a polling Waterloo, the biggest since 1992 when the polls got it wrong.

"I can't say I'd be surprised. I very much hope that polling companies do the market research industry justice by getting it at least in the right area. I think and hope the best that we in the industry can hope for is that we dodge a bullet, but it's entirely possible the bullets do start spraying our way."

Of the six companies that have been following the referendum, four, ICM, YouGov, Panelbase and Survation, use online panels. TNS speak to people face to face, while Ipsos MORI use phone interviews.

Experts believe the polls may be underestimating two possible factors. It is feared they may be missing a big surge in support for independence among people who rarely vote, underestimating the strength of the Yes vote.

However, there are also concerns the polls are blind to significant numbers of "shy Nos," people who are reluctant to tell pollsters they oppose independence.

Mr Boon said: "In Scotland now I believe it could be true to a limited degree there is a kind of patriotism spiral of silence going on. Perhaps some people think it is unpatriotic to say they are a No voter, maybe they are inclined to silence."

Stephen Fisher, an associate professor of Political Sociology at Oxford University, said the polls had overestimated the Yes or pro-change vote in 12 of 16 recent referendums around the world. He also said the polls may be underestimating the No vote because of a "spiral of silence".

Writing on the What Scotland Thinks website, he said: "It seems more likely that the headline poll figures are over- rather than under-estimating the vote for Scottish independence."