Not so much “polls apart” as “pollsters apart”. First the US statistical guru, Nate Silver, waltzes into festival Edinburgh and pronounces that there is “virtually no chance” of a Yes vote in the referendum.
But then up pops Mark Diffley of Ipsos Mori, claiming that no less than 44 per cent of Scots are still undecided about how they are going to vote – and that thus tide could still turn.
So who is right? First we need to unpack that 44 per cent. It is not the case that when Mori asks voters how they will vote that 44 per cent say they “don’t know”. Only 15-16 per cent fall into that camp. The 44 per cent also includes two other groups. One – a little under a fifth of all voters – consists of those who indicate a preference for Yes or No, but say they might change their minds. The other comprises those who while unwavering in their referendum preference, also say they are not “absolutely certain” they will vote.
In short, the 44 per cent are not Scots who have no idea at all how they might vote. Most have a view but might just possibly be open to persuasion or need a little encouragement to vote. Perhaps the real story is that as many as 56 per cent are apparently utterly determined in their beliefs.
Which brings us to Mr Silver. As the US guru noted, changing the balance of opinion when so many have already apparently made up their minds will not be easy.
Indeed as he quite correctly remarked, over the last 18 months (and indeed years) there has been little change in independence support.
Mr Silver also went on to insist that as referendum day approaches opinion could be expected to drift towards the status quo. That, however, is not what the campaign polls suggest happened in 1995 in Quebec; rather they exhibited a 4-5 per cent swing from No to Yes. “Separatists” can sometimes make gains in referendum campaigns.
In truth what pollsters need to be aware of above all is the fragility of their trade.
• John Curtice is Professor of Politics at Strathclyde University and editor of whatscotlandthinks.org.