GENERALLY speaking, opinion polls ask people what they think about things.

But one came along yesterday which asked people what they thought other people think. I think this illustrates how difficult it is becoming to interpret what the polls are telling us about September's referendum.

We have a had a flurry of surveys in recent days and - no surprises here - each side in the campaign is overjoyed. The statistics tell us the momentum is with the Yes side and the No camp at the same time - if you believe everything you read in campaign press releases, that is.

So what are the numbers?

The latest ICM poll at the weekend, conducted after George Osborne ruled out Alex Salmond's currency union proposal, put Yes on 37%, No on 49%, and Don't Know on 14%. Yes was unchanged compared with an ICM survey last month while support for a No vote rose 5%.

Good news for No, no? Well, yes, unless you look at ICM's "experimental" extra question asking people how they thought others might vote on September 18. It showed that, on average, people are expecting the referendum to produce a narrow victory for No by 53% to 47%.

The justification for asking the question is the idea of the wisdom of crowds, the idea that random groups are sometimes more prescient than individual experts. The result certainly cheered the crowd at Hope Street, where Yes Scotland chief executive Blair Jenkins said it showed the campaign was on a "winning trajectory".

Another new poll from Panelbase, who have tended in the past to find stronger support for Yes, put backing for independence at 37%; for staying in the UK at 47%, and the Don't Knows on 16%. The results suggested the gap had narrowed by two points in a fortnight but - as the No campaign pointed out - the poll was not as encouraging for Yes as Panelbase's September survey, which had the independence campaign one point ahead (though that poll used slightly different wording).

There have also been specific polls conducted in the north-east, across the UK and among oil workers which produced sufficiently contradictory findings to enthuse both sides.

They followed surveys last week from Survation and TNS BMRB which suggested, firstly, a small but significant backlash against the Chancellor's currency warning, and, secondly, a small narrowing of the gap earlier in the year. At the end of this latest statistical bombardment, both sides are insistent the public mood is moving in their direction. As ever in these situations, it's wise to turn to polling expert Professor John Curtice. His take, on the What Scotland Thinks website, is clear.

The row over whether an independent Scotland could reach a deal to share the pound with the rest of the UK has had little impact on voting intentions, when methodological changes with the ICM and Survation polls are taken into account, he believes. However, looking at all the polls over the past six months, support for a Yes vote has been higher this year than it was towards the end of last, rising from about 38% to 42% on average. The No campaign still enjoys a healthy lead, then, but there is growing evidence the gap is closing.