WHEN I see a poll showing that the No side has a 30-point advantage in the argument over Scotland's future, I have one reaction: it doesn't feel right.

You can thank me for the contribution later. In terms of profound insight, that's about as useful as it gets.

Notice the appliance of science. Observe the rigour from one who probably doesn't get out enough. His rebuttal of a careful, crafted survey by the respected YouGov organisation is based on a mere intestinal reaction. They say 59% will vote No and only 29% will vote Yes? The man on his way to the corner shop says it doesn't feel like the truth.

On the other hand, show me a poll with a 44%/43% split in favour of independence, like the latest Panelbase exercise, and I'll give you one of my sage nods. There's the country I recognise, the one still at odds with itself, opinions going toe to toe, day by day, as we work through arguments and seek a resolution.

Then throw more kindling on this little blaze. Here's the latest TNS BMRB survey. It has support for independence down to just 25%, a fall of eight points since February, the least support achieved by the self-determination proposition since 2007. The complications don't end there.

This poll says that Unionism, too, has lost a good deal of ground, falling from 52% to 47%. Most remarkable of all, those who still don't know, who have been bombarded with claim and counter-claim, are more numerous than before. In fact, their numbers have almost doubled, from 15% to 28%. When last I heard from my intuitions, they said no such thing. They said voters were moving, albeit by inches, towards Yes.

But then I have to weight my opinion, mostly for bias. Wishful thinking is potent stuff. First, the Panelbase survey that suits my guess has been conducted on behalf of the SNP. Secondly, it fails to fit a now-familiar pattern, the poll-of-polls awarding a handy win to No. You might not call the Panelbase result a rogue, but it is certainly an outlier.

Thirdly, it is probably true that I don't spend much time in the company of Unionists. When I meet the Daily Telegraph's prince of darkness, we talk about football. There might be a ferment in Britain's favour around the country to which I am blind. It could also be that, like the campaigns, I'm missing something.

One possibility involves the constituency excluded from this contest by Westminster fiat. Since additional autonomy within the UK - devo plus, max or whatever - is not on offer, how are those so inclined supposed to vote next year? How can they get the result they want? One poll says they'll stick decisively with the status quo; another that the ball is still in play; a third that ambivalence, mature indecision, is winning.

So I settle for the one undisputed fact, the one you might hear mentioned at any corner shop: the polls are all over the place. How come? YouGov, Panelbase and TNS BMRB are expert at what they do. These firms are not in the business of suiting their findings to their clients. Their methodology can be argued over, but each organisation has built its reputation on professionalism and accuracy. So how can there be such a disparity between their findings?

As we reported yesterday, Professor John Curtice of Strathclyde thinks a clue to the difference between Panelbase and YouGov might lie in the order in which questions are asked. If the popularity of the SNP as a government is established with an inquiry into election preferences rather than referendum choices, he speculates, Yes benefits. That might be true. But it generates as many questions as answers.

Two of those spring to mind. First, are we, collectively, still so uncertain, so pliant? This debate has been going for a while, to put it mildly. Does a small, inadvertent reminder to respondents of the popularity of the SNP truly have such a startling effect, even now? For that to be the case, we would have to be a pretty thick electorate. Contrary to the view of some politicians, I don't think that's the case. Yet TNS BMRB confirms a large, growing degree of uncertainty.

Secondly, if the professor is correct in believing that the sequence of questions and answers is vital, both campaigns are obliged to speculate. What will voters ask themselves next September, ballot paper in hand? The No faction has put a lot of effort into discrediting the SNP in general and Alex Salmond in particular. Better Together has tried to depict independence as one party's obsession. So what if voters buy the idea and act as they acted before the last Holyrood elections?

Panelbase had better luck with that campaign than most other pollsters, never mind (for here we draw a veil) pundits. The organisation did not predict an SNP majority government, for no-one did, but its surveys reported a movement in opinion long before the rest. In no form of logic does one previous result guarantee a future result, but the piece of history is tantalising. Failing to act like every other pollster has done Panelbase no harm in the past.

The question remains: how could reputable polling organisations be so much at odds? How do we, each of us, for or against, measure personal experience against survey data? It is a human habit to spend your time with like-minded people. Friends are friends because, most of the time, they confirm your beliefs about the world. Only in the workplace, perhaps, do we find our complacency challenged. That's where the best polls are conducted.

A full year ahead of the vote, Better Together, the coalition of Westminster interests, is urging its workers to beware complacency. Simultaneously, the campaign's leaders seem to want to create the impression that all is done and dusted. Yet they ignore the fundamental reality. One large minority of Scots want out of the Union; another large group - huge, according to TNS BMRB - has scarcely flocked to the flag. That "positive" British campaign is nothing to boast about.

In honesty, never mind in fairness, you could stand the argument on its head. Dissatisfaction with the Union is evident; enthusiasm for the alternative offered is muted. Only one conclusion is possible: independence as it is being described, independence as an idea diluted to the point of meaninglessness, is failing to win hearts and minds. There should be no surprise in that, but the lesson has yet to be learned.

A little more than a year remains. Unexpected things can and will happen. The one certainty is that voters will have to make a choice, ready or not. My feeling, no doubt born of hope and bias, is that they will converge on Yes, less because they fail to understand the benefits of Union than because they have plenty of experience of what Westminster government entails. Better Together has a pig-in-lipstick problem.

Yes Scotland's problem is less complicated, more stark. Those voters who have yet to be won are not hearing what they need to hear. Nationalists of all sorts need to ask themselves: why not?