This is an excellent example of how the London media persistently misrepresent the state of play shown by polls on independence, while maintaining a veneer of plausibility -
"But opposition leaders said Salmond knew that independence was unpopular: a series of opinion polls has consistently shown that about a third of Scots back independence, with support for remaining in the UK commanding majority support."
I have no problem with the claim that "about a third" of Scots back independence. That is indeed what recent polls have "consistently" shown - but only if the substantial number of don't knows and won't says are included in the calculation. The problem is that the only way the second part of the claim (that support for remaining in the UK commands majority support) can be said to be "consistently" true is by doing the complete opposite, and excluding the don't knows from the calculation. With don't knows included, the position suddenly looks a good deal less consistent. Here is the percentage No vote in each of the opinion polls this year...
Angus Reid, January - 50% - NOT A MAJORITY
TNS-BMRB, January - 48% - NOT A MAJORITY
Panelbase, January - 47% - NOT A MAJORITY
Angus Reid, February - 47% - NOT A MAJORITY
Ipsos-Mori, February - 55% - MAJORITY
TNS-BMRB, March - 52% - MAJORITY
So, far from the polls "consistently" showing a majority in favour of remaining within the UK, in fact only a third of the polls published this year do so. The Guardian have a choice - either they exclude don't knows, in which case the average support for Yes in recent months has been somewhat higher than a third, or else they include don't knows, in which case there is no clear majority for No. But they really can't have their cake and eat it.