POLLS have a way of confounding expectations, and the latest Herald/System Three survey on attitudes to our constitutional future is no exception.

The SNP had been on a seemingly relentless surge in recent months. On voting intentions for the Scottish Parliament it trailed Labour by 19 points in November. The gap fell to 11 points in February, and then to just a single point in March. By April, it was a dead heat, 40:40, or in the words of our headline, ''Deuce''.

A fortnight ago, the nationalists surged into a five-point lead and we were convinced that there was a fundamental shift taking place in the political landscape. Surely this would be reflected in the view of the people on the future governance of Scotland, and there would be a parallel surge in support for independence?

Not so. Those preferring independence remained at 34% - the same level as 18 months ago.

The surprise was reflected in The Herald's treatment of the finding, which focused on the 58% who rejected independence, and spoke of SNP leader Alex Salmond having a mountain to climb if he was to win a constitutional referendum.

The result would send the Mr Salmond homeward to think again, to paraphrase the song, and the result would be music to the ears of Mr Donald Dewar, the first good news for the Scottish Secretary from Herald/System Three polls in many months.

But once over the initial surprise that the SNP surge had not been matched by a commensurate increase in support for independence, it became possible to view the finding a little differently. The 34% in favour of independence is at the high end of our historic scale, being bettered only once, by a single point, in November 1991.

Just more than two years ago, in March 1996, the figure was as low as 25%, and the combined support for the alternative options of devolution or the status quo peaked at 72%. The ''anti-independence'' had never been lower than 61% - until now. Also, the number of ''don't knows'' had been as low as 2% and had never been above 6% - until now, on 8%.

Viewed on all these points, the situation may seem less bleak for the SNP, which can see it in terms of one of those sporting metaphors of which its leader is fond - game on and all to play for.

Certainly, the 40% who would vote SNP in the Scottish Election but do not support independence is a challenge for the party. But Labour must ponder the 29% of its supporters in favour of independence.

The wording and circumstances of such constitutional polls is crucial and it must be said that the SNP has never been particularly enamoured of the way The Herald poll has been phrased. The latest survey asked respondents if they wanted ''A Scottish Parliament, but with Scotland remaining part of the UK'' or ''An independent Scotland, separate from the UK''.

This is an honest attempt at clarity, but the SNP prefers the use of a phrase such as ''Independence in Europe'' and dislikes the negative overtones of the word ''separate'', pointing out that separatists is used by its political opponents as a pejorative term for the party.

Another concern is that a constitutional survey in which the independence/devolution question is asked up-front may colour people's subsequent answers on party voting intentions for Westminster or Holyrood, but the latter allegiances have to be investigated for the main result to be analysed.

For this reason, this week's voting intention results will not be formally recorded along with our regular monthly poll findings, but they were not out of line with our recent results. The SNP remained ahead of Labour by two points (down from five), while Labour appeared to bounce back in Westminster voting intentions, increasing its lead from 14 points to 21 points.

One consistent feature within the latest set of results was that a smaller proportion of those supporting a party for Holyrood was in favour of independence than among those supporting that party for Westminster.

Thus 62% of Westminster SNP voters want independence, compared to 56% (the higher total) of those voting SNP for Holyrood. Devolution is preferred by 33% of Westminster SNP supporters, but 40% of those who would vote SNP for the Scottish Parliament.

Of those supporting Labour at Westminster, 32% want independence, falling to 29% at Holyrood, with devolution support climbing from 61% to 65%. Among Conservative voters, 10% at Westminster favour independence, but 8% at Holyrood (devolution 79% and 80% respectively). Among Westminster Liberal Democrat voters 15% want independence, but 12% of Holyrood voters want the same (devolution 81% and 84%).

Westminster votes therefore clearly include a greater measure of protest pressure in favour of constitutional change, which dissipates in the context of the Scottish Parliament. But we know that there is a very marked shift in favour of the SNP in Holyrood voting intentions, presumably because with the Conservatives not seen as a threat or an option, the nationalists are the natural recipients of anti-Labour votes.

Perhaps the single surprise in the findings is among young voters, 24 and under, with 41% supporting independence, in contrast to 49% among 25 to 34-year-olds. Support for independence then falls steadily with age, sinking to just 20% among those aged 65 and over.

For young adults and the elderly it must already seem such an uncertain world that they are reluctant to add to that uncertainty with further constitutional change.