Nineteen and 41. In respect of this year’s looming General Election, and the number of seats to be won by the SNP, these are wildly different returns, and would have wildly different consequences.

Nineteen seats would mean that the Labour Party would win the election in Scotland, rather handsomely, with nearly 30 seats. Humza Yousaf, the First Minister, would almost certainly be gone, and Labour would become hot favourites to win the critical Scottish Parliament election of 2026, and see its Scottish leader Anas Sarwar strolling into Bute House to become First Minister. Independence would be in a deep sleep. It might never re-awaken.

Forty-one seats, on the contrary, whilst fewer than the 48 currently held, would represent a stunning victory for Mr Yousaf, against all the odds and against all the critics. Although Labour would increase its representation from one to 14 seats, it would be seen as a failure by Mr Sarwar, and indeed by Lib Dem leader Alex Cole-Hamilton (losing two, including his own back yard) and Douglas Ross’s wiped out Tories, who would have succumbed precisely to Mr Yousaf’s chosen strategy of making Scotland "Tory-free". Mr Yousaf’s internal critics would be firmly in their box, he would have momentum going into the 2026 poll, and independence would be back on the agenda.

Read more by Andy Maciver: Two key changes that could save the NHS

Read more by Andy Maciver: Three steps Scotland can take to reach net zero

In the last five days, these numbers have been the talk of the steamie, because two different "mega-polls" have predicted them as the outcome of the General Election. Both polls were conducted under the multi-level regression and post-stratification (MRP) model, designed to provide granular and accurate results down to a constituency level, and were both carried out by large, respected polling companies: Survation and YouGov.

Yet, alas, at least one of these polls is wrong. Logic suggests the prediction that the SNP will win 41 seats is the one which should raise eyebrows highest. The scale of that result runs contrary to almost all other polling which has been conducted over the last year, and several of the individual seat results appear difficult to understand. For instance, I am writing this week from the Western Isles, a seat won in 2019 by Angus Brendan MacNeil, for the SNP. Mr MacNeil has left the SNP and will stand as an independent, splitting the vote of his replacement SNP candidate and surely allowing Labour’s well-known, local journalist-turned-candidate Torcuil Crichton to win at a canter. And, yet, the poll predicts an SNP hold.

Nonetheless, Scottish Labour would be wise not to discount these returns altogether. The SNP has proven itself, over nearly two decades, to be a remarkably resilient political party. Its vote is sticky, with a floor of around one-third of the electorate for whom Labour is not offering anything sufficiently attractive to bother switching allegiance.

Even before this week’s MRP polls, Labour in Scotland was losing more polls, based on vote share, than it was winning. Why, after 17 years in government and with significant question marks over its performance, particularly since it entered the coalition with the Greens, does the SNP still beat Labour in most of the polls, most of the time?

An instructive answer on this may come from a third poll, conducted by the Diffley Partnership for the Holyrood Sources podcast, of which I am one of the hosts. We commissioned this poll ahead of our live edition reflecting on 25 years of devolution specifically to ask the rarer, more existential questions about our constitutional settlement. We know that the top concerns of the population are the economy (always first) and the health service (always second), but we also know that there remains interest and heightened tension around the constitutional question, and moreover we know that continued introspection on that question is crowding out debate on how to improve those primary concerns such as the economy and the NHS.

The Herald: Scottish Labour could look at decentralising power away from HolyroodScottish Labour could look at decentralising power away from Holyrood (Image: Newsquest)

Labour is curiously quiet on Scotland’s constitutional future, given the prescient leadership it displayed during the 1990s. Indeed, since the independence referendum, Labour has become decidedly Tory-ish on the matter, adopting an angry this-far-and-no-further approach. But our Holyrood Sources poll shows that if Labour takes a breath, and thinks calmly, it could inflict a mortal wound on the SNP.

Only 20 per cent of Scots support the current devolution settlement. And yet, when an enhanced devolution settlement is listed as an option, support for independence drops to 38 per cent - a lower level than we have seen for well over a decade. As we look at the detail, nearly half of Labour’s voters want either independence or enhanced devolution, and conversely nearly one-third of SNP voters would reject independence if enhanced devolution was offered.

There is a clear enough way through this for Labour, and it needs it. The party has begun to take some votes from the SNP, but the word "some" is critical here. It must be conscious of the fact that a greater deal of its increased vote share has come not at the expense of nationalists, but at the expense of the Tories, whose transactional unionist voters have moved en masse to the party they expect to be in Downing Street, returning the Tory Party to its natural core vote in the teens.

This is not enough to make Mr Sarwar First Minister. There are some early signs that he knows this. He is understandably wary about calling for much more devolution from Westminster to Holyrood, other than the special measures on immigration which he revealed on the Holyrood Sources podcast. If shifts in responsibility were performance-related, Holyrood would be defenestrated, not enhanced.

But whilst there remains a clear case for a Westminster-to-Holyrood transfer, the thistle he may grasp is to decentralise Scotland, from Holyrood to the cities, regions and islands. Last week, he mooted the concept of mayors for Scotland’s cities; an easy and obvious move given Labour’s history with mayoralties in England. He should go further, by creating more powerful regions, with strategic authority over life-affecting areas such as health, education, housing and transport, all with a mayor in charge.

Voters will rarely tell a pollster that constitutional change is more important than the economy or the NHS. But they will reward enthusiastic vision, especially if that vision unlocks the potential for improvement in the areas they really care about.

Mr Sarwar should be no more exhilarated by one poll than he should be disheartened by the other. Both, in their own way, highlight the enviable opportunity he has before him.