The latest poll on Scottish independence is out. George Eaton has the details on the New Statesman, but the headline numbers put the No campaign fractionally in front, 53 to 47 per cent (excluding undecideds).
The poll will have quelled union fears, coming after YouGov put the Yes camp in front for the first time on Saturday. But should it? What do the polls suggest overall?
There is no one way of measuring them. Everything depends on how much weight you give the latest polls. This is how the race looked before yesterday’s result if you just consider the polls published in the past fortnight.
The graph uses all the individual polls to create a trend-line that shows the current state of the race. It suggests the campaigns are within a percentage point of each other.
But yesterday’s poll has slightly widened the gap.
This shows the unionists fractionally ahead by 1-2 per cent, which matches the average of the four most recent polls (by TNS, Panelbase, YouGov and now Survation). But this is well within the margin of error.  The race is effectively a dead heat.
Some poll trackers, like the Daily Mirror’s, use linear trend-lines, which gives more weight to polls from earlier in the summer, and makes the race seem slightly less close.
Should these earlier polls, taken from a time when it seemed the unionists would win “by a decisive enough margin to settle the matter for many years to come”, be included?
It depends how decided you think the electorate is. If you think the sudden change in the polls is reminiscent of Cleggmania, you might want to give less weight to the most recent numbers.
This comes down to how much you think people are changing their minds. Most polls are reported with the undecideds excluded; this leaves out around 1 in 10 voters. But the changes of the past month have been greater than these voters can account for – people who declared themselves to be decided seem to be switching between the campaigns.
If you think they are, the two-month trend-line may seem more accurate.
The SNP do, however, have a record of building on their momentum. They trailed Labour for much of the run-up to the 2011 Scottish Parliament elections, but only pushed on once they caught them in the polls with six weeks to go.
By election day they were forecast to win in a landslide, which they did, in both types of the vote. 
Unionists should be wary of relying on old numbers. But they can dismiss any giddy predictions of a landslide for Salmond. It took his party six weeks to establish a double-digit lead over Labour in 2011 after they first tied in the polls.
The SNP haven’t the time to win convincingly, but they may have just enough time to win.
 Typically pollsters quiz around 1,000 people for a poll. This gives a very accurate measure of what millions of people think, but is inaccurate by ±3 per cent. For instance, when YouGov’s poll over the weekend put the Yes campaign in front, 51 to 49 per cent (excluding undecideds), it really meant Yes are between 48-54 per cent, and No between 46-52.
 The Scottish Parliament elect two MPs in two ways: by constituency and region. The system was unsuccessfully designed to prevent the SNP winning a majority.