Five points of margin on about 190,000 counted ballots comes to roughly 9,500 votes [1][1], so fewer than 5,000 people changing their minds would have produced the opposite headline [2]. Hold onto that figure, because it is the one both camps will misread this week.
The other number worth holding is 5.75, the distance in percentage points between Iceland's 8.00 per cent benchmark and the European Central Bank's 2.25 per cent deposit facility [12][3], which is to say the Icelandic rate runs about 3.6 times the euro-area one [4]. That gap is what the cost-of-living argument was reaching for [5]. Look at the route to it, though: reopen talks, close a deal that some analysts put at one to two years given how far the file had advanced before 2013 [3][2], ratify, and then arrive at what the campaign itself described only as the possibility of adopting the euro [5]. Four conditionals deep, a 5.75-point saving discounts hard, and it was being offered in exchange for something voters already own outright, namely the rules governing fishing and farming [6]. A contingent future coupon against a present asset, and written that way the outcome is ordinary.
The direction the campaign travelled is the part worth studying. One survey had 51.3 per cent in favour [8]; Gallup, nearer the vote, had 51.6 per cent opposed [7]; the count came in at 47.5 per cent in favour [1]. So the yes share finished 3.8 points below that earlier reading [6], and Gallup still understated the no side by 0.9 [7]. (The earlier survey's two figures sum to 99.7, which is the published arithmetic, not mine [8].)
This is probably wrong, but the more interesting reading is that the security case lost under about the friendliest conditions available to it. Iceland has no armed forces [9]. It voted while the US president was threatening to take over Greenland, an autonomous territory of Denmark [10], with Bruegel describing an Arctic in which China flexes economic muscle in partnership with Russia as melting ice opens new routes [11], and the governing party ran on precisely that exposure [4]. The counter-thesis deserves equal weight: 5 points sits inside a poll series whose margin moved 6.1 points between two readings [9], turnout in a small electorate can account for a move that size, and on that view Greenland did shift votes and simply did not shift enough of them.
Iceland applied in 2009, suspended in 2013 [2], and returned to the question about 13 years later [13], which is something like 17 years of not deciding [5], and a 52.5 per cent no forbids nothing about a third attempt. What would prove this desk wrong: a fresh application inside this parliament, which would show the referendum was a mood rather than a constraint; or a post-vote breakdown showing the yes share rose among the voters most exposed to Arctic risk rather than sliding 3.8 points [6] while security was the loudest argument in the field.
