“BOOM! Earth’s population could hit 12 billion by 2100”. That was the headline on Wired.com which greeted research by Patrick Gerland and others of the United Nations’ population division looking at the UN’s population projections to 2100. Britain’s Guardian newspaper said the study, published recently in the journal Science, “overturns 20 years of consensus on peak projection of 9 billion and gradual decline.” Climate News Network, a non-governmental organisation that tracks and summarises environmental articles, reckoned the study “has profound and alarming implications for political stability, food security and, of course, climate change.”

But hang on a second. The UN’s population division is the outfit that much of the world relies on for basic demographic information. If it had changed its forecasts and overturned 20 years of consensus, that would be a very big deal indeed. So has it? The answer is no. The headline projection in the Science study says the world’s population is likely to grow from 7.2 billion now to 9.6 billion in 2050 and to 10.9 billion in 2100 (not 12 billion). This projection is not new. It was first made by the UN itself in its 2012 estimates. (Before that, the UN had projected a population of 9.3 billion for 2050.) The Science study confirms, rather than changes it. The UN (and many other demographers) have already stopped projecting a peak population of 9 billion.

The distinctive feature of the new Science study is not what it says about the headline projections, but its approach to estimating the range of possible outcomes. At the moment, the UN works out its main projection (called the medium variant) based on countries’ fertility rates and life expectancies, the most important determinants of population growth. It then asks, what would our projection be if the fertility rate were 0.5 higher? That gives the high variant. What would it be if fertility were 0.5 lower? That is the low variant. This technique has the advantage of being clear and simple, but is also problematic because it is highly unlikely that all countries would have their fertility rates 0.5 higher or lower at the same time. It also produces a range of possible outcomes that looks excessively wide.

The new study uses more sophisticated statistical techniques called probabilistic projections. They enable the authors to assign probabilities to their projections. They reckon there is an 80% chance that the world population will increase to between 9.6 billion and 12.3 billion in 2100 and a 95% chance that the range will be 9 billion to 13.2 billion then. This does not mean they have increased their forecasts. Rather, they have used the same medium projection, with a narrower range of possibilities around it. The UN’s low variant for 2100 is now 6.7 billion; its high variant is 16.6 billion. The study therefore moderates the previously excessive uncertainty about the future size of the population. It reduces the error bars.

That is useful. Sometimes, it helps the authors confirm existing forecasts. Wolfgang Lutz of the Vienna Institute of Demography, for example, has said there is only a one-third chance of the global population doubling between 1997 and 2100; the new study puts the odds at one-quarter, roughly similar. Sometimes it provides different forecasts. Mr Lutz and his colleagues say there is an 85% probability that world population growth will stop by 2100; the Science study puts the odds at 30%.

But all such conclusions should come with a big health warning. Forecasting anything 85 years out is highly uncertain—and population projections are no exception. Nicholas Eberstadt of the American Enterprise Institute, a think-tank in Washington, DC, points out that “the basic trouble with all long-range population projections is that they are driven by assumptions about birth levels—and there is still no reliable method for predicting fertility levels a generation from now, to say nothing of a century hence.” Three generations will pass between now and 2100. Almost anything could happen in that time. And probabilistic projections may well not pick the biggest changes up.

They would not, surely, have forecast the spectacular declines in fertility in Bangladesh or Iran since 1980 (in both countries, from roughly 6 children per woman to about 2 now). At the moment Africa is the source of much new population growth and the authors assume that fertility rates will continue to fall more slowly there than they did in Asia and Latin America. But no one can be sure.

The new study refines the UN’s current projections and provides a lot of useful material. But population forecasts for 2100 should not be taken literally.