Science1 distinct publisher3 min readUpdated
Two provisional totals arrived within a thousand of each other. They measure different things over different weeks, and both point to dying concentrated in a fortnight.
The Scientist · Science desk

Compiled by The ScientistSomething wrong?How this is made
The two headline totals sit within a thousand of each other, which invites the reading that they corroborate. They do not. The continental estimate is all-cause excess mortality built from official statistics in 23 countries, with part of the east left out [5]. The eight-country count mixes measurement types: AFP describes its own tally as excess deaths or deaths directly linked to heat, depending on which authority published [2]. The reporting windows differ too, running roughly from late May into August for the national figures [2] against a single quoted month for the continental one [1]. That month carries an internal awkwardness of its own, since the mid-July to mid-August window AFP prints for the 32,000 does not contain either of the two weeks it names as the summer's deadliest [1][3][4]. Add the eight national tallies and the sum lands near 33,200 [1], so the arithmetic is sound. It is sound for eight countries.
The distribution matters more to a roster than the total does. Two weeks hold roughly 24,000 of the 32,000 [2]. Belgium's 2,570 fell inside a thirty-day window, an average of about 86 deaths a day [13][4]. The Dutch 968 arrived in fourteen days, about 69 a day [14][5]. Germany's 14,000 through 9 August is about 171 per million against a population of 82 million [6][7][3].
Where the load sits is not a mystery. The Robert Koch Institute says people over 75 suffered the most in Germany, as in other European countries [8]. That is the cohort held by residential care and home visiting services, whose cooling and night staffing were set by capital and rota decisions taken in cooler years.
Nearly every figure here is a floor. France has 7,300 to 22 July, and its public health authority expects that to rise because the long heat wave that started in late July has only just ended [9]. Belgium updates in September, and the added window covers the start of its worst wildfire in modern history [13]. England's 2,877 for May and June is 108 deaths short of the 2,985 record from 2022, about 3.6 percent below it [12][6], and the UK health and safety authority has said the 2026 total could be substantially higher once other hot spells are counted [12]. Austria's 396 June deaths have no precedent in data going back to 2017 [15]. Spain's near-4,500 is the highest for that stretch of the year since 2022 [10], recorded in a July that AEMET rates the joint-hottest for mainland Spain since records began in 1961 [11]. Portugal has more than 600 from mid-June to the end of July [16]. Scientists attribute the summer's heat waves and drought to man-made climate change [17]. Any plan sized on the August print is sized on an undercount.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
An AFP count of national provisional tolls, covering periods roughly between the end of May and August, found 33,000 excess deaths or deaths directly linked to heat across eight countries: Austria, Belgium, England, France, Germany, the Netherlands, Portugal and Spain.
According to EuroMomo's latest weekly estimates, the last week of June was the summer's deadliest, with nearly 16,000 excess deaths.
The first week of July was second, with some 8,000 excess deaths.
The UK health and safety authority estimates England saw 2,877 heat-related deaths across May and June's heat waves, against a historic tally of 2,985 from summer 2022, and warned the 2026 total could be substantially higher because of other periods of high heat.
Sciensano estimates Belgium saw 2,570 excess deaths between 18 June and 17 July; the toll is to be updated in September to cover 25 July to 16 August, which saw the beginning of the country's worst wildfire in modern history.
AFP's analysis of RIVM figures puts Dutch excess deaths between 22 June and 5 July at 968.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Named official agencies, all figures provisional
Every material number is attributed to an identifiable institution - EuroMomo, the Robert Koch Institute, French public health authorities, the Carlos III Health Institute, the UK health authority, Sciensano, RIVM, AGES, DGS and AEMET - and the reported national tolls sum consistently with the stated aggregate, which is a real internal check. Against that, the whole cluster rests on one wire story with no independent republication, EuroMomo publishes no country breakdowns to audit, no intervals or baselines are given, and the piece is explicit that the estimates will be refined. That supports a solid but capped evidence score.
No adoption surface in a mortality-statistics story
Adoption is not a meaningful dimension here and the supplied source contains no releases, deployments, benchmarks, pricing or usage disclosures to measure. The cluster reports provisional death tolls and the agencies that published them; inferring an adoption level from that would be fabrication.
Mildly overstated framing over sound underlying figures
The underlying agency numbers are reported soberly and with caveats, so this is not an inflated story - but the packaging runs slightly ahead of the data in two identifiable ways. The headline's 'at least 30,000 across a blistering summer' sits awkwardly against the body's 32,000 for a mid-July to mid-August window, and England is described as 'threatening to topple' a record its own figure is 3.6 percent below. Aggregating excess-death estimates and directly heat-attributed deaths across eight different date ranges into one 33,000 total also reads as more precise than the method allows. Small positive, not large.
Public-agency data, wire distribution, no commercial stake
The figures originate with national public health institutes and a non-commercial European mortality monitor, relayed by a news agency and a science aggregator; nothing in the supplied source shows a vendor, funder or party with a financial interest in the numbers. The residual incentive pressure is institutional rather than commercial: agencies control the release and revision cadence of their own provisional tolls, and the wire has an evident interest in a round continental headline number, visible in the 30,000-versus-32,000 framing.
Consistent agency data, single source, provisional throughout
Confidence is moderate: the arithmetic reconciles, every figure carries a named institutional source, and the two independent accounting paths converge within about a thousand. It is held down by the absence of any second publisher, the explicit provisional status of all tolls, at least three national figures flagged as certain or likely to rise, and the internal inconsistency in how the continental headline number is scoped.
science
A named ship and a published timetable turn the Arctic into a bookable lane1 distinct publisher
science
Honduras puts 234 of 298 municipalities on drought alert, and names its own end date1 distinct publisher
science
Belgium's peat fire took seven days and four countries' aircraft to contain1 distinct publisher
science
A 3.9C El Nino peak makes 2027 a stress test for plans built for the late 2030s1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
phys.org
1 article · August 22, 2026