The UK in figures

💷 How to read the debt figures ✕

Why today's interest bill was set years ago

The government borrows by selling gilts, and a gilt keeps the interest rate it was sold at until it is repaid. Most of the debt on the books today was sold between 2016 and 2021, when the market charged well under 1.5% for ten-year money. So the bill paid now is the average of those old rates — not today's. The chart of the two lines shows exactly that: the rate the market charges jumps around; the rate the whole debt pays creeps.

"Costs now" and "at today's rate"

Costs now is the IMF's figure for net interest paid in the last reported year, as a share of everything the government took in — the World map calls it Current debt. At today's rate is what that share would be if every pound of the debt were borrowed at this month's ten-year rate — the map's Forward debt. It is a what-if, not a forecast. The gap between them is the cost still on its way.

How the road between them is drawn — our arithmetic

Roughly one pound in fourteen of the debt is repaid and re-borrowed each year (the Debt Management Office's average life of the gilt portfolio is about 14 years, the longest in the G7). Each year, that slice moves from the old rate to today's, so the bill closes one fourteenth of whatever gap is left: 7% of the way in year one, about a third by year five, about half by year ten. The formula is printed on the tile with every term filled in. It is the reason Britain feels a rate rise more slowly than other countries — and for longer.

What the model leaves out, on purpose

It holds the debt at last year's size, so new borrowing to cover the deficit — which is all at the new rate — is not in it; that is why the OBR's own reckoner gives a bigger five-year figure than our line does, and why our line is a floor. About a quarter of gilts are index-linked and follow inflation, not the yield. "Costs now" is net interest and "at today's rate" is gross, so the true destination sits a little below the forward line. And the ten-year rate stands in for the whole range of maturities.

The ±1 point lines

The same model run with the ten-year rate one percentage point higher and one point lower. They are not a range we predict; nothing on this tile forecasts where rates go.

Where the numbers come from

Debt and revenue: the IMF's World Economic Outlook and Fiscal Monitor. The ten-year rate: the OECD's monthly long-term interest rates, same calendar month in each of the last ten years. Both are the series behind the World map's Debt layer, refreshed by the same daily job, so the map and this tile can never disagree. The cash figures in pounds are the OBR's.

🧾 How to read the income tax figures ✕

Why this site now divides costs by income tax payers, not by everybody

"£1.23 for every person in Britain" spreads a bill across babies, schoolchildren and everyone else who will never be sent it. Dividing by the people who pay income tax gives a larger and more honest number for the reader who does pay. Every "per income tax payer" figure on the Britain overseas tiles uses the one count on this tile, so if HMRC revises it they all move together.

"Income tax payer" is not the same as "taxpayer" — and we never say it is

Nearly everyone pays some tax: VAT on what they buy, duty on fuel and alcohol, council tax, National Insurance on wages. Nobody counts those people, and HMRC does not try. What HMRC does count is who is liable for income tax, and that is the only published head-count of people who pay a tax. So the figures here are exactly that, and the words on the tiles say "income tax payer" every time. Income tax is the government's largest single tax but it is well under half of what it raises, so the division is a way of sizing a cost, not a statement of who funded it.

Pensioners are in the count, because millions of them pay

It is easy to assume retired people are outside the tax system. HMRC projects that about 9.6 million income tax payers — nearly one in four — are over state pension age. The state pension itself is taxable income, and the full new state pension is £241.30 a week in 2026-27 — about £12,548 a year, some £22 short of the frozen £12,570 allowance — so almost any private pension on top is enough to create a bill. Children are genuinely outside: HMRC told Parliament there were about 13,000 income tax payers aged 16 and 17 in 2022-23.

"By the highest rate each one pays"

HMRC sorts people by their marginal rate — the rate on their top slice of income. A higher-rate payer still pays nothing on the first £12,570 and 20% on the next slice; they are counted once, as higher rate. The "savers rate" group are people whose only taxable income is savings interest or dividends. Scotland sets its own bands: HMRC files Scottish starter, basic and intermediate rate payers under "basic", and since 2024-25 Scottish advanced rate payers under "additional", so this is one UK-wide picture.

Measured and projected

The last year HMRC has actually measured is 2023-24, from its Survey of Personal Incomes — a sample of tax records. Everything later is a projection built on the Office for Budget Responsibility's March 2026 forecast, published on 15 July 2026. We divide by the 2026-27 projection because it describes the people paying now; the chart marks which bars are measured and which are projected.

Why there is no official number for people who work and pay no income tax

Asked in the Lords in July 2025 how many adults pay no income tax, the Treasury replied that the figure is "not currently held or published". HMRC's records cover people who come into contact with the tax system; somebody with a small job and no tax to pay can be invisible to it. The subtraction on the tile — adults minus income tax payers — is ours, and it uses a population figure and a payer figure for the same year on purpose.

How our estimate of low-paid workers is made, and where it is weak

The Office for National Statistics publishes two tables of how pay is spread out, and we read where the £12,570 allowance falls in each. Payroll records (PAYE Real Time Information) cover everybody on a payroll, add together the pay from all of a person's jobs, and give monthly pay at every fifth percentile. In July 2026 the 10th percentile was £893 a month and the 15th was £1,055; the allowance is £1,047.50 a month, which sits almost at the 15th — so about 15% of 30.25 million people, some 4.5 million. The earnings survey (ASHE, April 2025) gives annual pay, but only for jobs held for more than a year: 10th percentile £11,425, 20th £18,560, which places the allowance at about the 12th percentile — some 3.5 million. Between two published percentiles we draw a straight line; that step is ours. The two disagree for a reason. The payroll figure is a monthly rate, so it includes people in short or part-month jobs who will earn little over the year — and also some who will earn plenty in other months. The survey leaves out everyone who changed jobs in the year, who are disproportionately low paid, and it counts jobs rather than people. A whole-year answer most likely lies between them. Neither tells you who pays no income tax: someone paid £9,000 with a pension on top is a payer. The self-employed are a separate figure from a different source and an earlier year, which is why they are shown beside the employees and not added. One oddity worth knowing: the 40th percentile of part-time annual pay is exactly £12,570. Pay does not land on a round threshold by accident; a noticeable group of people are paid precisely the amount at which tax and National Insurance begin.

What was left out on purpose

How much each band pays in pounds per head; National Insurance, which has its own thresholds and is not paid at all above state pension age; and any judgement about who "contributes". A person who pays no income tax may be a carer, a student, a low-paid worker or a child, and may pay a good deal of other tax. This tile counts; it does not rank.

🇬🇧 How to read the Brexit figures ✕

Nobody has measured what Brexit cost. Everybody has estimated it

There is only one Britain, and it left. Every figure on the tile that says "Brexit cost X per cent of GDP" is a comparison between the Britain we have and a Britain that stayed in — and that second Britain has to be invented. The main way of inventing it is to build a "doppelgänger" out of other countries whose economies tracked Britain's closely before 2016 and see where the blend went afterwards. The Centre for European Reform did that in 2022 and got 5.5%. A team using the Bank of England's own survey of company finance directors got 6–8% by the end of 2025. The Office for Budget Responsibility does something different: it assumed in 2016 that trade would end up 15% lower and productivity 4% lower in the long run, and ten years on it has not changed either assumption. The tile prints the range, names each author, and prints the sceptics too. It does not pick one. What the reader should notice is that the serious estimates all have the same sign.

Why "the economy grew about as fast as the EU's" and "Britain fell behind" are both true

Between 2016 and 2024 the UK economy grew 10.2% and the EU's 12.9% — close. But the UK's output per person grew 4.4% against the EU's 11.5%, because Britain's population grew much faster than most of Europe's, mostly through non-EU migration. A country can add people and output at the same time and leave each person no better off. That per-head figure is the one that decides whether living standards rose, and it is the sharpest number on the tile. It is also why the Germans, whose per-head figure is worse than ours, are not a comfort: their problem is a different one, and a sceptic is right to say so.

What "universal" means, and what it does not

Two enormous things happened to every rich country in these ten years: the pandemic in 2020 and the energy shock in 2022. Every economy in the chart fell in 2020 and every European one had double-digit inflation in 2022. So a fall in British output in 2020, or a bad year for prices in 2022, proves nothing about Brexit by itself. The way to tell Brexit apart is to look for things that happened to Britain and not to its neighbours in the same storm: goods exports that fell while services exports rose, food prices that rose faster only for the foods that face EU border checks, sixteen thousand small firms that stopped selling to Europe, a migration boom made of care workers and students rather than European builders. The tile lists those separately from the global shocks, and that separation is the whole answer to "was Britain alone".

The £350 million a week, and the real number

The figure on the bus was the gross contribution before the rebate that was never sent and before the money that came back to British farms, regions and universities. The UK Statistics Authority called it "a clear misuse of official statistics" in 2017. The Treasury's own table puts the net contribution at £9.6 billion in 2016, about £185 million a week. That money is genuinely no longer sent — but the farms, regions and universities still had to be paid for from Westminster, and the tile prints what replaced each.

The migration figures changed underneath everyone

The ONS stopped counting migrants with a survey of passengers at ports in 2020 and now uses tax, visa and benefit records instead. The 2016 figures are on the old basis and the 2025 figures on the new one; they are not strictly comparable, and the new series keeps being revised — net migration for 2024 was first published as 431,000, then 345,000, then 331,000. The tile uses the ONS's current figure in every case and says which basis it is on. Treat a change of 50,000 either way as noise.

Fish: more money landed, in pounds that are worth less

UK boats landed £936 million of fish in 2016 and £1.16 billion in 2024. Prices in general rose about 30% over the same period, so in real terms the catch is worth about the same as it was — the record is a cash record. Meanwhile the fleet is 15% smaller and there are fewer fishermen. The quota the UK won on paper is real; the industry has not had the boats or the markets to use all of it, and in May 2025 the government traded the leverage it was about to gain — annual talks on EU access from July 2026 — for twelve more years of access, in return for a food-checks deal that is still not signed.

The European Court of Human Rights is not an EU court

The most common confusion in the whole subject. The Strasbourg court belongs to the Council of Europe, a separate body with 46 members that Britain helped found in 1949. Brexit ended the jurisdiction of the EU's court in Luxembourg over Great Britain (it continues for Northern Ireland under the Windsor Framework). It did nothing at all to the Human Rights court, and the UK–EU trade deal in fact commits Britain to staying in it.

What was left out on purpose

Any figure we could not trace to a primary source: the "£47 billion" and similar headline totals that circulate; a 2016-versus-2026 count of Britons living in the EU (no reliable series exists — Germany is missing from the UN data); the numbers of Britons taking EU citizenship in 2024 and 2025 (Eurostat has not broken them out); and the outcome of the second UK–EU summit in July 2026 beyond what the two sides' joint statement says. Where a claim is disputed the tile prints both sides and says who said what.

🛡️ How to read the defence figures ✕

Three measures of "defence spending"

The Ministry of Defence's budget (£62.0bn in 2025-26) is what Parliament votes the MoD. NATO's measure is wider — it adds things such as military pensions and, from 2027, the intelligence agencies — which is why the percentages ministers quote are higher than the MoD budget divided by GDP. The chart since 1960 uses SIPRI's measure, a third definition, so that every year is counted the same way.

Four headcounts

"All personnel" (182,980) includes reserves, trainees and others; "full-time trained" (127,060) is who could deploy today; "regulars" (137,480) is the number usually compared with 1990. Each line says which it uses.

What a weapon costs

The MoD publishes programme and contract totals, not prices, each in the money of its own year. "About £X each" is our division of the total by the number bought; some totals include development and support and some do not, so it is a guide to scale, not a price tag. The MoD does not publish missile or shell prices, so none is printed.

"If they do this, we do this"

The descriptions of Russia, China, Iran and North Korea are the UK government's own, quoted from its strategy documents and from MI5 — how Britain sees these states, which each disputes. "The game" under each is our reading of the logic, using the ideas in the box beneath it (deterrence by punishment and by denial, the grey zone, credible commitment), and is labelled as ours.

The threat level

Set by the Joint Terrorism Analysis Centre, independently of ministers, on a five-step scale. It says how likely an attack is, not what to do; the level printed is the one in force when the tile was reviewed — check MI5's page for today's.

Terrorist groups

The eleven listed are our selection from the 84 proscribed organisations, for their relevance to Britain. Each aim is the Home Office's description, not the group's. The IRGC is not on the terrorism list; it was designated in July 2026 under a separate law for state-backed groups.

Per income tax payer

As on every UK tile, money is divided by HMRC's count of income tax payers rather than by population, because they are the people paying for it. It is an illustration, not a bill: defence is paid for from all taxes and from borrowing.

What is not here, and why

Troop numbers for most individual overseas bases (published only in a spreadsheet we have not yet read); what each base costs (not published); a year-by-year path to 3.5% of GDP (none exists yet); the National Audit Office's verdict on the Defence Investment Plan (due this autumn); missile and shell prices (withheld). The tile will be updated as they appear.

🐒 How to read the Gibraltar figures ✕

Why the ledger looks lopsided, and why that is honest

Britain sells Gibraltar £4.1 billion of goods and services a year and buys £1.4 billion back — a surplus of £2.7 billion with a place of under forty thousand people. Against that, the only spending the Ministry of Defence has ever put a figure on is a £155 million facilities contract over seven years and a naval squadron costing about £1.5 million a year. The bars are on one scale so you can see the shapes, but they are different kinds of money: the surplus is private firms' revenue, mostly financial and insurance services, of which only the tax reaches the Exchequer; the contracts are tax spent. They are not added up, and the tile does not claim Gibraltar "makes" Britain £2.7 billion.

The garrison's cost is not published, and the tile says so

About 1,100 British personnel are on the Rock across the Navy, the RAF and the Royal Gibraltar Regiment, and the naval base is the only one Britain has at the mouth of the Mediterranean. No government has told Parliament what the whole presence costs in a year; the answers on record cover slivers — the two patrol launches, the estate contract. Rather than invent a total, the tile prints the slivers and marks the gap. The same thing is true of the Falklands, and it is the taxpayer's real grievance: not that the bill is large, but that it is unknowable.

The £500 million is a guarantee, not a gift

Gibraltar takes no grant from Britain — only Pitcairn, Montserrat and St Helena do. What Britain did in 2020 was guarantee up to £500 million of Gibraltar's borrowing during the pandemic, so it could borrow at Britain's credit rating; £425 million had been drawn by the time the guarantee was extended in 2023. That is a contingent liability: it costs nothing unless Gibraltar cannot repay, and Gibraltar is running a budget surplus. Whether the drawn sum has since been repaid is not in any statement we could reach.

What the treaty does and does not do

It removes every check at the land frontier with Spain and moves the EU's border to Gibraltar's port and airport, where Spanish officers carry out Schengen entry checks on the EU's behalf — the model used by French police at St Pancras. Gibraltar joins a customs union with the EU, brings in a transaction tax in place of VAT, and aligns on tobacco and state aid. It does not create free movement — Gibraltar's own government insists on the word "fluidity" — and Article 2 says the treaty is no basis for any claim to or denial of sovereignty. The military base is outside its scope entirely. Both the Gibraltar Government and the Opposition backed it; the Opposition did so with "profound misgivings" and won the right to a referendum before Gibraltar could ever leave it.

Signed is not ratified

The treaty was signed on 14 July 2026 and applied provisionally from the next day, which is why the fence is gone. But Parliament's 21-sitting-day objection period runs to 22 October 2026 and the European Parliament's consent vote is pencilled in for December. A treaty in provisional application can still be stopped; this one is being lived under before it is law.

Three referendums, one direction

In 1967 Gibraltarians voted 99% to stay British, and Spain closed the frontier for sixteen years. In 2002 they rejected joint sovereignty by 99% on an 88% turnout. In 2016 they voted 96% to remain in the EU on an 84% turnout — the first UK-wide vote Gibraltar had ever been allowed to take part in, because it has no MP. The 2026 treaty is the attempt to honour the second and third of those votes at once.

Why the population figure is a range

Gibraltar's 2022 census counts residents; its budget speech quotes 38,200 for 2025-26 on a broader basis. The tile uses "under forty thousand" where it can and avoids dividing by a number two publishers disagree on. The frontier-worker count, by contrast, is exact: 15,509 in December 2025, over half the jobs on the Rock, seven in ten of them Spanish.

What was left out on purpose

A total annual cost of British Forces Gibraltar (none published); the claim that Gibraltar insurers write "over 20% of UK motor insurance" (Gibraltar Finance's own figure, undated and uncorroborated by the regulators); the share of UK gambling duty paid by Gibraltar-licensed firms (not published — the duty is paid where the customer is, so it flows to HM Treasury whatever the licence says); raw vote counts for the three referendums (the Gibraltar Government publishes percentages); a count of daily frontier crossings; and a figure for the Campo economy newer than the House of Lords' 2017 evidence.

🏝 How to read the Falklands figures ✕

Why the two sides of the ledger are not added up

The left-hand bars are money the Exchequer spends — your taxes, paid to the Ministry of Defence. The right-hand bars are sales by British firms to the islands, and what Britain buys from them. A pound of sales is not a pound of tax back: the firm keeps most of it, and only the tax on its profit and its staff's wages reaches the Treasury. Worse, the two overlap — a large share of what the MoD spends on the garrison is spent with British contractors, so it appears on both sides at once. Netting them would produce a number that looked like an answer and was not one.

Why the cost figure is from 2015

The last time a minister gave Parliament an annual running cost for the garrison was July 2015: "around £85 million". In March 2026 the MoD declined even to say how many people are stationed there. So the figure on this tile is the newest official one, it is eleven years old, and the tile says so rather than inflating it or guessing. A seven-year support contract now being tendered — about £70 million a year for maintenance, catering and engineering alone — suggests the true running cost today is higher.

"Self-sufficient except for defence" — what that means

The islands' government raises its own taxes, sells its own fishing licences, and pays for its own schools, hospital, roads and even its local defence force. It receives no grant from the UK Treasury. Two things it does not pay for: the British garrison, and the diplomacy that keeps the sovereignty claim at bay. There is also one uncosted UK item: islanders who cannot be treated locally are referred to the NHS under a reciprocal agreement with no cap on numbers.

An income of £86,000 a head is not what islanders earn

GDP per head is that high because a very small population sits on a very large squid fishery. Most of the catch is landed by foreign-owned vessels and most of the profit leaves. Gross national income per head — what actually accrues to residents — is about £59,000, and a typical wage is lower again. It is a rich economy, not a population of rich people.

Who owns the oil

The Falklands government does. Its royalty (9%) and corporation tax (26%) on the Sea Lion field go to Stanley, not to Westminster; the Foreign Office has said in terms that the natural resources of every Overseas Territory belong to the territory. Britain's share is indirect — orders for British suppliers, and tax on any UK-listed company's profits — and an operator's forecast is not a receipt: first oil is expected in 2028, and no barrel has yet been sold.

Overseas Territory, not Commonwealth

The Commonwealth is a club of 56 independent countries. The Falklands are not a country and not a member: they are one of fourteen British Overseas Territories — places under British sovereignty that are not part of the United Kingdom itself. They govern their own domestic affairs; the UK keeps defence and foreign policy and appoints a Governor. That is why this tile sits under "Britain overseas" rather than under a Commonwealth heading.

What we left out on purpose

Whether the islands are "worth it" is a judgement, and the site does not make it. What being British means to someone living 8,000 miles from Britain is a feeling, and we cannot source a feeling. Everything on the tile is a published figure, a published fact or clearly labelled arithmetic on them; the weighing is yours.

🌴 How to read the Chagos figures ✕

So have we given the islands away or not?

No. A treaty was signed on 22 May 2025, and signing is not the end of the process — a treaty that needs a change in British law does not take effect until Parliament passes that law. The Bill got through the Commons, the Lords amended it in January 2026, and the Commons overturned those amendments on 20 January, dividing three times. Then it needed to go back to the Lords one last time, and that stage was listed as "date to be announced" and never scheduled. On 13 April 2026 ministers said the Bill could not finish, because the United States would not agree the updated 1966 exchange of notes that ratification depends on. It was not in the King's Speech in May 2026. No sovereignty has moved, no money has changed hands, and the base still operates under the arrangements Britain and America made in 1966, which run to 30 December 2036.

Is it dead, then?

The government will not say. On 30 July 2026 three separate written questions — including one asking simply whether it remains policy to transfer the territory — all got the same sentence back: that ministers "will continue to work with the US and Mauritius to find a way forward". On 1 September 2026 the Commons was told the United States "must be content with the agreement before we proceed in ratifying the treaty". So: signed, unratified, unfunded, no bill before Parliament, and not withdrawn. Treat this section of the tile as the part most likely to be out of date.

Why the same deal cost £3.4 billion and £35 billion

Because those are two ways of adding up the same instalments, and both were honest. The £3.4bn is a net present value: future payments discounted back to today's money at the Treasury's standard Green Book rates, on the reasoning that a pound payable in 2124 is not a pound now. The £34.7bn is what the cheques would total in cash. Between them, about £10bn is the figure if you adjust for inflation but do not discount. The head of the UK Statistics Authority did not say the government's number was wrong; he said it should not have been called "the net overall cost" without saying it had been discounted, and he noted the Office for Budget Responsibility thought the discount rate reasonable for valuing a lease. Higher totals have been quoted using different inflation and discount assumptions; the tile draws the three whose method is published, because a total you cannot check the workings of is not a fact.

An advisory opinion is not a court order — so why did it matter?

It is the heart of the disagreement. The International Court of Justice's 2019 opinion was advisory: it bound nobody, and critics say the government treated advice as a verdict. The government's answer is that the 2021 tribunal case showed where advice leads. There, a Law of the Sea chamber deciding a boundary between Mauritius and the Maldives — a case Britain was not party to — treated the International Court's findings as settling the legal status of the archipelago, and did it without Britain in the room. Ministers concluded a binding ruling was a matter of time, and that meanwhile the base's ability to hold radio spectrum, postal codes and fishing rights was being picked at through UN agencies. The strongest reply is that since 1998 Britain has formally excluded disputes about military activities from the compulsory jurisdiction of those tribunals — though that is a shield the tribunal itself decides whether to accept, and it does nothing about the International Court or the General Assembly. Whether all this was prudence or surrender is exactly what the argument is about, and the tile gives both sides their own words.

What "British Indian Ocean Territory" is

It is one of the fourteen British Overseas Territories — British sovereign land that is not part of the United Kingdom. Unlike the Falklands or Gibraltar it has no settled civilian population, because the people who lived there were removed between 1968 and 1973. That absence works both ways in the argument: the government says there is no population whose self-determination could be at stake, so the Falklands and Gibraltar are not a precedent; Chagossians say the absence is precisely the wrong being argued over, and that they were never asked. In February 2026 four Chagossian men landed on Île du Coin and refused to leave, and in March a court there struck down the rule that had kept them out — an order now suspended while the government appeals.

Why the tile does not say what "the Chagossians want"

Because they do not agree, and pretending otherwise would be the site putting words in people's mouths. The Chagos Refugees Group in Mauritius welcomed the signing; Chagossian Voices opposed it and wants a right of return that includes Diego Garcia itself, which no version of the deal offered. Both are quoted. What is not in dispute is that no Chagossian organisation was a party to the negotiation, that UN human rights experts said so publicly in June 2025, and that most of those who answered a Lords committee survey lacked confidence in the proposed trust fund.

What we left out on purpose, and what we took out

Whether the deal was right. That is a judgement about sovereignty, international law and an ally's goodwill, and it is not ours to make on your behalf. We also took several things out after a second check: two higher cost totals whose workings we could not obtain, a claim about how many supply ships sit in the lagoon, a comparison with what France pays for a base elsewhere, and a count of Gulf war bombing missions — all of them plausible, none of them traceable to a source we could open. Where a figure is missing from this tile it is usually because of that, and not because it was inconvenient.

🐧 How to read the Antarctic figures ✕

Doing science does not buy Britain its Antarctic vote — and we said it did

This is the correction that matters most, and the first draft of this tile got it wrong along with almost everyone else. The Antarctic Treaty has fifty-eight member countries and only twenty-nine of them decide anything. You will read everywhere that a country earns that status by "conducting substantial scientific research activity there". Those words are real, but they sit in Article IX(2), and Article IX(2) applies only to countries that joined the treaty later, by acceding to it. Britain is one of the twelve original signatories named in the treaty's preamble; its seat comes from Article IX(1), which attaches no condition whatever. So twelve countries hold a permanent vote and seventeen hold a conditional one. Britain would not lose its vote if it closed every station tomorrow. What it would lose is the standing to use it, the data to argue with, and any independent way of knowing what anyone else is doing down there — which is a real argument for the spending, just not the one usually made.

The mining ban does not expire in 2048

The other thing everyone gets wrong. The 1991 environmental protocol says any activity relating to mineral resources other than scientific research "shall be prohibited", and that sentence has no end date. What 2048 actually is: fifty years after the protocol came into force, at which point any country with a vote gains the right to ask for a review conference. Even if one were called, changing the ban would have to be adopted by a majority of the parties including three-quarters of the countries that had votes in 1991, then ratified by three-quarters of the voting countries including every single one of those 1991 countries, and it could not take effect at all unless a binding legal regime for mining were already in force. The one genuine pressure valve is different and less discussed: if such a change is adopted and has not come into force within three years, any country may then give two years' notice and walk out of the protocol altogether.

Why we print three different budgets

Because three are published and none of them agrees. A minister told Parliament in March 2025 that the allocation for 2024/25 was £116m. A Foreign Office minister had told a Commons committee in May 2024 that it was £35m of science plus "around another £60 million" of presence — and then said in the same breath "that is £95 million for presence alone", which cannot be right either way. The British Antarctic Survey's own website still says "about £50 million". They are almost certainly counting different things, but no source reconciles them, so we lead on the newest one given to Parliament and show you the disagreement rather than choosing quietly.

Why do so many countries want to be there at all?

Four reasons, and only one of them is science. Sovereignty: seven countries claim slices of Antarctica and three of those claims overlap on the Peninsula, where Britain, Argentina and Chile all claim the same ground; a tenth of the continent is claimed by nobody. The treaty freezes all of it, so nothing done now creates a right — but nothing has been given up either, and a station on your sector is what you would want to have if the freeze ever ended. A vote: twenty-nine countries decide what happens to a continent, and seventeen of them bought their seat by building a research station. Fish: krill and toothfish are the only things anyone is legally taking out, and the krill fleet is Chinese, Norwegian, Korean, Chilean and Ukrainian. Minerals: which nobody is legally taking out, though Russia keeps a survey ship mapping oil and gas prospects and calls the work science. The Foreign Office's own strategy, in December 2025, put the whole thing in one line: "The Antarctic region is subject to strategic competition, with some states seeking to assert influence by expanding their regional presence."

Whose territory is it?

Britain claims the British Antarctic Territory, and so, over parts of the same ground, do Argentina and Chile. The Antarctic Treaty does not settle who is right: it freezes the question, so that nothing anyone does down there counts for or against a claim, and no new claim can be made while the treaty holds. It is a British Overseas Territory in British law and an unresolved argument in international law, and both of those are true at once. Like the Falklands, it is not part of the Commonwealth, which is a club of independent countries.

Why the sea-ice chart has a hole in it

Because there is no published figure to put there. Antarctic sea ice hit its record low in 2023 and stayed at or near record lows for four years. Then, in October 2025, the American ice centre that publishes the analysis the whole world quotes suspended it because its funding was not renewed, and it only resumed at the beginning of this month. So no 2026 Antarctic minimum has been published, by them or anyone else, and rather than estimate one we have drawn the gap. What the centre did say, on 3 September 2026, is that Antarctic ice has been below the long-run average nearly every month since late 2016 but above the record lows of 2023 and 2024. Land ice — the ice that actually raises sea level — is measured differently and tells a clearer and grimmer story.

What we left out, and what we took out

We could find no named politician, think tank or commentator attacking this spending on the record in the last two years, so rather than invent a critic every point in the "case against" column comes from a Commons committee's own findings or from figures the British Antarctic Survey publishes about itself. Several numbers a reader might expect simply are not published: what the ship costs to run, what the Foreign Office spends on the territory, what South Georgia earns from fishing licences as a separate line, and how many people are in the territory at any one time. A second check also removed things the first draft had: an end date for the building programme, a funding split for the Thwaites glacier programme, and a tonnage for the ship's share of the carbon footprint — all plausible, none of them traceable to a source we could open.

🧳 How asylum is funded — the figures ✕

What a person seeking asylum receives

£49.18 a week (£7/day) in self-catered accommodation, or £8.86 a week (£1.26/day) where meals are provided — set by the Home Office, reviewed annually. Working is banned until a claim has waited 12 months, and then only in shortage occupations.

Hotels

Hotels are overflow, used because the decision backlog outgrew normal dispersal housing — block-booked, often budget or repurposed, frequently shared rooms; the star rating on an old sign says nothing about what's inside. They cost over £8m a day (~£3bn/yr), roughly six times dispersal housing — which is why successive governments have committed to ending their use.

Any belongings, such as mobile phones

Phones aren't government-issued. They're usually owned before the journey or cheap/charity handsets — and a smartphone is the one indispensable tool of a crossing (maps, translation, family contact), often the only possession that survives it. Smartphone ownership reflects the low cost of connectivity rather than personal wealth.

Who pays

The Home Office budget — i.e. general taxation — with first-year costs partly counted against the overseas aid budget under international accounting rules. The costs are genuinely large and worth debating: the National Audit Office has criticised multi-billion accommodation contracts. The policy argument is about the fix — faster decisions and cheaper dispersal versus deterrence — not about whether the bill exists.

Why people come — the documented drivers

The top small-boat nationalities map onto documented crises: Eritrea's indefinite military conscription, Taliban-ruled Afghanistan, Sudan's civil war and famine conditions, persecution in Iran, conflict and drought in Somalia. Around half of decided claims from these countries are granted — these are, in the law's own terms, largely refugee-producing states. Nationality is what's recorded; ethnicity isn't collected.

Eastern Europeans on the boats — a picture that has dated

True in 2022: Albanian nationals were the single largest group that year — about 12,700 people, 27% of all arrivals. Since the UK–Albania returns agreement (December 2022), Albanian crossings have fallen by over 99% — 25 people in the first quarter of 2025 — and Albania accounts for most completed returns. No European nationality features in today's top arrivals; the 2022 pattern no longer describes current crossings.

Sources: ONS long-term migration estimates, Home Office asylum & support statistics, Commons Library briefings, Full Fact. Figures are sourced in both directions. Refreshed quarterly.

🩺 How we work out the state of the economy ✕

"In recession" is a definition, not our opinion

A recession is two quarters in a row in which the economy shrinks — that is the rule everyone uses, and it is the rule we apply, to the ONS's own quarterly GDP figures. Run over the whole series back to 1955 it finds nine episodes, and they are the nine you would name: the mid fifties, 1961, 1973–74, 1975, 1980–81, 1990–91, 2008–09, 2020, and the short dip in late 2023. That agreement is the reason we are willing to use the word flatly rather than hedge it.

Why a count of measures, and not a score out of 100

A single "economic health score" would be easier to read and would be a number we invented: no statistics body stands behind it, and you could not check it. So instead each of the twelve measures is compared with itself a year earlier and marked better, worse or much the same, and the headline is simply how many fall in each pile. Every verdict shows the figure behind it, so if you disagree with one you can see exactly which number produced it.

"Much the same" is doing real work

Each measure has a band inside which a movement is treated as noise rather than a direction — a tenth of a point for a survey rate, one per cent for a monthly index. Without it a dial would flip between better and worse on a rounding difference, and the count would mean nothing.

The economy growing and jobs getting worse are not a contradiction

They happen together often, and both halves of this tile can be true at once. Output is measured quarterly and revised; the labour market turns first and is felt first. Watch particularly for the share of working-age people in a job falling while the number of people in work rises — that means the population grew faster than the jobs did, and it is why we lead on the rate.

Output per person is the one to watch

Total GDP can rise simply because there are more people. Output per person divides it by everyone in the country, and it is the line that tracks whether growth is reaching anybody. It is also the honest way to measure a recovery: after 2008 total output regained its old level years before output per person did.

How long recessions last is the part nobody quotes

The table under the chart shows, for every recession, how deep it went, how far unemployment rose afterwards, and how many quarters it took for output per person to get back to where it had been. That last column is the one worth reading: after 2008 it took the best part of eight years.

Two well-known indicators are missing, on purpose

The purchasing managers' index (S&P Global) and consumer confidence (GfK) would both belong here, and both are commercial data we are not entitled to republish. We would rather name that than quietly leave a gap: you can read them at their publishers.

What this is not

It is not a forecast, and there is no prediction anywhere in it — every rung is chosen from figures that have already been published, and each one is dated on the tile. It also does not say why anything moved. General information, not financial advice.

Sources: ONS quarterly national accounts, monthly GDP, Labour Force Survey, vacancy survey, claimant count, consumer price inflation, average weekly earnings, index of production, retail sales and business investment — Open Government Licence v3.0. The recession dating, the direction of travel and the count of measures are StreetOwl's own, by a fixed rule over those figures.

💡 How to read the industry figures ✕

It is the industry of someone's last job

Not the industry they are looking to join, and not where they work now. Somebody made redundant from a shop who is now applying for warehouse work still counts under shops & motor trade until they start the new job. It also means anyone who has never worked, or has not worked for years, appears in no industry at all — which is why these fourteen rates sit below the national unemployment rate and do not average out to it.

Read the rate down the column, the count across the row

The rate is the one that compares fairly between industries, because it already accounts for how big each one is: a large industry always has a large count. The count is what tells you whether a change is a lot of people or a handful. A percentage change on a small industry can be a few thousand people.

A falling count is not automatically good news

An industry that is shrinking has fewer people left in it to lose their jobs, so its count of unemployed can fall while the work disappears. That is why the rate for ten years ago is printed beside today's — if the count fell and the rate did not, the industry got smaller rather than safer.

Why there is no month-by-month figure

These come from the Labour Force Survey, a sample survey, and since 2018 each period is a rolling three months. Two periods next to each other share two months of the same respondents, so a "change since last month" is mostly the same people counted twice. We show the year and the ten years, which is also what ONS advises: read the long movements, not the short ones.

Ten years is as far back as we will go

ONS recoded industry from the 1992 classification to the 2007 one at the start of 2009 and says the effect "is significant for some of the industry sectors", with no adjustment made to the series. A twenty-year comparison would cross that and would not be comparing like with like. There is also a reweighting from January 2019 that puts a step in the ten-year figures — real, published, and worth knowing about before reading too much into any one of them.

What "small sample" means

ONS shades the estimates it considers based on a small sample; we carry that flag straight through rather than presenting every figure at the same confidence. Treat a flagged row as a direction, not a measurement.

And what it cannot tell you

Nothing about pay, nothing about which jobs are being created, and nothing about degree subjects — graduate outcomes by course are a different dataset entirely (the Department for Education's LEO figures), on a different definition, and cannot be put on the same scale as these.

ONS UNEM03, unemployment by previous industrial sector, Labour Force Survey. Published monthly under the Open Government Licence v3.0.

💼 What these numbers mean for your job and pay ✕

Two different unemployment numbers — read the right one

The unemployment rate comes from a survey (anyone jobless and actively looking, claiming or not) — the economist's number, national and slow. The claimant count is people claiming unemployment benefits — barely any of it is Jobseeker's Allowance these days; it's almost all Universal Credit with a requirement to seek work, plus a little New Style JSA. It's published monthly right down to ward level, which makes it the timely, local signal — that's the one this tile tracks for your area.

Real pay — pay measured against prices

Cash pay rises average 3.4%; with inflation at 2.8%, the average pay rise is worth +0.3% in real terms after prices (ONS measure). In any review or job offer, the comparison point is inflation, not zero: an offer below 2.8% is a reduction in real terms.

What a soft market changes in practice

Vacancies at their lowest since 2021 plus a rising claimant count means more applicants per job: moving jobs gets harder and riskier, employers regain negotiating power, and the premium for switching shrinks. It's the classic moment to consolidate — build the case for progression where you're already valued — and if you are job-hunting, expect longer searches and lean on the support that comes with the claim (work coaches, training budgets) rather than going it alone.

Why we show your ward, not just the nation

National rates hide huge local variation — neighbouring wards in the same borough can differ sharply, and your ward is the job market you actually live in. Ward-level claimant data updates monthly via Nomis, one of the very few economic numbers that exists at that scale.

So what does this mean for me, here?

Read the tile's local claimant figure against the national rate shown beside it. Lower than national generally means a tighter local labour market — fewer people chasing each vacancy, a little more negotiating room for you. Higher means more competition per job and typically longer searches. The single most useful thing is the direction over a few months, not any one month's number: a steady local rise is an early warning the area's employers are shedding, often visible before anything makes the news.

Which industries? This tile can't tell you — here's what can

The claimant count doesn't say which industries people left, and we won't pretend otherwise. For that, the ONS publishes employment by industry for your local authority (Business Register and Employment Survey, on Nomis's local labour-market profiles) — it shows which sectors dominate where you live, which is usually more telling than any national headline: a national retail downturn matters far more in a town where retail is the biggest employer.

What to look out for

Three signals worth a monthly glance: your local claimant trend (this tile), the national vacancies trend (fewer vacancies per jobseeker = cooling market), and pay against inflation — if the CPI figure on the Money tile outruns typical pay rises, "real" pay is falling even when payslips grow. None of these is advice; they're the same three dials economists watch, pointed at your postcode.

Sources: ONS Labour Force Survey & claimant count (live series), Nomis for local. General information, not financial or careers advice.

📅 Add your event ✕

In the full app this needs a signed-in account, and events are checked before other people see them. Your event stays in this browser — personal events are not synced to your account yet.

⚖️ How to read Citizens v the state ✕

What judicial review is

Not an appeal. A judicial review asks whether a public body was entitled to do what it did — whether it had the power, followed the right procedure, took account of what the law told it to, or reached a decision no reasonable body could reach. The court can quash the decision and send it back; it does not usually substitute its own. That is why a claimant can win and still get the same answer second time round, lawfully made.

Why so many case names begin with "R"

Because a judicial review is formally brought by the Crown on the citizen's behalf: R (on the application of Smith) v Secretary of State. The same thing is often written the other way about — Smith, R (on the application of) v Secretary of State — which is the form most of these judgments actually use. Both mean a citizen challenging the state. R v Smith, with no "on the application of", is something else entirely: that is the Crown prosecuting.

The two labels are ours, and they are read from the name only

"Judicial review" comes from that naming convention in the case title. "Public body" names the party after the "v" when it matches a list of public bodies we keep — a department, a council, a police force, a regulator, a statutory scheme. A case with no label is not necessarily private: it means we could not tell from the name, and we would rather say nothing than guess. Both labels describe who was in court, never who won.

What this tile cannot tell you, and why

It cannot tell you which rulings matter, what a case decided, or what it changes — because working any of that out means reading the judgment, and The National Archives' Open Justice Licence does not permit us to read inside judgments programmatically ("computational analysis", which the service describes as searching in bulk to identify, extract or enrich contents). So we publish what the service's own feed publishes and send you to the ruling itself. Nothing on this tile is a summary of a judgment, and nothing on it is legal advice.

What the date means

The day the court first published the document — not the day of the hearing, and not the day the ruling takes effect. A judgment handed down in July can appear in September.

Where the list does not reach

Find Case Law covers England and Wales, plus the UK Supreme Court and the Privy Council, which are UK-wide. In its own words it "does not contain judgments or decisions from Scottish, Northern Irish or Irish courts or tribunals". Within England and Wales we ask six courts only — the Supreme Court, the Court of Appeal (Civil Division), the Administrative Court, the King's Bench Division and two Upper Tribunal chambers — so criminal appeals, the family courts and the First-tier tribunals are not here either. A court that fails to answer is shown greyed out rather than dropped, because a missing court and a quiet court are different things.

Where to go next

The Supreme Court publishes a plain-English press summary and a video of the hand-down for its own judgments, and the judiciary publishes press summaries for High Court and Court of Appeal cases it considers to be of public interest. Both are linked from the judgment page this tile sends you to.

Source

Find Case Law, The National Archives — the service's own public feed, six courts, re-read every half hour. Contains information licensed under the Open Justice - Licence v2.0.

⚡ Electricity is not the same as energy ✕

What this tile is, and is not

These percentages are Great Britain's electricity generation in the current half-hour. Electricity is roughly a fifth of the energy the country actually uses — the rest is largely gas for heating and oil for transport. So a day when wind supplies half the grid is not a day when the country runs half on wind, and this tile would be misleading if it let you think otherwise.

Why "imported" has its own line

Interconnector cables link Britain to France, Norway, Belgium, the Netherlands, Denmark and Ireland. Power flowing in through them was not generated here, so folding it into a total would overstate what Britain produces. It moves with the weather and with prices abroad, and it can flow the other way when we have a surplus.

Coal at zero

Not a rounding artefact: Britain's last coal-fired power station closed in September 2024, ending 142 years of coal-fired electricity. A zero here is now the normal reading.

Why the two halves of this tile disagree

The live figures at the top are electricity, where low-carbon sources routinely supply more than half. The annual figures below are all energy, where fossil fuels still supply about three quarters. Both are true: Britain has decarbonised its electricity much faster than its heating and its transport, and the distance between the two numbers is the size of the job left.

Where the annual figures come from

DUKES — the Digest of UK Energy Statistics — published by DESNZ each July for the previous calendar year. The 2026 edition, covering 2025, was published on 30 July 2026. It is an annual publication, so these numbers will not move again until next summer, and the tile is dated accordingly.

The carbon intensity figure

Grams of CO₂ per kilowatt-hour for the current period, with NESO's own banding from very low to very high. It falls when wind and solar are strong and rises when gas is filling the gap, which is why it can change markedly within a day. Where the actual figure has not been settled yet, NESO's forecast is shown and the tile says so.

Live data: National Energy System Operator (NESO) carbon intensity API, Great Britain only — Northern Ireland is a separate grid, half-hourly, no adjustments by us. Annual data: DESNZ, Digest of UK Energy Statistics (DUKES) 2026, covering 2025, Open Government Licence.

🛢 How the fuel scorecard works ✕

"Shortage" means three different things, and the tile keeps them apart

A price shortage — fuel available, but at prices that force people to use less — is happening now, worldwide: diesel refining margins are running at eight to ten times their normal level. An allocation shortage — governments rationing by rule — is happening in some countries, not Britain. A physical shortage — homes or firms involuntarily cut off — has not happened anywhere, and for the UK the credible risk window is a cold snap between December 2026 and February 2027, on the gas side. Most arguments about "will there be a shortage" are two people talking about different rungs of this ladder.

Computed, not assigned

Each fuel is scored on three published indicators (import dependency, stock cover, and concentration of supply routes), plus one for who actually controls the emergency stock. Every indicator has fixed green/amber/red thresholds, set out once and drawn to scale as the coloured track behind each figure — the dark tick is where the UK sits. Green scores 0 points, amber 1, red 2; the fuel chips add them up, so lower is better. The thresholds are our editorial choice (published, fixed, dated); the values are not — each carries its source and its date.

Old figures say so

Most of these numbers have no live feed — they come from annual statistical digests and one-off official statements, and they age. Each figure shows its date; ones older than the indicator's shelf life are marked stale rather than being quietly passed off as current. The one live figure is gas storage, from GIE's AGSI+ transparency platform (the same feed the industry uses); the "days of winter demand" conversion is our own arithmetic and is labelled as ours wherever it appears.

Why Britain scores "Stressed" while the lights stay on

The UK's crude position is genuinely comfortable — half home-supplied, no chokepoint exposure. The stress is in what refines it and what stores it: diesel self-sufficiency around 45%, storage capacity for roughly 12 days against Germany's 89 and France's 103 — and far less than that actually in store, because Rough holds nothing — and no government-owned emergency stock at all — the obligation sits entirely on private company balance sheets. None of that means shortage tomorrow; it means less buffer than most neighbours if something else goes wrong.

Scored by StreetOwl's own rules engine from DESNZ DUKES 2026, IEA, Eurostat, GIE AGSI+, S&P Global and national sources — every figure dated and attributed on the tile. The same engine scores eight world regions on the World tab's Energy layer.

✉️ Send feedback ✕

Wrong number, broken widget, missing thing your area needs — all useful. Your current postcode view is included so we know what you were looking at.