Dataset pages: a two-paragraph introduction per dataset with links and citations

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Our World in Data — Total alcohol consumption per capita (WHO)"
+    url: https://ourworldindata.org/grapher/total-alcohol-consumption-per-capita-litres-of-pure-alcohol
+model: opus-5.5
+written: 2026-10-28
+---
+
+Alcohol is the legal drug every cannabis law is measured against. The [World Health Organization](https://ourworldindata.org/grapher/total-alcohol-consumption-per-capita-litres-of-pure-alcohol) estimates how many litres of pure alcohol each person aged 15 or over drinks in a year, sold legally or not.[@1] In 2020 [Romania](/country/ro/), [Georgia](/country/ge/) and [Latvia](/country/lv/) drank the most, 13 to 17 litres; in [Saudi Arabia](/country/sa/), [Somalia](/country/so/) and [Mauritania](/country/mr/) almost none. The median country drinks about 4 litres.
+
+The [alcohol chart](/charts/#alcohol) sets drinking beside each country's cannabis score, and shows where one drug is part of daily life while the other is a crime, or the other way round. Where cannabis is drunk at a festival, the atlas tells of it, as in [bhang at Holi](/culture/traditions/bhang-at-holi/); for the most common cannabis figure, see [cannabis use](/datasets/cannabis-use/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Google Books Ngram Viewer"
+    url: https://books.google.com/ngrams/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Words for cannabis rise and fall with the times. The [Google Books Ngram Viewer](https://books.google.com/ngrams/) counts how often each word appears in the books Google has scanned, as a share of all words printed each year; this site follows the names of the plant and the drug in English, German, French, Spanish, Italian and Russian since 1800.[@1] The curves can be read beside the atlas entries on [marijuana](/culture/language/marijuana/), [Indian hemp](/culture/language/indian-hemp/) and [hashish](/culture/language/hashish/), which writers tried at the [Club des Hashischins](/culture/traditions/club-des-hashischins/) in 1840s Paris.
+
+Books are counted by language, not by country, and the scanned books are not a random sample of what was printed.[@1] "Hemp" also means rope and cloth, so its curve is mostly about the crop. The atlas's [language](/culture/language/) section tells the story behind each word, from [ganja](/culture/language/ganja/) to [420](/culture/language/420/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Statistics Canada — Table 10-10-0165-01, government revenue from the sale of cannabis"
+    url: https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1010016501
+model: opus-5.5
+written: 2026-10-28
+---
+
+A legal market pays the state. [Statistics Canada](https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1010016501) reports what Canada's governments take from the legal cannabis trade each fiscal year: the profit of the provincial cannabis authorities, federal and provincial excise taxes, sales taxes and other revenue.[@1] The series runs from 2021/22 to 2024/25, the years after the [Cannabis Act of 2018](/history/canada-cannabis-act-2018/) and the [legal edibles of 2019](/history/canada-edibles-2019/).
+
+The [chart of the state as merchant](/charts/#state-merchant) breaks the revenue down by source. Other countries that sell cannabis legally appear in each profile's [cannabis industry](/context/production/) section; for how far legal sales have pushed out illegal ones in Canada, see [where people get their cannabis](/datasets/canada-sourcing/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Health Canada — Canadian Cannabis Survey, Health Infobase"
+    url: https://health-infobase.canada.ca/cannabis/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Legalisation is meant to move buyers from illegal sellers to legal shops, and Canada measures whether it has. Each year [Health Canada](https://health-infobase.canada.ca/cannabis/) asks people who used cannabis where they usually got it: a legal store or website, friends and family, an illegal seller, their own plants, or elsewhere.[@1] The 2018 survey came before the first legal stores opened under the [Cannabis Act](/history/canada-cannabis-act-2018/), so it has no legal figure. In 2019 37% bought legally; by 2023 it was 73%, while the share buying from illegal sellers fell from 28% in 2018 to 3%.
+
+The [sourcing chart](/charts/#sourcing) draws the shift year by year. The same questions in other countries would show how much of the market a law has reached; for that, see each profile's [recreational use](/context/recreational/) and [black market](/context/black-market/) sections, and for what the legal trade pays to the state, [government revenue from cannabis in Canada](/datasets/canada-revenue/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "EUDA — European Drug Report statistical bulletin 2026, drug law offences (DLO-03-2-0)"
+    url: https://www.euda.europa.eu/data/stats2026/dlo_en#displayTable:DLO-03-2-0
+model: opus-5.5
+written: 2026-10-28
+---
+
+How often the police record someone for using or carrying cannabis is the plainest measure of how a law is enforced. The [European Union Drugs Agency](https://www.euda.europa.eu/data/stats2026/dlo_en#displayTable:DLO-03-2-0) publishes these offences for cannabis alone, separate from dealing and growing for sale.[@1] Per 100,000 people, [Spain](/country/es/) records the most, about 596 in 2024, followed by [Belgium](/country/be/) and [Austria](/country/at/); [Malta](/country/mt/), which [legalised in 2021](/history/malta-legalisation-2021/), records almost none. In Spain, having or using drugs in public is punished with a fine of at least €601 under its [public safety law of 2015](/history/spain-public-safety-law-2015/), not as a crime.
+
+An offence is not a conviction, and countries count offences or offenders and define them differently, so a country is best compared with itself over time.[@1] The [chart of who is punished](/charts/#punished) follows the series, and each profile's [arrests and prisoners](/context/arrests/) section explains what an offence leads to there. Outside Europe, the nearest figure is UNODC's [police contact for drug possession](/datasets/police-possession/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Wikidata"
+    url: https://www.wikidata.org/
+model: opus-5.5
+written: 2026-10-28
+---
+
+In a few countries, people who want cannabis legal have founded parties of their own. [Wikidata](https://www.wikidata.org/), the open database behind Wikipedia, records parties that campaign on cannabis, with their country and the years they were founded and dissolved.[@1] It lists 22 parties in 9 countries, from [Australia](/country/au/), with Legalise Cannabis Australia and its state parties, and [New Zealand](/country/nz/), with the Aotearoa Legalise Cannabis Party, to [South Africa](/country/za/), with the Dagga Party, and the [United States](/country/us/).
+
+Wikidata is written by volunteers and far from complete: parties whose name does not mention the plant, such as Israel's Ale Yarok, are found only if their ideology is recorded.[@1] The [parties chart](/charts/#parties) shows when they were founded. In South Africa change came instead from the [Constitutional Court in 2018](/history/south-africa-constitutional-court-2018/); the atlas profiles campaigners such as [Keith Stroup](/culture/people/keith-stroup/) and [Marc Emery](/culture/people/marc-emery/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "UNODC — World Drug Report 2026, statistical annex"
+    url: https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html
+model: opus-5.5
+written: 2026-10-28
+---
+
+How many people use cannabis is the first thing most readers ask, and the [UN Office on Drugs and Crime](https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html) collects the answer from governments' national surveys: the share of people who used cannabis in the past year.[@1] This site applies each share to a population of the same ages, which gives the [number of users](/map/users/) on the map and the [chart of use](/charts/#use). All 220 places have a figure, and 93 have more than one survey, so many profiles can show use over time.
+
+The surveys are not made to be compared. They ask different age groups, in different years, and some include children, so a high share may only mean a younger or a more recent survey.[@1] Where a country has no survey, a regional median stands in, and it is marked as an estimate. For how people actually get cannabis in each country, see the [recreational use](/context/recreational/) and [black market](/context/black-market/) sections; for schoolchildren, see [use at 15 and 16](/datasets/youth-use/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Our World in Data — Human rights index (V-Dem)"
+    url: https://ourworldindata.org/grapher/civil-liberties-index-vdem
+model: opus-5.5
+written: 2026-10-28
+---
+
+How free people are from torture, political killings and forced labour, and how far they may move, worship, speak and associate, is estimated every year by the country experts of the [V-Dem Institute](https://ourworldindata.org/grapher/civil-liberties-index-vdem), from 0 to 1; Our World in Data publishes it as its human rights index.[@1] In 2025 [Estonia](/country/ee/), [Sweden](/country/se/) and [Denmark](/country/dk/) score highest, at 0.96, and [North Korea](/country/kp/), [Afghanistan](/country/af/) and [Eritrea](/country/er/) lowest.
+
+The series runs back to 1900, so a profile can show how a country's liberties rose or fell beside the history of its cannabis laws.[@1] A country can respect civil liberties and still treat cannabis as a crime: Sweden is among the highest here, and its [drug strategy](/history/sweden-drug-strategy-renewed-2006/) aims at a drug-free society. Compare the [Human Freedom Index](/datasets/hfi/) and each profile's [beyond the penalty](/context/consequences/) section.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "ClinicalTrials.gov — U.S. National Library of Medicine"
+    url: https://clinicaltrials.gov/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Trials are how cannabis medicines are tested on patients. [ClinicalTrials.gov](https://clinicaltrials.gov/), run by the U.S. National Library of Medicine, registers studies around the world; this site counts those with a cannabis or cannabinoid treatment that have actually started, once for each country with a site.[@1] The [United States](/country/us/) has the most, about 680, then [Canada](/country/ca/) and [France](/country/fr/); 52 countries have at least one.
+
+The register does not cover every trial in the world, and the current year is incomplete.[@1] The [trials chart](/charts/#trials) follows the count since 2000. How patients get cannabis today is in each profile's [medical cannabis](/context/medical/) section; for research beyond patients, see [research on cannabis](/datasets/research-papers/) and the [research](/context/research/) sections.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "CEPII — GeoDist database"
+    url: https://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=6
+model: opus-5.5
+written: 2026-10-28
+---
+
+Who once ruled a country can still be read in its laws: the British Empire wrote the name [Indian hemp](/culture/language/indian-hemp/) into the laws of its colonies, and many keep it today. This list lets the scores be compared by former ruler. [CEPII](https://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=6) records which countries ruled each country as a colony for a long time.[@1] Of the 204 places here, 55 were once ruled by the United Kingdom, 26 by France and 18 by Spain; 31 have no long colonial ruler recorded.
+
+The [empire chart](/charts/#empire) compares the scores of former colonies by their old ruler. Short or light rule is listed apart, and the data was compiled for trade studies, not for law.[@1] Each profile's history shows how its first laws arrived, and the atlas tells the story of [Russell Pasha](/culture/people/russell-pasha/), the British commandant of the Cairo police who told the League of Nations that hashish was a minor vice.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Transparency International — Corruption Perceptions Index"
+    url: https://www.transparency.org/en/cpi
+model: opus-5.5
+written: 2026-10-28
+---
+
+Where officials can be paid, a drug law is enforced unevenly. [Transparency International](https://www.transparency.org/en/cpi) scores how corrupt experts and business people judge each country's public sector, from 0, highly corrupt, to 100, very clean, from up to 13 surveys and assessments.[@1] In 2024 [Denmark](/country/dk/), [Finland](/country/fi/) and [Singapore](/country/sg/) are seen as cleanest, and [South Sudan](/country/ss/), [Somalia](/country/so/) and [Venezuela](/country/ve/) as most corrupt; the median of 180 countries is about 40.
+
+It says nothing about cannabis itself, but much about how a law is likely to be enforced: by the book, or by whoever pays.[@1] The [corruption chart](/charts/#corruption) sets it beside each score. Each profile's [black market](/context/black-market/) and [beyond the penalty](/context/consequences/) sections tell what enforcement looks like in practice; compare [government effectiveness](/datasets/government-effectiveness/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "World Bank — World Development Indicators"
+    url: https://databank.worldbank.org/source/world-development-indicators
+model: opus-5.5
+written: 2026-10-28
+---
+
+Five measures from the [World Bank's development indicators](https://databank.worldbank.org/source/world-development-indicators) describe how a country lives: the share of people using the internet, living in towns and cities and aged 65 or over; life expectancy at birth; and enrolment in higher education.[@1] The map shows the internet: nearly everyone is online in [Bahrain](/country/bh/), [Kuwait](/country/kw/) and [Saudi Arabia](/country/sa/), and under one in ten in [South Sudan](/country/ss/) and [Burundi](/country/bi/); the median country is at about 81%.
+
+Higher-education enrolment can pass 100% where many older students enrol.[@1] The chart of [what goes with it](/charts/#what-goes-with-it) sets these, along with [income per person](/datasets/gdp/), beside each country's score; how each law came about is in the [history](/history/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "ESPAD Group — ESPAD Report 2024"
+    url: https://www.espad.org/espad-report-2024
+model: opus-5.5
+written: 2026-10-28
+---
+
+Whether a teenager could get cannabis if they wanted to says more about the market than about the law. The [ESPAD](https://www.espad.org/espad-report-2024) school survey asked students aged 15 to 16 in 37 European countries the same question in 2024, and the median is 28% who say it would be fairly or very easy.[@1] [Denmark](/country/dk/) and [Germany](/country/de/) are highest, at 41%, with [Slovenia](/country/si/) close behind; [Moldova](/country/md/) and [Ukraine](/country/ua/) lowest, under 10%.
+
+Germany's survey was taken in 2024, the year of its [Cannabis Act](/history/germany-cannabis-act-2024/). The [chart of how easy it is to get](/charts/#easy-to-get) sets the answers beside each score; compare with [students who have tried cannabis](/datasets/espad-lifetime/) and each profile's [black market](/context/black-market/) section.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "ESPAD Group — ESPAD Report 2024"
+    url: https://www.espad.org/espad-report-2024
+model: opus-5.5
+written: 2026-10-28
+---
+
+The [European School Survey Project on Alcohol and Other Drugs](https://www.espad.org/espad-report-2024) has asked 15- and 16-year-olds the same questions every four or five years since 1995, which makes it the longest comparable series of cannabis use in Europe.[@1] In 2024, across 37 countries, the median is 11% who have ever tried cannabis. [Czechia](/country/cz/) is highest, at 24%, with [Liechtenstein](/country/li/) and [Slovakia](/country/sk/) next; [Moldova](/country/md/), [Georgia](/country/ge/) and [Romania](/country/ro/) are lowest, under 5%.
+
+Germany's figures are for Bavaria only.[@1] The [chart of teenage trends](/charts/#teen-trends) follows each country since 1995, wave by wave. Compare with [how easy students say it is to get](/datasets/espad-availability/) and with [use at 15 and 16](/datasets/youth-use/) outside Europe.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Gallup — Grassroots support for legalizing marijuana hits record 70% (2023)"
+    url: https://news.gallup.com/poll/514007/grassroots-support-legalizing-marijuana-hits-record.aspx
+model: opus-5.5
+written: 2026-10-28
+---
+
+No country has asked the question for longer. Since 1969 [Gallup](https://news.gallup.com/poll/514007/grassroots-support-legalizing-marijuana-hits-record.aspx) has asked Americans whether the use of marijuana should be made legal, and the yes answers rose from 12% that year to a record 70% in 2023.[@1] The series spans the years of [NORML's founding in 1970](/history/us-norml-founded-1970/), the decriminalisation wave of the 1970s, the [first legal shops in Colorado in 2014](/history/us-colorado-first-shops-2014/) and the federal move toward [Schedule III](/history/us-hhs-schedule-iii-2023/).
+
+The [opinion gap chart](/charts/#opinion-gap) sets public opinion against what the law allowed in each year. For moral views in 25 countries at once, see Pew's [is using marijuana morally wrong?](/datasets/pew-marijuana/); the [United States profile](/country/us/) has the history state by state.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "GBIF — Occurrence search, Cannabis sativa"
+    url: https://www.gbif.org/occurrence/search?taxon_key=2984538
+model: opus-5.5
+written: 2026-10-28
+---
+
+Cannabis is also simply a plant, wild in some places and grown in many more. The [Global Biodiversity Information Facility](https://www.gbif.org/occurrence/search?taxon_key=2984538) gathers every record of it shared by herbaria, museums, botanical surveys and citizen scientists, by the country where it was found.[@1] Records exist for 134 countries; the most come from [Russia](/country/ru/), the [United States](/country/us/) and [Sweden](/country/se/), and the oldest herbarium sheets go back centuries.
+
+The numbers say as much about who records plants as about where cannabis grows: recent records are mostly photographs from apps such as iNaturalist.[@1] For the plant's local varieties and the people who keep them, see each profile's [genetics and heritage](/context/genetics/) section and the atlas entry on [landraces](/culture/language/landrace/); for the crop grown legally, see [hemp grown](/datasets/hemp/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "World Bank — GDP per capita, PPP (constant 2021 international $)"
+    url: https://data.worldbank.org/indicator/NY.GDP.PCAP.PP.KD
+model: opus-5.5
+written: 2026-10-28
+---
+
+How rich a country is sets what it can spend on police, courts or a regulated market. The [World Bank](https://data.worldbank.org/indicator/NY.GDP.PCAP.PP.KD) measures income as everything a country produces in a year per person, at prices that make a dollar buy the same everywhere.[@1] In 2025 [Singapore](/country/sg/), [Ireland](/country/ie/) and [Luxembourg](/country/lu/) are richest, at over $125,000 a person, and [Burundi](/country/bi/), the [Central African Republic](/country/cf/) and [Somalia](/country/so/) poorest, near $1,000; the median is about $17,000.
+
+The chart of [what goes with it](/charts/#what-goes-with-it) sets income and other [development indicators](/datasets/development/) beside each score. Income does not decide the law: Singapore is rich and among the harshest, while [Uruguay](/country/uy/), far poorer, was the [first country to legalise, in 2013](/history/uruguay-legalisation-2013/). Where cannabis is a cash crop for poor farmers, each profile's [black market](/context/black-market/) section says so.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "World Bank — Worldwide Governance Indicators"
+    url: https://www.worldbank.org/en/publication/worldwide-governance-indicators
+model: opus-5.5
+written: 2026-10-28
+---
+
+A law is only as strong as the state that applies it. The [World Bank's governance indicators](https://www.worldbank.org/en/publication/worldwide-governance-indicators) estimate how well a state works: the quality of its public services and civil service, and how far it can carry out the policies it makes.[@1] On a scale of 0 to 100, [Singapore](/country/sg/), [Japan](/country/jp/) and [Luxembourg](/country/lu/) score highest in 2025, and [South Sudan](/country/ss/), [Haiti](/country/ht/) and [Sudan](/country/sd/) lowest; the median is about 52.
+
+It measures capacity, not freedom: an effective state can enforce a harsh law thoroughly, as Singapore does, or run a [legal market](/context/production/) well.[@1] The [chart of state capacity](/charts/#state-capacity) sets it beside each score. A weak state may have a strict law on paper and a [black market](/context/black-market/) in practice; see also the [corruption perceptions index](/datasets/corruption/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "FAO — FAOSTAT, crops and livestock products"
+    url: https://www.fao.org/faostat/en/#data/QCL
+model: opus-5.5
+written: 2026-10-28
+---
+
+Hemp is cannabis grown for fibre and seed, with little THC, and the [Food and Agriculture Organization](https://www.fao.org/faostat/en/#data/QCL) has recorded how much of it each country harvests every year since 1961.[@1] In 2024 the largest areas were in [France](/country/fr/), [North Korea](/country/kp/) and [China](/country/cn/), each about 20,000 hectares; 39 countries have a figure. The crop was once grown widely, as the atlas shows in [Italian hemp rope-making](/culture/traditions/italian-hemp-rope-makers/) and [Hmong hemp cloth](/culture/traditions/hmong-hemp-cloth/).
+
+The records are patchy: some countries report only fibre or only seed, some only the area, and many stop for years.[@1] Where fibre and seed overlap, the larger area is shown. The [hemp chart](/charts/#hemp) follows the crop since 1961, and each profile's [hemp and CBD](/context/hemp/) section gives the THC limit that separates hemp from the drug, such as the 0.3% the [United States adopted in 2018](/history/us-farm-bill-2018/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Cato Institute and Fraser Institute — The Human Freedom Index 2025"
+    url: https://www.cato.org/human-freedom-index/2025
+model: opus-5.5
+written: 2026-10-28
+---
+
+The index on this site scores one freedom; the [Human Freedom Index](https://www.cato.org/human-freedom-index/2025) of the Cato and Fraser Institutes scores freedom in general, from 0 to 10, from 87 indicators of the rule of law, movement, religion, expression, association, trade and regulation.[@1] For 2023 it ranks [Switzerland](/country/ch/), [Denmark](/country/dk/) and [New Zealand](/country/nz/) highest and [Syria](/country/sy/), [Iran](/country/ir/) and [Yemen](/country/ye/) lowest, across 165 jurisdictions; the median is about 6.9.
+
+The two measure different things, and a country can rank high for freedom in general while punishing cannabis hard. The [freedom chart](/charts/#freedom) sets the Human Freedom Index against this site's score country by country. None of it enters the score, which is built only from [six parts of cannabis law](/map/possession/). For a measure built by other experts, see [civil liberties](/datasets/civil-liberties/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Eurostat — Monthly minimum wages (earn_mw_cur)"
+    url: https://ec.europa.eu/eurostat/databrowser/view/earn_mw_cur/default/table
+model: opus-5.5
+written: 2026-10-28
+---
+
+A street price means little without a wage beside it. [Eurostat](https://ec.europa.eu/eurostat/databrowser/view/earn_mw_cur/default/table) publishes the national gross minimum wage for a month of full-time work, in euros, as it stood each January.[@1] In 2026 it is highest in [Luxembourg](/country/lu/), at about €2,700, [Ireland](/country/ie/) and [Germany](/country/de/), and lowest in [Ukraine](/country/ua/), [Moldova](/country/md/) and [Albania](/country/al/); 31 countries have a figure.
+
+Countries without a national minimum wage, such as Austria, Denmark, Finland, Italy and Sweden, are left out.[@1] The chart of [work for a gram](/charts/#work-for-a-gram) divides the [street price of cannabis](/datasets/prices-europe/) by the wage, to show how many minutes of work a gram costs; a fine for possession, given in each profile's penalties, can be read the same way.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "The GDELT Project"
+    url: https://www.gdeltproject.org/
+model: opus-5.5
+written: 2026-10-28
+---
+
+News coverage shows where cannabis is argued over. [The GDELT Project](https://www.gdeltproject.org/) monitors news worldwide; this site counts the cannabis and marijuana articles from each country's outlets as a share of all the news GDELT follows.[@1] The [United States](/country/us/) accounts for by far the most, followed by the [United Kingdom](/country/gb/) and [Mexico](/country/mx/); 147 countries have a figure.
+
+The share is of all news worldwide, not of the country's own news, so large media markets dominate, and outlet coverage is uneven.[@1] The chart of [cannabis in the news](/charts/#in-the-news) follows the series since 2017; compare with what readers look up in [views of the cannabis article on Wikipedia](/datasets/wikipedia-views/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "OpenStreetMap Wiki — Tag:shop=cannabis"
+    url: https://wiki.openstreetmap.org/wiki/Tag:shop%3Dcannabis
+model: opus-5.5
+written: 2026-10-28
+---
+
+Where cannabis is sold openly, shops appear on the map. In [OpenStreetMap](https://wiki.openstreetmap.org/wiki/Tag:shop%3Dcannabis), the open map of the world, volunteers tag dispensaries, licensed shops, coffeeshops and CBD shops alike as cannabis shops.[@1] In October 2026 the [United States](/country/us/) had about 4,700 tagged, [Canada](/country/ca/) about 1,600 and [France](/country/fr/) about 700; 63 countries have at least one.
+
+The count says as much about mappers as about shops: [Thailand](/country/th/), which had thousands of shops after [its dispensaries opened in 2022](/history/thailand-dispensaries-open-2022/), has only a few hundred tagged.[@1] Each count is a dated edition, so the trend builds up as it is repeated. What may be sold, and by whom, is in each profile's [cannabis industry](/context/production/) and [recreational use](/context/recreational/) sections; the atlas explains the [dispensary](/culture/language/dispensary/) and [the coffeeshop](/culture/traditions/the-coffeeshop/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Pew Research Center — In 25-country survey, Americans especially likely to view fellow citizens as morally bad (2026)"
+    url: https://www.pewresearch.org/religion/2026/03/05/in-25-country-survey-americans-especially-likely-to-view-fellow-citizens-as-morally-bad/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Laws follow what people believe, and in 2025 the [Pew Research Center](https://www.pewresearch.org/religion/2026/03/05/in-25-country-survey-americans-especially-likely-to-view-fellow-citizens-as-morally-bad/) asked adults in 25 countries whether using marijuana is morally acceptable, morally unacceptable, or not a moral issue.[@1] The share who call it morally wrong runs from 91% in [Indonesia](/country/id/), 83% in [Nigeria](/country/ng/) and 82% in [Türkiye](/country/tr/) down to 19% in [Canada](/country/ca/), 23% in the [United States](/country/us/) and 24% in [Australia](/country/au/). The median country sits near half.
+
+The three countries at the top treat cannabis as a crime, Indonesia with death still possible for aggravated sale, while Canada, at the bottom, has allowed it since its [Cannabis Act of 2018](/history/canada-cannabis-act-2018/). The survey cannot say which came first, the law or the belief. The [opinion chart](/charts/#opinion) sets the answers beside each country's score, and for one country over half a century see [Gallup's question in the United States](/datasets/gallup-us/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "UNODC — World Drug Report 2026, statistical annex"
+    url: https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html
+model: opus-5.5
+written: 2026-10-28
+---
+
+Beyond Europe, the widest measure of enforcement is how many people the police bring into formal contact, as suspects, arrests or cautions, for using or possessing drugs, as governments report it to [UNODC](https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html).[@1] It covers 104 countries. Per 100,000 people, [Sweden](/country/se/) reports the most, about 741 in 2024, then [Spain](/country/es/) and [Sri Lanka](/country/lk/); several countries report none at all.
+
+All drugs are counted together, because cannabis is not reported separately.[@1] Some include administrative fines and some do not, so the comparison is rough. For cannabis alone in Europe see [offences for cannabis use](/datasets/cannabis-offences/); for what a contact with the police leads to, see each profile's [arrests and prisoners](/context/arrests/) and [beyond the penalty](/context/consequences/) sections.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Our World in Data — Population sources"
+    url: https://ourworldindata.org/population-sources
+model: opus-5.5
+written: 2026-10-28
+---
+
+Population is the background to every other figure: it turns counts into rates and shares into numbers of people. The site takes it from [Our World in Data](https://ourworldindata.org/population-sources) and the World Bank, a figure every ten years from 1900 and the latest one, for all 220 places.[@1] [India](/country/in/) and [China](/country/cn/) each have over 1.4 billion people; [Tuvalu](/country/tv/) and [Nauru](/country/nr/) about ten thousand.
+
+It weights the population-weighted world average, shown beside the plain average of countries, and sorts the rankings by size.[@1] It also turns survey shares into the [number of cannabis users](/datasets/cannabis-use/), drawn in the [treemap of users](/charts/#users-treemap), and gives rates per person for [offences](/datasets/cannabis-offences/), [shops](/datasets/osm-shops/) and [research](/datasets/research-papers/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "EUDA — European Drug Report statistical bulletin 2026, price, purity and potency"
+    url: https://www.euda.europa.eu/data/stats2026/ppp_en
+model: opus-5.5
+written: 2026-10-28
+---
+
+Cannabis has grown stronger, and potency is one of the arguments in every debate about the law. The [European Union Drugs Agency](https://www.euda.europa.eu/data/stats2026/ppp_en) collects the mean THC content of herb and resin tested at retail level, mostly from seized samples.[@1] Across 28 countries the median is about 11%; the strongest recent herb or resin is reported from [Finland](/country/fi/), [Spain](/country/es/) and [Estonia](/country/ee/), at 15–17%.
+
+Samples and laboratory methods differ between countries, so a country is best compared with itself over time.[@1] The [chart of THC per euro](/charts/#thc-per-euro) sets potency against the [street price](/datasets/prices-europe/). Where a country sets a limit for legal products, it is in the profile's [hemp and CBD](/context/hemp/) section; the plant's chemistry is explained in the atlas under [CBD](/culture/language/cbd/) and [indica and sativa](/culture/language/indica-and-sativa/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "EUDA — European Drug Report statistical bulletin 2026, price, purity and potency"
+    url: https://www.euda.europa.eu/data/stats2026/ppp_en
+model: opus-5.5
+written: 2026-10-28
+---
+
+What a gram costs on the street says how available cannabis is despite the law. The [European Union Drugs Agency](https://www.euda.europa.eu/data/stats2026/ppp_en) collects the mean retail price of herb and resin from police purchases, seizures and surveys in each country.[@1] Among 29 countries the median is about €10 a gram; the highest recent prices are in [Norway](/country/no/), [Malta](/country/mt/) and [Cyprus](/country/cy/), the lowest in [Croatia](/country/hr/), at about €3.
+
+Dutch prices are for cannabis bought in [coffeeshops](/culture/traditions/the-coffeeshop/), not on the street, and prices are not adjusted for inflation.[@1] The [price chart](/charts/#price) follows each country since 2002, and the chart of [THC per euro](/charts/#thc-per-euro) combines price with [potency](/datasets/potency-europe/). For what a gram means for a worker, see the [minimum wage](/datasets/minimum-wage/) and the chart of [work for a gram](/charts/#work-for-a-gram).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Pew Research Center — Religious composition by country, 2010–2020"
+    url: https://www.pewresearch.org/religion/feature/religious-composition-by-country-2010-2020/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Faith shapes drug laws, and sometimes cannabis is part of the faith. The [Pew Research Center](https://www.pewresearch.org/religion/feature/religious-composition-by-country-2010-2020/) estimates each country's religious make-up from more than 2,700 censuses and surveys; this site takes the largest group.[@1] Of 191 countries, Christians are the largest group in 117, Muslims in 51, the religiously unaffiliated in 11, Buddhists in 8, Hindus in 3 and Jews in 1.
+
+The [religion chart](/charts/#religion) compares the scores of countries by their largest faith. The atlas follows cannabis inside religions: the [Rastafari sacrament](/culture/traditions/rastafari-sacrament/), [Shivaratri at Pashupatinath](/culture/traditions/shivaratri-at-pashupatinath/), [Santa Maria in Santo Daime](/culture/traditions/santa-maria-in-santo-daime/) and the Sikh [shaheedi degh](/culture/traditions/shaheedi-degh/). Each profile's [culture and traditions](/context/culture/) section tells how cannabis sits in the country's life.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "OpenAlex"
+    url: https://openalex.org/
+model: opus-5.5
+written: 2026-10-28
+---
+
+Where cannabis may be studied, papers follow. [OpenAlex](https://openalex.org/), an open catalogue of scholarly works, lets this site count the papers whose title or abstract mentions cannabis, marijuana or a cannabinoid, by the countries their authors work in, every year since 1950.[@1] Per million people, small places with a single active group can rank highest, so the map is best read for large countries and for change over time.
+
+Modern cannabis science began when [Raphael Mechoulam](/culture/people/raphael-mechoulam/) and Yehiel Gaoni [isolated THC in 1964](/history/israel-mechoulam-thc-1964/), and [Lumír Hanuš](/culture/people/lumir-hanus/) later helped find the body's own cannabinoids. A paper with authors in several countries counts for each.[@1] Each profile's [research](/context/research/) section says who may study cannabis there; for medicines tested on patients, see [clinical trials](/datasets/clinical-trials/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "UNODC — World Drug Report 2026, statistical annex"
+    url: https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html
+model: opus-5.5
+written: 2026-10-28
+---
+
+Seizures show where cannabis moves and where the authorities stop it. [UNODC](https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html) collects the kilograms of herb and resin each government seizes in a year; this site covers 167 countries from 2015 to 2024.[@1] The largest totals come from [Namibia](/country/na/), [Tanzania](/country/tz/) and [Paraguay](/country/py/), each over a million kilograms in 2024, figures that may include plants destroyed in the field.
+
+A seizure is not a measure of use: a transit country can seize far more than its people smoke, and a large total may mean a strong police force as much as a large trade.[@1] Plants, oil and seeds are left out. Each profile's [black market](/context/black-market/) section explains where the country's cannabis comes from, and the history records large finds such as the [transit seizure at Singapore's Pasir Panjang in 2026](/history/singapore-pasir-panjang-transit-seizure-2026/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "EUDA — European Drug Report statistical bulletin 2026, treatment demand"
+    url: https://www.euda.europa.eu/data/stats2026/tdi_en
+model: opus-5.5
+written: 2026-10-28
+---
+
+Some countries punish people who use cannabis; others send them to treatment, and some do both. The [European Union Drugs Agency](https://www.euda.europa.eu/data/stats2026/tdi_en) reports how many people entered drug treatment in a year and how many came mainly for cannabis.[@1] In 2024 cannabis was the main reason for 66% of those entering treatment in [Hungary](/country/hu/), 52% in [Germany](/country/de/) and 45% in [France](/country/fr/) in 2023, and for about a tenth in [Estonia](/country/ee/), [Bulgaria](/country/bg/) and [Slovenia](/country/si/).
+
+Some people come of their own accord, others are sent by a court or the police, and countries count treatment differently.[@1] A high share can reflect courts sending people to treatment as much as harm from cannabis. The chart of [punish or treat](/charts/#punish-or-treat) sets treatment beside [offences](/datasets/cannabis-offences/); each profile's [beyond the penalty](/context/consequences/) section explains what diversion means there.

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "EUDA — Drugs in municipal wastewater in Europe, source data 2025"
+    url: https://www.euda.europa.eu/data/repository/drugs-municipal-wastewater-europe-source-data-2025_en
+model: opus-5.5
+written: 2026-10-28
+---
+
+Surveys depend on what people admit; sewage does not. The [SCORE group and the European Union Drugs Agency](https://www.euda.europa.eu/data/repository/drugs-municipal-wastewater-europe-source-data-2025_en) measure THC-COOH, which the body makes from cannabis, in the wastewater of cities over one week a year, in milligrams a day per 1,000 people.[@1] The highest recent loads are in the [United States](/country/us/) and [Canada](/country/ca/), around 600, followed by the [Netherlands](/country/nl/) at about 200; the median of 33 countries is about 72.
+
+The load depends on how much and how often people use, so it is not a count of users, and one city may stand for a whole country.[@1] It is still the one measure that cannot be shaped by what people are willing to say about an illegal drug. The [wastewater chart](/charts/#wastewater) follows cities since 2011; compare with the survey figures in [cannabis use](/datasets/cannabis-use/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "Wikimedia Foundation — Analytics API (pageviews)"
+    url: https://doc.wikimedia.org/generated-data-platform/aqs/analytics-api/
+model: opus-5.5
+written: 2026-10-28
+---
+
+What people look up shows what they are curious about. The [Wikimedia Foundation](https://doc.wikimedia.org/generated-data-platform/aqs/analytics-api/) publishes how often each article is loaded; this site follows the article on cannabis as a drug in 26 language editions, every month since July 2015.[@1] Each language is linked to the country where most of its readers live; the Japanese, German and Italian editions rank highest.
+
+A language is not a country: German Wikipedia is read in Austria and Switzerland too, and page loads are not unique readers.[@1] The [chart of who looked it up](/charts/#looked-it-up) follows each edition month by month, to be read beside changes in the law such as [Germany's Cannabis Act in 2024](/history/germany-cannabis-act-2024/). For coverage in the news rather than readers' curiosity, see [cannabis in the news](/datasets/news/); for the words themselves, the atlas entry on [cannabis](/culture/language/cannabis/).

@@ -0,0 +1,11 @@
+---
+sources:
+  - title: "UNODC — World Drug Report 2026, statistical annex"
+    url: https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html
+model: opus-5.5
+written: 2026-10-28
+---
+
+Use among 15- and 16-year-olds is the one cannabis figure built for comparison: the same age group, mostly asked in school surveys such as Europe's [ESPAD](/datasets/espad-lifetime/), and reported by governments to [UNODC](https://www.unodc.org/unodc/en/data-and-analysis/world-drug-report-2026-annex.html).[@1] It covers 51 countries, and the median is 8.5% who used cannabis in the past year. The highest shares include [Liechtenstein](/country/li/) and the [United States](/country/us/), at about a quarter; among the lowest are [Mexico](/country/mx/) and [Egypt](/country/eg/), under 2%.
+
+The price of comparability is coverage. Countries that survey other age bands, Germany, France and Spain among them, are left out, and survey years run from 2003 to 2024.[@1] A survey of pupils also misses young people who are not in school. Read it beside [how easy students say cannabis is to get](/datasets/espad-availability/) and beside the adult figures in [cannabis use](/datasets/cannabis-use/).

@@ -429,3 +429,13 @@ const contextIntros = defineCollection({
 });
 
-export const collections = { countries, units, segments, changes, prices, articles, culture, datasets, glossary, contextIntros };
+// The same for each outside dataset's page (/datasets/<slug>/): datasets/intros/<id>.md.
+const datasetIntros = defineCollection({
+  loader: glob({ base: "./datasets/intros", pattern: "*.md" }),
+  schema: z.object({
+    sources: z.array(source).min(1),
+    model: z.string().min(1).optional(),
+    written: z.coerce.date().optional(),
+  }).strict(),
+});
+
+export const collections = { countries, units, segments, changes, prices, articles, culture, datasets, glossary, contextIntros, datasetIntros };

@@ -41,6 +41,9 @@ export const datasetPages = Object.values(DATASETS).map((d) => {
     const mapRow = d.map ? series.findLast((r) => r[d.map.field] != null) : null;
     const value = words ? null : valueOf(c, d, row, name);
+    // a figure in words left empty (no long colonial rule recorded) is said as such
+    const head = words && row?.[name] == null ? ["None recorded", ""] : headline(c, d, row);
     return {
-      c, row, year: row?.year ?? null, head: headline(c, d, row), details: fill(c, d, row, d.profile?.details),
+      // a weekly series' year (2026.79) is shown as its year
+      c, row, year: row?.year != null ? Math.floor(row.year) : null, head, details: fill(c, d, row, d.profile?.details),
       value: Number.isFinite(value) ? value : null,
       mapValue: mapRow && (!d.map.since || mapRow.year >= d.map.since) ? d.map.value(mapRow, c) : null,
@@ -53,7 +56,10 @@ export const datasetPages = Object.values(DATASETS).map((d) => {
   const answers = words ? [...rows.reduce((m, r) => m.set(r.head[0] || r.head[1], (m.get(r.head[0] || r.head[1]) ?? 0) + 1), new Map())]
     .map(([label, n]) => ({ label, n })).sort((a, b) => b.n - a.n) : [];
-  // the highest and the lowest, where countries can be compared (survey ages differ for cannabis use)
+  // the highest and the lowest, where countries can be compared (survey ages differ for cannabis use),
+  // by the figure the map shows (a rate per person, a recent year) where there is a map view
   const comparable = !words && field?.format !== "rank" && d.id !== "cannabis_use";
-  const ranked = comparable ? rows.filter((r) => r.value != null && !r.c.territory).sort((a, b) => b.value - a.value) : [];
+  const measure = (r) => (d.map ? r.mapValue : r.value);
+  const shown = (r) => (d.map ? d.map.format(r.mapValue) : r.head[0]);
+  const ranked = comparable ? rows.filter((r) => measure(r) != null && !r.c.territory).sort((a, b) => measure(b) - measure(a)).map((r) => ({ ...r, shown: shown(r) })) : [];
   const mapView = d.map ? viewByMetric[`data:${d.id}`] ?? null : null;
   const mapPage = mapView ? `map/${mapView.slug}/` : (d.pages ?? []).map((p) => p.path).find((p) => p.startsWith("map/")) ?? null;
@@ -61,5 +67,5 @@ export const datasetPages = Object.values(DATASETS).map((d) => {
     d, slug, group: groupOf(d), rows, colouring: col, words, answers, comparable,
     highest: ranked.slice(0, 3), lowest: ranked.slice(-3).reverse(),
-    median: ranked.length >= 5 ? median(ranked.map((r) => r.value)) : null,
+    median: ranked.length >= 5 ? median(ranked.map(measure)) : null,
     series: rows.filter((r) => r.years > 1).length,
     mapPage, chart: CHART_OF[d.id] ?? null,

@@ -11,4 +11,8 @@ import { siteMap } from "../../lib/world-map.js";
 import { tierColor } from "../../lib/tiers.js";
 import { places } from "../../lib/countries.js";
+import { getEntry, render } from "astro:content";
+import { references } from "../../lib/cite.js";
+import { sourceTuple } from "../../lib/sources.js";
+import References from "../../components/References.astro";
 
 export function getStaticPaths() {
@@ -18,4 +22,9 @@ export function getStaticPaths() {
 const { page } = Astro.props;
 const { d, rows } = page;
+// the dataset's introduction (datasets/intros/<id>.md), with its references numbered in the order cited
+const intro = await getEntry("datasetIntros", d.id);
+const { Content: Intro, remarkPluginFrontmatter } = intro ? await render(intro) : {};
+const refs = references("ref");
+if (intro) for (const i of remarkPluginFrontmatter.citeOrder ?? []) refs.cite(sourceTuple(intro.data.sources[i - 1]));
 const base = import.meta.env.BASE_URL;
 const num = (n) => n.toLocaleString("en-GB");
@@ -28,5 +37,6 @@ const profileHref = (c) => `${base}country/${c.slug}/#ds-${d.id}`;
 // a figure written as the dataset writes it: the median of the countries
 const name = d.worldwide ? null : headField(d);
-const written = (v) => (v == null ? null : name === "per" ? d.map.format(v) : figure(null, d, { [name]: d.fields[name]?.decimals != null ? +v.toFixed(d.fields[name].decimals) : +v.toFixed(v < 10 ? 2 : v < 100 ? 1 : 0) }, name));
+const round = (v) => (Math.abs(v) < 1 ? +v.toPrecision(2) : +v.toFixed(v < 10 ? 2 : v < 100 ? 1 : 0));
+const written = (v) => (v == null ? null : d.map ? d.map.format(round(v)) : figure(null, d, { [name]: round(v) }, name));
 const years = d.fileInfo.years && (d.fileInfo.years[0] === d.fileInfo.years[1] ? `${d.fileInfo.years[0]}` : `${d.fileInfo.years[0]}–${d.fileInfo.years[1]}`);
 const figures = d.worldwide
@@ -54,7 +64,7 @@ const description = `${d.name}, country by country: ${d.about.split(/(?<=\.)\s/)
     </section>
 
+    {/* the introduction, or until one is written, what the dataset counts */}
     <div class="prose ctxp-intro dsp-intro">
-      <p>{d.about}</p>
-      {d.map?.lede && d.map.lede !== d.about && <p>{d.map.lede}</p>}
+      {Intro ? <Intro /> : <p>{d.about}</p>}
     </div>
     <dl class="dsp-facts">
@@ -135,6 +145,6 @@ const description = `${d.name}, country by country: ${d.about.split(/(?<=\.)\s/)
         {[["Highest", page.highest], ["Lowest", page.lowest]].map(([label, list]) => (
           <div>
-            <h2 class="dsp-ends-h">{label}</h2>
-            <ol>{list.map((r) => <li><a href={profileHref(r.c)}><span aria-hidden="true">{r.c.flag}</span> {r.c.name}</a><b>{r.head[0]}</b>{r.year != null && <small>{r.year}</small>}</li>)}</ol>
+            <h2 class="dsp-ends-h">{label}{d.map && <span> · {d.map.label}</span>}</h2>
+            <ol>{list.map((r) => <li><a href={profileHref(r.c)}><span aria-hidden="true">{r.c.flag}</span> {r.c.name}</a><b>{r.shown}</b>{r.year != null && <small>{r.year}</small>}</li>)}</ol>
           </div>
         ))}
@@ -174,5 +184,6 @@ const description = `${d.name}, country by country: ${d.about.split(/(?<=\.)\s/)
 
     <section class="section prose dsp-origin" aria-labelledby="dspOrigin">
-      <h2 id="dspOrigin">How the file was made</h2>
+      <h2 id="dspOrigin">{Intro ? "What is counted, and how the file was made" : "How the file was made"}</h2>
+      {Intro && <p>{d.about}</p>}
       <p>{d.origin}</p>
       {Object.keys(d.notes).length > 0 && (
@@ -181,4 +192,6 @@ const description = `${d.name}, country by country: ${d.about.split(/(?<=\.)\s/)
       <p>The file belongs to {d.credit ?? d.publisher} and keeps its terms; cite the publisher when you use it.</p>
     </section>
+
+    {intro && <References refs={refs} archive />}
   </main>
 </Base>