Every claim has
a paper trail.

FIELD NOTE 01

Denmark changes the mix fastest.

Denmark's fossil share falls from 67.9% to 12.0%, or 4.00 percentage points a year. It ranks first among the 58 eligible grids. Wind and solar add 16.7 TWh.

Interpretation limit. This is a screened cohort, not every country. A percentage-point ranking rewards starting with more fossil generation; smaller grids can move faster.

Share change in percentage points = 100 * (4.240/35.450 - 26.320/38.750) = -55.962073
Speed = that change / 14 elapsed years
Absolute fossil change = 4.240 - 26.320 = -22.080 TWh
Demand / clean / trade identity: 1.640 - (18.780) - (4.940) = -22.080 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

FIELD NOTE 02

Britain's decline has three parts.

The UK's 198.1 TWh fall balances with 70.2 TWh less demand, 97.3 TWh more clean generation and 30.5 TWh more net imports.

Interpretation limit. These are accounting contributions, not independent causes. Imports have an electricity mix of their own; the chart does not measure consumption emissions.

Share change in percentage points = 100 * (95.690/281.330 - 293.740/382.060) = -42.869776
Speed = that change / 14 elapsed years
Absolute fossil change = 95.690 - 293.740 = -198.050 TWh
Demand / clean / trade identity: -70.200 - (97.320) - (30.530) = -198.050 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

FIELD NOTE 03

A cleaner share can hide more fossil output.

China's fossil share falls 17.6 points, yet fossil output rises 2,893 TWh. Clean generation grows 2,972 TWh while demand grows 5,865 TWh. India shows the same directional pattern.

Interpretation limit. Lower fossil share is not evidence of lower absolute emissions. This analysis measures electricity output, not economy-wide energy use or a causal policy effect.

Share change in percentage points = 100 * (6239.000/10072.600 - 3346.040/4207.160) = -17.591723
Speed = that change / 14 elapsed years
Absolute fossil change = 6239.000 - 3346.040 = 2892.960 TWh
Demand / clean / trade identity: 5865.020 - (2972.480) - (-0.420) = 2892.960 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

FIELD NOTE 04

New clean supply also replaces old clean supply.

Germany adds 153.9 TWh of wind and solar, but nuclear generation falls 140.6 TWh. The net gain from all clean sources is only 28.0 TWh.

Interpretation limit. This balance does not estimate the counterfactual without nuclear retirements. France is a useful comparison with a very different starting mix, not a controlled experiment.

Share change in percentage points = 100 * (202.930/476.870 - 378.240/624.210) = -18.040417
Speed = that change / 14 elapsed years
Absolute fossil change = 202.930 - 378.240 = -175.310 TWh
Demand / clean / trade identity: -106.940 - (27.970) - (40.400) = -175.310 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

FIELD NOTE 05

The leader changes with the unit.

The US records the largest absolute fossil decline in the cohort, at 334.5 TWh, but ranks 26 by share-change speed. Coal falls 1,195 TWh while gas rises 877 TWh.

Interpretation limit. Coal-to-gas substitution stays inside the fossil category. A fossil-output metric cannot by itself quantify the emissions benefit or cost of that substitution.

Share change in percentage points = 100 * (2547.950/4387.260 - 2882.490/4113.940) = -11.990297
Speed = that change / 14 elapsed years
Absolute fossil change = 2547.950 - 2882.490 = -334.540 TWh
Demand / clean / trade identity: 261.150 - (607.860) - (-12.170) = -334.540 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

FIELD NOTE 06

Starting clean leaves less room to fall.

Norway starts with just 4.2% fossil electricity and reaches 1.6%. Brazil starts at 12.4% and ends at 10.2%, yet adds 11.9 TWh of fossil output as its grid grows.

Interpretation limit. A slow share decline can coexist with a very clean grid. Hydro output varies by year; changing the baseline tests sensitivity, not statistical confidence.

Share change in percentage points = 100 * (2.510/156.120 - 5.180/123.140) = -2.598856
Speed = that change / 14 elapsed years
Absolute fossil change = 2.510 - 5.180 = -2.670 TWh
Demand / clean / trade identity: 6.910 - (35.650) - (-26.070) = -2.670 TWh

Pinned original source CSV · Local original snapshot · Independent Decimal verification with original raw records · Explore these grids

The complete 58-grid ranking.

Ranked by percentage-point decline per elapsed year, 2010–2024. Countries and territories are screened for a 2010 generation floor of 20 TWh and complete annual core data. This is not a global ranking of every national grid.

Sources, definitions and limits.

Data retrieval: 2026-09-03T04:34:42.266108+00:00. Archived data commit: f419e8f3a5e92f796f4f2e49ef5da187510d2f94. The CSV is pinned and checksummed. The current Ember methodology PDF is an explanatory reference, not a claim that the 2025 data vintage used the identical 2026 revision.

Source hierarchy: Our World in Data combines Ember (2025) and Energy Institute (2024), giving Ember priority for overlapping values. Annual observations can include source estimates. No row-level estimate flags or uncertainty intervals are available in this export. No statistical causal effects are estimated.

Generation is electricity produced domestically, not capacity or all energy. Demand is generation plus net imports. Clean is renewables plus nuclear, including bioenergy with feedstock-dependent climate impacts. Shares divide by domestic generation and exclude imports.

Missing cells remain missing. A calculated “other clean” residual is labeled separately and does not fill source columns. The primary dataset has 870 eligible country-years, 339 missing detailed cells, and complete core totals. Both annual accounting identities balance at the source precision in all 870 rows.

• owid-owid-energy-data.csv · local preserved copy
• owid-owid-energy-codebook.csv · local preserved copy
• owid-README.md · local preserved copy
• ember-metadata.yml · local preserved copy
• owid-electricity-mix.py · local preserved copy
• ember-methodology-current.pdf · local preserved copy

Machine-readable source manifest with licensing and hashes · All 220 candidate grids and exclusion reasons · 2010 / 2015 / 2020 baseline sensitivity