# Power in Bloom: technical notes These notes give the detail behind the methodology on the page. Two tables come with them: - `states.csv` lists the 197 present-day states, with each state's UN region, the dates it existed as recorded in the dataset, the earlier states counted with it, and its counts across the century. - `continuity.csv` lists every earlier state the story counts under a present-day state, and the ones it leaves out, each with its dates, the reason, a source and any doubt. ## Sources The leaders come from a dataset built for this project from English Wikipedia and Wikidata: each office's list of holders, each person's own article and Wikidata's records of who held which office. Regions follow the UN geoscheme. Portraits come from Wikimedia Commons and flags from Flagpedia. ## What is counted - **Unit.** One person in one present-day state in one decade. People are told apart by their Wikidata identifier. - **Decades.** Ten decades, each from September 17 to September 17, starting in 1926. A decade includes its first day but not its last. A person counts in a decade if they held office there for at least one day of it. - **Several terms** in the same state and decade count once. So does a term that crosses a change of state (for example, Helmut Kohl as chancellor of West Germany and then of Germany). - **Offices.** The heads of state view holds heads of state and offices that are both; the heads of government view holds heads of government and offices that are both. All leaders is the two views combined, with each person counted once per state and decade. - **The century total** counts each person once per state, however many decades they served. A person who led two states counts twice; it is not a count of individuals worldwide. - **Dates.** A term with no end date runs to September 17, 2026. A date recorded only to the month or the year is placed on the first day of that month or year. - **Not counted:** how long anyone served, and how much power an office carried. ## Who is included - **Acting, interim and caretaker leaders** who held the office during a vacancy or a transition, and **regents** who exercised the role of a vacant throne, are included. Their own records keep the flag `acting`, and a woman's card says "Acting or interim" or "Regent". - **Stand-ins** are not in the dataset at all: people who acted for a holder who stayed in office and was absent, ill, on leave, campaigning or suspended. - **Disputed leaders**, such as the heads of rival governments, are left out. - **Dependent territories, colonies and governments in exile** are left out, apart from the wartime governments of occupied states that the Allies recognised. ## States and continuity The story uses a fixed set of 197 present-day states (`states.csv`). A state's earlier eras under another name or system count as that state, as the French Third Republic counts as France. A few states that no longer exist count under the present-day state that continued them: West Germany under Germany, the Federation of Malaya under Malaysia, North Vietnam under Vietnam, and the 1958–1961 union of Egypt and Syria, the United Arab Republic, under Egypt. `continuity.csv` gives the reason and source for each and flags two doubts: - whether Vietnam continued North Vietnam or was a new state; - that the United Arab Republic's leader counts only under Egypt. Serbia and Montenegro follow a separate rule. From April 27, 1992, when the Socialist Federal Republic of Yugoslavia ended, each republic's own presidents and prime ministers count, including acting presidents during a vacancy, such as Nataša Mićić in Serbia (2002–2004); the leaders of the federation they shared, the Federal Republic of Yugoslavia (later Serbia and Montenegro), do not count for either. A term that began earlier counts from that day, and a person who held both a federal and a republic office, such as Slobodan Milošević or Vojislav Koštunica, counts once per state and decade. Before that date the leaders of Yugoslavia's republics are outside the project's scope, and their plots say so. This is a rule for the comparison, not the date either country became independent; both did so in 2006. The dataset keeps the federal leaders under the federation, and the republic offices with their own titles and dates; `continuity.csv` records where sources disagreed. The Soviet Union and the Socialist Federal Republic of Yugoslavia governed many republics, so their leaders are not given to any one successor. Other dissolved states, such as Czechoslovakia, East Germany, South Vietnam and the two Yemens, are left out too. China and Taiwan follow a geographical convention, not a statement about sovereignty. Leaders of the Republic of China count under Taiwan from December 7, 1949, when its government retreated to Taiwan. Its mainland years stay in the dataset but are not shown. Leaders of the People's Republic count under China from October 1949. An empty plot in the Decade rooms says why it is empty when the data can tell: - the state was not yet independent, by its dates in the dataset; - its territory belonged to a state outside the data, from a hand-written table rather than Wikidata's successor links; - otherwise, only that no qualifying leader is recorded. ## Gender Each person's gender is the value of Wikidata's "sex or gender" property, frozen when the data was gathered (October 2026) so that later edits do not change the counts. The dataset holds two values, female and male; a person with no value on Wikidata and no article that states one is left blank rather than guessed. Every woman's record was checked by hand against her own Wikipedia article. Two leaders without either source, Turiddu Foschi (San Marino, 1930) and Wolde Tzaddick (Ethiopia, 1936–1942), are recorded as men by a project ruling, and no leader in the story has a blank gender. ## Checks - Every figure in the story is computed from the same data. A second count, written separately in Python, recounts each office view and decade, and the story's tests require the two to agree. - The build runs 70 automated checks on the counts, for example that no person appears twice in one state and decade, and that the combined view is exactly the two offices joined. - The sums of `states.csv` equal the century totals in the story. - The dataset itself was audited against sources it did not use. Every woman's record was checked by hand. Men's records were checked by a sample and by comparing every term with the person's own article, so a dating error is more likely to remain in a man's record than in a woman's.