What the figures measure, how they were assembled, and how carefully they should be compared across years.
Each figure is the percentage of people aged 3 and over who reported, in that census, that they could speak Irish. It is self-reported ability, not a test of fluency and not a measure of how often the language is used.
Census questions and definitions changed over time, so comparisons between distant years should be interpreted cautiously. The CSO notes that results such as 1926 and 2022 are not directly comparable, and a new question series began in 1996.
Northern Ireland recorded Irish only in the all-island censuses of 1851 to 1911 and again from 1991, so the six counties show as "no data" in between. Its 2001 and 2021 censuses appear under 2002 and 2022 to line up with the Republic; the county card says so when that applies. Antrim and Down for 1851, 1861 and 1901 are estimated, because Belfast straddles both counties and was not split between them in those returns.
Some Northern Ireland county figures are themselves estimates. The 1991, 2001 and 2011 values were derived from ward-level returns, and NISRA's 2021 county figures are estimated from census areas, as the North does not publish this on a county basis directly.