Bars with a hatched end run past the edge of the chart — the scale stops at +150% so the three eras stay comparable. Six of the seven 1930s bars break the scale; only one modern figure does, and it belongs to a country under invasion.

Sources. 1908–13 and 1903–13: Correlates of War, National Material Capabilities v7.0, variable milex (total military expenditure, army and navy combined), thousands of current pounds sterling. 1930–38 and 1933–38: the same dataset, thousands of current US dollars. 2015–25 and 2020–25: SIPRI, constant 2024 US dollars. Percentage changes are calculated from the published values.

The 1908–13 panel is total spending, not army budgets. An earlier version of this chart showed army expenditure only, which is not comparable with the total-spending figures used for the other two eras. The six-power total of +52% over these five years independently matches David Stevenson’s published estimate of “about 50 per cent”.

These series are not perfectly like for like. All three panels now show total military spending, but only the modern column is inflation-adjusted; the two historical columns are in current prices. Currency changes too — the Correlates of War series is denominated in pounds before 1914 and dollars afterwards, so there is no continuous line across the two eras. Read this as scale, not as measurement.