The standing army.
Of every 1,000 people alive, how many are soldiers? The record keeps that number for two hundred and one years, and it is a mountain range with a long western slope: 9.11 per 1,000 in 1816, 31.95 at the 1945 summit — one person in 31 — and 2.64 in 2016, one person in 379 — the lowest armed share the two-century record holds. The money walks the other way: world military spending stands at $2.77 trillion in 2025, the largest figure in its own series. That divergence — fewer soldiers, dearer each: $61,292 spent per soldier in 1988, $99,581 by 2016 — is this report's finding, and it is drawn as two panels in two units, never as one converted line. Every figure is computed from the records at render time.
published 2026-08-29 · personnel and population 1816–2016 · spending 1988–2025 · edition 1 — updated only by a new dated edition · sources: Correlates of War National Material Capabilities v6.0 · OWID military expenditure
Two centuries of the armed share.
One number, year by year from 1816 to 2016: soldiers per 1,000 people, summed across every state the file lists. Two mountains stand in it — 1918 and 1945 — and after the second, a descent that had not stopped at the record's last year.
Fewer soldiers. Dearer each.
Since 1988 the manpower record and the spending record point in opposite directions. They are drawn side by side, each in its own unit on its own axis — never merged, never converted, never composited into an index. The refusal to blend them is what makes the divergence readable.
Eight states, two mountains.
The personnel column for eight states across the full record. The world wars stand as mountains; every one of the eight had its largest army in the past — 5 of them in 1945 alone. The largest standing force in the file's last year, China's 2,333 thousand, is less than a fifth of the largest the column ever held (12,500 thousand, Russia, 1945).
Whose columns, and what this page refuses.
Whose columns. Every personnel and population figure is the Correlates of War National Material Capabilities file, version 6.0 — its milper column (military personnel, thousands) and its tpop column (total population). Soldiers per 1,000 divides one by the other, summed over every state the file lists in a year. Spending is Our World in Data's military-expenditure series, in the unit its header states. Those two publishers are the page's whole sourcing.
The state-system caveat. The per-1,000 denominator is the file's state system, not the planet. The system grows across the two centuries: early years sum the few sovereigns the file recognizes, and much of the world's population lived outside that system before the twentieth century. The line is exact by construction — soldiers per 1,000 people of the system — and this caption is the difference between that and "of the world."
What this page refuses, and why. The NMC file carries its own military spending column back to 1816, and this page refuses to draw it: its currency basis changes across two centuries, and a line through changing units would be a picture of monetary history, not armies. Spending appears only inside the one consistent OWID series, and per-soldier arithmetic happens only in the years both records cover — 1988 to 2016 — stated on the exhibit. No composite "militarization index" is computed from the two directions: two panels stay two panels, because a number that merged them would have no unit at all.