the war atlas · numbers
Arithmetic is a state technology.
A tax roll, a land survey, a census and a payroll are all the same object: a list that must be summed, carried and checked by somebody who was not there when it was written. Everything below is about what made that possible, and about who tried to stop it.
A guide to the labels and evidence
What each one made possible
628 · notation[T2]
Positional notation with zero
Arithmetic on arbitrarily large numbers with a fixed set of symbols, and written calculation instead of a physical abacus. The place is what makes the algorithm possible.
The distinction that matters: zero as a PLACEHOLDER is older and appears in several systems; zero as a NUMBER you may operate on is the innovation.
820 · method[T2]
Algebra as a systematic method
Solving a class of problems by procedure rather than solving each problem by ingenuity. A method is transmissible; a trick is not.
1202 · notation[T2]
Hindu-Arabic numerals in Latin Europe
Written calculation that leaves an auditable trace. An abacus computes and forgets; a written sum can be CHECKED BY SOMEONE ELSE LATER.
🔴 THE ADOPTION LAG AFTER 1202 IS THE MEASUREMENT, and see innovation_resistance — it is one of the very few lags with a dated cause.
1494 · accounting[T2]
Double-entry bookkeeping
An internal consistency check. Every entry appears twice with opposite sign, so the books can DISAGREE WITH THEMSELVES and be caught — error detection built into the notation.
⚠ DO NOT RECORD 1494 AS AN ORIGIN. Pacioli describes practice already running in Venice; the origin is undated here and should stay that way rather than inheriting a publication date. Same class of error as UNESCO's inscription year standing in for a construction date.
undated · accounting[T3]
Age recording as an administrative act
A population that can state its own ages — which is the substrate of every conscription list, tax roll and pension scheme.
T3 and deliberately undated. It is a category the confound forced into existence, not a documented event.
The one lag with a named cause
A diffusion lag is normally a number with no author: the idea took ninety years to cross a border and nobody knows why. A dated prohibition turns that into a measured interval with somebody’s name on it. It is the only such case in this table.
1299 · Florence[T3]
Hindu-Arabic numerals in Latin Europe, forbidden
the Arte del Cambio — the Florentine bankers' and money-changers' guild
what it said
FRAUD PREVENTION: the new figures were held to be alterable — a 0 could be made a 6 or a 9, and there was no equivalent of the Roman system's redundancy to catch it.
what it also protected
A professional monopoly on calculation. A guild whose members are the people who can compute has an interest in the method staying difficult. ⚠ KEPT IN A SEPARATE COLUMN FROM THE STATED REASON ON PURPOSE — both may be true, and merging them decides the question by formatting.
outcome · It did not hold indefinitely; the numerals are in general commercial use in Italy well before the printing of Pacioli in 1494. The INTERVAL is the measurement.
⚠ Unverified. This is the load-bearing example for the argument above and it is the one most often repeated without a citation — which is exactly the profile of a claim thinner than its repetition. It is published here as a research task, not as a finding, and nothing else on this page depends on it.
Two variables that are one variable
Not correlated. Partly identical — and the overlap is invisible because the two arrive from different datasets under different names.
numeracy (ABCC / age heaping) is partly identical with administrative reach[T2]
Age heaping measures excess frequency at multiples of five. But knowing your own exact age is partly a function of whether the state ever MADE you say it — conscription lists, tax rolls, parish registration, pension acts. Both sides of the A'Hearn exchange accept that culture and state capacity influence age awareness and accurate reporting.
A'Hearn 2022 regresses the Whipple index on literacy, region, gender, urban residence and age group: only 15% of the variation is attributed to LITERACY, against 31% to REGION.
what can still be asked · ORDER, NOT MAGNITUDE. Ask whether administrative innovations LAG numeracy innovations in the same culture. A sequence question does not require the two to be disentangled, and gate_reversed_diffusion already fires on the wrong direction. If administration consistently LEADS, the substrate model is wrong and the state is manufacturing numeracy, not consuming it.
education (Seshat) is collinear by construction with administration (Seshat)[T2]
Across 51 variables and nine complexity characteristics — 414 societies, 30 regions, 10,000 years — a SINGLE PRINCIPAL COMPONENT captures about three-quarters of the observed variation. Administration, information systems, scale, money and governance are close to ONE variable wearing nine hats.
~75% of variance on one principal component
what can still be asked · Do not regress one on the other inside Seshat; it is regressing a thing on itself and will produce a beautiful, meaningless result. Use Seshat for PRESENCE/ABSENCE and prerequisite structure, and take the ordering question elsewhere.
What a system selects for
The interesting axis is not how much education a society had. It is what the system certified, and who was allowed to certify it.
| model | certified by | decouples | seen by a literacy count? |
|---|---|---|---|
| examination certified | the state | office from BIRTH | yes |
| corporate degree | a self-governing corporation | certification from THE STATE | yes |
| chain of transmission | a named teacher | certification from ANY INSTITUTION | yes |
| apprenticeship | the product | competence from LITERACY | no — invisible |
Apprenticeship is certified by whether the product works. A society running mostly on it reads as uneducated to every instrument ever built for counting readers.
Borrowing, in three rungs
The Norse took Islamic silver and refused the script and the creed stamped on it. What happened next is measurable rather than interpretive, and the sample is in the hundreds.
rung 1 · veneer
substance adopted, form destroyed
silver taken; coins cut into hacksilver
rung 2 · creolization
standard adopted
local weights calibrated to the foreign unit
rung 3 · substitution (aesthetic)
form adopted as symbol
the cubo-octahedral shape migrates from weights into balances and then into personal adornment, including arm-rings
The best thing to trade is the one you must be taught to use
Furs are the highest value-to-weight cargo on that list and teach nothing. The Rus sold furs, wax, honey and captives to the Abbasids and bought silver; the exports taught the buyers essentially nothing, and the import rebuilt Scandinavian weight standards from the base unit up. Knowledge ran opposite to value.
Every row is derived from the atlas database. Confidence tiers travel with the claim: T1 primary or dated act, T2 standard consensus, T3 contested or recalled. NOTHING BELOW T3 IS PUBLISHED.
Built only from public-domain and CC BY 4.0 material plus general historical fact. Seshat and Cliopatria are CC BY-NC-SA and are NOT used here.