Historical Map Source Audit

Superseded on the start date. This audit argued for 1820 and the game shipped with it for a time, but the map was later re-cut from Victoria II's state regions and the canonical start moved to 1830-01-01. The sourcing research below still stands and is why several 1830 rollbacks exist (Algiers, Belgium, Texas, Gran Colombia, Ottoman Syria, Congress Poland). For how the map is actually built today, see docs/VIC2-PIPELINE.md.

Recommendation (as written for 1820)#

Grand Century should lock 1820-01-01 as its canonical start date.

That date gives the game a distinct identity rather than placing it on the standard Victorian starting line. The Congress of Vienna settlement is only five years old, the post-Napoleonic Concert of Europe is still new, the Spanish monarchy still legally spans most of mainland Spanish America, independence wars are actively redrawing that empire, and European control over much of Africa, Asia, and the western half of North America remains partial, layered, or contested.

1820 also creates stronger opening play:

• Gran Colombia, Chile, the United Provinces, Paraguay, the two Haitian states, and revolutionary movements coexist with Spanish royalist strongholds. Brazil is still joined to Portugal as a constituent kingdom, not yet an independent empire.
• Egypt is an autonomous Ottoman vassal whose forces control the Hejaz, while the Levant remains directly Ottoman. This creates an imperial relationship with competing agency instead of two simple sovereign color fills.
• The East India Company governs major territories but shares the subcontinent with dependent princely states and independent powers such as the Sikh Empire.
• Qing authority in Mongolia, Tibet, and Xinjiang, Dutch reach in Indonesia, Anglo-American claims in Oregon, and Indigenous ground control all require multiple relationship types rather than modern borders.
• The German Confederation and the Italian states preserve the small-state diplomacy and roster density that the current map lacks.

The open GeoPolHist database supplies entity and relationship dates from 1816 onward. Exact 1820 vector references also exist for Germany and Qing China, although their licenses limit them to validation. Public-domain period maps can support an independent reconstruction.

The exact day matters. All initial entities, relationships, wars, and borders must describe 1 January 1820, not a loose blend of the 1820s. Later atlases are corroborating evidence only and cannot introduce states that did not yet exist.

Source classes#

Natural Earth or another modern basemap may supply coastlines, rivers, and a neutral editing mesh. It must not determine 1820 ownership. Modern ADM0 polygons are the main reason historical maps acquire anachronistic countries and distorted borders.

Chronological corrections to the 1820 audit#

The following are high-confidence corrections to proposed 1820 entities or ownership. GeoPolHist is useful here as a machine-readable first pass, but its broad labels such as "part of" and "colony" should be implemented through owner, overlord, occupation, and claim relationships rather than copied as one flat owner field.

These corrections come principally from the GeoPolHist status series, checked against the contemporary and scholarly regional sources below. GeoPolHist's status model is evidence for relationships and dates, not precise ground control.

1820 initial roster and relationship strategy#

The initial roster should maximize historically meaningful actors, not count modern successor states. A polity belongs on the map when it had a government, ruler, recognized corporate identity, or durable military and fiscal organization that gameplay can express. Dependency does not make a polity disappear.

Use three linked layers:

1. Playable polity: the actor with its own court, cabinet, military, treasury, or revolutionary command. Sovereign states, autonomous vassals, major princely states, and durable confederated members can all qualify.
2. Imperial relationship: overlord, tributary, subsidiary alliance, personal union, protectorate, or federal membership. This is diplomacy and obligation, not a replacement map color.
3. Ground situation: direct administration, occupation, rebellion, competing claim, Indigenous control, or low-confidence frontier. This determines local gameplay without pretending every frontier was surveyed.

High-confidence opening blocs#

This structure creates country count and flavor through real 1820 institutions. It also makes the start dynamic: independence movements can consolidate, composite monarchies can fracture, imperial dependents can resist or integrate, and frontier claims can become surveyed borders through play.

Regional reconstruction anchors#

Germany#

Use member states, not a German national polygon. The German Historical GIS 1820 layer contains 41 polygons representing 38 of the 41 sovereign members of the German Confederation within the later German Empire. It excludes Austria, Luxembourg, and Liechtenstein and treats Schleswig with Denmark. Its geometry can validate a fresh reconstruction but cannot ship under the standard license. Because its snapshot is 31 December 1820, every change during that year must be checked before using it for 1 January.

The public-domain 1835 Abridgement of Universal Geography is a later corroborating roster. Its enumeration includes Austria and Prussia's federal lands, Bavaria, Saxony, Hanover, Württemberg, Baden, the Hessian and Saxon duchies, Mecklenburg, Oldenburg, Brunswick, Nassau, the Anhalt, Schwarzburg, Reuss, Lippe and Hohenzollern states, Liechtenstein, and Bremen, Hamburg, Lübeck, and Frankfurt. Use it to catch missing small states, but never back-port its count or borders without checking changes since 1820.

Minimum validation rule: there must be no unified Germany, and the four free cities must not disappear merely because they are small.

Italy#

The same 1835 geography provides a later public-domain roster: the Kingdom of the Two Sicilies, Papal States, San Marino, Tuscany, Lucca, Modena, Parma, Austrian Italy, Monaco, and the Kingdom of Sardinia. These all have valid 1820 counterparts, but borders and rulers should be reconstructed from sources representing 1 January 1820.

Minimum validation rule: there must be no unified Italy. Lombardy-Venetia belongs to the Austrian imperial system; Sardinia-Piedmont includes Savoy and Nice; the Papal States cross central Italy; and the Bourbon Two Sicilies control the south and Sicily.

India#

Build three visually and mechanically distinct categories:

1. Company-administered presidencies and provinces.
2. Princely or protected states with their own ruler and an EIC/British overlord relationship.
3. Independent states outside Company paramountcy.

The 1835 India map, 1836 India map, and scholarly Historical Atlas of South Asia explicitly support this distinction, but the later maps cannot supply 1820 borders without a change audit. GeoPolHist identifies Hyderabad and Awadh as dependent/vassal entities and the Sikh Empire and Kalat as sovereign entities in this period. Kashmir passed from Afghan to Sikh control in 1819.

Minimum validation rule: no modern India outline and no single British fill. Hyderabad, Awadh, the Sikh Empire, Nepal, Bhutan, Kalat, Mysore, Travancore, and major Rajput and Maratha successor states need explicit review.

Central Asia#

The 1834 Central Asia map, compiled from 1829 material and printed in 1839, gives a strong public-domain visual anchor. Its catalog summary identifies Persia, a multi-part Afghanistan, Baluchistan under Kalat, and the khanates or emirates including Khiva and Bukhara. GeoPolHist additionally supports Kokand, Bukhara, Khiva, and Afghanistan as distinct sovereign entities from 1816.

The public-domain 1835 Persia, Arabia, Tartary, and Afghanistan map is a useful second view, but the Library of Congress explicitly notes its incomplete and inaccurate geography. Use agreement between sources for core polity placement and encode uncertain steppe and desert limits as frontiers rather than precise surveyed borders.

Minimum validation rule: no Russian fill over the later Russian Central Asian empire. Khiva, Bukhara, Kokand, Afghanistan, Persia, and Kalat must survive as separate entities.

Arabia and the Ottoman-Egyptian border#

At 1820-01-01, Egypt should have an Ottoman-overlord relationship while exercising its own government and military. The Levant remains directly Ottoman. The Hejaz belongs in an Egyptian-control/Ottoman-suzerainty relationship after the Egyptian campaign against the First Saudi State. Nejd is still under Egyptian occupation, not the restored Saudi polity that appears in 1824. Oman and Yemen require separate polities, and the Gulf coastal sheikhdoms later called the Trucial States should not be merged into a modern UAE. The General Maritime Treaty was signed after the start date in January 1820. The 1835 regional map provides a later public-domain visual check, while GeoPolHist supplies dates and relationship types.

Minimum validation rule: the Ottoman Empire, Egypt, Nejd, Oman, Yemen, and the Gulf sheikhdoms cannot be collapsed into modern national polygons. Control and suzerainty must be separate fields.

West and Central Africa#

No source supports a clean modern-style sovereign tessellation across the entire region. The public-domain georeferenced 1829 Africa map can align period labels and routes, while the Library of Congress African maps guide points to early Liberian maps that record settlements and Indigenous political districts.

Blank space on a European map is not unowned land. A Library of Congress map-history analysis shows how an 1833 colonization map omitted Indigenous towns and polities that its own source text knew existed. It also discusses locally made maps whose claimed domains were contested. The correct game model is named polity cores, settlement and trade networks, influence zones, and uncertain or disputed frontiers. Do not create an UNC country as a substitute for research.

Minimum validation rule: Sokoto and its emirate relationships, Bornu, Asante, Dahomey, Oyo's successor landscape, Futa states, Segu/Kaarta, coastal states, Kongo successor polities, Lunda, Luba, and other regionally documented powers must be checked before any generic fill is accepted. Their exact extents need regional evidence and confidence metadata, not modern borders projected backward.

Southeast Asia and Indonesia#

GeoPolHist supports Burma, Siam, and Vietnam as distinct sovereign states. It records Cambodia under overlapping Siamese and Vietnamese relationships in 1820, which requires relationship and control layers rather than a simple independent-state fill.

For Indonesia, the Dutch National Archives' official overseas map guide identifies the Ministry of Colonies, topographic, and comprehensive Netherlands East Indies map series beginning in 1814-1816. These are the right archival collections for island-by-island adjudication. The archive's research guide also distinguishes centrally governed Java and Madura from outer possessions, confirming that Dutch imperial reach was not one uniform administrative condition. GeoPolHist preserves Aceh as sovereign until the late nineteenth century.

Minimum validation rule: no modern Indonesia polygon owned uniformly by the Netherlands. At minimum, distinguish core Dutch administration, dependent or treaty-linked courts, independent states such as Aceh, and areas where the map evidence supports only a claim or sphere of influence.

North American frontier#

Use the Newberry Atlas of Historical County Boundaries for dated US state and territorial boundaries. Do not extend it beyond its scope into an international-control source.

The 1820 southwest remains Spanish: Texas, Nuevo Mexico, and Alta California are not Mexican, Texan, or American entities yet. The later public-domain 1835 map of Texas is useful for physical alignment and retrospective comparison, not for ownership. Russian America remains Russian. Oregon requires simultaneous British, US, Spanish, and Russian claims as applicable on 1 January, with Indigenous ground control rather than a single owner. Use the Newberry Atlas for exact US territorial organization on the start date.

Minimum validation rule: Alaska is Russian; California, Nuevo Mexico, and Texas are Spanish; and Oregon has no homogeneous owner at the January 1820 start.

Reconstruction protocol#

Every political polygon or relationship should carry this provenance record:

asOf: exact ISO date
entity: stable internal polity ID
relationship: sovereign | direct-administration | vassal | protectorate | occupation | claim
overlord: optional polity ID
source: item-level URL and page/map identifier
sourceDate: date represented by the source, not only publication date
confidence: high | medium | low
license: public-domain | ODbL | validation-only | permission-required
notes: contradiction, interpolation, or disputed-boundary explanation

Recommended build order:

1. Freeze 1820-01-01 and publish an authoritative entity/relationship roster before drawing.
2. Use a modern physical basemap only for coastlines, rivers, and the editable mesh.
3. Reconstruct one region at a time from public-domain period sources, checking roster and status against GeoPolHist and the restricted academic datasets.
4. Represent dependencies, occupations, revolts, shared claims, and uncertain frontiers explicitly. A single owner field cannot reproduce this world.
5. Preserve the source and confidence record beside every authored feature so later corrections are local and auditable.
6. Add map assertions for the minimum validation rules above, including prohibited anachronisms such as unified Germany/Italy, independent 1820 Uruguay, US Alaska, direct-British all-India, and uniform Dutch Indonesia.

Bottom line#

The weak map is not primarily a polygon-count problem. It is a time model and provenance problem. Locking 1820-01-01, separating sovereignty from control and overlordship, and rebuilding from source-backed regional overlays will increase country count and flavor without replacing one false clean map with another. The result should feel specifically post-Napoleonic: revolutions in motion, composite monarchies under strain, a young Concert of Europe, and imperial frontiers that are still political systems rather than fixed lines.