# Sanctuary Knowledge Vault V3 — Civilisation Dataset Published: 6 September 2026 Sanctuary Ready should preserve more than emergency checklists. The Civilisation Dataset is the long-duration knowledge layer: a curated, source-labelled offline library that keeps practical knowledge, history, education and culture available when normal internet access is unavailable. This guide defines the *shape* of the archive. Large third-party files remain at their official sources and are resolved into a build plan when a drive is prepared. ## Design rules 1. Keep the web interface simple. The public page shows vault size, knowledge categories and a small number of featured archives. The complete file-level catalogue belongs in manifests and the advanced catalogue page. 2. Keep Sanctuary-authored survival material readable without specialist software. 3. Use Kiwix/openZIM and other legally distributable sources rather than repackaging copyrighted commercial reference libraries. 4. Preserve source, archive date, licence, attribution and checksum data beside the content. 5. Keep encyclopaedic summaries, primary sources and Sanctuary guidance visibly distinct. 6. Treat offline snapshots as historical/reference material. Current emergency, medical, legal, conflict and travel guidance must still be checked against live official sources whenever available. 7. Do not put private household plans, personal messages or sensitive user data in shared vault images. 8. Build in redundancy: field manual in plain text/Markdown, Kiwix archives for large collections, offline maps, and a paper fallback for the most important procedures. --- # Knowledge domains ## 01 — Emergency & survival - Sanctuary Ready Field Manual - 72-hour / 2-week / 4-week readiness plans - household, grab-bag and communications templates - first aid and medical continuity - water, food, sanitation, shelter, heating/cooling and power - flood, storm, fire, smoke and radiological guidance - local Gateshead contacts and community support routes ## 02 — Medicine & human biology Priority topics: - anatomy and physiology - first aid and resuscitation - common diseases and symptoms - infection and hygiene - maternal and child health - medicines/pharmacology reference - mental health and psychological first aid - nutrition - public health and epidemiology Offline medical material is reference/education, not a replacement for a clinician or current NHS guidance. ## 03 — General encyclopedia English Wikipedia is the main broad reference layer. Keep one variant appropriate to the drive: - **Mini:** introductions + infoboxes, smallest footprint - **Nopic:** full article text, no images, best knowledge density - **Maxi:** full visual archive ## 04 — Human history & primary sources Keep both context and evidence: - dedicated Wikipedia History archive on small drives - full Wikipedia on larger drives - Wikisource primary documents - public-domain/open historical books where rights allow - Sanctuary Human History Vault chronology/index Coverage should include prehistory; Africa; West, South, East and Southeast Asia; Europe; the Americas; Oceania; migration; trade; technology; religion; philosophy; law; empire; colonialism; slavery; disease; environment; conflict; industrialisation; world wars; decolonisation and the contemporary world. ## 05 — Mathematics Preserve arithmetic through advanced material: - numeracy and arithmetic - algebra - geometry - trigonometry - calculus - probability/statistics - discrete mathematics - applied mathematics ## 06 — Physical & natural sciences - physics - chemistry - astronomy - geology - meteorology and climate - biology - ecology - genetics and evolution Open textbooks and simulation/learning resources are preferred additions after the core encyclopaedia. ## 07 — Engineering, mechanics & repair - hand tools and workshop safety - mechanical systems - vehicle basics - electrical/electronics fundamentals - batteries and low-voltage systems - appliance repair - bicycle repair - diagnostics and maintenance - iFixit/open repair guidance ## 08 — Computing & communications - operating-system documentation - Linux manuals - programming-language references - networking - databases - cybersecurity fundamentals - electronics/computing interfaces - amateur/community communications concepts within local law - developer Q&A/documentation where archive size allows ## 09 — Agriculture, growing & food - soil - seeds and propagation - horticulture - crop planning - composting - pests/disease - animal husbandry basics - food safety - cooking - food preservation - nutrition ## 10 — Construction, utilities & infrastructure - basic carpentry - masonry - roofing/weatherproofing - plumbing concepts - drainage - sanitation - water storage/treatment principles - household electricity safety - heating and insulation - structures/materials This library is educational. Dangerous electrical, gas and structural work should still be carried out by competent people. ## 11 — Geography, maps & navigation - offline Gateshead/Tyneside maps - wider UK maps - OpenStreetMap-derived regional data - topography/geography - compass/navigation fundamentals - Wikivoyage as general place reference Emergency routing must never depend on stale archived travel information. ## 12 — Language & communication - English Wiktionary - grammar and writing reference - major language dictionaries/phrase resources as capacity allows - sign/visual communication references where legally distributable ## 13 — Education & teaching - Wikibooks - Wikiversity - open textbooks - mathematics/science learning - children’s literacy/numeracy - teacher/reference material - experiments and simulations that work offline ## 14 — Law, government & civic knowledge - constitutional/government structure - basic rights and civic processes - historical laws and public documents - current-law snapshots clearly dated Never present an archived legal text as guaranteed current law. ## 15 — Literature, philosophy, religion & culture - public-domain/open literature - philosophy - religious texts with provenance/licence information - mythology - visual arts/history - music theory/history - cultural heritage Project Gutenberg material must retain its rights notices; US public-domain status does not automatically mean a work is public domain in the UK. ## 16 — Children & family - age-appropriate reading - basic science and maths - offline activities - child first-aid source material - family routines during prolonged disruption ## 17 — Community operations - volunteer coordination templates - capacity and skills directories (public/consented data only) - repair/reuse procedures - food and essentials logistics - charging/information-point operations - local-node operation and update procedures --- # Storage profiles The resolver should calculate the current byte total from the official source at build time. Archive sizes change, so the capacities below are budgets and selection rules rather than promises that an old hard-coded list will always fit. ## 64 GB — Essential Vault Purpose: household/portable knowledge drive. Priority: 1. Sanctuary Field Manual and local plans 2. compact English Wikipedia 3. dedicated human-history archive 4. dictionary/language 5. compact medical + first-aid reference 6. selected repair/growing/education content 7. Gateshead/UK offline maps when available Keep at least ~10% free for updates, indexes and user notes. ## 128 GB — Core Vault Purpose: strong household/community reference. Priority: - full-text/no-image Wikipedia - Wikisource primary texts - Wiktionary - open textbooks/courses - medical reference - repair and practical knowledge - regional maps - Sanctuary manuals and history guide ## 256 GB — Community Vault Purpose: shared community/Node library. Priority: - full Wikipedia with images where current size permits - primary sources + language - medicine - education - repair/engineering - agriculture/food - maps/navigation - selected computing material ## 512 GB — Deep Community Vault Purpose: substantial shared resilience/learning node. Add: - wider medical collection - wider computing/developer documentation - larger repair/engineering set - science and mathematics learning - literature/history books - more maps and regional reference ## 1 TB — Civilisation Vault Purpose: long-duration local knowledge archive. Add broad coverage of every domain above, including full visual Wikipedia, primary-source collections, technical/computing resources, medicine, education, repair, books and maps. Maintain free space for refreshes because ZIM files are replaced rather than incrementally patched. ## 2 TB — Archive Vault Purpose: preservation/node edition. Add: - larger book/literature collections - developer Q&A/documentation - additional language encyclopaedias/dictionaries - deeper science/technical collections - broader maps - legally distributable cultural/educational media - previous archive versions only where deliberate historical preservation is valuable Do not fill the last bytes of a drive merely because space exists. Reserve capacity for future updates, indexes and integrity files. --- # Build and integrity procedure For every physical drive or Node image: 1. Resolve the newest approved archives from the official catalogue. 2. Read actual remote byte sizes before download. 3. Stop the build if the selected profile exceeds its safe capacity budget. 4. Download by official direct link or torrent. 5. Store SHA-256/checksum files where the source publishes them. 6. Verify downloads before marking the vault complete. 7. Store `MANIFEST.json`, `CHANGELOG.txt`, archive dates and licences on the drive. 8. Run a sample open/search test in Kiwix. 9. Label the physical media with vault version and build date. 10. Keep the Sanctuary Field Manual and essential contacts independently readable even if Kiwix fails. # Web-page rule The public Sanctuary page stays curated: - one compact vault-size selector - knowledge-domain overview - a few featured core archives - build/download actions - advanced catalogue on a separate route The page should **not** render every file, mirror or checksum as a card. Detailed archive data belongs in the Advanced Catalogue and machine-readable manifests.