Preservation Digital preservation changes the scale at which the social and cultural history of Mithila, Vajji and Anga can be reconstructed. Manuscripts dispersed among families, institutional records, fragile newspapers, cassette recordings, photographs, field notes, performance video and born-digital literary material can now be copied and connected without physically gathering every original in one repository. But the possibility of copying also creates a dangerous illusion: a scan on a website is not automatically an archive, and a file on a hard disk is not automatically preserved. This chapter therefore treats digital preservation as an institutional process joining conservation, selection, description, technical integrity, rights, access and long-term stewardship. It uses official archival initiatives in India and Nepal as comparative evidence while keeping the regional problem in view: the materials needed for Mithila–Vajji–Anga history are scattered across public institutions, private collections, religious establishments, periodicals, performance communities and transborder networks. The durable objective is not merely to digitise more objects, but to make their provenance, relationships and future intelligibility survive changes of software, storage and institutions. 115.1 Digitisation and digital preservation answer different historical problems Digitisation creates a digital representation of an analogue object; preservation keeps a trustworthy representation usable across time. A manuscript photographed once may become easier to consult immediately, yet the photographs can still be lost through disk failure, account closure, corrupted files, obsolete formats or missing metadata. Preservation therefore begins where one-time scanning ends. The National Mission for Manuscripts has long joined documentation and conservation to digitisation, while the National Archives of India’s mass-digitisation programme explicitly couples online access with preservation of physical and digitised records. The distinction is essential for regional archives because copying without stewardship can multiply files while leaving cultural evidence just as vulnerable. 115.2 The physical original remains evidence after a digital surrogate is created High-resolution imaging can reduce handling and widen access, but it cannot make the source object historically irrelevant. Paper, palm leaf, ink, binding, watermarks, marginalia, sequence, repairs and damage patterns carry evidence that may not be captured adequately in a derivative image. Audio tape and film likewise retain physical information about recording technology and editing. A responsible programme therefore links the digital surrogate to the identity and location of the source and records the conditions of capture. Digitisation should reduce unnecessary handling while preserving the possibility of returning to the original when material questions require it. 115.3 Selection is the first act of canon formation in a digital archive No institution can digitise everything at once. Decisions about which manuscript bundle, newspaper title, author archive, photograph set or performance recording receives attention determine what later researchers can discover easily. Selection may favour celebrated writers, visually attractive objects, well- connected owners or material that is technically easy to scan. Such convenience can reproduce older exclusions. A regional preservation policy should therefore document its selection criteria and deliberately HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II include fragile, vernacular, women’s, Dalit, occupational, local-periodical and everyday-life materials alongside prestigious literary and religious collections. Figure 456 — Preservation as a chain of responsibilities. 115.4 Provenance is the bridge between a file and a historical source A filename such as image001.jpg conveys almost nothing about historical meaning. Provenance records who created or held the source, where it came from, who digitised it, when and under what conditions, and what changes were made afterward. This information makes a digital object citable and auditable. For family- held Panji materials, private letters, unpublished notebooks or recordings supplied by performers, provenance can be more important than elaborate cataloguing because it preserves the chain through which the evidence entered the archive. Without it, later users may possess pixels while losing context. 115.5 Metadata should describe relationships as well as individual objects Regional cultural sources rarely exist as isolated units. A folio belongs to a manuscript; a manuscript may belong to a copied textual tradition; an issue belongs to a periodical run; a song recording belongs to a performance occasion; a photograph may depict a person identified elsewhere in correspondence. Metadata that records only title, date and file type misses these relationships. Digital preservation becomes historically powerful when collections expose links among person, place, language, script, genre, institution, event, edition and related object. Such relational description also allows one archive to connect meaningfully with another without physically merging ownership. 115.6 Language and script metadata are indispensable in the Mithila–Vajji–Anga region Search systems designed around one dominant language can make a multilingual archive effectively invisible. Maithili may occur in Tirhuta, Devanagari and historical scripts; Sanskrit, Persian, Hindi, Urdu, Bengali, Nepali, Angika and Bajjika may coexist in a single regional collection. Catalogue records should therefore separate language from script and retain original-script titles alongside normalised or transliterated 11631163 GAJENDRA THAKUR access forms. This avoids treating a change of script as a change of language and supports search by readers with unequal script competence. 115.7 Preservation masters and access copies should have different functions A preservation master is created for durability and future reprocessing, while an access derivative is optimised for delivery. A large lossless image may be unsuitable for a mobile reader but valuable for future scholarship; a compressed JPEG or PDF can be convenient for everyday use but should not become the only surviving copy. The same principle applies to audio and video. Keeping master and derivative roles explicit allows an institution to regenerate access files when web standards change without repeatedly rescanning fragile originals. Figure 457 — Preservation masters and access derivatives serve different functions. 115.8 Checksums convert silent corruption into detectable change Digital files can alter without obvious visual warning. Storage errors, failed transfers or accidental overwriting may corrupt data while leaving directory listings intact. A cryptographic checksum functions as a compact fingerprint calculated from file content. Recomputing it later can show whether the file has changed. Checksums do not prevent damage, but they make integrity testable. For small institutions, this is one of the highest-value preservation practices because it requires little infrastructure yet transforms trust from assumption into something that can be audited. 115.9 Redundant copies must be independent enough to survive the same disaster Keeping three copies on one computer is not meaningful redundancy. Fire, theft, ransomware, electrical damage or administrative deletion can remove them together. Preservation systems therefore spread copies across different storage devices, accounts or locations and test them periodically. A practical regional strategy may combine an institutional server, an offline copy and a geographically separate repository or trusted preservation service. The design should assume that any single storage location can fail. HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II 115.10 Format choice is a long-term historical decision File formats differ in openness, documentation, compression and dependence on proprietary software. Preservation practice generally favours well-documented, widely implemented formats and retains enough technical metadata to support later migration. Yet no list of preferred formats is permanent. The deeper principle is to monitor technological change and preserve the ability to convert without losing significant properties. A TIFF image, WAV recording or standards-based text file is useful because it keeps more recoverable information than a heavily compressed or proprietary derivative, not because any format can guarantee immortality. 115.11 File naming is not metadata, but bad file naming can destroy order Meaningful identifiers help humans and systems keep related files together, but long descriptive filenames are fragile substitutes for catalogues. A good preservation identifier should be stable, unique and independent of changing titles or interpretations. Folder and filename conventions then support, rather than replace, structured metadata. This matters especially when digitisation is distributed among volunteers or multiple centres: inconsistent naming can separate pages from manuscripts or recordings from their descriptions before they ever enter a repository. 115.12 Web archives preserve relationships that ordinary file downloads miss A website is an assemblage of pages, stylesheets, images, scripts, fonts, audio, video links and external references. Saving only visible text can destroy navigation, chronology and contextual relationships. Web archiving therefore requires capture strategies that preserve linked resources and record crawl dates and limitations. For long-running regional journals and cultural projects, periodic whole-site capture is valuable because a future historian may need to reconstruct not only what a page said but how readers reached it, what sections existed and which external services the site depended on. 115.13 Born-digital records require preservation before they appear old Email, word-processing files, spreadsheets, digital photographs, source code, database exports and social- media posts can disappear faster than paper because they depend on accounts, devices and software. Their archival value is often recognised only after platforms have changed. Editors, researchers and cultural organisations should therefore schedule exports and deposits during active use rather than waiting for institutional closure. The history of twenty-first-century Maithili and Angika culture will be unusually incomplete if born-digital working records are treated as disposable. 115.14 Version history can itself be evidence A digital text may be corrected repeatedly without leaving the visible traces that handwritten revisions leave on paper. Version control, dated snapshots and changelogs can recover part of this history. For scholarly editions, websites, dictionaries and databases, preserving successive states allows researchers to distinguish correction from silent rewriting and to study how knowledge infrastructures evolved. Git-based repositories are especially useful for text and code because they store changes as an intelligible sequence rather than as unrelated copies. 11651165 GAJENDRA THAKUR 115.15 Databases must preserve exportable data, not only a functioning interface A searchable website can fail even when its underlying data remain valuable. Proprietary database systems, abandoned plugins or expired hosting can make records inaccessible. Preservation therefore requires regular exports in documented, portable forms together with schema documentation explaining fields and relationships. A Panji search tool, bibliographic catalogue or historical-place database should be preservable independently of the particular web interface through which users currently query it. 115.16 OCR expands discovery but should never erase uncertainty Optical character recognition can turn page images into searchable text, making large periodical and book collections vastly easier to investigate. Yet OCR accuracy varies with typeface, paper quality, language and script. Historical Tirhuta, damaged Devanagari or mixed-language pages may produce serious errors. The responsible model keeps the page image as evidence, stores OCR as a derived layer and allows corrections without pretending that machine transcription is identical to the source. Searchability and fidelity should reinforce rather than replace one another. 115.17 Handwritten-text recognition is promising but corpus-dependent Machine learning can increasingly recognise handwriting when sufficient training data exist, but minority scripts and local hands are poorly served by generic models. Regional projects can contribute by creating carefully transcribed training sets linked to images and by documenting confidence. Such work has value beyond automation: aligned image–transcription corpora become scholarly resources in their own right. The danger is to publish uncertain automated readings as authoritative text. 115.18 Audiovisual preservation has a shorter technological horizon than paper conservation Cassettes, VHS tapes, optical discs and early digital video combine chemical decay with hardware obsolescence. Playback equipment disappears, while each use may damage fragile media. Priority digitisation should therefore identify at-risk recordings before more stable paper sources simply because audiovisual rescue can be time-sensitive. Preservation-quality capture should retain the original signal as fully as feasible, while smaller files serve streaming users. Performer, event, date, location and consent metadata are as important as technical format. 115.19 Oral-history preservation requires consent to travel with the recording An interview is not ethically complete when only the audio file survives. Future users need to know what the speaker agreed could be preserved, quoted, published or restricted. Consent may distinguish public access from research-only access, impose time limits or require anonymisation. These conditions should be stored with the recording so that later migrations do not sever ethical obligations from the media. Community trust is itself preservation infrastructure. 115.20 Rights uncertainty can produce a digital dark archive Institutions often possess material they can preserve but cannot confidently publish. Copyright, donor restrictions, privacy and culturally sensitive content may require tiered access rather than either unrestricted release or no digitisation. Preservation copies can be maintained securely while rights are investigated, with metadata exposed where appropriate. The important distinction is between preservation authority and HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II publication authority: uncertainty about public display should not automatically justify allowing a fragile source to decay. 115.21 Traditional knowledge may require community governance rather than maximal openness Open access is valuable but not universally appropriate. Ritual knowledge, genealogical records, sacred images or community-specific materials may have customary restrictions not captured by ordinary copyright law. Ethical digital preservation therefore needs consultation about description, audience and reuse. A repository can preserve material while applying culturally informed access controls. The goal is durable stewardship, not extraction of every source into an unrestricted global database. 115.22 National-scale infrastructures show the value of federated discovery India’s National Digital Library and cultural portals demonstrate a model in which resources remain distributed among institutions while users search through a common discovery layer. This is especially relevant to Mithila–Vajji–Anga, where sources are divided among state archives, university libraries, private collections and institutions in both India and Nepal. A federated regional catalogue may be more realistic than attempting to centralise every digital object. Shared identifiers and metadata standards can make dispersed custody compatible with unified discovery. Figure 458 — A distributed regional archive depends on shared standards. 115.23 The National Archives of India illustrates preservation at mass scale The National Archives of India states that its mass digitisation programme, begun in 2021, serves both online access through Abhilekh-Patal and preservation of physical and digitised holdings. By April 2026 it reported more than 204 million digitised pages uploaded. The scale is far beyond any regional volunteer archive, but the underlying lesson is transferable: digitisation must be tied to repository systems, metadata, quality control and continuing institutional responsibility. 11671167 GAJENDRA THAKUR 115.24 Nepal’s National Archives digital repository creates a transborder preservation opportunity Nepal’s National Archives, established in 1967, now operates a digital repository presenting archival publications, books and reports. For Mithila history this matters because the cultural region crosses the modern border and many relevant sources are preserved in Kathmandu and other Nepalese institutions. Cross-border discovery standards, reciprocal cataloguing and stable citations can allow scholars to reconstruct connected histories without assuming that one national repository contains the whole record. 115.25 Disaster planning must include floods, humidity and infrastructure failure Preservation planning in the Gangetic and Tarai environment cannot be copied mechanically from temperate-climate institutions. Flooding, monsoon humidity, insects, unstable electricity, heat and transport disruption affect both originals and digital storage. Digitisation can reduce risk by creating off-site copies, but scanners, servers and backup media are themselves vulnerable. Disaster plans should therefore combine physical conservation, environmental controls where feasible, offline backups and geographically separated replicas. 115.26 Small institutions need preservation routines that survive staff turnover Complex systems that only one technically skilled person understands are fragile institutions. Documentation should explain directory structure, passwords and custody rules, backup schedules, checksum procedures, metadata fields, renewal dates and restoration steps. Tasks should be simple enough to be repeated by successors. For a small journal, family archive or cultural association, institutional memory may be a greater risk than storage capacity. Preservation succeeds when routine work can continue after the original digitiser leaves. 115.27 Community digitisation can enlarge the archive if quality controls remain visible Portable cameras and smartphones allow families and local organisations to contribute material that a central archive could never collect systematically. The resulting diversity is valuable, but image quality, cropping, sequence and description vary. Community programmes should therefore publish minimum capture guidelines and preserve information about who supplied each object. A lower-quality image with strong provenance can still be historically valuable, while an excellent image with unknown origin may be difficult to use responsibly. 115.28 Digital preservation should produce citations stable enough for scholarship Researchers need persistent identifiers, clear collection titles, item-level metadata and access dates so that evidence can be checked later. A URL alone is weak when web structures change. Where formal persistent- identifier systems are unavailable, institutions can still maintain stable item IDs and redirect old addresses. Citation design is not cosmetic: it connects preservation to reproducible historical argument. HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II 115.29 Preservation metrics should measure recoverability, not merely upload totals Page counts and terabytes are easy to publicise, but they do not reveal whether files are described, fixity- checked, replicated, rights-cleared or recoverable after a failure. Better metrics include the proportion of objects with complete metadata, successful restore tests, checksum verification rates, independent copy counts and unresolved rights status. The purpose of measurement should be to expose risk, not to reward the fastest accumulation of scans. Figure 459 — Six questions for testing preservation readiness. 115.30 A durable regional archive is a federation of responsibilities No single website, university, government programme or private collector can preserve the entire cultural record of Mithila, Vajji and Anga. The realistic model is a federation: owners retain custody where appropriate; repositories provide durable storage; scholars improve description; communities govern sensitive knowledge; technologists maintain formats and integrity; and public interfaces make permitted material discoverable. Digital preservation succeeds when these responsibilities are explicit enough to survive technological and generational change. Table 115.1 — Evidence and preservation requirements for major digital-archive domains Evidence domain Preservation master Essential metadata Primary risk Manuscripts & rare Lossless page images Collection, item, Missing provenance; print + source retained folio/page, language, incomplete sequence script, owner, capture Newspapers & Lossless images/PDF Title, issue, date, OCR error; missing periodicals master + OCR page, issues; brittle web derivative editor/publisher, copies place Photographs High-resolution Photographer, Unidentified lossless image subject, place, date, subjects; lossy original format, rights recompression 11691169 GAJENDRA THAKUR Evidence domain Preservation master Essential metadata Primary risk Audio Preservation-quality Performer/speaker, Carrier decay; WAV or equivalent event, date, place, hardware recorder, consent obsolescence Video High-quality Event, participants, Codec/container preservation file + date, place, technical obsolescence; original carrier where source, rights compression loss possible Websites & born- Periodic captures + URL/repository, Link rot; platform digital texts native files + version creator, dates, closure; silent history dependencies, rights overwriting Databases & tools Documented data Field definitions, Interface survives export + schema + identifiers, relations, while data become source code where software/version non-portable possible