Full chapter text
Evidence
Relational map of event, trace, record, silence, classification, source criticism and revisable history
Event/process → trace → selection → recording → preservation/destruction → classification → archive → discovery → source
criticism → corroboration → counter-position → interpretation → public claim → revision
History is not merely the victor’s narrative; the archive is not merely a neutral storehouse; silence is not in itself complete
evidence; genealogy does not search for a pure origin but examines contingent formation, power, classification and rupture; yet
power does not make evidence arbitrary. A historical claim remains bound by the provenance and date of its sources, authorial
position, context, textual transmission, independent corroboration, counter-evidence, material traces and revisable public
argument. The Parallel view of history adds marginal voices without abolishing evidentiary standards.
Problem
The first illusion in writing history is that the past lies ready-made like a book and the historian merely reads it. In fact, the past
leaves uneven traces of events, processes, memories, remains, documents, objects, landscapes, languages, oral traditions and
institutional records. Some survive, some are destroyed, some are deliberately preserved, and some persist by accident. The
historian works not with “everything” but with remnants of remnants.
The second illusion is that an archive is a neutral warehouse. Whom records were made about, who wrote them, in what
language or form, for what administrative purpose, how long they were retained, and who had access—all these shape the
historical material. Records of taxation, crime, land, census, justice, schools, the military, caste, religion, public health or trade
may have been created for the work of the state, not for future historians.
The third difficulty is silence. Groups that wrote little, whose documents were destroyed, who stood outside literate
administrative languages, or whose experience was not deemed “record-worthy” may appear less often in the archive. Yet the
absence of a record does not justify saying, “This must have happened.” Archival silence is a clue, not an automatic licence to
invent a narrative.
The fourth difficulty concerns the meaning of genealogy. In ordinary usage, genealogy may mean a family pedigree or lineage
list; in its philosophical and historical sense, however, it does not seek a fixed origin. It examines the historical formation,
conflict, rupture, classification and institutionalisation of what today appears normal, natural or timeless. Its purpose is not to
prove that “everything is false.”
The fifth difficulty is originism. Finding one pure beginning for an institution, caste, word, practice, boundary, title, community
or idea does not mean that its entire history has been understood. A name can remain while its meaning changes; an institution
can persist while its function changes; an old word can become a modern category.
The sixth difficulty is presentism. If modern caste lists, national borders, religious identities, linguistic standards, legal categories
or political terms are projected unchanged onto the past, the historical actors’ own words and classifications disappear. The
historian must translate, but must also declare the limits of translation.
The seventh difficulty is the relation between memory and history. Memory is a serious source of experience, especially for
histories of violence, displacement, family, labour, women’s lives, caste, disaster and locality. But memory changes over time,
takes narrative form, and absorbs collective language. Oral history should therefore be treated neither as “false” nor as superior
to documentary records.
The eighth difficulty is the forged or altered source. Inscriptions, copperplates, manuscripts, genealogies, letters, photographs,
newspapers, digital scans, screenshots, audio and video can all contain interpolation, false attribution, re-dating, cropping,
transcription error, OCR error, synthetic media or deliberate forgery. Provenance is not an optional convenience; it is
foundational to history.
The ninth difficulty is scale. Modern archives may contain millions of pages, digital files, census tables, satellite images, social-
media posts, databases and born-digital records. More data does not automatically mean more truth; selection bias, missingness,
metadata, duplication, algorithmic ranking and inaccessible formats create new problems.
The tenth difficulty is the relation between power and truth. Power creates records, defines categories, restricts access,
standardises language and amplifies some voices. But it does not follow that historical truth is only a product of power and
therefore arbitrary. Power itself is constrained by material traces, contrary records, external sources, chronological
inconsistency and resistant consequences in the world.
The eleventh difficulty is the history of the margins. “The mainstream wrote it, therefore it is false; the margin said it, therefore it
is true” is merely an inverted privilege, not Parallel history. Suppressed or under-recorded sources must be sought, but their
claims too must be tested for date, context, corroboration, self-interest, transformations of memory and counter-evidence.
The twelfth difficulty concerns sameness of name and identity. The same name, title, surname, gotra, place-name or book-title
may refer to different persons or periods. It is risky either to merge people merely because names match or to separate them
merely because they appear separately. Chronological, textual, genealogical, palaeographic, institutional and geographical
evidence must be considered together.
The thirteenth difficulty is “filling the gap”. Narrative pressure tempts the historian to turn a missing decade, missing person,
missing cause or missing manuscript relation from possibility into certainty. The strict rule of the Parallel view of history is that
where evidence is absent, the historian should explicitly write “unknown”, “probable”, “disputed” or “not yet established”.
The problem of this chapter is therefore this: how can one construct a view of history that exposes the power structure of the
archive, uses genealogy to reveal how categories are formed, listens to silence and marginal experience, yet does not make
historical fact arbitrary; that accepts plurality of evidence without treating all evidence as equally weighty; and that makes its
conclusions revisable public arguments?
Core Proposition
First principle: the past is unalterable; historiography is revisable. Events cannot change, but available evidence, questions,
methods and interpretations can. When we say “history has changed”, we usually mean not that the past changed but that the
structure of knowledge about it was revised.
Second principle: the source is not the event. An order, letter, inscription, report or autobiography is a perspective upon an
event. The historian must distinguish source claim, source production and underlying event.
Third principle: the archive is a selection machine. Record creation, retention, cataloguing, declassification, destruction,
digitisation and the search interface all produce historical visibility.
Fourth principle: archival bias is not evidentiary emptiness. A biased source may still provide evidence of administrative
language, fear, purpose, classification, conflict and unintended detail. Do not discard a source merely because it is biased; change
the question asked of it.
Fifth principle: genealogy is not a science of origins. Instead of a pure beginning, it studies contingent formation, rupture,
reclassification, appropriation and institutionalisation.
Sixth principle: genealogy is not opposed to truth. A category may have been historically constructed and still have real effects in
the present; “constructed” does not mean “imaginary”.
गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २
Seventh principle: continuity of a name does not prove continuity of meaning. Usage, semantic field, legal definition, social
practice and self-identification must be examined across time.
Eighth principle: absence of evidence and evidence of absence are different. Where a record would normally have been created
or preserved and is systematically absent, that absence may indicate something; but a silent archive does not automatically
establish a negative fact.
Ninth principle: the weight of evidence begins with provenance. Where did the source come from? Is it an original, copy or
edition? What is the chain of custody? How was its date established? Are there signs of alteration? These are the first questions.
Tenth principle: dates may be layered—date of event, date of writing, date of copying, date of publication and date of
digitisation. Collapsing them into one produces chronological error.
Eleventh principle: authorship may be probabilistic. Internal style, external attribution, manuscript colophon, citation history,
palaeography, material analysis and historical possibility may justify classifications such as confident, contested or unknown.
Twelfth principle: independent corroboration is not mere accumulation of numbers. Ten websites copying the same erroneous
book do not constitute ten pieces of evidence. Examine the dependency graph: are the sources genuinely independent, or are
they echoes of a single ancestral claim?
Thirteenth principle: seeking counter-evidence is obligatory. Selecting only sources that support one’s thesis is confirmation bias.
A serious answer to the strongest contrary source improves the quality of historiography.
Fourteenth principle: material evidence can stand as a counter-position to texts. Archaeology, numismatics, architecture,
landscape, palaeoclimate, material composition, GIS and genetics—within their proper limits—can support or challenge a textual
narrative.
Fifteenth principle: quantitative data are not neutral, but they are useful. Census, revenue, prices, mortality, migration,
landholding, elections and school enrolment must be read with changes in series definitions, coverage, missingness and category
shifts.
Sixteenth principle: oral history should not be dismissed merely as a “late source”. Its weight should be judged through interview
context, prompting, memory horizon, social desirability, repeated motifs, convergence across interviews, documentary anchors
and silence.
Seventeenth principle: an image or photograph is not automatically a transparent window. What lies outside the frame, staging,
caption, date, photographer, audience, editing and reuse must be examined. Image provenance is as necessary as textual
provenance.
Eighteenth principle: a map is both measurement and power. Boundaries, toponyms, scale, projection, survey purpose,
administrative classification and blank spaces make history; a map should not be treated as a complete representation of
territorial reality.
Nineteenth principle: genealogy or Panji is a relational source, not a census. It may provide information on marriage, lineage,
prestige, social memory and textual transmission; but coverage, later insertion, exclusion, status claims, copy-date and regional
variation must be examined.
Twentieth principle: colonial categories must not be retrojected into the past. A name, caste, tribe or language fixed by census or
ethnography is not thereby proved to have existed in the same sense earlier; equivalent terms in precolonial sources must be
examined independently.
Twenty-first principle: alongside the state archive, use family, monastery, temple, mosque, maṭha, school, merchant, village,
newspaper, literary, oral, archaeological and environmental archives. Multiple archives expose one another’s blind spots.
Twenty-second principle: marginal sources should not be reduced to token citation. They should change the architecture of the
narrative by bringing in histories of labour, women, Dalits and other disadvantaged groups, peasants, marginal languages,
displaced persons, children, the domestic economy, environment, body and everyday governance.
Twenty-third principle: declare the evidentiary level—direct or contemporary, near-contemporary, later tradition, secondary
analysis, conjecture. The reader should be able to see the level of confidence.
Twenty-fourth principle: do not hide disagreement. If scholars disagree about a date, author, place or event, state the principal
alternatives, their evidentiary bases and one’s own tentative judgement.
Twenty-fifth principle: negative language is a discipline. “There is no evidence yet”, “this source alone is insufficient”, “it is
possible”, “not decisive”—these are not signs of weakness but of epistemic honesty.
Twenty-sixth principle: translation is not merely secondary evidence; it is interpretation. Examine the original term, its
transliteration and semantic range, the translator’s choices and the risk of an anachronistic equivalent.
Twenty-seventh principle: citations should be traceable. Provide page, folio, inscription number, archive shelfmark, file ID or
version wherever possible, so that the reader can re-check the claim. “According to one source” is not scholarly evidence.
Twenty-eighth principle: a digital copy is not a complete substitute for the original. Resolution, colour, cropping, missing folios,
compression, metadata, OCR, scan order and interface problems must be checked; in critical cases, the physical object or an
authoritative facsimile may be necessary.
Twenty-ninth principle: an AI summary is not a source. AI search, translation and OCR can assist, but historical claims require
the underlying source. Because of the risks of hallucinated citations, conflated persons, invented dates and false quotations, the
provenance chain must remain under human supervision.
Thirtieth principle: the Parallel view of history is ultimately the historical form of multi-evidential rationalism—see the power of
the archive, seek traces from the margins, open categories through genealogy, but bind claims to objective resistance,
corroboration, counter-position and revision.
Principal Arguments
First argument—Unequal survival of traces. The past does not give every actor an equal archive. Royal and state orders may
survive in abundance while domestic labour leaves few records; educated elites write letters while illiterate workers may leave
oral traces. Source density is therefore not a direct measure of social importance.
Second argument—Administrative purpose of records. A revenue register may record landholding, productive units or taxable
categories, but not the full picture of family life, ecological use or customary rights. Reading a source against its original
administrative task can yield information “against the grain”.
Third argument—Classification performs reality. When an administration standardises a category through census, school, law,
police or land records, people may alter strategies and identities in response. Classification does not merely describe society; it
can generate social feedback.
Fourth argument—Genealogy reveals contingency. An institution that now appears inevitable may have emerged from
alternatives, conflicts, failed experiments and temporary compromises. Knowing this can reveal possibilities for reform in the
present.
Fifth argument—Critique of origin myths. The prestige of a community, nation or institution may rest on a sacred origin.
Genealogy examines the documentary history of the origin claim, later redactions, political use and competing memories. The
aim is not insult but historical differentiation.
Sixth argument—Corroboration. The self-interest of one source may be balanced by another independent source. Convergence
among a court record, newspaper, private letter, material trace and oral memory can increase confidence; divergence is itself
something to analyse.
गजेन्द्र ठाकु र
Seventh argument—Patterns of silence. The absence of one or two files may be accidental; systematic deletion, a missing series,
restricted catalogue, repeated euphemism or anonymous actors may indicate institutional silence. To use silence as evidence,
one needs a model of what records would ordinarily be expected.
Eighth argument—Microhistory. A small event, lawsuit, village, family, object or unusual document may reveal a wider system.
But the representativeness of the micro-example must be stated; an extraordinary case is not the average of an entire society.
Ninth argument—Serial history. Repeated records can show long-term patterns, but changes in forms or definitions may break
the series. Statistically smooth continuity does not prove that a historical category remained stable.
Tenth argument—Counterfactual restraint. “What if this had not happened?” can be an analytical tool, but without source-based
constraints it easily becomes fiction. The counterfactual space of possibility should be bounded by historically known
institutions, resources and actors’ options.
Eleventh argument—The plurality of memory-truth. Memory may err in factual chronology but still reveal histories of fear,
humiliation, belonging, narrative templates and later politics. Even “mistaken memory” is a real fact about memory culture.
Twelfth argument—Material resistance. A document may proclaim that a canal succeeded while sediment, an abandoned
channel, crop data and complaint files tell a different story. The material world can contest textual claims.
Thirteenth argument—The multilingual archive. The same event may be recorded under different names and genres in Sanskrit,
Maithili, Persian, Urdu, Hindi, English, Nepali or an administrative local language. Monolingual searching creates its own silence.
Fourteenth argument—Edition history. A modern printed edition of a famous work embodies editorial choices, manuscript
selection, punctuation, normalisation and omitted variants. Treating “printed” as identical to “original” erases textual history.
Fifteenth argument—Citation genealogy. A historiographical myth often spreads from one older secondary statement into many
books and websites. Tracing a claim backwards through its citations is an important modern form of genealogy.
Sixteenth argument—Archival access as power. Restricted catalogues, expensive travel, language barriers, fragile materials,
private custody and digitisation choices influence the agenda of scholarship. Open access can increase epistemic equality, but
must be balanced with privacy and cultural restrictions.
Seventeenth argument—Historical scale. One event can show different causalities at local, regional, imperial, ecological, familial
and gendered scales. A single-scale explanation oversimplifies; nested scales are a special strength of Parallel history.
Eighteenth argument—Causal pluralism. In history, one cause is rare. Institutions, economy, belief, ecology, technology,
contingency, individuals and networks interact. Causal chains should be made explicit in the evidence; chronology alone should
not be mistaken for causation.
Nineteenth argument—Revision is not relativism. A new source may alter an old conclusion; this revisability makes history
stronger, not weaker, much as revision strengthens science. A doctrine that never changes when evidence changes may be a
creed rather than historiography.
Twentieth argument—Public accountability. Because history is tied to political identity, the sources, uncertainty and method
behind claims should be public. Whether a popular narrative or an official textbook, no account is wholly exempt from scholarly
evidentiary standards.
Pūrvapakṣa
First pūrvapakṣa: the archive is made by power; therefore archive-based history merely reproduces power, and objective truth
is impossible.
Second pūrvapakṣa: genealogy shows every “truth” to be an effect of power; therefore no independent criterion of evidence
remains.
Third pūrvapakṣa: marginal voices may survive only in oral memory centuries later; demanding documentary corroboration is
an elite bias.
Fourth pūrvapakṣa: because victors write history, justice requires simply accepting the opposite account.
Fifth pūrvapakṣa: sources are always contradictory; in the end the historian’s ideology decides which narrative to choose.
Sixth pūrvapakṣa: the political needs of modern identity matter more than historical complexity; a simple origin myth may be
useful for communal unity.
Seventh pūrvapakṣa: a family genealogy or Panji is a sacred community record; external criticism is a cultural insult.
Eighth pūrvapakṣa: colonial records are detailed while local oral and textual sources are uneven; therefore it is practical to treat
the colonial archive as the principal truth.
Ninth pūrvapakṣa: digital search has democratised archives; catalogue bias and problems of access are now nearly over.
Tenth pūrvapakṣa: AI can read millions of documents and detect patterns; slow human source criticism will soon become
irrelevant.
Eleventh pūrvapakṣa: a broad historical insight should not be blocked merely because an exact date or author cannot be found;
inference is necessary.
Twelfth pūrvapakṣa: local exceptions and variant manuscripts are too much detail for national history; school and public
narratives should be simple.
Thirteenth pūrvapakṣa: material evidence is “hard fact” while text and memory are “soft”; archaeology is therefore
automatically more reliable.
Fourteenth pūrvapakṣa: statistics can eliminate subjective narrative; large datasets are better than small, biased sources.
Fifteenth pūrvapakṣa: historiographical disputes are endless; the public needs one authoritative version, not uncertainty.
Uttarapakṣa
Reply 1: the archive may be made by power, but it is not wholly arbitrary. Records from different centres of power, physical
traces, foreign and neighbouring archives, private papers, dates and material constraints challenge one another. Critique of the
archive extends the search for truth; it does not abandon it.
Reply 2: genealogy reveals the historical conditions of truth-claims; it does not logically prove that “all claims are false”.
Genealogical claims themselves depend on evidence, chronology, textual reading and counter-evidence. To avoid self-
contradiction, genealogy must therefore accept evidentiary standards.
Reply 3: corroboration does not mean making an oral source seek permission from an elite text. Independent oral accounts,
material landscapes, family objects, demographic patterns, local vocabulary and later institutional effects can also corroborate.
Criteria may vary with the source type; intellectual honesty does not.
Reply 4: “the opposite of the victor = truth” is mirror-image dogmatism. Oppressed actors too can have memory errors, factional
interests and later myths. Just history applies evidentiary standards to all actors while recognising unequal archival survival.
Reply 5: ideology influences the choice of questions and the language of interpretation, but it does not abolish evidentiary
constraint. Historians may interpret differently, yet impossible chronologies, forged documents, mistranslations and
contradicted data can still be detected. There are degrees of freedom in disagreement, not total freedom.
Reply 6: political usefulness is not a licence for historical falsification. A community whose dignity depends on myth remains
permanently fragile. Complex, layered history can mature identity; uncertainty is not an insult.
Reply 7: the cultural dignity of a community record should be respected; in scholarly use, however, copy-date, internal
chronology, variants, interpolation, coverage and social function should be examined. Criticism does not erase the record; it
enriches its historical meaning.
गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २
Reply 8: the density of the colonial archive is itself an effect of administrative power. Detailed records are valuable, but “detail =
truth” is invalid unless categories, incentives, translation, informants and survey purpose are examined. Local and vernacular
evidence provides correction.
Reply 9: digitisation adds a new layer of selection. What has been scanned becomes more visible, while what has not been
scanned may fall outside the search universe; OCR errors, metadata, ranking, paywalls, broken links and takedowns create new
silences. Digital abundance makes archival criticism more necessary, not less.
Reply 10: AI provides scale, not judgement. Entity matching, OCR and clustering are useful, but errors remain possible with
uncommon scripts, homonyms, sarcasm, damaged text, source dependence, forged citations and historical semantics. Human-
readable provenance and verification are indispensable.
Reply 11: a hypothesis is legitimate if it is marked as a hypothesis. The problem begins when possibility becomes fact under
narrative pressure. Disciplined history keeps conjecture, probability and confidence level distinct.
Reply 12: public narrative may be simple, but it must not be false. If the principal disputes within a complex issue are omitted,
readers receive false certainty. Even a short sentence such as “scholars disagree on this point” can preserve precision in a school
text.
Reply 13: material evidence also requires interpretation—stratigraphy, dating, context, contamination, reuse and sampling all
matter. The binary between “hard” and “soft” evidence is misleading; different kinds of evidence have different error modes.
Triangulation is stronger.
Reply 14: big data can magnify coverage bias. People outside the record system may be missing; categories may change;
duplicates and data-cleaning decisions matter. Numbers strengthen reasoning; they do not replace it.
Reply 15: public trust is created not by performing certainty but by transparent method. An authoritative institution can say,
“Confidence is high here; this point is disputed; this point needs further research.” That is epistemic maturity.
Method of Genealogy
The first stage of genealogy is to identify the present category—for example, “caste”, “language”, “tribe”, “boundary”, “crime”,
“normal”, “tradition”, “author”, “nation”, “religion”, or “female/male role”. Begin by asking: in which institution does this word
operate today, and with what meaning?
The second stage is to trace usage backwards. Before succumbing to the temptation to find the oldest occurrence, create clusters
of usage across different periods. Keep spelling, language, legal definition, self-use, outsider-use and administrative use distinct.
The third stage is to locate discontinuities. Which law, census, educational change, print culture, war, land system, religious
reform, technology, administrative reorganisation, market or migration changed the meaning of the category?
The fourth stage is to recover forgotten alternatives. Seek rival models, local customs, failed reforms and temporary
compromises to the arrangement that eventually prevailed. Genealogy breaks the illusion of inevitability.
The fifth stage is actor vocabulary. Use modern analytical categories when necessary, but quote and explain the historical actors’
own terms. Avoid retrospective certainty of the form, “What they really were was modern X.”
The sixth stage is change of scale. A category may appear stable at the state level but porous locally; household practice may
differ from official rule. Genealogy should work across nested scales.
The seventh stage is an institutional map. Who records the category—the school, court, revenue office, census, temple, guild,
clan, newspaper, university or archive? The form of recording can change the category’s social existence.
The eighth stage is an incentive map. What benefits or harms follow from accepting a category—employment, tax, marriage,
representation, land, legal protection, prestige or stigma? This helps explain the feedback effects of classification.
The ninth stage is material anchoring. How is a textual category embodied in settlement patterns, dress, architecture, ritual
objects, occupational tools or land use? Do not presume a direct equivalence in every case.
The tenth stage is citation genealogy. Trace a famous claim in modern scholarship backwards to its first identifiable source, and
determine whether later writers read the original or merely copied a secondary account.
The eleventh stage is to avoid overstating rupture. Genealogy seeks discontinuities but must also recognise evidence of
continuity. Sweeping claims such as “everything was created after 1871” must be rejected where earlier evidence exists.
The twelfth stage is to examine the social work of the origin myth. A myth may be factually weak yet still perform present
functions of identity, legitimacy, hierarchy and solidarity. The historian should write separately about factual truth and social
function.
The thirteenth stage is relation rather than essence. Instead of asking, “What is this community really?”, ask which relations,
institutions, boundaries, marriage practices, languages, laws, markets and memories have repeatedly reshaped it.
The fourteenth stage is self-genealogy. Historians should also examine the history of their own categories—terms such as
“medieval”, “classical”, “tribe”, “feudal”, “vernacular” and “religion”: where did these scholarly terms come from, and how well
do they fit the local case?
The Philosophy of the Archive
The archive should be thought of at three levels: the making of records, the survival of records and the finding of records.
Creation bias, preservation bias and retrieval bias are different. If one looks only at documents that have reached the reading
room, the first two levels disappear from view.
File structure is itself evidence. Note-sheets, enclosures, forwarding chains, marginalia, redactions, missing annexures and file
movement can reveal administrative decision-making. Reading only the final order is incomplete.
Catalogue description is a secondary layer. The archivist’s title, date range, subject keywords and collection boundaries shape
the scholar’s path of discovery. Faulty metadata can render a source practically invisible.
Collection provenance is necessary. Were private papers donated, purchased or seized? Did the donor select what to preserve?
Were sensitive letters removed? Was original order retained, or was the collection rearranged thematically? An archival
collection has a biography of its own.
The history of destroyed archives should also be written. Fire, war, humidity, insects, neglect, political destruction, privacy
purges and routine retention schedules may explain why a source series has vanished.
Privacy is part of archival ethics. Personal medical, sexual, family, caste-related, minor-related and victim records may possess
historical value, but possible harm to living persons or descendants, legal restrictions and informed access must be considered.
An open archive does not mean unrestricted exposure.
Differentiated access may also be appropriate for sacred or community knowledge. Scholarly curiosity is not automatically the
highest right, yet blanket secrecy can obstruct public history. Negotiated protocols, anonymisation, delayed release and
summary access are possible alternatives.
The colonial archive is a “state eye”, but one full of cracks. Petitions, vernacular annexures, witness testimony, complaints,
translation disputes and local reports can contest official categories from within. This is the basis of reading against the grain.
The judicial archive contains many kinds of actor speech, but court procedure reshapes speech: what questions were asked,
what was admissible, which words the interpreter chose, the conditions of confession, and the outcome on appeal all matter. A
transcript is not raw voice.
The newspaper archive is a trace of the public sphere, but editorial policy, ownership, censorship, circulation geography,
sensationalism, reprint networks and missing issues must be examined. One headline is not public opinion.
गजेन्द्र ठाकु र
A family archive can provide intimate sources—letters, diaries, photographs and account books—but survival is selective and
self-fashioning can be strong. “Private” does not guarantee “honest”.
An institutional minute book may display final consensus while dissent disappears into informal discussion. Drafts, agendas,
correspondence, resignation letters and press reports should be combined to reconstruct institutional politics.
Once an oral interview is recorded, a new record has been produced. The interviewer, question guide, camera, transcription,
translation, consent, editing and repository metadata all influence future interpretation. The production of the oral-history
archive is itself a documentable process.
Versioning is essential for digital archives. A webpage, social post or collaborative document can be updated or deleted; without
timestamp, hash, web capture, file metadata, platform context and reply chain, a later snapshot can mislead.
Avoid archive fetishism. Finding an unpublished document is not automatically a revolutionary discovery; significance depends
on comparison, authenticity, representativeness and relation to existing scholarship.
Evidence Criticism and Multi-Evidential Reason
The first question of evidence criticism is authenticity: is the object or document genuine, or a later copy or forgery? Examine
material, script, watermark, ink, paper, seal, file provenance, digital metadata and internal anachronism.
The second question is integrity: is the source complete or fragmentary? Are folios missing, photographs cropped, audio edited,
tables truncated or OCR pages skipped? Do not give an incomplete source the appearance of completeness.
The third question is contemporaneity: a source made at the time of an event and a narrative written centuries later have
different evidentiary roles. Later sources are not useless; they may be important for memory and the history of tradition.
The fourth question is competence: was the author in a position to observe the event? Is an administrator’s summary based on
hearsay? Is the chronicler distant? Even an eyewitness has a limited perspective.
The fifth question is incentive: praise, blame, legal defence, petition, tax reduction, royal legitimation, community prestige,
commercial promotion or propaganda—what stakes shaped the production of the source?
The sixth question is genre: inscriptional panegyric, court deposition, poem, satire, ritual manual, land deed, census or diary.
Genre expectations determine how literally a text should be read. Do not treat a poem as a survey report or a praise inscription
as neutral biography.
The seventh question is audience: a private letter, public proclamation, secret memorandum, legal petition and school textbook
are written for different audiences; self-presentation changes accordingly.
The eighth question is language: examine the contemporary meaning of key terms, polysemy, translation, scribal abbreviations
and technical legal or ritual usage. The first modern dictionary meaning is not necessarily the historical meaning.
The ninth question is transmission. In the chain original → copy → edition → translation → quotation → website, each stage can
introduce error. Where possible, go back to the earliest accessible witness.
The tenth question is independence: are the sources copying one another? A newspaper may reproduce an agency report;
chronicles may share a common source; websites may all repeat the same encyclopedia. Apparent convergence may simply be
dependency.
The eleventh question is scope: which region, community and period does the source cover? One village case does not represent
all of Mithila, one court file an entire community, or one inscription the daily life of a whole dynasty.
The twelfth question is missingness: who is systematically absent from the record system—women’s unpaid work, informal
labourers, nomads, the poor, children? The mechanism of missingness should be written into the narrative.
The thirteenth question is negative evidence: establish the institutional routine by which an expected record would normally be
created before inferring from its absence. “No mention has been found” may describe the limits of research, not historical
absence.
The fourteenth question is quantitative consistency: check totals, units, currency, calendars, boundary changes, population
denominators and price conversions. Numbers do not possess aesthetic authority merely by being numerical.
The fifteenth question is chronological fit: was the person alive? Did the institution exist? Was the title contemporary? Was the
technology available? Impossible chronology is a warning sign of forgery or misattribution.
The sixteenth question is geographical fit: account for shifts in place-names, same-name villages, changing boundaries, routes,
river courses and time-specific administrative districts, using maps appropriate to the period.
The seventeenth question is material-context fit: a text may describe a rich urban centre while archaeology appears sparse. That
is not immediately a contradiction; examine excavation coverage and genre, and record the tension.
The eighteenth question is the counter-source: seek the opposing party’s record, a neighbouring jurisdiction, a foreign traveller,
local oral memory, tax records or material traces as independent tests of the claim.
The nineteenth question is cumulative confidence: several medium-quality independent pieces of evidence can together yield
high confidence; one spectacular but dubious source cannot.
The twentieth question is the defeater: what new evidence would make you change your conclusion? Writing this in advance
prevents the historian from constructing an unfalsifiable narrative.
The twenty-first question is citation precision: give the exact location for a claim—page, folio, inscription, file number, interview
ID or timestamp—wherever possible. A bibliography alone is not enough.
The twenty-second question is quotation context: a ten-word quotation can convey the opposite meaning from the paragraph or
genre as a whole. Avoid quote-mining; if the text is translated, identify the translator or editor.
The twenty-third question is replication: can another researcher reach the same source and roughly reconstruct the reading?
Where an extraordinary claim depends on a private inaccessible source, a facsimile, transcription or independent witness may
be necessary.
The twenty-fourth question is uncertainty class: high, medium or low confidence; disputed, conjectural or unknown. Use
linguistic markers in the narrative. Not every sentence should speak with the same tone of certainty.
The twenty-fifth question is the revision log: in a digital scholarly project, preserve a changelog for corrections, new sources,
date revisions and merged or separated identities. Do not conceal revision.
Indian Philosophical Dialogue
Indian epistemology is not a direct precursor of modern historiography, but its fine-grained reflection on means of knowledge,
testimony, memory, inference, error and counter-position offers comparative vocabulary useful to historians. The comparison
must remain careful: identifying Navya-Nyāya with modern source criticism would be an oversimplification.
In the Nyāya tradition, questions about the validity of means of knowledge, the reliability of verbal cognition, the competence or
defects of speakers, memory and the conditions of inference can enter into dialogue with historical source evaluation. Calling a
historical document “verbal testimony” does not make it automatically valid; parallel questions about speaker, context and
transmission still arise.
Mīmāṃsā offers sophisticated hermeneutics concerning text, sentence, context, reconciliation of conflict and the meaning of
injunctions. For the historian, this encourages reading a text not as an isolated quotation but within its textual system, genre and
context. Yet religious textual authority and historical factuality have different aims.
Buddhist reflections on momentariness, causation and the critique of memory can question easy assumptions about continuity,
and thus enter into dialogue with genealogy’s critique of fixed essences in categories. But a philosophical ontology does not by
itself establish any particular historical claim.
गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २
Jain anekāntavāda offers an ethical and epistemic stimulus for recognising the partiality of perspectives: state, subaltern,
gendered, ecological and material evidence may reveal different aspects of history. But this does not mean that “all perspectives
are equal”; a discipline akin to conditional predication in syādvāda can qualify certainty.
The pūrvapakṣa–uttarapakṣa structure of Sanskrit commentary offers a parallel with historiographical review: present the
opposing position in its strongest form before answering it. Straw-man revisionism is not scholarship.
Indian inscriptions, copperplates, manuscripts, praśastis, grants, genealogies, poetry, Dharmaśāstra, travel accounts, Persian,
Arabic and Urdu chronicles, colonial records and vernacular print each demand a distinct evidentiary grammar.
A praśasti is a source for political self-representation; literal military and royal facts require corroboration. Hyperbole is not a
reason to discard it, because it also yields a history of ideals of kingship, vocabularies of patronage, genealogy and religious
legitimation.
Land grants and copperplates may provide data on land, beneficiaries, administrative titles, religious institutions and geography,
but forgery and later confirmation are possible. Palaeography, formulae, seals, findspot, plate sequence and parallel charters
should be examined.
In manuscript history, treating “one work = one text” is risky. Recensions, scribal correction, interpolation, regional copies,
commentary, marginal notes and the apparatus of modern editions all matter.
The arrival of print can change textual stability and standardisation. A previously fluid manuscript tradition may later acquire a
fixed author, title or orthography; the genealogy of print knowledge is therefore important.
Colonial survey, census and ethnography created an influential archive of India’s languages, castes, tribes and customary
categories. These records described social realities, but also produced feedback through administrative classification; continuity
from the precolonial period should not be assumed without independent evidence.
Nationalist historiography arose partly as a counter-position to colonial stereotypes, but it can itself create counter-myths of a
civilisational golden age or homogeneous community. Parallel history criticises both archives.
Subaltern Studies enlarged the place of peasant and ordinary actors’ agency, rumour, insurgency and fragments against elite
archives. The Parallel view retains this achievement while demanding transparent criteria for evidentiary weighting.
Feminist historiography changed the source base by bringing household, body, care, sexuality, law, kinship, labour and violence
into view. In official archives, women may appear only as “cases”; reading against the grain can reveal both actor experience
and structure.
Dalit and Bahujan history expands the archive through autobiography, oral history, movement literature, local print, legal
petitions, educational records and documentation of caste violence where official and elite textual archives are silent. Dignity-
based history neither fetishises victim testimony nor dismisses it.
Environmental history adds forest reports, rainfall, river maps, crops, settlements, animal records, satellite imagery and
landscape as a material archive, opening the limits of human-centred political chronicles.
Indian historiography offers influential examples such as D. D. Kosambi’s combination of materialism, archaeology, numismatics
and field observation, Romila Thapar’s source and genre criticism, and the fine evidentiary disciplines of epigraphy and
palaeography. Their interpretive disagreements do not diminish the methodological importance of these traditions.
The Indian formula of the Parallel view of history is: plurality of evidence + pūrvapakṣa + criticism of word and context + social
dignity + critique of the archive. Ancient philosophy should not be declared identical to modern historiography; it should be
treated as a comparative methodological resource.
The Parallel Perspective of Mithila, Vajji and Anga
The histories of Mithila, Vajji and Anga do not fit within a single modern administrative boundary. The India–Nepal border, river
valleys, linguistic regions, ancient and medieval political units, colonial districts and modern states all have different spatial
frames. The historian must first define the “region” in a time-specific way.
Panji and genealogical materials relating to Mithila can be important relational archives, but their present copy-date, branch,
interpolation, coverage, social function, exclusions and prestige claims must be examined. A later social category should not
automatically be projected backwards as medieval reality.
The problem of multiple persons or works sharing the same name demands special caution in regional intellectual history.
Avoid conflation through name-similarity; examine textual chronology, manuscript witnesses, patrons, places, citation chains,
language and style, together with independent genealogical evidence.
In the history of Maithili literature, manuscripts, oral performance, early print, journal publication, authorial recollection,
family archives, library catalogues and ISBN-era metadata constitute different evidentiary layers. When saying “first
publication”, specify whether one means print, composition, performance or manuscript date.
Without cross-script searching across Tirhuta, Devanagari, Kaithi and Persian–English administrative scripts, the archive may
remain incomplete. Where OCR is limited, palaeographic and manual reading are necessary; a change of script should not be
mistaken for cultural disappearance.
Archives in Nepal and India can supplement one another. Cross-border families, land, trade, intellectual exchange and political
movements may be incompletely documented in either country alone. Bilateral archival mapping is a fundamental method for
Parallel regional history.
For the ancient history of Vajji and Vaiśālī, literary, Buddhist and Jain, archaeological, inscriptional and later-memory sources
belong to different periods and genres. Modern democratic ideals should not be projected unchanged onto ancient gaṇa
institutions; institution-specific evidence must be examined.
For ancient and medieval claims concerning Anga and Campā, place-identification, river geography, archaeological context and
textual chronology must be kept distinct. Modern district names should not be treated as ancient territorial boundaries.
The history of rivers—the Kosi, Kamala, Bagmati and the Ganga system—requires a combined archive of maps, revenue surveys,
engineering reports, flood memory, settlement shifts, sediment and geomorphology. An administrative map alone cannot
represent the history of a dynamic river.
The history of earthquakes, floods, famines and epidemics requires different kinds of evidence: official casualty and revenue
reports, newspapers, relief files, diaries, oral memory, architecture and demographic data. Both numbers and experience are
necessary.
Colonial gazetteers are valuable syntheses, but not final authorities. Examine their sources, district officers’ reports,
ethnographic classifications, statistical years and copied legends; a later website repeating a gazetteer-derived claim should not
be counted as an independent source.
Land and revenue archives reveal property relations, but may under-record customary use, women’s claims, sharecropping,
commons and informal tenancy. Court cases, oral history, village records and field patterns should be added.
In histories of caste and community, modern association pamphlets, commemorative volumes, genealogies, censuses, colonial
ethnography, marriage networks, legal cases and local memory have different goals. Treat competition for prestige as part of
source context, not as an automatic enemy or friend of fact.
In linguistic history, survey labels, literary standards, census “mother tongue”, home speech, school language and political
demands are different. Numbers can mislead without a genealogy of the “language/dialect” category.
In the history of local religious sites, maṭhas, mosques and temples, inscriptions, endowments, architectural phases, oral legends,
pilgrimage print, land records, repair inscriptions and archaeological evidence may offer different chronologies. Sacred
narrative may be respected while historical dating is kept distinct.
गजेन्द्र ठाकु र
For women’s history, marriage songs, folktales, letters, legal cases, school registers, health and census records, memoirs,
photographs, textiles and crafts, and oral-history archives expand the evidentiary field. Name the silence of the elite male textual
archive as an evidentiary gap.
Migration history can combine passports and permits, railway records, recruitment, remittances, letters, phone records, oral
memory, associations, diaspora publications and social media to reveal continuity and discontinuity. Preservation of born-digital
sources is an immediate necessity.
A long-running e-journal and digital collection such as Videha is itself a future archive. If issue dates, stable URLs, file checksums
and versions, author metadata, correction notes, image provenance, audio/video transcripts, backups and redundancy are
preserved, born-digital Maithili intellectual history will become more verifiable.
The Parallel principle of regional history replaces one “grand narrative” with layered chronologies—political, social, linguistic,
intellectual, economic, ecological, gendered and local. But every layer remains bound to an evidence map; regional pride is no
exception to source criticism.
Digital Archives, OCR and AI
Digitisation improves preservation and access, but the quality of the digital surrogate should be documented. Resolution such as
300 or 600 dpi, colour targets, complete foliation, recto-verso order, scale, filename conventions, metadata and checksums form
part of the foundation for future scholarship.
OCR creates searchable corpora, but error rates can be high with uncommon scripts, old type, damaged pages, ligatures and
multilingual text. OCR text is a derivative layer; the facsimile should remain the authoritative visual witness.
Crowdsourced OCR correction can be useful, but version control, reviewer identity, confidence and links to the original image
are necessary. Silently “correcting” text can erase variant evidence for later researchers.
AI can propose hypotheses for transcription of damaged manuscripts, but output should not enter an edition without model
confidence, alternatives and palaeographic review. Critical reading remains indispensable.
AI entity resolution can wrongly merge different persons with the same name. Birth and death dates, place, title, patron,
relations, source IDs and uncertainty require manual adjudication; a false merge becomes a historiographical hallucination.
A knowledge graph is useful only if edge provenance is retained. For a relation such as “A influenced B”, what is the evidence—a
direct citation, a later scholar or stylistic inference? Relation type and confidence must be explicit. Graph visualisation should
not create an illusion of certainty.
Embedding-based and semantic search can improve discovery, but opaque ranking and majority-language bias may suppress
rare sources. Exact search, hierarchical browsing, faceted metadata and random or coverage sampling should remain available
alongside them.
In an archive mediated by AI summaries, source boundaries can disappear. A single summary sentence may combine multiple
documents; users should be able to see claim-level citations and links. “AI read 10,000 pages” is not a substitute for an evidence
trail.
Synthetic images and audio create two historical problems: contemporary misinformation and future authenticity. Where
possible, capture-device metadata, cryptographic provenance, trusted repositories and chain of custody should be preserved.
When a web archive captures a dynamic page, comments, embedded media, external links, login state and personalisation may
be missing even when a timestamp exists. The limitations of the captured page should be recorded in metadata.
Ethical social-media archiving requires caution. A public post may be historically relevant, but mass collection raises questions
of privacy, harassment, expectations of deletion, minors’ data and doxxing. Research access can therefore be tiered.
File-format obsolescence produces born-digital silence. Migration plans, open formats, checksum audits, redundant backups,
media refresh and emulation strategies should be part of institutional archive policy.
Digital preservation is not merely “backup”; it requires fixity + metadata + format management + access + governance +
geographic redundancy. Ransomware and credential loss are also archival risks.
Verification of AI-generated citations can be partly automated through DOI, ISBN and catalogue lookup, page-existence checks,
quotation search and date consistency. But absence may reflect limits of machine databases; a human escalation path should
remain.
Digital correction ethics requires versioned correction, errata and changelogs rather than quietly overwriting the original file.
The history of “how an error was corrected” can itself become future evidence.
Open access must be balanced with cultural rights. Releasing a high-resolution manuscript may be beneficial, yet donor
restrictions, sacred material, private letters, copyright and data concerning vulnerable persons require differentiated treatment
and proportionality to public interest.
AI should be treated not as “the historian” but as a historical assistant. Hypothesis generation, clustering, OCR, translation and
named-entity extraction can be accelerated; significance, ethical access, source weighting, causal inference, wording of
uncertainty and accountability remain responsibilities of human scholarly judgement.
Contemporary Applications
Scenario 1: a viral claim says, “This tradition has remained completely unchanged for a thousand years.” Examine the genealogy
of usage, the earliest securely dated source, intervening changes and modern standardisation. If there is continuity, describe it
specifically; remove the overclaim “unchanged”.
Scenario 2: a copperplate is found in a family archive. Test authenticity through metal, script, formula, seal, provenance and
expert comparison. Do not accept it merely because it supports family prestige, and do not discard it merely because suspicion
exists.
Scenario 3: a colonial gazetteer says, “Community X always did Y.” Add earlier vernacular and legal records, the contemporary
administrative motive, field variation and later self-description to test the word “always”.
Scenario 4: an oral witness recalls a riot or disaster eighty years later. Cross-check exact dates against newspapers and files; treat
sensory, emotional and local-sequence memory as a distinct kind of evidentiary value, and record whether leading questions
were used.
Scenario 5: two people have the same name and an AI system merges them into one biography. Use chronology, place, patron,
family, publications and manuscript evidence to split or merge them. If the matter remains unresolved, keep separate entities
with explicit uncertainty.
Scenario 6: twenty Wikipedia or website pages repeat the same unsourced date. Trace the citation genealogy backwards. If no
earliest authoritative source can be found, popularity is not evidence.
Scenario 7: an old photograph carries an incorrect modern caption. Reverse the provenance through the print’s verso, collection
context, photographer, architecture, clothing and event records; give a date range rather than false precision.
Scenario 8: a map claims a village boundary has been unchanged since 1900. Examine map edition, survey year, administrative
purpose, river shifts and later redrawing; state coordinate uncertainty when using a modern GIS overlay.
Scenario 9: one Panji copy contains a lineage and another does not. Examine the lineage of the copies, scribal hand, page
insertion, regional branch, date and independent marriage records; assign a confidence grade to the presence or absence.
Scenario 10: a school textbook quotes a national hero without a source page. Search for the exact text in the earliest edition. If it
is a paraphrase or translation, remove quotation marks or identify the source properly.
गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २
Scenario 11: an inscription is known only through a transcript in a later book because the original is lost. Examine the book’s
accuracy, any estampage or photograph, earlier readings and palaeographic notes; qualify the claim as “a lost inscription
reported in…” rather than presenting the inscription as directly extant.
Scenario 12: popular legend identifies an archaeological site with an exact ancient city. Test excavation, stratigraphy,
inscriptions, ceramic chronology, toponymy and travel distance. The legend is a source for cultural history, not sufficient
evidence for identification on its own.
Scenario 13: a census series from 1881 to 2011 appears to show category growth. Examine changes in census definitions,
boundaries, enumeration, language self-report, caste or category definitions and missing years; do not present a raw graph that
conceals these discontinuities.
Scenario 14: an archive file is missing. Check the record-retention schedule: was it routinely destroyed or is there a suspicious
gap? Search parallel office copies, index registers, dispatch numbers and court annexures. Do not write “the state destroyed the
evidence” without a basis.
Scenario 15: a private diary describes a secret plan contrary to official policy. Test diary date and authenticity, the author’s
access, corroborating minutes or letters, and subsequent outcomes. A private source is not automatically true.
Scenario 16: oral tradition assigns a famous poet to one village while a manuscript colophon points to a different region.
Examine the age and earliest record of the tradition, manuscript provenance and patronage network. Do not declare a definitive
birthplace if the conflict remains unresolved.
Scenario 17: OCR misreads a key word in a digitised manuscript, and a secondary article builds a thesis upon the error. Recheck
the image, issue a corrected transcription and correction to the article, and document the citation genealogy of the digital error.
Scenario 18: a social-media archive has been deleted. Use contemporaneous web captures or screenshot hashes, independent
captures, platform metadata and reply context. A screenshot alone is easily fabricated, so confidence should be graded
accordingly.
Scenario 19: a government press release calls a scheme “the first ever”, while archives, newspapers or departmental annual
reports show an earlier precursor. Define what “first” means—first name, scale, law or technology.
Scenario 20: a heritage campaign links a modern identity to an ancient term. Examine the earliest use of the term, semantic
change, institutional continuity and geographical scope. A legitimate modern identity need not depend upon a false claim of
unbroken ancient sameness.
Scenario 21: a translated chronicle uses the modern word “nation” for an original term that may mean kingdom, people or
realm. Provide a translation note, identify the original lexeme and context, and avoid anachronistic nationalisation.
Scenario 22: an AI-generated bibliography combines a real author with a fake title. Verify against library catalogues, DOI
systems, WorldCat or the author’s bibliography; remove unverifiable entries. Model output is not a citation.
Scenario 23: a community is offended by critical source analysis. Distinguish respectfully between cultural reverence and
historical dating; make method and uncertainty public, and state conclusions without insult or derision.
Scenario 24: a new document contradicts a historian’s cherished thesis. Do not suppress it. Test authenticity, revise the thesis,
and issue a changelog or addendum. Here the self-critique of Parallel Philosophy becomes practical.
Scenario 25: a museum label presents a disputed date as certain. Use compact uncertainty language such as “c.”, “traditionally
attributed”, or “scholars disagree”, and provide a QR-linked route to fuller evidence where appropriate.
Chapter Conclusion
The Parallel view of history changes the centre of history without abolishing the standard of truth. Alongside state archives, it
adds Panji, letters, oral memory, material traces, language, ecology, women’s and disadvantaged experience, and digital records
—but it tests the error modes of every source.
Genealogy does not seek the “true origin”; it reveals the formation, rupture, institutionalisation, forgotten alternatives and
feedback of present categories. A historically constructed reality is not imaginary; it is contingent and real.
The archive begins before the archive room: which records were created, which survived, which were catalogued, and which
were digitised. Silence may be the result of this entire process.
Silence matters, but it is not permission to fill a gap with a desired story. Disciplined inference rests on expected-record patterns,
parallel sources, material traces, convergence of oral evidence and explicit uncertainty.
The moral force of histories from the margins becomes scholarly force when they discover neglected sources, change the
questions, break elite categories, and also submit their own claims to counter-position and evidence.
In the digital and AI age, archives are more searchable but also more fragile and mediated. OCR error, metadata, ranking,
synthetic media, link rot, version overwrite and AI hallucination are new subjects of archival criticism.
For the histories of Mithila, Vajji and Anga, cross-border archives, multilingual scripts, river and land records, Panji,
manuscripts, print, oral memory and born-digital collections such as Videha supplement one another. No single family of
sources can carry the entire history of the region.
The formula of Parallel history is this: power makes archives, but archives are challenged by evidence, counter-positions and
objective remains. History is neither a pure mirror nor arbitrary fiction; it is constrained reconstruction.
The final discipline is simple: where evidence about a person, date, text, lineage or event is not decisive, do not manufacture a
story. Write clearly, “unknown at present” or “disputed”. Respect for the limits of evidence is a fundamental ethic of
historiography.