{ "title": "COMPASS codebook", "version": "0.2", "date": "2026-10-02", "status": "Expanded research protocol; pilot validation is planned for project months 7–12.", "scope": "One framework for tracing the translation of geographical knowledge across connected maps, inscriptions, images, objects and historical episodes.", "units": [ { "name": "Census witness", "rule": "One identifiable surviving map in a specified manuscript, edition or state. Record holding, qualifying cross-cultural operation and evidence. Families organize linked witnesses and do not inflate the count." }, { "name": "Linguistic tier", "rule": "One inscription or explicitly defined component. Record language, script, reading, source form and mediating form where supported." }, { "name": "Visual tier", "rule": "One graphic element, with composite and component records linked. Establish its meaning within the relevant convention profile." }, { "name": "Material tier", "rule": "One map object linked to other states through evidenced production, revision or use episodes." }, { "name": "Power relation", "rule": "One actor-action relation or condition in a historical episode. This relational layer connects to the three tiers; it does not assign power to whole traditions." } ], "coding_rules": [ "Establish census eligibility before intensive coding. Baseline maps can inform convention profiles but remain outside the cross-cultural map totals.", "Separate visible observation, proposed interpretation, source relationship and historical consequence. Each can have a different evidence grade.", "Use a common vocabulary with tier-specific applicability. TRL is linguistic; a graphic label is coded as an inscription.", "Assign one primary operation to a defined feature and dimension. Add linked assignments only when they express distinct operations; record alternatives as alternatives.", "Record partial illegibility, missing images and unresolved identity explicitly. Empty fields never mean absence.", "Keep a witnessed source relationship distinct from a family resemblance, a candidate intermediary and a contextual parallel.", "Never infer a power relation from an operation alone. A changed name does not identify the person who changed it or their authority.", "Aggregate by stated unit and denominator. Report assessed records, unknowns, source coverage and family dependence; distinguish maps, elements and episodes." ], "operation_codes": [ { "code": "RET", "name": "Retention", "tiers": [ "L", "V" ], "definition": "An identified feature is carried into a new mapping context without change in the dimension being assessed: wording, meaning, shape or another explicitly defined attribute.", "include": "Match the source and target feature and state what persists. In the linguistic tier, distinguish an unchanged form from a change of script.", "exclude": "Shared appearance without a source relationship is correspondence, not yet retention. Retention alone does not make a baseline map cross-cultural.", "evidence": "Source and target locators; evidence for their relationship; the particular dimension retained.", "co_codes": "L-TRL or V-ADP can coexist only when they describe different dimensions or stages, each with its own assignment.", "example": "Decision example: an outline, sign or name kept from an identified source without a change of script, meaning or graphic convention may receive RET. Resemblance to an unidentified exemplar does not establish retention." }, { "code": "TRL", "name": "Transliteration", "tiers": [ "L" ], "definition": "A name or expression passes between writing systems through a rendering of its sound or written form.", "include": "Record both strings in their original scripts, a reading, language and script, and any evidenced intermediary. Distinguish historical spelling from modern normalization.", "exclude": "Use TRA for transfer of meaning. A name inside a graphic symbol is still a linguistic element; V-TRL is not used.", "evidence": "Paired inscriptions with exact locators; a defensible reading of each form; evidence for the transmission link.", "co_codes": "TRA may apply to a translated component of a mixed name. Split the components instead of assigning incompatible codes to one undivided string.", "example": "Decision example: a source name rendered by correspondence of written or sound forms may receive L-TRL once both readings and their relationship are supported. A change of script alone is insufficient. On Ricci’s world map of 1602, America rendered as 亞墨利加 (Yamolijia) is L-TRL." }, { "code": "TRA", "name": "Translation", "tiers": [ "L", "V" ], "definition": "Meaning is rendered through another language or an explicitly corresponding graphic convention.", "include": "For L-TRA, record source and target meanings. For V-TRA, establish that different signs communicate the same function or category in their respective convention profiles.", "exclude": "A change in appearance alone is ADP. A different referent is SUB. A resemblance between signs does not establish equivalent meaning.", "evidence": "Paired elements, contextual readings and the relevant convention profiles; a reason to identify semantic correspondence.", "co_codes": "TRL or ADP may describe a separate component or dimension, with linked assignments explaining the distinction.", "example": "Decision example: a generic place-name component rendered for its meaning, such as Latin insula by a word meaning island, receives L-TRA, while a phonetic proper-name component in the same label receives L-TRL." }, { "code": "SUB", "name": "Substitution", "tiers": [ "L", "V" ], "definition": "A source name, referent or element is replaced by a different one without preserving its semantic identity.", "include": "Identify what was replaced, what took its place and the relevant location or function.", "exclude": "Use TRA when different conventions express the same meaning, and ADP when the inherited feature persists in an altered form. Do not infer suppression from replacement alone.", "evidence": "A bounded source-target pairing and an argument for replacement rather than an unrelated addition.", "co_codes": "Use a separate OMI or INV only if a distinct operation is evidenced; do not count the two sides of the same substitution twice.", "example": "Decision example: replacement of one named authority by another in the same cartouche can be coded SUB when the source relationship is established." }, { "code": "ADP", "name": "Adaptation", "tiers": [ "L", "V" ], "definition": "An inherited element is altered to fit a receiving language, convention, audience or use while retaining its referent or function.", "include": "State the inherited feature, the alteration and the convention or use that makes the change intelligible.", "exclude": "Difference alone does not establish adaptation to a local convention. Use ND if the receiving convention cannot be established.", "evidence": "Source and target locators, identified continuity, and a convention profile or contextual source for the proposed adjustment.", "co_codes": "RET or TRA can apply to another dimension. ASM belongs to the enclosing composite, not automatically to each adapted feature.", "example": "Decision example: a confirmed inherited sign redrawn to fit an evidenced receiving convention may receive V-ADP. A change of page orientation alone does not establish that convention or justify this code. On the Teozacoalco map of 1580 (AM-024), horseshoe tracks join the footprints marking roads, a visual adaptation of an Indigenous cartographic convention." }, { "code": "ASM", "name": "Assemblage", "tiers": [ "L", "V" ], "definition": "Elements from two or more identified source contexts are brought into one inscription, graphic unit or composition.", "include": "Define the composite unit and link each contributing element to a source or explicitly delimited source tradition.", "exclude": "Complexity, mixed appearance or several languages alone do not establish assemblage. Do not invent direct ancestors for contextual parallels.", "evidence": "At least two supported contributions with their locators; separate grades for each link.", "co_codes": "Component records may carry RET, TRL, TRA, SUB or ADP. Preserve the parent-child link so the composite is not counted as an extra independent feature.", "example": "Decision example: a map combining a traceable coastal outline with a separately sourced inland network may receive V-ASM at composition level; each component keeps its own source link and grade." }, { "code": "INV", "name": "Innovation", "tiers": [ "L", "V" ], "definition": "An element is introduced relative to a specified source set, with no counterpart in that bounded set.", "include": "Name the source set examined, the addition and the search boundary. Explain why adaptation or substitution is insufficient.", "exclude": "This code does not mean first invention in world history. An unidentified source or unreadable source area cannot establish innovation.", "evidence": "A defined, inspected source set; target locator; coverage sufficient to support the absence of a counterpart.", "co_codes": "ASM may describe the enclosing composition. Do not code the same change as both INV and ADP without distinguishing dimensions.", "example": "Decision example: a newly introduced legend may receive L-INV only relative to an identified, sufficiently examined source set. An unidentified immediate exemplar cannot support an unrestricted novelty claim." }, { "code": "OMI", "name": "Omission", "tiers": [ "L", "V" ], "definition": "An element present in an identified source is absent from the corresponding, assessable part of the target.", "include": "Locate the source feature and the target area where it would be expected. Check cropping, damage, scale and selection of the surviving image.", "exclude": "A blank, damaged or unphotographed area does not establish omission. OMI does not by itself establish censorship, refusal or intent.", "evidence": "An identified source relationship and sufficient target coverage; an explicit account of relevant material loss.", "co_codes": "A power assignment such as P-SEL needs separate evidence for an actor and decision. Avoid double counting a replacement already coded SUB.", "example": "Decision example: a source legend missing from an intact corresponding target panel can receive OMI; a missing photograph of that panel cannot." }, { "code": "ND", "name": "Not determinable", "tiers": [ "L", "V" ], "definition": "The evidence has been examined but cannot support a stable operation assignment for this unit.", "include": "State the question examined, the evidence consulted, the unresolved alternatives and what would resolve them.", "exclude": "ND does not mean no translation. Use not_assessed for work not yet undertaken, and proposed for a candidate awaiting review.", "evidence": "An assessment record describing the examined evidence and the reason determination failed.", "co_codes": "Do not attach ND and a settled operation to the same question. Different features or separately defined dimensions can have different statuses.", "example": "Decision example: an examined inscription whose operation cannot be resolved after source checking may receive L-ND. This differs from a material or power relationship that has not yet been assessed." } ], "material_codes": [ { "code": "M-REP", "name": "Reproduction", "definition": "A new physical witness reproduces a mapped representation.", "include": "Identify the two witnesses and evidence of copying, printing or another reproductive process.", "exclude": "Similarity alone does not establish direct copying. A digital photograph is a documentation object, not automatically another historical map.", "evidence": "Source and result witness IDs, locators, dates or ranges, production evidence and link status.", "co_codes": "Localization or transformation may be recorded for a distinct aspect, with an explanation.", "example": "Decision example: a manuscript map engraved for print may receive M-REP only when the drawing used and the printed state are identified. Similarity of outline alone does not establish copying." }, { "code": "M-LOC", "name": "Localization", "definition": "A map is given a new material setting or use through translation, annotation, layout, binding or related changes.", "include": "Specify the change, its audience or setting where evidenced, and the earlier state used.", "exclude": "A modern collection location or a different language alone does not establish a historical act of localization.", "evidence": "Witness records and evidence for the changed setting or use; keep presumed purpose separate from documented reception.", "co_codes": "Reproduction may describe manufacture of the new witness; linguistic and visual assignments describe its features.", "example": "Decision example: an identified map rebound and annotated for a documented new use can carry M-LOC for that episode." }, { "code": "M-TRF", "name": "Transformation", "definition": "The map changes format, medium or organization in a way that changes how it can function.", "include": "Describe the before-and-after forms and the specific functional consequence supported by evidence.", "exclude": "Do not infer a changed function merely from a new owner or minor physical variation.", "evidence": "Source/result descriptions, material inspection or catalogue evidence, and a separately supported account of function.", "co_codes": "Use alongside M-REP or M-LOC only when each names a distinct, evidenced aspect of the transition.", "example": "Decision example: transfer from a bound sequence to a wall display requires evidence for both the new physical format and the proposed use." } ], "power_rules": [ "A single episode can contain several relations involving different actors.", "Record actor, action, affected actor or resource, episode and source in the actor and setting fields. Do not assign power relations to whole cultures.", "A colonial setting does not itself establish coercion; collaboration does not establish equality or consent.", "Separate the existence of a relation, its proposed effect on a translation operation, and its subsequent outcome. Each claim requires its own evidence.", "Distinguish contemporary evidence from later testimony or reconstruction. Do not project later use back into production.", "Each power assertion has its own evidence grade. A map’s census admission grade does not establish its power relations.", "Use these questions to interpret documented episodes. They do not form a separate code family or a numerical power score." ], "required_fields": [ { "field": "map_or_illustration_id", "description": "Census witness ID, or a clearly separate illustration ID when the physical witness is not yet identified." }, { "field": "element_id / parent_id", "description": "Stable inscription, graphic element or composite identifier; preserve component relationships." }, { "field": "locator", "description": "Folio, sheet and image region. Store image identity and coordinate system so a crop can be found again." }, { "field": "observation", "description": "What can be read or seen, before the interpretation. Preserve original script and mark uncertain readings." }, { "field": "source_links", "description": "Candidate or established source objects, their locators and relation status. A contextual parallel is not a direct ancestor." }, { "field": "convention_profile_id", "description": "Versioned account of the conventions needed to interpret the element, with its sources and exceptions." }, { "field": "assessment_id / target_id", "description": "One scholarly assertion about one target and defined dimension. Link revisions rather than overwriting them." }, { "field": "assigned_code / candidate_code", "description": "Record a settled assignment separately from a proposed alternative; never count a candidate as verified coding." }, { "field": "assessment_status", "description": "not_assessed, proposed, assessed or adjudicated. Record disagreement and resolution separately." }, { "field": "evidence / grade", "description": "Bibliographic or archival source, page/folio/image region, relevant basis and the CROSS evidence grade for this claim." }, { "field": "reasoning / alternatives", "description": "Explain the inference, its boundaries, and any live competing interpretation." }, { "field": "coder / date / version", "description": "Researcher ID, assessment date and codebook version. Codes are specialist assignments; record any machine-assisted transcription or discovery input, tool/version and expert verification separately." }, { "field": "episode / actors / outcomes", "description": "For historical actions, identify phase, participants and affected actors or resources. Grade relation, effect and outcome separately." } ], "evidence_grades": { "documented": "An identifiable source explicitly supports the particular relation or action.", "derived": "The relation follows from linked documentary or object evidence through an explicit, checkable reconstruction.", "inferred": "Indirect evidence supports a plausible interpretation; alternative explanations remain material.", "uncertain": "The available evidence is ambiguous, conflicting or insufficient to sustain a stable interpretation." }, "assessment_states": { "not_assessed": "The relevant assessment has not been undertaken. No positive or negative finding follows.", "proposed": "A candidate interpretation has been recorded for investigation or specialist review.", "assessed": "A researcher has examined the cited evidence and recorded a supported assignment or explicit ND.", "adjudicated": "A recorded disagreement has been reviewed; preserve both original assignments and the reason for the resolution." }, "relation_states": { "established": "The recorded relationship is supported at its stated grade.", "candidate": "A relationship to test.", "contextual_parallel": "Relevant correspondence without an asserted line of descent.", "not_determinable": "Examined evidence does not resolve the relationship." }, "workflow": [ { "step": "Identify and admit", "description": "Locate the surviving witness and record the operation that qualifies it for the census. Record family leads separately. The current census contains 107 admitted maps; the project database target is more than 500." }, { "step": "Select and contextualize", "description": "Build the intensive corpus of at least 70 map families. Set the corpus at month 18. Assemble source packets and versioned convention profiles; keep baseline material outside the counted corpus." }, { "step": "Pilot and revise", "description": "During months 7–12, independently double-code 24 maps. Review the actor, setting and episode fields on 12–15 episodes within this pilot. Preserve disagreements, revise rules and release COMPASS 1.0 at month 12." }, { "step": "Code and connect", "description": "For each selected map, record inscriptions, graphic elements and material relationships. Link source objects, historical episodes and modern assignments through stable identifiers. Full coding runs in months 13–36." }, { "step": "Review", "description": "Double-code one fifth of the maps in the intensive corpus, including the 24 pilot maps. Report agreement by dimension and retain adjudication history." }, { "step": "Analyze and release", "description": "Trace routes and transformations across the linked records. Release versioned data, codebook and source locators, subject to image rights and repository conditions. Report documentary gaps with each analysis." } ], "power_episode_template": { "episode_id": "Stable identifier for the encounter or phase.", "map_id": "Identifier of the particular map witness.", "episode_date": "Date or range, with uncertainty and dating source.", "stage": "Production, revision, use or circulation; multiple stages require separate episodes when relations change.", "actors": "Identified people, groups or institutions and their evidenced roles. Record an actor as unknown when attribution is unresolved.", "assertions": [ { "assertion_id": "Stable identifier.", "assessment_status": "not_assessed | proposed | assessed | adjudicated", "relation_status": "present | not_determinable | evidence_against after assessed or adjudicated; otherwise null", "claim_scope": "relation | encounter_condition | observed_attribution", "claim": "Concise description of what is established or remains unresolved.", "actor_from": "Actor exercising or initiating the action; unknown is permitted.", "action": "The particular evidenced action or condition.", "actor_to_or_resource": "Affected actor, map, information, access, recognition or claim.", "direction": "Identify who acts upon whom or what. Use undetermined when evidence cannot establish direction.", "compass_operation_link": "Optional identifier of the linguistic, visual or material operation affected. Do not infer a causal link from co-occurrence.", "evidence": [ { "source": "Bibliographic or archival reference and URL where available.", "locator": "Page, folio, inscription, passage or image feature.", "source_date": "Date of the evidence, distinguished from the episode date.", "basis": "Brief paraphrase of the supporting or contrary evidence." } ], "evidence_grade": "documented | derived | inferred | uncertain; null when no relation claim can be assessed", "reasoning": "Explain how the evidence supports the assertion, including any inferential steps.", "alternative_interpretation": "Relevant competing explanation or uncertainty.", "outcomes": [ { "status": "claimed | observed | unknown", "description": "A specific asserted, evidenced or unresolved result, without a ranked scale.", "source_and_locator": "Required for a claimed or observed result.", "evidence_grade": "Grade this outcome separately from the relation." } ], "contextual_dimension": "A descriptive topic from actor_setting_questions where useful. Record the historical action in ordinary language." } ], "place": "Place of the episode, with uncertainty and source.", "coding_provenance": { "coder": "Name or stable researcher ID.", "coded_on": "Date of assessment.", "provenance": "Specialist assignment. If machine assistance supplied a transcription or discovery input, record the input, tool, version and expert verification separately; machines do not assign accepted COMPASS or power codes.", "review_status": "Draft, independently coded, or adjudicated; retain original assignments and the resolution separately.", "review_notes": "Preserve disagreements and revisions." } }, "crm_mapping_file": "cidoc_crm_mapping.json", "sources": [ { "title": "CROSS Part B2: COMPASS methodology and work plan", "type": "project_protocol" }, { "title": "CIDOC CRM 7.1.3 official definition", "url": "https://cidoc-crm.org/html/cidoc_crm_v7.1.3.html" } ], "material_assessment_rule": "When evidence has been examined but a material relationship cannot be resolved, set assessment_status to assessed, relation_status to not_determinable and assigned_code to null. Material relationship types remain M-REP, M-LOC and M-TRF; uncertainty is a separate status, not a fourth relationship type.", "machine_assistance": "OCR/HTR, vision-language models (VLMs), image retrieval and other methods can propose transcriptions, feature locations, place matches and source leads. The choice of method remains open. Machine outputs are inputs for review; specialists determine witness identity, historical relationships and COMPASS codes. Accepted records enter a structured database with explicit identifiers, controlled values and deterministic validation. Unverified suggestions remain separate from accepted data.", "acronym_expansion": "Coding of Map-translation Practices Across Scripts and Substrates", "example_revision_date": "2026-10-09", "example_revision_note": "2026-10-09: examples drawn from the China study replaced by decision examples; the study is withdrawn from this release pending specialist verification.", "actor_setting_questions": [ { "name": "Commissioning and direction", "question": "Who initiated, funded or prescribed the work?", "inclusion_evidence": "A commission, payment, instruction, correspondence or explicit contemporary attribution identifies the initiating or directing actor and the work concerned.", "interpretation_limits": "Funding does not establish control over every decision. Record instruction, funding and commissioning as distinct actions where the evidence permits." }, { "name": "Constraint and coercion", "question": "Whose freedom to participate, leave or withhold knowledge was restricted?", "inclusion_evidence": "Attested captivity, forced removal, threats, compulsory service or comparable restrictions connected to the encounter.", "interpretation_limits": "Distinguish coercive circumstances from evidence that drawing or disclosure itself was forced. A colonial setting, employment or unequal status alone does not prove coercion." }, { "name": "Selection and revision", "question": "Who could accept, reject, alter or retain geographical information?", "inclusion_evidence": "Instructions, documented revisions, editorial exchanges or other evidence assign a selection or revision decision to an actor.", "interpretation_limits": "Translation and intermediary roles alone do not establish decision-making authority. An observed difference between maps does not identify who caused it or why." }, { "name": "Authorisation and validation", "question": "Who could authenticate the map or make it admissible?", "inclusion_evidence": "Signatures, certificates, inspections, privileges, official acceptance or rejection establish a validating act or authority.", "interpretation_limits": "A dedication, seal or royal emblem alone does not establish official authorization. Distinguish authentication, permission to publish and acceptance as evidence." }, { "name": "Attribution and recognition", "question": "Whose contribution was acknowledged, claimed, reassigned or suppressed?", "inclusion_evidence": "Inscriptions, paratexts or comparisons between attributable versions document how a contribution was credited or represented.", "interpretation_limits": "Record visible credit separately from intentional suppression. Missing names do not prove erasure. The actor assigning credit may remain unknown; interpretation as an exercise of power requires a further argument." }, { "name": "Access and circulation", "question": "Who could obtain, retain, copy, distribute or restrict the map or its information?", "inclusion_evidence": "Custody records, permissions, bans, secrecy instructions, copying restrictions or conditions of exchange establish access or circulation practices.", "interpretation_limits": "Present-day ownership does not establish early modern control. Distinguish physical custody, access to information and permission to reproduce." }, { "name": "Claims and entitlement", "question": "Whose rights or jurisdiction did the map assert or contest?", "inclusion_evidence": "A petition, litigation file, treaty, explicit inscription or documented use connects the map to a specific claim or counterclaim.", "interpretation_limits": "A boundary, projection or territorial label alone does not prove a political purpose. Record the claimant, intended audience and object of the claim; do not assume the claim succeeded." }, { "name": "Negotiation and refusal", "question": "Who imposed conditions, challenged a demand, withheld information or secured a concession?", "inclusion_evidence": "Recorded bargaining, refusal, counterclaims, conditions of participation or corroborated withholding establish a particular negotiating act.", "interpretation_limits": "Participation does not establish consent, reciprocity or equality. Omission from a map does not establish deliberate withholding." } ], "contextual_fields_revision": "2026-10-05: actor, setting and evidence questions aligned with the project description.", "workflow_revision": "2026-10-08: intensive corpus stated as at least 70 map families, as in the proposal." }