{
 "version": 1,
 "researched": "2026-10-01",
 "overview": [
  {
   "id": "H001",
   "heading": "The city is a palimpsest, not a stack of political eras",
   "text": "Carthage should be read as a changing distribution of settlements, public works, burial grounds, water systems and shoreline facilities. “Punic”, “Roman”, “Vandal” and “Byzantine” are useful historical labels, but they do not describe one uniform layer spread across the peninsula. Construction can remove earlier deposits, bury them in levelling fills, reuse foundations or redirect circulation. Excavation then creates another visible arrangement: low walls from different phases may stand together because their overlying deposits were removed at different times. The present view therefore records ancient actions, preservation, excavation and presentation. The four photographed areas reveal different parts of that process, while the other atlas sites supply comparisons rather than additional photographed proof.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 94–97"
    },
    {
     "source": "mills2013",
     "locator": "; printed pp. 11–12, 16–17"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H002",
   "heading": "Foundation tradition and early settlement",
   "text": "The Dido/Elissa narrative comes through mediated literary witnesses, including Justin’s epitome of Pompeius Trogus and accounts preserved by later authors. The conventional 814/813 BCE foundation belongs to that literary chronology. It must not become the exact excavation date of a wall, a harbour or an urn deposit. Hoyos discusses both the tradition and archaeological arguments for settlement around the late ninth or early eighth century. Early ceramics, inscriptions and radiocarbon results have different dating problems: an heirloom can be older than its burial, and a dated bone is meaningful only within a controlled context. The atlas therefore distinguishes a traditional foundation marker from an approximate early-settlement interval. Neither identifies a surviving founder’s house.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 2–3, 8"
    },
    {
     "source": "docter2009",
     "locator": "; printed pp. 180–181, table and note 12"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H003",
   "heading": "146 BCE and the uneven beginning of Roman Carthage",
   "text": "Rome destroyed the Punic city in 146 BCE, but a historical event is not a universal archaeological horizon. Burnt material may have been cleared, redeposited or incorporated in later construction. Nor did the defeat erase Punic-speaking communities elsewhere in North Africa. The Gracchan colonial project of 122 BCE, Caesarian initiative and Augustan expansion are separate episodes. Hoyos places an existing colony by 40 BCE while discussing uncertainty about its exact beginning and Octavian’s supplement of settlers in 29 BCE. A date slider should therefore show an interval for refoundation and a distinct marker for later colonial development, instead of one switch that instantly converts every district into a Roman plan.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 88–91"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, BM00S2 redeposited Punic destruction material"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H004",
   "heading": "The imperial city joined separate urban systems",
   "text": "Roman Carthage’s grid, civic centre, residential quarters, performance buildings, harbours and water supply formed connected systems. Byrsa’s enormous platform changed the hill; the villas used the street grid on the eastern slopes; the theatre organized access and spectators; the cisterns, aqueduct and baths managed water and fuel. A map explains their distribution, but chronological agreement does not establish physical contact or one building campaign. A reservoir could supply a later installation, and a bath’s service floor could remain useful after its upper halls failed. Spatial analysis should connect infrastructure only where a publication documents the link. The present map’s anchor distances are orientation aids; they cannot determine the slope or hydraulic gradient of an ancient conduit.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 94–97"
    },
    {
     "source": "lezine1968",
     "locator": "; printed pp. 5–6"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H005",
   "heading": "Late antiquity: change, continuity and reuse",
   "text": "The Vandal capture of 439 CE and Byzantine reconquest of 533 CE are political markers. They must not supply dates for every collapsed vault or modified doorway. Published accounts document repairs, churches, industrial activity, burials and altered circulation across different districts. Lézine explicitly disputed a simple Vandal destruction explanation for the Antonine Baths. Mills distinguishes roof collapse, disturbed material and reused bricks at Bir Massouda; Potthoff links Christian burial to particular basilicas and commemorative practices. These are local sequences with their own evidence. An object’s material, a religious interpretation and the date of its final deposition should remain separate records, particularly when an older stone or tile was reused.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 118, 130–133"
    },
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–73"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12"
    },
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H006",
   "heading": "698 CE was not the disappearance of all activity",
   "text": "Hoyos distinguishes the final loss of Byzantine Carthage in 698 from the survival of buildings, scattered residents and later rural production. He also stresses the distance between later conquest narratives and contemporary observation. Medieval quarrying and the movement of architectural materials changed the city’s fabric over centuries. A reused column at another site has a different custody history from a column still standing in its original position. Modern excavation and conservation further determine what can be seen. The chronology accordingly ends neither in a blank after 698 nor in an assertion of uninterrupted urban life: it separates political decline, residual habitation, spoliation, investigation and the dated photographic visit.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 141–142"
    },
    {
     "source": "lezine1968",
     "locator": "; printed p. 73"
    }
   ],
   "status": "Sourced synthesis",
   "limitation": ""
  },
  {
   "id": "H007",
   "heading": "What this collection actually covers",
   "text": "All 264 supplied photographs were screened on six labelled contact sheets. That screening establishes subject categories and a visit-level reading, not a complete building catalogue. The existing camera-based grouping is 57 harbour, 53 villas, 41 theatre, 112 Byrsa and one unassigned image; these counts include signs, displayed objects and modern scenes. Selected images were then inspected at larger resolution, including the Byrsa orientation board and the MAGON wine panel. Camera coordinates locate the camera, not necessarily the subject. The EXIF dates read 18 January 2025; timezone and accuracy are unverified. A directory named “2026 carthage” is not a capture date. Every photograph retains an audit row, and no image is treated as evidence of a deposit that its view does not show.",
   "refs": [
    {
     "source": "photoaudit",
     "locator": "Six contact sheets, photo metadata and selected full-resolution checks"
    }
   ],
   "status": "Collection audit",
   "limitation": ""
  },
  {
   "id": "H008",
   "heading": "Coverage and historiography",
   "text": "This extension uses targeted readings from eight publication and a bibliographic check of a ninth, within the existing bounded source bibliography. It does not claim to have read every book. Older architectural studies remain valuable for observations now difficult to reproduce, but their reconstructions and datings need to remain attributable. Institutional project accounts provide orientation and excavation history; they do not replace published context records. Publication plates have been cited rather than copied into the application. The new figures are original drawings from cited numerical records and phase statements. Source disagreements are retained alongside the visualization so that a clear graphic cannot turn an uncertain argument into an apparent fact.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 3, 70–73"
    },
    {
     "source": "mills2013",
     "locator": "; printed pp. 12, 17"
    },
    {
     "source": "khader",
     "locator": "; title and first American edition 1987",
     "locator_status": "Published page locator not supplied; use the named work, chapter or ancient passage."
    }
   ],
   "status": "Research coverage audit",
   "limitation": ""
  }
 ],
 "sites": {
  "harbour": [
   {
    "id": "H101",
    "heading": "Two harbours, several histories",
    "text": "The circular basin and central island are among the collection’s most recognizable landscapes. Their form makes the late Punic naval interpretation intelligible, but the present waterline, visible piers and paths are not a preserved photograph of the fleet base in 146 BCE. Hurst treats the Punic arsenal, subsequent harbour development and Roman island buildings as separate phases. Ancient literary description, excavated structures and a museum model are different witnesses: their agreements can strengthen an interpretation, while their differences need to remain inspectable. In the user photographs, water and modern shore buildings provide orientation; excavated masonry records surviving fabric; museum miniatures present proposed completions of missing buildings.",
    "refs": [
     {
      "source": "hurst2010",
      "locator": "; printed pp. 53, 55–56"
     },
     {
      "source": "photoaudit",
      "locator": "IMG_1930, IMG_1948, IMG_3951, IMG_3955–3960"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H102",
    "heading": "Roman rebuilding did not preserve the naval base unchanged",
    "text": "The Roman island complex belongs to a different system of use. Hurst discusses a later second-century colonnaded setting, a temple and an octagonal building, while examining the port’s wider commercial and state-linked activities. The association with Rome’s food supply is a historical interpretation requiring comparison with warehouses, cargo, texts and the wider harbour, not a function visible in one column base. The atlas keeps alternative emphases together rather than labelling the entire circular harbour “the annona port”. Dredging, shoreline movement and successive construction also mean that a modern basin outline cannot delimit each ancient phase. A phase-specific plan would require the publication’s plans to be georeferenced and its proposed geometry to be distinguished from excavation measurements.",
    "refs": [
     {
      "source": "hurst2010",
      "locator": "; printed pp. 53, 55–56, 65"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H103",
    "heading": "Read masonry and models at different scales",
    "text": "IMG_3951 shows wall faces, a lowered enclosure, upright supports, wet ground and a circular feature in the background. It does not by itself establish the enclosure’s function or the dates of its repairs. IMG_3955–3957 and IMG_3958–3960 show two museum-model groups with strikingly different architectural arrangements. Comparing them is useful precisely because they are interpretive objects: roofs, complete colonnades and shipsheds can be supplied where the excavated record is incomplete. No dimensional measurement should be transferred from the photograph of a model to a ruin without the model’s scale and source documentation. A useful annotation first describes the visible joint or surface, then proposes a phase with a publication locator and an explicit degree of uncertainty.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Full-resolution IMG_3951; model groups IMG_3955–3960"
     },
     {
      "source": "hurst2010",
      "locator": "; printed pp. 55–56"
     }
    ],
    "status": "Image observation and qualified interpretation",
    "limitation": ""
   },
   {
    "id": "H104",
    "heading": "What a harbour stratigraphy view should show",
    "text": "A harbour sequence needs dredging cuts, basin deposits, construction fills, foundations, floors and later robbing as separate entities. Waterborne sediments and disturbed material should not inherit the date of the oldest object in them. Hurst’s coastline discussion shows why the marine setting requires its own evidence rather than a modern shoreline overlay. For this collection the implemented harbour graphic is an ordinal reading of published phase groups. It supplies no buried sediment thicknesses, water depths or direct contact lines between unrecorded contexts. Future work should start with the excavation section, the individual context identifiers and the datum used in the published survey. Those controls can then support a measured profile and a more defensible map of changing harbour geometry.",
    "refs": [
     {
      "source": "hurst2010",
      "locator": "; printed p. 65"
     },
     {
      "source": "mills2013",
      "locator": "; printed p. 17"
     }
    ],
    "status": "Source-based visualization proposal",
    "limitation": ""
   }
  ],
  "villas": [
   {
    "id": "H201",
    "heading": "A burial landscape became a residential district",
    "text": "The official site account places a Punic necropolis of the third and second centuries BCE beneath the later residential area. It then distinguishes early Roman occupation aligned with the urban grid from subsequent affluent houses. Hoyos describes the residential remains near the Odeon and a bird-decorated mosaic in a comparatively well-preserved house. This is district history, not a universal stratigraphic column: tombs need not lie directly under every photographed room, and mosaic floors can belong to successive refurbishments. The reader therefore separates the necropolis, street organization, domestic construction and later changes. The site’s conventional name “Roman Villas” should not make every exposed feature Roman or every house one contemporary building.",
    "refs": [
     {
      "source": "villas",
      "locator": "Overview, first occupation and residential development"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed p. 97"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H202",
    "heading": "The photographs record rooms, floors and water installations",
    "text": "IMG_1962–1996 and IMG_3964–3979 include wall runs, paved or tessellated floors, openings into underground spaces, circular mouths and the paved route through the park. A circular opening may suggest a water installation, but its original use and alterations require a matched plan or excavation record. Similarly, a visible mosaic does not reveal its bedding sequence or prove that it remains in its original setting. Close views are especially useful for distinguishing what is present—different block sizes, mortar, plaster edges and broken flooring—from the interpretation of those traces. The contact-sheet assessment supports the visited area; it does not automatically identify each room as part of the House of the Aviary or supply a date for its repair.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Contact sheets 1–2 and 4; IMG_1962–1996, IMG_3964–3979"
     },
     {
      "source": "mills2013",
      "locator": "; printed pp. 16–17"
     }
    ],
    "status": "Image observation and qualified interpretation",
    "limitation": ""
   },
   {
    "id": "H203",
    "heading": "The MAGON panel is a warning about names",
    "text": "IMG_1987 is a modern MAGON wine-interpretation panel. Its text discusses grape treading and pressing; the mosaic illustration is captioned as from El Jem and the museum of Sousse. The panel’s presence in the villas visit does not establish a winepress at the camera position, nor does the word MAGON identify the excavated Magon Quarter. The panel is evidence of modern interpretation and of the images chosen to explain ancient production. It should be linked to its subject and illustrated object separately from its installation location. This distinction is important for any collaborative system: a visitor’s helpful contextual image can enrich a site record without proving that the depicted artefact was found there.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Full-resolution IMG_1987, text and illustration caption"
     }
    ],
    "status": "Direct image observation",
    "limitation": ""
   },
   {
    "id": "H204",
    "heading": "A domestic sequence needs interfaces, not just a floor plan",
    "text": "For a house, the most informative relationships are often a doorway blocked by later masonry, a floor running against a wall, a drain cut through an earlier surface or a new mosaic above earlier bedding. A plan represents positions; a Harris matrix represents the order of recorded events; a chronology assigns dates to those events. The collection supplies oblique views suitable for identifying research questions and candidate interfaces. It does not supply all three records. The current sequence consequently shows district-level stages with their sources. A building-level reconstruction should first match the photograph to a numbered plan, record the surviving versus restored fabric, and distinguish a secure physical relation from a proposed functional explanation.",
    "refs": [
     {
      "source": "mills2013",
      "locator": "; printed pp. 16–17"
     },
     {
      "source": "harrismatrix",
      "locator": "About the Matrix, sections and surfaces"
     }
    ],
    "status": "Source-based visualization proposal",
    "limitation": ""
   }
  ],
  "theatre": [
   {
    "id": "H301",
    "heading": "A performance building with a modern performance history",
    "text": "The theatre is a semicircular monument and should be kept distinct from Carthage’s oval amphitheatre and its nearby Odeon. Hoyos proposes an Augustan origin and later alterations; the surviving form and modern use do not themselves settle its earliest phase. The official agency stresses how little ancient fabric survived relative to the whole monument and records adaptation for performances from 1906. This is central to the user photographs: the seating bowl, stage equipment, railings and access routes form a functioning modern venue in an archaeological setting. A learner needs to know which elements are documented ancient survivals, which were rebuilt and which were added for present-day use. An attractive complete outline can otherwise conceal the loss of original material.",
    "refs": [
     {
      "source": "hoyos2021",
      "locator": "; printed pp. 96–97"
     },
     {
      "source": "theatre",
      "locator": "Overview: limited ancient remains and 1906 adaptation"
     },
     {
      "source": "photoaudit",
      "locator": "IMG_3985–4006"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H302",
    "heading": "Seats and joints are observable; ancient phases need controls",
    "text": "IMG_3999 and IMG_4001–4003 show joints, different stone surfaces and damaged seating blocks. Such details may guide a fabric survey, but wear, colour and a cement-like joint are not sufficient dating criteria. A block can be ancient and reset in a modern restoration; a repair can preserve the line of an ancient seat without preserving its original surface. IMG_3985–3995 also show the relationship between the seating and modern performance infrastructure. The most useful initial annotation states the visible condition and identifies a repeatable region. A proposed ancient or modern date should then cite a conservation record, dated excavation photograph or published plan. The system leaves unknown repairs unresolved instead of colouring every seat according to a presumed period.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Contact sheets 4–5; close seating views IMG_3999, IMG_4001–4003"
     },
     {
      "source": "theatre",
      "locator": "Overview: scope of ancient surviving structures"
     }
    ],
    "status": "Image observation and qualified interpretation",
    "limitation": ""
   },
   {
    "id": "H303",
    "heading": "The displayed objects have separate biographies",
    "text": "The theatre visit includes sculpture and sarcophagus-relief photographs, notably IMG_4007–4014. Their current display near a monument is not proof that they formed part of its original decoration. A sarcophagus has a production date, burial context, excavation history and current custody; these may be quite different from the date of the wall behind it. The reliefs can support a study of visible imagery, but identification of a specific myth, deceased person or workshop requires object records. The reader treats these as displayed objects with unresolved original provenance. Their discussion belongs beside the theatre’s history, while their archaeological origin remains an independent question.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Contact sheet 5, IMG_4007–4014"
     },
     {
      "source": "potthoff2017",
      "locator": "; printed pp. 174–176, burial and object associations"
     }
    ],
    "status": "Image observation and research question",
    "limitation": ""
   },
   {
    "id": "H304",
    "heading": "An accessible scan is a lead, not an ancient reconstruction",
    "text": "Open Heritage 3D lists terrestrial LiDAR for dataset qkhk-s475, collected in 2012 and published in 2020, with a CC BY-NC-SA label and a rating stating that it lacks georeferencing. Its title says Amphitheatre, while its description says Theater and refers to festivals. Viewer navigation timed out during this investigation, so the monument identity and viewer operation remain unresolved. The dataset is retained as a research lead, not placed at an atlas anchor. Even after its identity is checked, a scan represents recorded surfaces at the acquisition date. It cannot recover the Roman stage building or separate restoration phases without additional documentation.",
    "refs": [
     {
      "source": "oh3d",
      "locator": "Dataset attributes, site description, collection date, license and reuse score"
     }
    ],
    "status": "Audited research lead",
    "limitation": "No scan downloaded; no GIS placement or reconstructed ancient mesh is asserted."
   }
  ],
  "byrsa": [
   {
    "id": "H401",
    "heading": "One hill contains several differently preserved histories",
    "text": "Byrsa joins a powerful foundation tradition to an archaeological history of burial, industry, housing and monumental rebuilding. The French mission’s account identifies archaic graves, metalworking and a Punic housing development destroyed in 146 BCE, as well as later Roman remains. That combination does not turn the whole hill into one exposed Punic citadel. Its summit and slopes were transformed differently. The Roman colony used the hill as the centre of its urban grid, and a great platform supported a later civic and religious ensemble. Archaeologists could recover some earlier slope structures beneath levelling while summit structures had been removed. The contrast between preserved low rooms and massive later supports is therefore more instructive than a simple “oldest layer at the bottom” slogan.",
    "refs": [
     {
      "source": "efrbyrsa",
      "locator": "Research account: Punic finds and Roman platform"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed pp. 94–96"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H402",
    "heading": "Read the Punic quarter against the Roman platform",
    "text": "The close Byrsa photographs, especially IMG_4027–4064, reveal low wall networks, upright dressed stones, rubble infill, plaster-like red surfaces and larger projecting masses. Those observations are useful, but no date is inferred merely from a masonry style. The published interpretation of the preserved Punic slope quarter and later Roman levelling provides the historical framework. A defensible overlay would first match each visible wall to a numbered excavation plan, identify restoration and test whether the asserted relation occurs at that exact location. Until then, the graphic represents phase groups. The loss of a summit building and the burial of a slope house are different processes, even if both accompany the same redevelopment.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Contact sheets 5–6; full-resolution IMG_4029"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed p. 95"
     }
    ],
    "status": "Image observation and source-based synthesis",
    "limitation": ""
   },
   {
    "id": "H403",
    "heading": "Signs, a reconstructed view and a modern cathedral",
    "text": "IMG_4021 contains an orientation plan, the Dido foundation story and a reconstruction credited to J.-C. Golvin. It labels some building locations as probable. These distinctions matter: the reconstruction combines evidence with completion, while the plan also guides a modern visitor among ruins, museum space and the cathedral. The broad paved summit and cathedral views in IMG_2048–2059 record the present monumental landscape, not the surviving floor of one precisely identified Roman temple. The institutional Byrsa page gives an 1846 cathedral date that conflicts with Hoyos’s late-nineteenth-century account; this reader adopts neither that date nor an unsupported date for every modern surface. Published excavation and restoration records must mediate the comparison between the current view and ancient functions.",
    "refs": [
     {
      "source": "photoaudit",
      "locator": "Full-resolution IMG_4021, board text and reconstruction credit"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed p. 95"
     },
     {
      "source": "efrbyrsa",
      "locator": "Research account, cathedral-date inconsistency"
     }
    ],
    "status": "Image reading and source audit",
    "limitation": ""
   },
   {
    "id": "H404",
    "heading": "A city model must preserve removed and missing evidence",
    "text": "A Byrsa visualization should distinguish at least three states: excavated survival, published inference and a proposed reconstruction. The platform can be discussed as a transformation without inventing the height of every earlier house or the complete appearance of Eshmun’s sanctuary. The name of a temple supplied by literary tradition cannot substitute for a securely identified foundation. Likewise, a Roman pier visible today may have lost its relationship to surrounding fills after excavation. The implemented ordinal sequence explains burial, levelling and later monumental construction, while the image cards let the user inspect the evidence and annotate questions. A future section should use measured levels and original plan coordinates rather than deriving depth from perspective in a tourist photograph.",
    "refs": [
     {
      "source": "hoyos2021",
      "locator": "; printed pp. 2–3, 94–96"
     },
     {
      "source": "lezine1968",
      "locator": "; printed p. 3, plan precision warning"
     }
    ],
    "status": "Source-based visualization proposal",
    "limitation": ""
   }
  ],
  "tophet": [
   {
    "id": "H501",
    "heading": "A deposit sequence and a cause-of-death debate",
    "text": "The Tophet must be approached through urn contexts, burnt remains, associated monuments and excavation records. Harden’s accessible extract describes the urn series from the eighth to second centuries BCE; its original article is dated 1937, despite a 2016 online publication stamp. The broad sequence is useful, but the familiar Tanit I–III labels require an excavation-specific crosswalk before their boundaries are visualized as layers. Hoyos surveys the interpretive dispute, while Smith and Schwartz represent opposing arguments about age estimation and sacrifice. An urn’s stratigraphic position, the age of the cremated individual and the cause of death are three questions. A photograph of a stela does not settle them. No supplied photograph is securely assigned to this sanctuary in the present grouping.",
    "refs": [
     {
      "source": "harden1937",
      "locator": "Accessible extract and bibliographic record"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed pp. 17–18"
     },
     {
      "source": "smith2013",
      "locator": "Published sacrificial interpretation"
     },
     {
      "source": "schwartz2017",
      "locator": "Methodological challenge"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H502",
    "heading": "What to visualize before choosing an interpretation",
    "text": "Start with the excavation sector, urn number, deposit relation, associated pottery and recorded position of the stela. Plot chronological constraints separately from interpretations of ritual. Keep reused or displaced monuments distinct from monuments securely associated with an urn. If several osteological analyses concern the same sample, they should appear as competing analyses of that sample rather than independent archaeological discoveries. The current reader supplies a broad sanctuary chronology and a discussion of the dispute; it does not fabricate an urn-by-urn matrix or transplant dates from a generic Tanit scheme onto unrecorded material.",
    "refs": [
     {
      "source": "harden1937",
      "locator": "Extract and footnotes on stratigraphy and disused stelae"
     },
     {
      "source": "schwartz2017",
      "locator": "Age estimation and inference"
     }
    ],
    "status": "Source-based visualization proposal",
    "limitation": ""
   }
  ],
  "baths": [
   {
    "id": "H601",
    "heading": "Service architecture survives where the grand halls do not",
    "text": "Lézine explains the architectural response to the coastal location: the service level had to remain above the sea, with public bathing rooms above and differing access levels around the building. This is a functional arrangement within a construction, not an older Punic stratum beneath a Roman bath. The official agency gives inauguration in 157 or 159 CE and completion in 162; Lézine’s historical summary also records a construction date of 145 and restoration in 389. These dates refer to different operations and should be stored separately. No supplied image is securely assigned to the bath complex, so the visualization is publication-based and explicitly separated from the four photographed visit areas.",
    "refs": [
     {
      "source": "lezine1968",
      "locator": "; printed pp. 5–6, 71"
     },
     {
      "source": "baths",
      "locator": "Overview, inauguration and completion"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H602",
    "heading": "Room 18 supports a real, limited column",
    "text": "On printed page 70, Lézine records Room 18 from the floor datum: ash and charcoal to +0.62 m, red clay to +0.84 m, earth with charcoal streaks to +0.95 m, a compact charcoal layer at +1.06 m, and rubble with windblown fills above. The original drawing in this reader scales only the intervals whose limits are reported. It shows the +1.06 m layer as a horizon because its thickness is not given, and leaves the intervening description and upper extent open. The datum is the room’s radier, not sea level. The column is an account of one historical excavation record; it cannot be extended across the entire baths or attached to a modern camera point.",
    "refs": [
     {
      "source": "lezine1968",
      "locator": "; printed p. 70, Salle 18; visually checked"
     }
    ],
    "status": "Published deposit description",
    "limitation": "No modern survey, context IDs or absolute deposit dates are supplied."
   },
   {
    "id": "H603",
    "heading": "Collapse dates require an argument, not a political label",
    "text": "Lézine rejected a simple deliberate Vandal demolition and argued for a sequence of structural failure and later use. His narrower argument uses the restoration inscription of 389 and a coin deposit whose latest coin is dated 425. The official overview instead places the frigidarium collapse between 388 and 392. Those accounts cannot be silently merged into one exact date. Moreover, a coin’s mint date normally supplies a terminus post quem for its deposition: it is not automatically an ante quem for a collapse. Lézine’s inference must therefore remain attributable and its premises inspectable. The current chronology displays both reported readings as alternatives and uses neither to date the Room 18 deposits automatically.",
    "refs": [
     {
      "source": "lezine1968",
      "locator": "; printed pp. 71–72"
     },
     {
      "source": "baths",
      "locator": "Overview, frigidarium collapse"
     }
    ],
    "status": "Competing interpretations and dating audit",
    "limitation": ""
   },
   {
    "id": "H604",
    "heading": "Later use is part of the bath history",
    "text": "The recorded deposits and altered stairs explain why the baths cannot be reduced to construction followed by sudden abandonment. Lézine distinguishes service use, changes after the great cold hall failed, sixth-century reorganization and later exploitation of the ruins. The agency likewise describes partial use and quarrying while admitting that the exact final abandonment date is unknown. A building model should allow room-level phases and gaps: an upper vault can collapse while a lower space remains useful. Measured geometry could eventually show this, but the present publication-based column and ordinal history already communicate the distinction without inventing a complete three-dimensional bath complex.",
    "refs": [
     {
      "source": "lezine1968",
      "locator": "; printed pp. 70–73"
     },
     {
      "source": "baths",
      "locator": "Overview, partial use and unknown desertion date"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ],
  "magon": [
   {
    "id": "H701",
    "heading": "Housing, shoreline fills and redevelopment",
    "text": "The official account describes a residential district built from the fifth century BCE on older fills near the shore, followed by major redevelopment from the early second century. Smaller houses were combined into larger, decorated dwellings, and reclaimed ground changed the relation between streets and the shore. These are processes to visualize—fill, subdivision, amalgamation and changed access—rather than a single unchanging plan. The district’s modern name does not demonstrate residence by an individual named Magon. It is also unrelated to the installation location of the MAGON wine panel in IMG_1987. No supplied photograph has been securely matched to the Magon excavation plan. The reader consequently offers a comparative site history, with ancient geometry left unmeasured.",
    "refs": [
     {
      "source": "magon",
      "locator": "Overview, fills and second-century redevelopment"
     },
     {
      "source": "photoaudit",
      "locator": "IMG_1987 interpretation-panel identity"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ],
  "massouda": [
   {
    "id": "H801",
    "heading": "Several investigations occupy one modern area",
    "text": "Bir Massouda is an especially strong example of why a site outline should not be treated as one trench. The project account distinguishes earlier residential excavations, a Christian basilica investigation and the later Phoenician-Punic programme. Mills separates BM00S1, the basilica area, from BM00S2, a set of eight sondages intended to reach earlier deposits below the Roman grid. Kaufman’s metallurgical argument concerns excavated residues and analyses, not resemblance between a surface feature and a furnace. These records can share a district while retaining different context systems, research questions and dates. No photograph in the supplied collection is securely matched to a Bir Massouda trench.",
    "refs": [
     {
      "source": "massoudaproject",
      "locator": "Project account: investigations and area"
     },
     {
      "source": "mills2013",
      "locator": "; printed p. 12, BM00S1 and BM00S2"
     },
     {
      "source": "kaufman2016",
      "locator": "; printed pp. 33–35"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   },
   {
    "id": "H802",
    "heading": "An ordinal relation and a named legacy chronology",
    "text": "The project’s image caption explicitly places a Late Punic floor above a Late Punic cistern in Trench 2 and describes Roman-to-Early-Modern layers above them. That is a usable published physical relation, although it is not a full context matrix. Mills’s Table 1.3 supplies a different product: an initial BM00S2 phase scheme. Its Archaic and Late Archaic ranges overlap; its “Vandal” phase begins around 400, before the political capture of 439; and its later bands leave a gap. The table is reproduced faithfully, labelled as the author’s scheme, beside the city chronology. These irregularities are meaningful audit questions and must not be repaired silently to make the graphic neater.",
    "refs": [
     {
      "source": "massoudaproject",
      "locator": "Composite image caption, Trench 2"
     },
     {
      "source": "mills2013",
      "locator": "; printed p. 12, Table 1.3"
     }
    ],
    "status": "Published relation and legacy scheme audit",
    "limitation": ""
   },
   {
    "id": "H803",
    "heading": "Residual material can reverse a naïve visual reading",
    "text": "Mills records primary roof collapse, material disturbed by robbing and earlier bricks incorporated as makeup at BM00S1. At BM00S2 he describes redeposited material from Punic destruction alongside later structures and deposits. An old tile in a later fill therefore does not turn that fill into an early occupation surface. His methodological account distinguishes contexts, subgroups, groups and phases, and warns about residuality. This hierarchy is the right model for the application: photograph regions point to visible material, context records describe deposition, relations establish order, and dated evidence constrains events. The measured heights of the Antonine Baths cannot be imported into Bir Massouda merely because both belong to Carthage.",
    "refs": [
     {
      "source": "mills2013",
      "locator": "; printed pp. 12, 16–17"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ],
  "didon": [
   {
    "id": "H901",
    "heading": "A long urban sequence with limited reconstructions",
    "text": "DAI’s project at Rue Ibn Chabâat documents a central quarter investigated from Punic occupation through Roman construction to a Byzantine rotunda. The Roman public building’s function remains uncertain; a cult interpretation for the rotunda is probable rather than certain. The project also distinguishes original masonry from a reconstructed corner in its presentation. The present atlas uses an approximate street anchor, not a surveyed park footprint, and no supplied image has been identified as a Didon view. The ordinal diagram therefore explains the published succession while carrying neither measured depths nor a complete plan. The rotunda here must remain distinct from the separate underground rotunda at Damous el Karita.",
    "refs": [
     {
      "source": "didon",
      "locator": "Research account and original/reconstructed fabric captions"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ],
  "malga": [
   {
    "id": "H1001",
    "heading": "Water infrastructure has its own chronology",
    "text": "Hoyos discusses La Malga’s great reservoirs and their connection to the Zaghouan aqueduct from around 160 CE, while qualifying their earlier supply by rain or springs. Storage capacity, construction date and the beginning of an aqueduct connection are different attributes. The cisterns should also be distinguished from the Borj Jedid reservoirs discussed in relation to the Antonine Baths. A hydraulic visualization needs measured conduit levels and gradients; distances between modern atlas anchors are insufficient. The current reader supplies the historical infrastructure connection and records the absence of securely matched supplied photographs. It does not interpolate an ancient water route or infer that every cistern was built in one campaign.",
    "refs": [
     {
      "source": "hoyos2021",
      "locator": "; printed p. 96"
     }
    ],
    "status": "Qualified infrastructure synthesis",
    "limitation": ""
   }
  ],
  "amphitheatre": [
   {
    "id": "H1101",
    "heading": "Separate the oval arena from the photographed theatre",
    "text": "The official agency describes an initial first-century phase, a second-century enlargement and later reoccupation, including a probable defensive use. Hoyos instead proposes an Augustan or soon-after beginning. These datings remain named alternatives. The oval arena is not the semicircular theatre photographed in this collection. The nineteenth-century chapel and column commemorate Perpetua and her companions; a commemorative location is not independent excavation proof of the exact ancient execution place. The reader keeps the modern commemoration as its own stage and does not identify the sarcophagi displayed during the theatre visit as amphitheatre finds. The Open Heritage 3D record’s mixed terminology is also retained as an unresolved external-resource issue.",
    "refs": [
     {
      "source": "amphagency",
      "locator": "Overview, construction phases, reoccupation and 1893 commemoration"
     },
     {
      "source": "hoyos2021",
      "locator": "; printed p. 96"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ],
  "damous": [
   {
    "id": "H1201",
    "heading": "A cemetery basilica changed through several building stages",
    "text": "Potthoff describes Damous el Karita as a cemetery basilica on earlier Roman burial grounds, probably first built at the end of the fourth century. He reports a three-stage construction interpretation ending in the later Byzantine period. A trefoil chapel, clustered burials and graffiti belong to the argument for veneration near a holy grave; the saint is not securely named. The nearby subterranean rotunda has been read as a burial monument, a baptistery or a shrine to Julian of Antioch. These alternatives should be displayed together, and the rotunda must not be confused with Didon’s. No supplied image is securely assigned to this basilica. The implemented succession is a source-derived site synthesis, not a fabricated context matrix.",
    "refs": [
     {
      "source": "potthoff2017",
      "locator": "; printed pp. 174–176"
     }
    ],
    "status": "Sourced synthesis",
    "limitation": ""
   }
  ]
 },
 "methods": [
  {
   "id": "M001",
   "heading": "Choose the graphic by the question",
   "text": "Use a chronology to ask when an event or phase is dated; a measured profile to ask which deposits were recorded at a particular vertical position; a relation graph to ask what physically precedes what; and a phase plan to ask how spaces changed. A political timeline is not a Harris matrix. Equal-height boxes in an ordinal graph carry no thickness, and proximity on a map carries no stratigraphic relation. The present reader implements three distinct products: city/site chronology, published phase-group relations, and a column drawn from Room 18’s recorded local heights. None supplies an unrecorded whole-city cross-section.",
   "refs": [
    {
     "source": "harrismatrix",
     "locator": "Sections versus sequence"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 17"
    }
   ],
   "status": "Implementation design grounded in method",
   "limitation": ""
  },
  {
   "id": "M002",
   "heading": "Build the archaeological record before the matrix",
   "text": "For future excavation-level work, record each deposit, cut, surface and standing structural unit with its original context identifier. Store physical relations as directed edges from later to earlier, with a relation type, source locator and review state. Keep inferred relations separate from recorded contacts. Detect self-links, cycles and incompatible equalities before rendering; reduce redundant edges only in the displayed matrix, preserving the original relation table. A gap should remain a gap. The phase graphs in this release use named groups because the consulted summaries do not supply the necessary complete context tables. Their app identifiers are navigation keys and are not ancient excavation context numbers.",
   "refs": [
    {
     "source": "harrismatrix",
     "locator": "About the Matrix"
    },
    {
     "source": "hematrix",
     "locator": "Greater transparency and reinterpretation"
    }
   ],
   "status": "Project data-design proposal",
   "limitation": ""
  },
  {
   "id": "M003",
   "heading": "Dates constrain events; they do not replace relations",
   "text": "A context should retain the dated object or sample, dating method, association, residuality risk and lower or upper constraint. A coin minted in 425 cannot make its deposition occur before 425; an inherited object can substantially predate burial. Calibrated radiocarbon ranges should retain their probability level and calibration information, while typological dates should name the reference scheme. Bayesian modelling is feasible only after the sampled events and their relative order are controlled; no posterior distributions are computed here. The current chart distinguishes dated events, source ranges, approximate display windows and literary tradition. The BM00S2 legacy table remains a separate scheme so that overlapping categories do not become apparently precise city eras.",
   "refs": [
    {
     "source": "mills2013",
     "locator": "; printed pp. 12, 17"
    },
    {
     "source": "docter2009",
     "locator": "; printed pp. 180–181"
    },
    {
     "source": "hematrix",
     "locator": "Bayesian chronological modelling"
    }
   ],
   "status": "Project dating-design proposal",
   "limitation": ""
  },
  {
   "id": "M004",
   "heading": "Register photographs and historical plans",
   "text": "Begin with an immutable image checksum and a region in image pixels or normalized coordinates. Identify the photographed subject independently of the camera position. To place a plan in GIS, record its original scale, north, datum, coordinate system, control points, transformation and residuals. Keep ancient measured fabric, restored fabric and reconstructed outlines in separate layers. An oblique photograph needs camera and geometric controls before its pixels can be treated as map coordinates. Lézine’s warning that incomplete excavation limited the precision of his plans is part of the plan’s provenance. A modern OSM park polygon is useful orientation, but it cannot supply those missing controls.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed p. 3"
    },
    {
     "source": "annotation",
     "locator": "SpecificResource and FragmentSelector"
    }
   ],
   "status": "Project registration workflow",
   "limitation": ""
  },
  {
   "id": "M005",
   "heading": "Model present surfaces before proposed ancient states",
   "text": "A present-condition model can be made from controlled overlapping photographs or scans. The acquisition plan should document scale, control targets, camera settings, coverage, occlusions and accuracy checks; the output needs coordinates, units and rights. It records visible surfaces at a particular date. A reconstructed ancient state is another model, with per-element evidence and alternative completions. Mesh vertices must not gain ancient phase labels automatically from colour or material appearance. The 264 visitor images are useful for interpretation and annotation, but no calibration, scale targets or coverage test establishes them as a metric photogrammetric survey. The published LiDAR lead may reduce acquisition work after its identity and permissions are checked; its absent georeferencing still needs to be resolved.",
   "refs": [
    {
     "source": "hephotogrammetry",
     "locator": "Guidance summary and contents"
    },
    {
     "source": "charter",
     "locator": "Principles 2–6"
    },
    {
     "source": "oh3d",
     "locator": "Dataset attributes and identity conflict"
    }
   ],
   "status": "Feasible model-production workflow",
   "limitation": ""
  },
  {
   "id": "M006",
   "heading": "Keep the publication and contribution workflow auditable",
   "text": "The implemented history links to the existing shared image workspace so that a reader can inspect a full image, select a region and contribute a sourced observation. Contributions retain pending or reviewed status; an accepted comment does not rewrite a published claim. A future proposed relation should name its two context targets and its evidence, then enter review before it changes the graph. Export the source statements, chronology records, original relation edges and figure definitions together. Display missing evidence and competing interpretations in the reading interface, while keeping the application’s internal file organization in the research notes. This makes the model useful for learning without concealing what remains to be established.",
   "refs": [
    {
     "source": "hematrix",
     "locator": "Transparency and interoperable data"
    },
    {
     "source": "annotation",
     "locator": "Annotation bodies and targets"
    }
   ],
   "status": "Implemented workflow and extension proposal",
   "limitation": ""
  }
 ],
 "events": [
  {
   "id": "T01",
   "site": "city",
   "label": "Traditional foundation",
   "start": -814,
   "end": -813,
   "kind": "tradition",
   "claim": "H002",
   "note": "Literary chronology; not a measured settlement date.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 2–3, 8"
    },
    {
     "source": "docter2009",
     "locator": "; printed pp. 180–181, table and note 12"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T02",
   "site": "city",
   "label": "Early settlement evidence",
   "start": -850,
   "end": -750,
   "kind": "display",
   "claim": "H002",
   "note": "Broad reading window, not a calibrated confidence interval.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 2–3, 8"
    },
    {
     "source": "docter2009",
     "locator": "; printed pp. 180–181, table and note 12"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T03",
   "site": "city",
   "label": "Roman destruction of Punic city",
   "start": -146,
   "end": -146,
   "kind": "event",
   "claim": "H003",
   "note": "Historical event; not a universal deposit date.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 88–91"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, BM00S2 redeposited Punic destruction material"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T04",
   "site": "city",
   "label": "Gracchan colonial project",
   "start": -122,
   "end": -122,
   "kind": "event",
   "claim": "H003",
   "note": "Project and land settlement do not establish a completed new urban centre.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 88–91"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, BM00S2 redeposited Punic destruction material"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T05",
   "site": "city",
   "label": "Caesarian colony: foundation debated",
   "start": -46,
   "end": -40,
   "kind": "display",
   "claim": "H003",
   "note": "Caesarian initiative and existence by 40 BCE; exact start debated.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 88–91"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, BM00S2 redeposited Punic destruction material"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T06",
   "site": "city",
   "label": "Octavian supplements settlers",
   "start": -29,
   "end": -29,
   "kind": "event",
   "claim": "H003",
   "note": "Separate episode from the colony’s first foundation.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 88–91"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, BM00S2 redeposited Punic destruction material"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T07",
   "site": "city",
   "label": "Vandal capture",
   "start": 439,
   "end": 439,
   "kind": "event",
   "claim": "H005",
   "note": "Political transition; not the date of every building alteration.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 118, 130–133"
    },
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–73"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12"
    },
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T08",
   "site": "city",
   "label": "Byzantine reconquest",
   "start": 533,
   "end": 533,
   "kind": "event",
   "claim": "H005",
   "note": "Campaign marker, not the construction date of every church.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 118, 130–133"
    },
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–73"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12"
    },
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T09",
   "site": "city",
   "label": "Final loss of Byzantine city",
   "start": 698,
   "end": 698,
   "kind": "event",
   "claim": "H006",
   "note": "Later habitation and quarrying continue in the source account.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 141–142"
    },
    {
     "source": "lezine1968",
     "locator": "; printed p. 73"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T10",
   "site": "city",
   "label": "Save Carthage research campaign",
   "start": 1972,
   "end": 1991,
   "kind": "range",
   "claim": "H008",
   "note": "Campaign launch and French Byrsa investigations, not a single excavation everywhere.",
   "refs": [
    {
     "source": "efrbyrsa",
     "locator": "Campaign 1972; French research 1974–1991"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T11",
   "site": "city",
   "label": "Supplied photographic visit",
   "start": 2025,
   "end": 2025,
   "kind": "event",
   "claim": "H007",
   "note": "18 January 2025 in camera EXIF; timezone unverified.",
   "refs": [
    {
     "source": "photoaudit",
     "locator": "Six contact sheets, photo metadata and selected full-resolution checks"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T100",
   "site": "harbour",
   "label": "Late Punic naval interpretation",
   "start": -200,
   "end": -146,
   "kind": "display",
   "claim": "H101",
   "note": "Approximate window for the late Punic harbour reading.",
   "refs": [
    {
     "source": "hurst2010",
     "locator": "; printed pp. 53, 55–56"
    },
    {
     "source": "photoaudit",
     "locator": "IMG_1930, IMG_1948, IMG_3951, IMG_3955–3960"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T101",
   "site": "harbour",
   "label": "Roman monumental island",
   "start": 175,
   "end": 300,
   "kind": "display",
   "claim": "H102",
   "note": "Later second-century rebuilding; endpoint is a display limit, not abandonment.",
   "refs": [
    {
     "source": "hurst2010",
     "locator": "; printed pp. 53, 55–56, 65"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T102",
   "site": "byrsa",
   "label": "Late Punic slope housing",
   "start": -200,
   "end": -146,
   "kind": "display",
   "claim": "H401",
   "note": "Broad display window; no date is transferred to a specific photo wall.",
   "refs": [
    {
     "source": "efrbyrsa",
     "locator": "Research account: Punic finds and Roman platform"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 94–96"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T103",
   "site": "byrsa",
   "label": "Roman levelling and redevelopment",
   "start": -44,
   "end": 200,
   "kind": "display",
   "claim": "H402",
   "note": "Broad window combining initial levelling and later monumental work.",
   "refs": [
    {
     "source": "photoaudit",
     "locator": "Contact sheets 5–6; full-resolution IMG_4029"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 95"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T104",
   "site": "villas",
   "label": "Punic necropolis",
   "start": -300,
   "end": -146,
   "kind": "display",
   "claim": "H201",
   "note": "District-level third/second-century orientation.",
   "refs": [
    {
     "source": "villas",
     "locator": "Overview, first occupation and residential development"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 97"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T105",
   "site": "villas",
   "label": "Roman residential development",
   "start": 1,
   "end": 300,
   "kind": "display",
   "claim": "H201",
   "note": "Broad window; individual houses and floor repairs have separate dates.",
   "refs": [
    {
     "source": "villas",
     "locator": "Overview, first occupation and residential development"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 97"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T106",
   "site": "theatre",
   "label": "Roman theatre: proposed early origin",
   "start": -27,
   "end": 100,
   "kind": "display",
   "claim": "H301",
   "note": "Hoyos’s Augustan proposal; broad display interval, not a secure construction range.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 96–97"
    },
    {
     "source": "theatre",
     "locator": "Overview: limited ancient remains and 1906 adaptation"
    },
    {
     "source": "photoaudit",
     "locator": "IMG_3985–4006"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T107",
   "site": "theatre",
   "label": "Modern performance adaptation begins",
   "start": 1906,
   "end": 1906,
   "kind": "event",
   "claim": "H301",
   "note": "Documented adaptation; present seating is not wholly ancient fabric.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 96–97"
    },
    {
     "source": "theatre",
     "locator": "Overview: limited ancient remains and 1906 adaptation"
    },
    {
     "source": "photoaudit",
     "locator": "IMG_3985–4006"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T108",
   "site": "tophet",
   "label": "Urn series in Harden’s account",
   "start": -800,
   "end": -146,
   "kind": "display",
   "claim": "H501",
   "note": "Broad eighth-to-second-century sequence, not an urn-specific date.",
   "refs": [
    {
     "source": "harden1937",
     "locator": "Accessible extract and bibliographic record"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed pp. 17–18"
    },
    {
     "source": "smith2013",
     "locator": "Published sacrificial interpretation"
    },
    {
     "source": "schwartz2017",
     "locator": "Methodological challenge"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T109",
   "site": "baths",
   "label": "Construction date in Lézine",
   "start": 145,
   "end": 145,
   "kind": "event",
   "claim": "H601",
   "note": "Separate reported operation from inauguration.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 5–6, 71"
    },
    {
     "source": "baths",
     "locator": "Overview, inauguration and completion"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T110",
   "site": "baths",
   "label": "Alternative inauguration years",
   "start": 157,
   "end": 159,
   "kind": "range",
   "claim": "H601",
   "note": "The source gives 157 OR 159; the bar does not assert activity throughout the interval.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 5–6, 71"
    },
    {
     "source": "baths",
     "locator": "Overview, inauguration and completion"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T111",
   "site": "baths",
   "label": "Completion",
   "start": 162,
   "end": 162,
   "kind": "event",
   "claim": "H601",
   "note": "Official account; Lézine discusses completion of decoration.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 5–6, 71"
    },
    {
     "source": "baths",
     "locator": "Overview, inauguration and completion"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T112",
   "site": "baths",
   "label": "Restoration inscription",
   "start": 389,
   "end": 389,
   "kind": "event",
   "claim": "H603",
   "note": "An inscription date, not the date of each repair visible today.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–72"
    },
    {
     "source": "baths",
     "locator": "Overview, frigidarium collapse"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T113",
   "site": "baths",
   "label": "Collapse: agency reading",
   "start": 388,
   "end": 392,
   "kind": "range",
   "claim": "H603",
   "note": "AMVPPC alternative; incompatible with treating 389 as a secure lower bound without discussion.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–72"
    },
    {
     "source": "baths",
     "locator": "Overview, frigidarium collapse"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T114",
   "site": "baths",
   "label": "Collapse: Lézine’s inference",
   "start": 389,
   "end": 425,
   "kind": "interpretation",
   "claim": "H603",
   "note": "Author’s proposed bracket; coin-based ante quem inference is audited, not endorsed.",
   "refs": [
    {
     "source": "lezine1968",
     "locator": "; printed pp. 71–72"
    },
    {
     "source": "baths",
     "locator": "Overview, frigidarium collapse"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T115",
   "site": "magon",
   "label": "Housing and redevelopment",
   "start": -500,
   "end": -146,
   "kind": "display",
   "claim": "H701",
   "note": "At least two phases; broad site overview.",
   "refs": [
    {
     "source": "magon",
     "locator": "Overview, fills and second-century redevelopment"
    },
    {
     "source": "photoaudit",
     "locator": "IMG_1987 interpretation-panel identity"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T116",
   "site": "massouda",
   "label": "Published floor-over-cistern sequence",
   "start": -300,
   "end": -146,
   "kind": "display",
   "claim": "H802",
   "note": "Late Punic orientation; no measured context-specific construction date.",
   "refs": [
    {
     "source": "massoudaproject",
     "locator": "Composite image caption, Trench 2"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, Table 1.3"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T117",
   "site": "didon",
   "label": "Punic-to-Byzantine sequence",
   "start": -800,
   "end": 600,
   "kind": "display",
   "claim": "H901",
   "note": "Reading window for separate groups, not uninterrupted use of a single building.",
   "refs": [
    {
     "source": "didon",
     "locator": "Research account and original/reconstructed fabric captions"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T118",
   "site": "malga",
   "label": "Aqueduct connection around 160",
   "start": 150,
   "end": 170,
   "kind": "display",
   "claim": "H1001",
   "note": "Around 160 in Hoyos; endpoints are a visualization convention.",
   "refs": [
    {
     "source": "hoyos2021",
     "locator": "; printed p. 96"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T119",
   "site": "amphitheatre",
   "label": "Foundation: Hoyos’s proposal",
   "start": -27,
   "end": 50,
   "kind": "interpretation",
   "claim": "H1101",
   "note": "Augustan or soon after; broad display window for an earlier dating.",
   "refs": [
    {
     "source": "amphagency",
     "locator": "Overview, construction phases, reoccupation and 1893 commemoration"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 96"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T120",
   "site": "amphitheatre",
   "label": "Foundation: agency reading",
   "start": 1,
   "end": 100,
   "kind": "interpretation",
   "claim": "H1101",
   "note": "First-century architectural dating; displayed beside Hoyos’s alternative.",
   "refs": [
    {
     "source": "amphagency",
     "locator": "Overview, construction phases, reoccupation and 1893 commemoration"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 96"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T121",
   "site": "amphitheatre",
   "label": "Modern Christian commemoration",
   "start": 1893,
   "end": 1893,
   "kind": "event",
   "claim": "H1101",
   "note": "Commemoration does not independently prove the execution location.",
   "refs": [
    {
     "source": "amphagency",
     "locator": "Overview, construction phases, reoccupation and 1893 commemoration"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 96"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T122",
   "site": "damous",
   "label": "First basilica: probable late fourth century",
   "start": 375,
   "end": 400,
   "kind": "interpretation",
   "claim": "H1201",
   "note": "Conventional display window for “end of the fourth century”.",
   "refs": [
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  },
  {
   "id": "T123",
   "site": "damous",
   "label": "Byzantine construction stages",
   "start": 533,
   "end": 700,
   "kind": "display",
   "claim": "H1201",
   "note": "Broad reading window; no context-specific end date is asserted.",
   "refs": [
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "date_convention": "Civil years: negative = BCE, positive = CE; no year zero"
  }
 ],
 "legacy_phases": [
  {
   "id": "BP1",
   "phase": "1",
   "label": "Archaic",
   "start": -800,
   "end": -450,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP2",
   "phase": "2",
   "label": "Late Archaic",
   "start": -600,
   "end": -450,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP3",
   "phase": "3",
   "label": "Middle Punic",
   "start": -450,
   "end": -301,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP4",
   "phase": "4",
   "label": "Late Punic",
   "start": -300,
   "end": -146,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP5",
   "phase": "5",
   "label": "Roman I",
   "start": -75,
   "end": 100,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP6",
   "phase": "6",
   "label": "Roman II",
   "start": 101,
   "end": 400,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP7",
   "phase": "7",
   "label": "Vandal",
   "start": 400,
   "end": 530,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP8",
   "phase": "8",
   "label": "Byzantine",
   "start": 531,
   "end": 700,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP9",
   "phase": "9",
   "label": "Medieval",
   "start": 801,
   "end": 1900,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  },
  {
   "id": "BP10",
   "phase": "10",
   "label": "Modern",
   "start": 1900,
   "end": 2000,
   "source": "mills2013",
   "locator": "Table 1.3, printed p. 12"
  }
 ],
 "sequences": [
  {
   "id": "byrsa",
   "title": "Byrsa: preservation through transformation",
   "kind": "Published phase-group synthesis",
   "claim": "H402",
   "nodes": [
    {
     "id": "B1",
     "label": "Later excavation and presentation",
     "detail": "The present exposure joins selected remains after deposits have been removed."
    },
    {
     "id": "B2",
     "label": "Roman platform and monumental construction",
     "detail": "The rebuilt summit supports later civic and religious buildings."
    },
    {
     "id": "B3",
     "label": "Roman levelling and burial of selected slope remains",
     "detail": "A process that removes some earlier structures and preserves others beneath fill."
    },
    {
     "id": "B4",
     "label": "Late Punic slope houses",
     "detail": "Preserved quarter; not equivalent to an intact summit citadel."
    }
   ],
   "edges": [
    {
     "later": "B1",
     "earlier": "B2",
     "basis": "historical",
     "label": "later investigation of"
    },
    {
     "later": "B2",
     "earlier": "B3",
     "basis": "physical",
     "label": "construction on prepared platform"
    },
    {
     "later": "B3",
     "earlier": "B4",
     "basis": "physical",
     "label": "buries selected slope remains"
    }
   ],
   "refs": [
    {
     "source": "photoaudit",
     "locator": "Contact sheets 5–6; full-resolution IMG_4029"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 95"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  },
  {
   "id": "harbour",
   "title": "Harbour: compare interpretations without inventing contacts",
   "kind": "Historical phase reading",
   "claim": "H102",
   "nodes": [
    {
     "id": "P1",
     "label": "Current basin and archaeological presentation",
     "detail": "Modern waterline and visible exposure."
    },
    {
     "id": "P2",
     "label": "Roman island redevelopment",
     "detail": "Colonnaded complex, temple and octagonal building discussed by Hurst."
    },
    {
     "id": "P3",
     "label": "Late Punic naval complex",
     "detail": "Arsenal interpretation and reconstructed shipsheds."
    }
   ],
   "edges": [
    {
     "later": "P1",
     "earlier": "P2",
     "basis": "historical",
     "label": "later presentation of"
    },
    {
     "later": "P2",
     "earlier": "P3",
     "basis": "historical",
     "label": "later redevelopment of harbour area"
    }
   ],
   "refs": [
    {
     "source": "hurst2010",
     "locator": "; printed pp. 53, 55–56, 65"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  },
  {
   "id": "villas",
   "title": "Villas: a district sequence",
   "kind": "Historical phase reading",
   "claim": "H201",
   "nodes": [
    {
     "id": "V1",
     "label": "Later domestic building and refurbishment",
     "detail": "Individual houses and mosaics require their own records."
    },
    {
     "id": "V2",
     "label": "Roman street grid and occupation",
     "detail": "Spatial organization in the new city."
    },
    {
     "id": "V3",
     "label": "Punic necropolis in the area",
     "detail": "Burial landscape; not proven directly below every photographed house."
    }
   ],
   "edges": [
    {
     "later": "V1",
     "earlier": "V2",
     "basis": "historical",
     "label": "develops within grid"
    },
    {
     "later": "V2",
     "earlier": "V3",
     "basis": "historical",
     "label": "later district use"
    }
   ],
   "refs": [
    {
     "source": "villas",
     "locator": "Overview, first occupation and residential development"
    },
    {
     "source": "hoyos2021",
     "locator": "; printed p. 97"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  },
  {
   "id": "massouda",
   "title": "Bir Massouda: published Trench 2 relation",
   "kind": "Published physical relation; incomplete group sequence",
   "claim": "H802",
   "nodes": [
    {
     "id": "BM1",
     "label": "Roman–Early Modern deposits",
     "detail": "Composite later group described in the project caption; no individual contexts imported."
    },
    {
     "id": "BM2",
     "label": "Late Punic floor",
     "detail": "Caption explicitly places the floor over a cistern."
    },
    {
     "id": "BM3",
     "label": "Late Punic cistern",
     "detail": "Its full construction and disuse sequence is not supplied."
    }
   ],
   "edges": [
    {
     "later": "BM1",
     "earlier": "BM2",
     "basis": "physical",
     "label": "above, in published caption"
    },
    {
     "later": "BM2",
     "earlier": "BM3",
     "basis": "physical",
     "label": "floor over cistern"
    }
   ],
   "refs": [
    {
     "source": "massoudaproject",
     "locator": "Composite image caption, Trench 2"
    },
    {
     "source": "mills2013",
     "locator": "; printed p. 12, Table 1.3"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  },
  {
   "id": "didon",
   "title": "Didon: separate functions and degrees of certainty",
   "kind": "Published architectural succession",
   "claim": "H901",
   "nodes": [
    {
     "id": "D1",
     "label": "Byzantine rotunda",
     "detail": "Cult interpretation probable; not Damous el Karita’s rotunda."
    },
    {
     "id": "D2",
     "label": "Roman public building",
     "detail": "Function unresolved."
    },
    {
     "id": "D3",
     "label": "Punic quarter",
     "detail": "Detailed reconstruction constrained by preservation."
    }
   ],
   "edges": [
    {
     "later": "D1",
     "earlier": "D2",
     "basis": "physical",
     "label": "later superimposed construction"
    },
    {
     "later": "D2",
     "earlier": "D3",
     "basis": "physical",
     "label": "Roman over Punic levels"
    }
   ],
   "refs": [
    {
     "source": "didon",
     "locator": "Research account and original/reconstructed fabric captions"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  },
  {
   "id": "damous",
   "title": "Damous el Karita: cemetery and basilica development",
   "kind": "Published site synthesis",
   "claim": "H1201",
   "nodes": [
    {
     "id": "K1",
     "label": "Later Byzantine construction stage",
     "detail": "Rotunda functions disputed."
    },
    {
     "id": "K2",
     "label": "Early Byzantine chapel / redevelopment",
     "detail": "Burials cluster near a venerated grave; saint unidentified."
    },
    {
     "id": "K3",
     "label": "Probable late-fourth-century basilica",
     "detail": "Initial church date is a published probability."
    },
    {
     "id": "K4",
     "label": "Earlier Roman burial grounds",
     "detail": "Earlier cemetery beneath the later religious complex."
    }
   ],
   "edges": [
    {
     "later": "K1",
     "earlier": "K2",
     "basis": "historical",
     "label": "later construction stage"
    },
    {
     "later": "K2",
     "earlier": "K3",
     "basis": "historical",
     "label": "redevelopment"
    },
    {
     "later": "K3",
     "earlier": "K4",
     "basis": "physical",
     "label": "built on earlier burial grounds"
    }
   ],
   "refs": [
    {
     "source": "potthoff2017",
     "locator": "; printed pp. 174–176"
    }
   ],
   "contract": "Phase-group nodes, not excavation contexts. Ordinal order, not depth. Solid = published physical relationship; dashed = historical succession."
  }
 ],
 "column": {
  "id": "antonine-room18",
  "title": "Antonine Baths · Room 18",
  "claim": "H602",
  "source": "lezine1968",
  "locator": "printed p. 70, Salle 18",
  "datum": "0.00 m = the room’s radier in Lézine’s record; not sea level",
  "layers": [
   {
    "id": "L1",
    "label": "Ash and charcoal",
    "bottom": 0,
    "top": 0.62,
    "color": "#8e8272",
    "raw": "Du zéro du radier à +62"
   },
   {
    "id": "L2",
    "label": "Red clay",
    "bottom": 0.62,
    "top": 0.84,
    "color": "#ad6f54",
    "raw": "jusqu’à +84"
   },
   {
    "id": "L3",
    "label": "Earth and charcoal streaks",
    "bottom": 0.84,
    "top": 0.95,
    "color": "#b3a582",
    "raw": "jusqu’à +95"
   }
  ],
  "horizons": [
   {
    "id": "L4",
    "label": "Compact charcoal layer",
    "height": 1.06,
    "color": "#3c4840",
    "note": "Position reported; thickness unspecified"
   }
  ],
  "upper": "Rubble and windblown fills above; upper extent unspecified",
  "gap": "The interval between +0.95 and the +1.06 m horizon is not independently characterized in this terse record.",
  "units_note": "The mixed notation +62, +84, +95 and +1,06 is normalized here to metres relative to the stated floor datum. No absolute dates are assigned."
 },
 "assessment": {
  "conclusion": "The prompt is sound as a source-led investigation and visualization task. “Fully discuss” needs a bounded corpus, a coverage statement and a distinction between site history, visible fabric and context-level stratigraphy. Images can motivate and illustrate an explanation, but cannot supply missing excavation records.",
  "corrections": [
   "Identify photographed subjects separately from camera locations, installation sites and illustrated objects.",
   "Keep historical chronology, archaeological relative order, measured profiles and reconstruction as different data products.",
   "Audit original publication dates, legacy phase schemes, contradictory interpretations and the difference between a dating constraint and an exact event.",
   "Implement the supported figures now; express unrecorded depths, contacts, contexts and functions as gaps.",
   "Specify acceptance by image coverage, claim locators, source fidelity, graph validity and working interactions."
  ],
  "implementation": "The new reader implements the supported chronology, ordinal relations, legacy BM00S2 scheme and historical Room 18 column. It links to existing persistent image annotation. Measured site-wide geometry, a full Harris matrix and an ancient 3D reconstruction remain dependent on additional controlled records."
 },
 "selected_images": {
  "harbour": [
   "IMG_1930",
   "IMG_1948",
   "IMG_3951",
   "IMG_3955",
   "IMG_3958"
  ],
  "tophet": [],
  "byrsa": [
   "IMG_2048",
   "IMG_4021",
   "IMG_4027",
   "IMG_4029"
  ],
  "baths": [],
  "villas": [
   "IMG_1962",
   "IMG_1987",
   "IMG_3964"
  ],
  "theatre": [
   "IMG_3985",
   "IMG_3999",
   "IMG_4001",
   "IMG_4007"
  ],
  "magon": [],
  "massouda": [],
  "didon": [],
  "malga": [],
  "amphitheatre": [],
  "damous": []
 },
 "coverage": {
  "photographs": 264,
  "contact_sheet_screened": 264,
  "full_resolution_this_extension": 4,
  "photographed_areas": [
   "harbour",
   "villas",
   "theatre",
   "byrsa"
  ],
  "local_publications_targeted": [
   "hurst2010",
   "kaufman2016",
   "docter2009",
   "hoyos2021",
   "kelsey1979",
   "lezine1968",
   "mills2013",
   "potthoff2017"
  ],
  "bibliographic_only": [
   "khader"
  ],
  "corpus_pdf_unique": 44,
  "claim": "Targeted investigation of a bounded corpus; not a complete reading of all publications or a building-by-building excavation catalogue."
 }
}