Sources and coverage

Ricci catalogue coverage

Checked 6 November 2026. sources.json contains 37 records. These are not 37 original maps. The file separates physical copies, digital surrogates, historical photographs, facsimiles, reconstructions and lost attested editions. works.json supplies the two work nodes used by the relations.

Seven physical holdings have verified IIIF manifests: Kyoto, Miyagi’s 1602 print, National Archives of Japan, Bologna, Austrian National Library, Tohoku, and Miyagi’s later manuscript. Eighteen exact manifests are archived. Twelve belong to sections of one cut and remounted National Archives copy. The first image service from each holding returned HTTP 200; exact responses and timestamps are in image-service-checks.json and the seven *-image-info.json files. No large original images were downloaded.

Remaining gaps include the full historical edition sequence (particularly the alleged 1596 stele and 1609 double-hemisphere version), current provenance of the Schiller/Robinson print, a possible second Vatican holding, the Vatican later-print fragment, Waseda’s source photograph, Japanese manuscript copies beyond the institutions above, and exact digital book witnesses for early lost maps. Aleni, Sambiasi and Verbiest derivatives belong in related work families; they are not silently counted as Ricci editions here.

All unknown manifest/image endpoints are null. Historical catalogue titles and date inconsistencies are retained in uncertainties. The inventory is systematic but is not a claim that every worldwide copy has been found.

Download this source-survey report

The Verbiest and Aleni research file contains 39 records across 23 named holding institutions or catalogue providers, checked on 6 November 2026. It distinguishes physical maps, complementary sheets, book maps, printing blocks, and a photographic facsimile. Six records have retrieved IIIF manifests. Eight direct image URLs returned HTTP 200 with an image/jpeg content type during small streamed checks. No full-resolution map files were downloaded.

This is a documented research inventory, not a claim that every surviving copy has been found. A failed request is recorded as a retrieval failure, never as proof that images do not exist. The row count is not a census of independent complete maps: the file includes incomplete exemplars, a provisional grouping of complementary sheets, contextual objects, and a book whose map presence is still unconfirmed.

sources.json is a JSON array. Each row includes the requested title, creator, dates, institution, shelfmark, source record, manifest, image service, image/thumbnail URL, rights, evidence, access status, relations, uncertainty, and source URLs. Additional fields identify entityType, recordKind, grouped component identifiers, source publication statements, and image-page representations. copyDate stays null when only a work date or edition date is known. BnF structured records are preserved in the SRU XML and normalized JSON files. Actual manifests are retained as JSON, rather than reconstructed from URL templates.

The core Verbiest holdings and variants are:

BnF's GE C-5803 record needs particular care. Its publication statement uses 1860, but its notes explicitly mention Japanese printing in 1933 and the Seoul-Keijo blocks; it also records Nakamura's1934 donation. The two component dimensions are those of hemispheres. This evidence does not justify two complete-map entries or an exact 1860 date for the surviving impression. BnF GE B-1275 is an incomplete western hemisphere, acquired from Marcel in 1845. CHINOIS 1914 has a1763 Jesuit acquisition inscription. CHINOIS 9231 has a Bréquigny provenance and one missing sheet.

The Aleni records are:

Vatican images are explicitly limited to free personal/study use; online or printed publication requires rights permission. Ryukoku's secondary-use policy explicitly requires prior application for image/data reuse. These four records are marked publicDisplayMode: external-link-only and rightsVerification: explicit-restriction. Their image services remain in research metadata for scholarly access, but their images must not be bundled as public site assets. The small research contact sheet is internal visual-verification material, not a licensed asset. Other museums' public availability is not treated as an open reuse licence. National Museum of Korea's specific Type 1 attribution permission is recorded separately.

The following searches and exclusions define the current gaps:

• BnF public SRU was queried for both romanized titles. It returned additional unrelated maps, including Sambiasi's ca.1648 Kunyu quantu, GE AA-2565, which is not assigned to Verbiest. Verbiest's CHINOIS 1923 map is distinct from the prose book commonly called Kunyu tushuo.
• National Library of China's Tianxue chuhan record fid411999011140 is a serious Aleni image lead. Direct access returned 403. A second reported Fujian book, fid411999026205, reportedly lacks maps. Neither is promoted to a verified map-image entry here.
• Fu Ssu-nien Library's official digitization lists confirm three Zhifang waiji holdings and two Kunyu tushuo catalogue entries. The page13 list identifies a1623 manuscript and an undated printed copy; page6 identifies manuscript barcode 174555. Public item images were not resolved; cited scholarship describes the relevant book database as onsite. A Kunyu tushuo entry is labelled Ming despite Verbiest's chronology, so that date requires catalogue correction or clarification before reuse.
• Waseda manuscripts ル02 01068 and 文庫08 B0138 are verified institutional search records, the latter dated 1796. Their map content is not verified and they are excluded from the map table. Other reported Waseda manuscripts remain leads.
• Meiji's 1774 Nagakubo Sekisui manuscript has a retrieved public IIIF manifest, but the catalogue explicitly says the original maps are not bound in it. It is excluded from map holdings. The actual manifest is retained to document that exclusion.
• Kyoto's RB00022879 is an anonymous nineteenth-century Japanese oval map. Its identical title alone does not establish Verbiest authorship, so it is excluded. First-hand research reports Zhifang waiji at Kyoto F1-77 without maps and 2B-43 with maps; current institutional records were not located in this bounded search.
• The standalone Aleni sheets at Biblioteca Nazionale Braidense, AB XV 34, and Veneranda Biblioteca Ambrosiana, S.P.II.308, remain important gaps. Their forms and shelfmarks are documented in Cams and Papelitzky's first-hand study, but no current institutional object record/IIIF was verified. They are not disguised as institutionally verified rows. That study also provides leads to Tōyō Bunko V-5-A-28, Rome Central Library 72-C494, Ryukoku 490.9-36-W-1, Nichibunken GA/74/Ga, and Tōhoku 3/8156/1. Map presence and public images should be checked at each institution.
• NCL Taiwan's apparently separate Verbiest hit proved to be a union-catalogue description of BnF FRBNF 40733999. It is excluded as a duplicate, not counted as an NCL holding.
• Public catalogue/domain searches also targeted the British Library, Bodleian, Leiden, Berlin Staatsbibliothek, Stanford, Harvard, and Princeton. They did not produce a verified additional item within this run. This is a search limitation, not a negative holdings assertion. British Library/Phillipps and Royal Geographical Society copies are mentioned in institutional or scholarly descriptions and need direct current catalogue verification.
• The NLA checklist confirms its object, but the current catalogue search did not expose an individual digital item. The Tianjin original underlying the 1925 photo album, additional Korean 1860 impressions, and later 1931/1933 impressions require more copy-specific provenance work.

No auction listing, dealer description, Wikipedia assertion, guessed shelfmark, guessed IIIF identifier, or later facsimile is used as proof of an original institutional map copy. Work dates, edition dates, approximate copy dates, source metadata conflicts, and retrieval failures are kept distinct.

Download this source-survey report

Other Jesuit-connected cartographies: coverage and counting

This is a source-based research inventory beyond Ricci, Aleni and Verbiest. It contains 152 catalogue rows and 106 conservatively countable map units. It is not an exhaustive census of every surviving Jesuit-connected map. The corpus separates works, editions, surviving copies, atlas containers, constituent maps, compound-map sheets, later impressions, facsimiles and a lost presentation attestation.

How to count these records

Use countable or its identical alias countableUnit. Do not count JSON rows, images or manifests as maps. entityCategory has four values: map, atlas-container, map-sheet and lost-attestation. The 16 atlas/group containers, 23 image-sheet or fragment records, one lost attestation, five historical copy attestations without verified current copy identity, and the non-Jesuit Bering appendix are excluded from the 106 map units. Countable units include distinct physical copies and manuscript drafts. They are not a count of independent intellectual works.

Two d’Anville records explicitly group numbered image parts. The twelve detailed Tartary sheets form one compound map; the nine detailed Tibet sheets form another. Each sheet remains available as an image part through imageParts, partOfUnit, unitGroupId and its exact IIIF canvas/service. The general overview of Tartary and general overview of Tibet remain separate maps. The sampled Tartary heading identifies a numbered “Feuille particulière”; the Tibet heading identifies a numbered sheet within the general map. The saved study images and full manifest support this grouping decision. No mosaic has been fabricated.

The BnF Kangxi container is different. BnF calls it an ensemble of thirteen maps, and each child catalogue describes one separately titled regional carte. Its container is excluded, while its thirteen independently catalogued regional maps remain countable. A few maps can be geographically joined, but that alone does not turn every map in an atlas into one physical map. The NPM Qianlong example is only one sheet from a much larger thirteen-row map and is therefore a non-countable map-sheet. Bologna’s surviving portion of Schall’s star map is likewise retained as a fragment, not counted as another complete map.

Primary evidence and image access

The research retains three complete institutional IIIF manifests: Boym, Vatican Borg.cin.531, Schall, Vatican BAVOA.233, and d’Anville, Göttingen PPN340023538. Their image-service IDs are copied from the manifests. There are 61 exact service-bearing image records. Many rows inherit the same bound-volume manifest; 83 manifest-bearing rows do not mean 83 distinct manifests.

Zürich’s Martini catalogue publishes the German 1656 edition, shelfmark T 3, Public Domain Mark and its IIIF manifest. The web tool exposed the manifest’s map ranges and the complete institutional table of contents. Sixteen map canvas IDs were recorded from those ranges. For Iaponia Regnum, the exact JPEG link from the institutional map viewer was retained; a canvas ID was not invented from that filename. Direct raw-manifest fetches returned HTTP 403. The retained evidence therefore distinguishes a published, observed manifest from a successfully downloaded raw manifest.

A bounded image-access follow-up located and followed the exact institutional JPEG medium link for each of all seventeen Zürich maps, including Honan through its table-of-contents range link. All seventeen rows now have imageUrl values sourced independently from the library viewers. These are JPEG download links, not invented IIIF services. The repeated raw-manifest and manifest-published METS requests still returned HTTP 403; direct HTTP requests to all seventeen JPEG links also returned 403 from this environment. The link evidence and the unsuccessful byte-fetch evidence are retained separately in martini-jpeg-link-inventory.json, the follow-up web snapshots, and martini-jpeg-http-checks.json. imageServiceUrl remains null for these maps. The overall catalogue now has 84 rows with an observed image or thumbnail link, without asserting that every link was successfully downloaded.

Rome’s Ruggieri archive supplies an institutional live image index. Its viewer code constructs legacy IIP JPEG URLs from the exact index filenames. The research follows that same code and retains the code and index as evidence. Two representative JPEG requests returned HTTP 200. The complete index has 79 digitized surfaces, including texts, blank versos and alternative drafts; it does not contain 79 maps. Its manuscript surface numbers are retained separately from a shelfmark, which the captured archive presentation does not provide.

BnF’s thirteen Kangxi records were saved individually, with their exact shelfmarks and linked Gallica reproductions. The catalogue identifies later manuscript annotations and colouring as copy features. Sample Gallica IIIF requests returned HTTP 403, so no guessed manifest or service URL has been supplied. A failed fetch is not evidence that an institution lacks IIIF. The Sambiasi record is treated the same way.

The NPM exhibition identifies 平圖021068 as a 1932 printing from Qianlong 25, 1760, copperplates. Four bounded exhibition images returned HTTP 200. That physical impression is not silently redated to 1760. Delaware provides a book-map exhibition record and scanned PDF rather than a verified individual IIIF service.

evidence/fetch-*-log.json files record attempted URLs, UTC times, statuses or exceptions, content types and saved filenames. The evidence folder contains only metadata, source code needed to resolve published image links, small study images and one short institutional historical study. No massive originals were downloaded. build_inventory.py reproduces sources.json and summary.json locally and checks required fields, unique IDs, URL shapes, evidence-file existence and counting exclusions. These structural checks do not prove untested HTTP access.

Attribution and date limits

Chinese collaboration is recorded where primary institutional descriptions identify it. Rome names an anonymous Chinese contributor to Ruggieri’s atlas and identifies the original Chinese Liautum map as a source within the collection. Schall’s entry preserves Xu Guangqi’s leadership, Wu Ming-chu’s production role in the Bologna description, and unnamed Chinese astronomical observers. Kangxi’s survey is described as a Qing imperial enterprise with Chinese, Manchu, Tibetan and European participants. The Yellow River headwaters sheet preserves the specifically named earlier and later official expeditions. Matteo Ripa is an engraver of the copperplate version and was not a Jesuit; Bonjour is not relabelled as Jesuit merely because he appears among the survey contributors.

The public Vatican catalogue for Boym is shelfmark-only. The manuscript’s title was read from its title-page image; attribution to Boym and probable composition in 1651–1652 are separately sourced to the institutional ARSI study. Eighteen unfolded map states were visually identified. Folio labels are used when a title has not been transcribed. Bindings, partly folded states and repeated scans are not added as new maps.

Sambiasi’s dates cannot safely be standardized across copies. BnF currently dates its impression to circa 1648, while older institutional descriptions use earlier work-family dates. The 1639 imperial presentation is retained as a lost-object attestation with a possible-prototype relation, not identified as the BnF or Ghent physical copy. The historical six-copy census is attributed to Ghent’s library exhibition and is not asserted as a complete 2026 census.

Remaining coverage gaps

This inventory does not yet establish every original Sambiasi impression’s current shelfmark and access endpoint, every Martini language/edition/state, all Ruggieri draft dependencies, Boym’s other dispersed manuscripts, or every original Kangxi and Qianlong manuscript, woodcut, copperplate impression and later derivative. The British Library, National Palace Museum, First Historical Archives of China and other repositories need copy-level follow-up. The Kangxi Ripa copperplate edition is documented as a related branch, but no new unverified copy record has been fabricated for it.

Further Jesuit cartographic families, including Semedo-related maps, Brancato/Brancati’s mission geography, Gerbillon’s travel maps, later Gaubil and Hallerstein material, and Benoist’s separate world-map tradition, remain research leads rather than unsupported catalogue entries. This bounded inventory should be presented as an evidence-backed study corpus with explicit gaps, not as “all Jesuit maps.”

Vatican endpoints provide study access but the viewer reserves publication rights. Zürich explicitly uses Public Domain Mark, and Spain’s Defence copy uses CC BY 4.0 with a required provenance line. Other objects retain an unknown or institution-specific digital-image reuse status where the captured record does not state a licence. Availability of an image endpoint has not been converted into permission for unrestricted republication.

Download this source-survey report