Seminar I · What does an armorial source prove?

✦ A methods seminar · original sources required

The difference between seeing and knowing

Heraldry is unusually attractive to researchers because many of its most important documents are pictures. A shield can be recognised before a sentence is understood. The danger is that visual recognition feels like proof. It is possible to identify a coat of arms correctly and still be completely wrong about its bearer, legal significance, date, genealogy or place of origin.

Begin with a rule borrowed from historical scholarship more broadly: every assertion must be attached to a source that could, in principle, have known the asserted fact. A modern painting can be excellent evidence of how a designer represented medieval arms in 2020, yet poor evidence of what a knight painted in 1200. An armorial entry can establish that a compiler recorded a bearer and blazon in a particular text, but may not establish that the bearer had the legal entitlement the compiler assumed.

Four basic concepts give this seminar its structure. Provenance is the history of an object and its custody. Genre is what kind of thing it was created to be. Witness is the specific surviving instance of a text, seal or image. Authority describes the basis on which a source speaks, whether as an official decision, contemporary observation, family assertion or later synthesis. These terms are related but not synonyms.

The same arms in four documents

Imagine a silver shield with a red chevron surviving in four places: a thirteenth-century seal matrix, a sixteenth-century illuminated armorial, a late-seventeenth-century register and a modern web illustration. Are these four independent pieces of evidence for the same identity?

We need to establish how the four depictions came into being before counting their agreement. The illuminated armorial may have copied an earlier roll, while the government register may have recorded a later claimant’s statement rather than verified medieval descent. If the modern illustration then copied that register, its agreement is exactly what we would expect from copying. Counting all four as independent confirmations would conceal the unresolved attribution behind a larger number of pictures. That is a source dependence error.

Build a witness table. Record creation date, medium, location, original text, whether observed directly, possibility of copying, named bearer, means of attribution and known changes in later transmission. Only then ask whether the sources corroborate one another. A picture and its caption on the same folio are not always independent witnesses. Nor is a catalogue summary independent of the physical item it describes.

Find the archive, not the clip art

The BnF’s guide to the Armorial général de France identifies textual manuscript series and painted series. This makes it possible to specify whether a research claim comes from a verbal registration or an image. The Torre do Tombo’s identification of the Livro do Armeiro-Mor allows a Portuguese example to be located by book and folio. A scholarly statement such as ‘I saw it on Wikimedia’ is not an adequate substitute for either record.

Digital facsimiles transform access but introduce an extra layer. The scan is not the medieval manuscript; the photograph may omit the binding, crop its edges, adjust colours, or hide alterations visible under different illumination. Metadata may be missing or wrong. A durable citation should include the original repository and shelfmark as well as the digital permalink and the date consulted.

A critical edition should distinguish what is visible from what has been reconstructed. Missing tincture may be supplied from another witness; the edition must label it as supplied. An illegible name should not be normalized silently. If an image has been digitally redrawn from a manuscript, the redraw is an editorial object, not a neutral copy.

Evidence matrices

For every proposed claim, build an evidence matrix:

When a source is strong for one column but weak for another, say so. The answer may be provisional. Responsible uncertainty is scholarship, not timidity.

A seal dossier with deliberately separate witnesses

The seal evidence workshop begins with two Met objects: an armorial wax impression, 1981.534.7, and a bronze matrix, 1981.291. The first has a broad date range and qualified cultural attribution. The second is catalogued as nineteenth- or twentieth-century manufacture in thirteenth-century style. Both records and their photographs were consulted on 9 October 2026.

Begin by recording the physical witness you are discussing. In the wax photograph, the impression and the handwritten mount are distinguishable surfaces. In the matrix photograph, the cut face and its accompanying impression are distinguishable objects. The current museum catalogue is a third kind of witness in each case. One accession provides a documentary address; it does not make every photographed surface contemporary with every other surface.

The National Archives’ seals research guide explains matrices, impressions, and documentary attachment in its British collection context. Use that distinction to formulate questions about the Low Countries objects, while taking their specific dates and cultural attributions from their own records. A general institutional guide does not provide the missing original document for either museum object.

Next build a date inventory. Record the wax impression’s catalogue range in full, with the cultural question mark. For the matrix, separate manufacture date from stylistic reference. Give the pictured orange impression a separate row, without a supplied date or material. If you later find a documented sealing event, record its date in another field rather than silently replacing the manufacture date.

Finally, examine the authority claim. A seal’s use in authenticating a document does not make a loose photograph equivalent to the document, its sealing clause, and its custody history. A shield and a name need a defensible connection to a person before they can become evidence in a genealogy. The workshop’s unverified mount inscription cannot provide that connection simply because the handwriting looks old.

Submission: Write 750 words explaining which claims the supplied photographs support, which you attribute to the catalogues, and which require a further witness. Propose a repository search for one missing connection. Specify what a negative result would leave unresolved. The comparison notebook can hold your observations and export them with both accession records.

Graduate document exercise

Study the BnF guide and the Portuguese national archive’s description. Construct a sample catalogue record for each, using actual verified shelfmarks where provided and leaving unknown fields explicitly blank. Specify institution, collection, physical witness, dated or estimated creation, source language, script, original blazon, editorial transcription, English translation, image rights and related witnesses. Distinguish the source from its webpage.

Next, write 800 words answering whether a colored armorial image, by itself, can prove a heraldic entitlement. Do not answer as a slogan. Compare at least three different jurisdictions from our source atlas and explain how the relevant legal or customary framework changes the question.

Evaluation

An outstanding submission distinguishes visual identification from entitlement, states the chain of custody, acknowledges dependence between sources, cites an exact witness where possible, and separates observed facts from editorial conjecture. A failing submission bases descent on a surname search, reproduces images without rights review, or invents an archival shelfmark.

Core sources

• BnF / Gallica, genealogical sources and Armorial général
• Torre do Tombo, Livro do Armeiro-Mor catalogue
• Związek Szlachty Polskiej, Herbarze bibliography