# Watermark research update, prepared 5 October 2026

This note is the preparation text for the watermark increment. It is not an attribution, a mould identification, or a publication date. The drawing remains unattributed. The custodian is named by first name on the dossier; the surname used in the research file is not repeated here.

# Watermark research update — 5 October 2026

**No conclusive exact mould match has been established.** The photographed fragment shows features compatible with published lily-related designs and a possible WR reading, providing useful comparison leads. Its complete motif, letters, variant and relationship to any particular mould remain unresolved. This is a comparison programme, not a dated paper identification.

## What the photographs support

The supplied transmitted-light photographs show a marginal group of bright, curved and lobed lines against the paper’s closely spaced linear structure. The fragment recurs in several views. A structural wiremark interpretation is reasonable, but a photograph can contain lighting effects and overlapping drawing media: only continuously visible segments should be recorded. The recent images improve documentation; they do not supply an orthogonal, calibrated map of the complete sheet.

The owner’s annotated photograph labels two features W and R. Those annotations record a proposed reading, not letters established by this study. If eventually confirmed, WR would be part of the wire design impressed into the paper during manufacture, not an artist’s written signature. BM as an artist’s initials is not supported either. The alternatives should not be decided by turning an ambiguous image until it resembles preferred letters.

The owner now reports **10 laid lines per centimetre**, chain intervals **25–25–30 mm**, and a sheet **203 × 153 mm**. Earlier case material used approximately **206 × 153 mm** and other laid-line counts. These are reports from different measurements, not a new calibrated measurement by this study. Their differences must remain visible until the sheet is measured consistently. Similar chain spacings in historical paper are compatible but are not enough to identify a mould, ream or workshop.

## Why show the fragment upside down as well?

In the supplied photograph, the lobed fragment **appears inverted relative to the conventional orientation of the lower lily inside the shield in the published comparison examples**. A 180-degree rotation is therefore useful as a viewing test. It supports examining a lower-lily-in-shield hypothesis alongside the earlier crown/ornamental interpretation. It does not establish the original orientation of the sheet, the identity of the signs, or which missing elements existed.

The gallery shows the untouched photograph and its exact 180-degree rotation side by side. They are two presentations of one capture. Rotation and reflection are different operations: a photograph taken through the opposite side can reverse a wiremark laterally, whereas a rotation turns it upside down. Simulated reflections in the research archive are controlled viewing experiments, not captured verso photographs and not signs of authenticity. Imaging side and wire side must be recorded independently. [NGV database, orientation and classification context](https://www.ngv.vic.gov.au/explore/collection/conservation/watermarks-rembrandt/).

The Worcester Cathedral Library’s **A124i, 1616, figure 11** provides a photographic control for the arrangement of lily, crowned shield and lower 4/WR. It helps locate the relevant parts of a design. It has not been aligned at physical scale to the present sheet, and no shared wire geometry has been measured. [Library publication and image](https://worcestercathedrallibrary.wordpress.com/2024/08/07/watermarks-in-the-a-series-of-diocese-accounts-in-the-library-at-worcester-cathedral/).

## Comparative watermark studies

These publications classify or illustrate lily and WR designs in other bodies of paper. They are not pages for single museum objects. Those object pages are in the section “Artwork records that report a watermark”, after the Ter Borch catalogue. The labels below must remain attached to the publications that use them. **Fleur-de-lys II** in the Viola da Gamba Society appendix and **STRASBOURG LILY 2** in Rasch are descriptive classifications in those sources. They have not been mapped here to a specific Hinterding variant such as **B.y**, **A.4WR.c.a**, **E’.a.a** or **D’.a**. A Strasbourg bend is also a different design family from a lily in a shield, even though WR can occur with different emblems.

| Published comparison | What it contributes | What it cannot establish for the present sheet |
|---|---|---|
| **VDGS, Fleur-de-lys II**, pp. 272–273 | Six subdivisions, **II/1/a–b, II/2/a–b, II/3/a–b**, with paired forms and measured structure. The scheme on p. 272 shows WR below a crowned shield without an intermediate 4 represented. | A type scheme is not a radiograph. Absence of 4 in the present sheet would not by itself prove that a 4 was cut off. The terminology is not automatically a Rembrandt-watermark code. |
| **Och 1185**, II/1/a and /b | Upper and lower segments recorded on different leaves of an oblong quarto; the /a example uses ff. 9 and 10, the paired /b example ff. [76] and [75]. | Historical precedent for divided marks, not proof of the present sheet’s original sheet format. |
| **Ob MS Mus. Sch. E.406**, II/2/a and /b | Segments on ff. 3/4 and 10/13, with chains around 29.5–31.5 mm. | Broad spacing similarity is not a distinctive mould identity. |
| **Och 373**, II/3/a and /b | Segments on ff. [17]/[16] and [18]/[19]; the /a lower segment has chains 23.5/25/25 mm, while the /b upper segment has 27.5/23.5/25.5 mm. | The different intervals and geometry must be tested, not collapsed into a generic 25 mm “match”. |
| **Rasch, Geminiani, entry 1B1a, p. 8** | A **1719** Walsh–Hare impression on paper described as STRASBOURG LILY 2 with 4/WR, an **IV countermark**, and 26 mm vertical chains. | A later documented use warns against dating by motif. The title’s 1716 is not the impression’s date or a date of paper manufacture. |
| **Maas 2023, fig. 12, p. 234** | A manually traced crowned-shield lily, 4 and WR from a Cape document dated **8 May 1686**, reported as 55 × 105 mm. | It is a tracing, not diagnostic radiography; the date is the document’s use, not the mould’s manufacture. |
| **Hinterding, in *Rembrandt Now* (2022), pp. 109–110** | Strasbourg Lily **D’.a** is listed among impressions of *Jan Six*. The chapter asks whether watermark evidence should redate impressions. | It does not identify the NGV impression’s paper, and a plate date of 1647 does not date that impression. The rest of the volume was not read. |
| **Worcester Cathedral Library A124i, 1616, figure 11** | A library photograph of a diocesan account, used above as a control for lily, crowned shield and lower 4/WR. | Not a museum artwork page. It has not been aligned at physical scale, and no shared wire geometry has been measured. |

Sources: [VDGS appendix](https://vdgs.org.uk/indexmss/08%20Watermarks.pdf), [Rasch’s university-hosted study](https://geminiani.sites.uu.nl/wp-content/uploads/sites/417/2020/05/The-Sonatas-of-1716-Copies.pdf), [Maas 2023 PDF](https://research-portal.uu.nl/ws/files/207813285/725534.pdf), [Hinterding, in *Rembrandt Now*](https://www.nationalgallery.org.uk/media/5t0bre5n/rembrandt-now-online-pdf-170322.pdf), [Worcester Cathedral Library](https://worcestercathedrallibrary.wordpress.com/2024/08/07/watermarks-in-the-a-series-of-diocese-accounts-in-the-library-at-worcester-cathedral/). The VDGS PDF’s contents list uses different pagination from the actual section headings; the cited pages above are the numbers printed on the consulted pages, PDF pages 22–23. The NGV essay and the NGV watermark database explain Strasbourg marks, variants, twins and orientation. They are not an examination of this sheet. [NGV essay](https://www.ngv.vic.gov.au/essay/rembrandts-watermarked-papers-at-the-ngv/), [NGV database](https://www.ngv.vic.gov.au/explore/collection/conservation/watermarks-rembrandt/).

WR is historically associated with Wendelin Richel/Rihel, but the motif and monogram circulated more widely. It cannot identify a single active papermaker, artist or date. Object records that leave WR attached to a hunting horn, rather than to a lily, are kept with the artwork records below.

## Published examples in the Ter Borch family corpus

The earlier search of online metadata found no explicit object-specific Strasbourg-lily description. **That bounded result is superseded by the subsequently consulted printed catalogue**, which supplies seven pertinent entries. They demonstrate that lily/WR-related papers occur in the family corpus; they do not show that the present sheet shares a mould or came from the family’s supplies.

| Kettering 1988, part I | Object | Source-reported paper details |
|---|---|---|
| **GSr 62, p. 52**; PDF 90 | **A-1211**, Gerard I, *Two horses playing* | Lower half of fleur-de-lis with WR; horizontal chains 24–26 mm; Piccard XIII, 1297 ff., Strasbourg c. 1600. |
| **GJr 12, p. 98**; PDF 136 | **A-788**, *Four views of a horse-drawn sleigh*, 29 January 1631 | Fleur-de-lis in crowned shield with WR below; horizontal chains 22 mm. |
| **GJr 50, p. 122**; PDF 160 | **A-1134**, *Vegetable market in the evening*, 1634 | Upper half of crowned-shield fleur-de-lis; vertical chains 24–26 mm; Heitz 293, Heawood 1766 and Piccard 1316 comparisons. |
| **GJr 93, p. 154**; PDF 192 | **RP-T-1888-A-1797**, landscape sketchbook | Crowned lily/WR on specified leaves; vertical chains 22–24 mm; folio 36 dated 1633. Other leaves have different paper marks. |
| **H 17, p. 204**; PDF 242 | **A-949**, Harmen mentioned on object; 2 August 1648 | Upper half of crowned-shield lily; vertical chains 24–25 mm; Piccard XIII, 1333, Münster 1652 comparison. |
| **H 19, p. 206**; PDF 244 | **A-911**, 27 September 1648 | 4 and WR, **possibly** belonging to a fleur-de-lis; vertical chains 25 mm. |
| **M 46, p. 312**; PDF 350 | **A-1095**, Moses mentioned on object; copy after Testa, 1659 | Crowned-shield fleur-de-lis, 4, WR, **IHS countermark**; vertical chains 26 mm; similar to Heawood 1785, recorded in 1670. |

The catalogue’s terms “same”, “similar” and “possibly” are different evidentiary grades. Comparative watermark dates such as 1652 or 1670 are not automatic earliest dates for the drawings. The absence of technical images prevents us from checking the cited comparisons independently. **M 18 / A-1327, p. 296, is not a positive lily example:** Kettering suggests a possible lamb fragment in a crowned shield and leaves Moses/Gesina/Harmen alternatives. It should no longer lead the lily comparison request.

Only **part I** of Kettering has been supplied and consulted. The watermark plates, **pp. 776–802 in part II**, remain to be recovered. The part-I catalogue text is positive bibliographic evidence; the missing plates are not images already inspected. [Rijksmuseum publication record](https://id.rijksmuseum.nl/30038077). The accompanying reference file supplies links to each object and page locators.

## Artwork records that report a watermark

These are pages for individual works. Each reports a watermark on that object. They are not the research articles or the comparative studies above. A mark reported on another sheet is not a mould match for the present sheet.

The proposed NGV comparator is the specific *Jan Six* impression **work 23466 / accession 1242-5**. Its record reports the partial variant **B.y** and chains of 27–28 mm. An exact institutional master was not downloaded. The custodian’s screenshot is not an independently verified master, and a plate dated 1647 does not date that impression’s paper. Hinterding’s 2022 chapter, in the comparative studies above, also lists Strasbourg Lily D’.a in impressions of *Jan Six*. That is a reason to identify this impression, not a reason to change the gallery’s label. [NGV object record](https://www.ngv.vic.gov.au/explore/collection/work/23466/).

Ackland Art Museum **2017.1.86**, Esaias van de Velde, *A Cluster of Trees near a Ruin*, is catalogued about 1620–25. The record reports a partial lower Strasbourg lily, 4/WR and vertical chains of 24–26 mm. It is a non-Rembrandt comparison request. No technical image of its mark was inspected here. [Ackland record](https://peck.ackland.org/artworks/a-cluster-of-trees-near-a-ruin).

Two further object records are warnings, not lily candidates. The Glasgow Whistler catalogue leaves a partial WR open between a hunting horn in a shield and a Strasbourg lily. Getty entry 82, Jacques de Gheyn II, documents a hunting horn above WR. That passage was read in an indexed extract, and a usable full PDF was not retrieved. [Glasgow object record](https://www.etchings.arts.gla.ac.uk/catalogue/etchings/etching/?catno=K235&filename=K2350106), [Getty catalogue, entry 82](https://www.getty.edu/publications/resources/virtuallibrary/0892364807.pdf).

## Exact-match method: three JHNA articles

Three articles in the *Journal of Historians of Netherlandish Art* set out how a watermark in Rembrandt’s etchings is classified and then tested for an exact mould match. None of them examines the present sheet, and none reports a result for it. The pictures that bear on this case are named below. Each link opens that picture on the article. The plates are not copied here. The articles are CC BY-NC 4.0, and some 2020 photographs are credited to Sound and Vision Publishers BV.

### Decision trees, 2020

[C. Richard Johnson Jr., “Decision Trees for Watermark Identification in Rembrandt’s Etchings,” JHNA 12, no. 2 (2020)](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/), [DOI 10.5092/jhna.12.2.5](https://doi.org/10.5092/jhna.12.2.5). The article covers the WIRE project’s decision trees: a sequence of visible features that takes a watermark type to a variant and a subvariant, then a check against a radiograph and the chain-line spacing. It is demonstrated on seven prints in the Frick Collection. The part that applies to a lily in a shield is printed pp. 37–39. There the article treats Strasbourg Lily as frequent in Rembrandt’s etchings and counts sixteen variants.

[Figure 51](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-51) is a labelled sketch of a complete Strasbourg Lily, drawn by Margaret N. Canfield at Cornell. It is the picture used here to name the parts a full mark would have: crown, shield, petals, the 4 and letters below, and the chain line. Only the parts actually in the present photograph can be compared. Crown bands and jewels, shield sides, petal-to-shield contact, the presence and placement of 4 and letters, and the relation to chain lines are presently **unknown**.

[Figure 52](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-52) is the WIRE decision tree from type to subvariant for Strasbourg Lily in Rembrandt’s etchings, drawn by Isabella M. Dobson at Cornell. It is the picture that lists the questions. A feature outside the photograph, or hidden by the drawing, is not an observed absence, so no terminal branch is selected by answering “no” to what cannot be seen. Note 39, on printed p. 54, adds a separate limit: the presence of initials is not a feature suited to separating Strasbourg Lily variants, because variant A has subvariants with initials and subvariants without them. A possible WR reading, even if it were confirmed, would not by itself choose a branch.

[Figures 53](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-53), [54](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-54) and [55](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-55) are a finished identification, not comparators for this sheet. They identify Strasbourg Lily **E’.a.a** on the Frick *Landscape with Three Gabled Cottages* (1916.3.30). Figure 53 sets the watermark under investigation beside the published E’.a.a; the right-hand photograph is credited to Sound and Vision. Figure 54 is the branch that reaches variant E’.a. The article says that five questions in figure 52 identify the mark in figure 53, and that the “P” side of the shield overlapping the chain line identifies subvariant E’.a.a. The apostrophe means a chain line down the centre. Figure 55 aligns the chain spaces; the upper image is Sound and Vision, and the overlay is by Canfield. Hinterding’s date of about 1652 belongs to that identified watermark on that print. It is not a date for the present sheet. The article’s other plates — Foolscap, Paschal Lamb, IHS, and *The Flight into Egypt* — are other watermark types. They are not pictures of this fragment.

### Overlay comparison, 2021

[C. Richard Johnson Jr., William A. Sethares and Margaret Holben Ellis, “Overlay Videos for Quick and Accurate Watermark Identification, Comparison, and Matching,” JHNA 13, no. 2 (2021)](https://jhna.org/articles/automated-watermark-identification/), [DOI 10.5092/jhna.2021.13.2.1](https://doi.org/10.5092/jhna.2021.13.2.1). The article covers a way to mark the same points on two watermark images and play one over the other, so differences in the wires and the nearby chain lines become visible. The passages consulted are paragraphs 12–29 for the method, paragraph 42 for the marked library, paragraph 43 for a missing point, and appendix 3 for the alignment-error calculation.

The pictures that apply to a cut sheet are the second case study, [figures 11–17](https://jhna.org/articles/automated-watermark-identification/#fig-11). They compare a full Foolscap with Five-Pointed Collar with a fragment of that type, and the overlay includes the chain lines beside the mark. That is why the present fragment is not automatically unusable, and also why it is not yet matched. The article says its decision tree is for full watermarks, because the distinguishing features are spread through the design. A fragment still has to be shown to agree with one candidate and to disagree with the near alternatives. Those pictures are Foolscap watermarks, not a lily in a shield, so they are not comparators for this sheet. The skipped-point rule there is a numbering rule for a missing collar intersection on a Foolscap. It is not a licence to score an unseen feature on this sheet as a difference.

Figures 2–10 are a Strasbourg Bend case. [Figure 8](https://jhna.org/articles/automated-watermark-identification/#fig-8) is that type’s decision tree, and figure 10 is the overlay that separates the twins D’.a.a and D’.a.b. A Strasbourg bend is a different design family from a lily in a shield, even where WR occurs with both. Those pictures are named so that the distinction stays visible. They are not candidates for the present fragment.

### Exact matches with dated documents, 2026

[C. Richard Johnson Jr. and Andrew C. Weislogel, “Assisting Creation Dating of Rembrandt Prints using Exact Watermark Matches Between the Prints and Seventeenth-Century Dutch Documents,” JHNA 18, no. 1 (2026)](https://jhna.org/articles/assisting-creation-dating-of-rembrandt-prints-using-exact-watermark-matches-between-the-prints-and-seventeenth-century-dutch-documents/), [DOI 10.5092/jhna.2026.18.1.3](https://doi.org/10.5092/jhna.2026.18.1.3). The article covers exact watermark matches between Rembrandt prints and dated seventeenth-century Dutch documents, as a way to assist the dating of when an impression was printed, not when the plate was invented. Its examples are Foolscap papers in prints and in the Zeeland Archives. The passages consulted are p. 3 §4, p. 7 §16, p. 9 §§20–22, and p. 26, note 22.

As the article uses Briquet, an exact match between a watermark on a sheet of interest and one on a dated sheet can support assigning that date, and the wires of a watermark remained undamaged on the mould for about two years for popular paper sizes. That implies about a two-year maximum gap in manufacture between identical marks. Twin moulds, used in rotation, carry nearly identical marks and are not the same mould. The article then adds a further period for storage and distribution between manufacture and use, and gives four years as an example of that further period (note 22). It tests a proposed match with an overlay, not with a similar motif. The two-year figure is not a universal life for every wire, and it is not a date for this drawing. The article reports no result for this sheet.

[Figures 1 and 2](https://jhna.org/articles/assisting-creation-dating-of-rembrandt-prints-using-exact-watermark-matches-between-the-prints-and-seventeenth-century-dutch-documents/#fig-1) are photographs of a modern laid mould and of how a watermark wire is fastened to it. They show the object an exact mould match would have to agree with. They are not a picture of this fragment. [Figures 4 and 5](https://jhna.org/articles/assisting-creation-dating-of-rembrandt-prints-using-exact-watermark-matches-between-the-prints-and-seventeenth-century-dutch-documents/#fig-4) show a point-marked Foolscap with Five-Pointed Collar and an overlay against a second image of that type. [Figure 8](https://jhna.org/articles/assisting-creation-dating-of-rembrandt-prints-using-exact-watermark-matches-between-the-prints-and-seventeenth-century-dutch-documents/#fig-8) is Rembrandt’s *The Flute Player*, the print whose paper supplies one of those Foolscap examples. These are illustrations of the method on another watermark family. They are not placed beside the present fragment.

The custodian’s side-by-side comparison and apparent 3D screenshot are preserved as proposals. Neither supplies a registered wire-for-wire overlay at common physical scale, an instrument record, a calibrated height axis or raw topographic data. The drawing’s inked forms remain prominent in the apparent relief. An intensity-to-height visualization can make a photograph look three-dimensional, but its height would encode brightness rather than paper thickness. Without a documented procedure we cannot identify the software or affirm that this was the method used. It does not demonstrate removal of surface ink or a “felt-side authenticity effect”. [ImageJ’s explanation of intensity-based surface plots](https://imagej.net/ij/plugins/surface-plot-3d).

The accompanying process diagram shows this study’s actual route: acquire the sheet, test orientation, classify motif and letters, discriminate variants, then test mould geometry. It is **not figure 52**, not a replacement WIRE tree, and it assigns the present sheet no WIRE code. [watermark-decision-path.svg](/research/rembrandt-school/young-draughtsman/dossier/notes/watermark-decision-path.svg).

## Evidence for the lead, evidence limiting it, and the next test

**Support:** a repeatedly photographed structural-looking fragment; a lobed shape worth testing against lily geometry; published lily/WR designs in several relevant corpora; documented divided watermark examples; and concrete Ter Borch object/page references.

**Limits and alternatives:** unresolved letters and complete motif; missing crown/shield diagnostics; uncalibrated and oblique captures; non-exclusive spacing; competing WR-bearing emblems; later documented uses; and no verified, registered comparison demonstrating diagnostic wire correspondence. Old paper carrying a later drawing remains possible even if a historic paper type is eventually identified. The present limitation is not a completed negative test excluding every candidate.

The most useful next step is a **paper conservator-led full-sheet structural record**, with scale, sides, all edges, chain/laid geometry and documented imaging settings. Begin with controlled transmitted light where it resolves the structure. If media conceal it, the conservator should choose an appropriate method, such as low-energy X-radiography, after assessing the sheet’s condition. Beta/electron radiography has particular handling/contact requirements and is not interchangeable with a photographic filter. [Rijksmuseum project](https://www.rijksmuseum.nl/en/research/our-research/conservation-science/science/the-watermarks-project), [Bunting’s first-hand account](https://cool.culturalheritage.org/byform/mailing-lists/cdl/1999/0792.html).

A focused enquiry to **Aurora Belli** is proposed to ask about existing radiographs and further references for the seven Kettering objects, together with the part-II plates. No enquiry has been sent and no availability or acceptance of a private object is assumed. Comparison should include alternatives and twin moulds, preserve the registration and residuals, and state which diagnostic features agree, disagree or remain unreadable. Authorship remains a separate question.

## Decision path

This is the study’s evidence-acquisition sequence, not an extracted branch of WIRE and not a classification result. The diagram is published with the dossier as [watermark-decision-path.svg](/research/rembrandt-school/young-draughtsman/dossier/notes/watermark-decision-path.svg). It is an acquisition sequence, not a WIRE tree. The WIRE Strasbourg-lily tree is [Johnson 2020, figure 52](https://jhna.org/articles/decision-trees-for-watermark-identification-in-rembrandts-etchings/#fig-52), discussed with the other two JHNA articles above.

1. **Acquire:** complete, scaled sheet, both sides, all edges, documented chain and laid lines. Current outcome: insufficient for calibrated comparison.
2. **Orient:** test 0° and 180° against emblem geometry. Record camera side/wire side separately. Current outcome: 180° useful under the lower-lily-in-shield interpretation; conditional.
3. **Classify:** identify surviving motif, then read any wire letters. Current outcome: lily-related lead; WR unconfirmed. Retain crown/other ornamental or emblem possibilities.
4. **Discriminate:** use crown bands, jewels, shield curves, petal contacts, initials and chain intersections. Current outcome: too many unknowns to choose a WIRE variant.
5. **Test mould:** compare identified technical masters, global registration, diagnostic wire deformations and complete neighbouring chain geometry; include twins and alternative candidates. Current outcome: no conclusive exact match established.
6. **Interpret dates separately:** manufacture/use date of comparator, possible storage or later reuse, date of the mark on the present sheet, and authorship are separate findings.

At any step **UNKNOWN** is a valid endpoint. It is not **NO** and not an invitation to reconstruct missing features. Do not display a highlighted path to B.y/E’.a.a/D’.a or any other terminal class for the present sheet.

## Examination plan

The immediate question is the paper’s structural mark, not the artist’s identity. Ask a paper conservator to assess the sheet before selecting acquisition or handling methods. The research files do not replace that assessment.

| Priority | Record needed | Purpose and interpretive boundary |
|---|---|---|
|1|Complete recto/verso, orthogonal, in-plane millimetre scale; maximum/minimum dimensions; all four edges|Resolve 203 versus 206 mm and coordinate orientation. No inference of quarto format or chronology from dimensions alone.|
|2|Whole-sheet transmitted-light master where safe and effective; RAW/16-bit archival files; unedited acquisition settings|Record continuously visible wire segments, distances to edges, laid density in several fields, and the complete chain sequence. Do not fill gaps concealed by the drawing.|
|3|If the medium obscures structure, conservator-selected technical imaging, potentially low-energy X-radiography|Acquire structure independently of surface brightness. Beta/electron techniques may involve contact, pressure or vacuum and require a separate suitability assessment. No domestic handling prescription.|
|4|Verified comparator masters with accession, technical-image ID, camera/radiograph side, physical scale and capture method|Retrieve NGV 1242-5; request Worcester A124i, VDGS/Oxford examples and the seven Kettering objects. Published tracings and schemes are for classification leads, not submillimetre proof.|
|5|Specialist-reviewed global registration, visible diagnostic landmarks, chain/laid relationships and wire defects; report disagreements|Test exact mould, twin, related or incompatible geometry. Preserve source images, transforms and residuals; no local warping to force an apparent match.|

Proposed enquiry: Aurora Belli, named on the [Rijksmuseum Watermarks Project](https://www.rijksmuseum.nl/en/research/our-research/conservation-science/science/the-watermarks-project). Ask whether relevant technical images or references exist for A-1211, A-788, A-1134, A-1797, A-949, A-911 and A-1095. Ask about Kettering part II, pp. 776–802. No enquiry has been sent. The former A-1327 lead is a possible lamb, not a verified lily.

Report an uncertain or negative result as carefully as a positive one. Even an exact mould relationship would not, by itself, identify a ream, ownership, a workshop, an artist, or when the drawing was executed on that paper.

## Audit of claims and remaining gaps

Current synthesis: 2026-10-05. All substantive current watermark notes are included here and in the update, evidence table, source registry, bibliography and examination plan. Historical notes are preserved in the private research archive with dates and supersession labels. They must not be treated as current findings when later sources correct them.

## What “similarities and leads” means here

The positive observations are the repeated bright lobed fragment, nearby laid structure, and a plausible geometric interpretation that can be tested against published lily designs. A possible WR reading is a hypothesis, not another established similarity. The historical literature adds precedents for partial marks and divided sheets, relevant object-specific Ter Borch references, and broad chain spacings comparable with the owner’s reports. Those contextual facts do not constitute geometric matches to the present sheet.

An exact match has not been demonstrated. Nor have all published candidates been subjected to a complete negative test. The surviving geometry may ultimately be informative, but the present images do not let us discriminate same mould, twin mould, related motif or a different emblem reliably.

## Corrections of specific claims

| Received or previous claim | Current assessment | Evidence and consequence |
|---|---|---|
| A three-lobed flower is always a pendant below letters, never above them | Unsupported universal rule; the cited Briquet number does not document it | Briquet 7104 is a lily in a circle, Ferrara 1560. The inspected complete designs distinguish a lily inside a shield from 4/WR below it. Do not use the citation to force an orientation. |
| the present sheet necessarily preserves the lower WR pendant | Unverified | The photographed lobed form may instead belong to the lower lily inside a shield. The letters and full anatomical arrangement remain unresolved. |
| Rotating 180° proves WR, or recto/verso reversal authenticates the sheet | Unverified inference | Rotation and reflection differ. Side information must be established; neither transform is a chronological test. |
| Missing 4 and crown prove a quarto cut | Unverified | VDGS documents divided marks and a scheme without 4. the present sheet’s original full-sheet dimensions and lost design have not been recovered. Current sheet size alone does not establish bibliographic quarto format. |
| “10 lines/cm and 25–25–30 mm” prove the same paper as NGV Jan Six | Compatible but non-diagnostic | Owner reports differ from the NGV record’s 27–28 mm chain intervals. A few intervals are not a complete chain pattern. Earlier case data reported 17 laid lines/cm and a height of 206 mm; the newer report is 10 and 203 mm. None is adopted as an independent measurement here. |
| NGV control is watermark ID 126508 | Unverified identifier | The earlier inspected technical-image URL used Ff126499; no equivalence with 126508 was verified. Keep work 23466 and accession 1242-5 as the object identity. No original technical image was downloaded. |
| Jan Six is a single 1647 watermark standard | Incorrect abstraction | Identify the individual impression and its paper. NGV reports B.y; Hinterding’s 2022 chapter lists D’.a in Jan Six impressions. Plate date and impression date differ. |
| All moulds lasted a maximum of two years, so the present sheet was made around 1647 | Overstatement and unsupported date inference | The 2026 article uses an approximate undamaged-wire estimate under specified conditions and adds storage/distribution considerations. A drawing date would remain separate even after paper identification. |
| “3D” removes the ink and proves identical curvature | Unverified | No calibrated topographic data, scale, instrument, acquisition protocol or registration residuals supplied. Inked drawing features remain in the visualization. An apparent height rendering cannot recover physical geometry lost or obscured in its input. |
| No relevant lily/WR examples exist in the Ter Borch literature | Earlier bounded search superseded | Kettering part I now supplies seven pertinent entries. This is positive source evidence but not a mould or authorship link. |
| Moses A-1327 is a Strasbourg-lily example | Not supported by the fuller source | Kettering M 18 describes a possible lamb fragment in a crowned shield and allows Gesina/Harmen alternatives. Do not count it among the seven positives. |
| Dresden C 1421 lacks a watermark, excluding common paper | Does not follow | The earlier institutional report of no watermark does not exclude a marked full sheet from which a small unmarked piece was cut. No paper relationship has been established either. |
| Gravell SLD.051.1 is the present sheet’s mark | Not established | Earlier dossier research identifies a different Folger document, dated 1651, with its own dimensions and chains. Motif similarity is not object or mould identity. The record was not newly re-retrieved here. |

The underlying source locators and reading scope appear in the source note. The laboratory specification explains what data could resolve these questions.

## What was supplied, what was found, and what was read

The 2021 overlay PDF, Cornell description, five original 2020 figures and the 2022 National Gallery volume were downloaded from the identified scholarly/institutional servers. The relevant passages were read; the whole 140-page National Gallery volume and the whole Kettering catalogue were not exhaustively studied entry by entry. The full Kettering part-I PDF is retained in the research delivery; part II is absent.

## Present data gaps

1. A single calibrated full-sheet structural capture of the present sheet; recorded recto/verso, wire side, scale, all edges and uncertainty.
2. A dependable tracing of only visible continuous wire segments; no reconstruction of obscured letters or missing crown.
3. Original comparator masters, particularly the specific NGV impression and the seven Kettering objects. The NGV image request returned HTTP403; it has not been bypassed or replaced by a guessed image.
4. Kettering part II, especially the watermark section, pp. 776–802. The part-I references are not a substitute for its plates.
5. Direct inspection of the original Churchill 401, Heawood 1746/1780/1785 and the specific Hinterding volume-I p. 38 and Ash–Fletcher p. 11 cited in owner material. Some are source-reported comparisons; that is not our independent verification of each original plate/page.
6. The full Getty European Drawings 3 PDF and *Looking at Paper* PDF: requests returned non-PDF HTML and no usable PDFs were saved. Bibliographic links remain, with scope limits.
7. Full texts of Dexter 1999 and the 2026 *Imago Mundi* lead, plus the Whistler technical-image library. They remain leads and are not used to identify the present sheet.
8. the custodian’s original acquisition/processing files for the apparent 3D image. It remains possible he has further data, but none has been supplied here.

## How a result would change the case

An identified motif would narrow a search. A securely read WR would support a paper-mark classification, not an artist’s signature. A variant would narrow the control group. A reviewed exact mould relationship with dated documentary sheets could constrain a manufacturing/use context. Each of these requires evidence beyond the previous step. None would alone settle ream, ownership, workshop, authorship, sitter, execution date or authentication. An incompatible diagnostic feature or material could reject a candidate; a compatible result leaves multiple hypotheses open.

The best next investment is acquiring better structural data, rather than further transformation of the same uncalibrated photographs. The proposed query to Aurora Belli now has concrete inventory numbers and source pages; no message or request for treatment has been sent.