Historians Publish Facial Reconstruction Claiming to Show Anne Boleyn's Appearance

On May 2, 2026, headlines across British media posed a familiar question with renewed urgency: "Is this the real face of Anne Boleyn?" [1]. A team led by Professor Hassan Ugail at the University of Bradford's Centre for Visual Computing, working with independent historian Karen L. Davies, announced they had used AI-powered facial recognition to identify a Hans Holbein the Younger sketch — previously catalogued as "An Unidentified Woman" — as a previously unknown portrait of Henry VIII's second wife [2]. The study, published in npj Heritage Science, a peer-reviewed Nature portfolio journal, generated immediate international coverage and equally immediate criticism [3].

The claim is bold. But the evidentiary path from a 16th-century chalk drawing to a definitive identification is anything but straightforward.

What the Researchers Actually Did

The Bradford team developed what they call a "Working Likeness Methodology" — a framework that combines computational facial recognition with documentary and art-historical analysis [3]. Using a deep learning model based on AdaFace embeddings, a type of neural network trained to measure similarity between faces, they compared digital copies of Holbein's preparatory sketches held in the Royal Collection [2].

The central problem they faced is that no confirmed lifetime portrait of Anne Boleyn exists. To work around this, the researchers built a biometric family network: they compared facial geometry across sketches against authenticated portraits of Anne's daughter Elizabeth I, her first cousins, and other Boleyn relatives, looking for family clustering patterns [2] [4].

Their conclusion: a sketch catalogued as RCIN 912190, the "Unidentified Woman," shows stable facial similarity to an authenticated William Scrots portrait of Elizabeth I and aligns closely with contemporary written descriptions of Anne [3]. The sketch traditionally labelled as Anne Boleyn (RCIN 912189), Davies argues, actually conflicts with eyewitness accounts — showing light hair and a full chin, where contemporaries described dark hair and a slender neck [4].

Professor Ugail told reporters: "We essentially ran facial recognition across the entire corpus of Hans Holbein, and when we actually look at all this together, looking at family clustering, the unidentified woman seems to cluster with the royal family. We think it might take the place of Anne Boleyn" [1].

No Bones About It: The Absence of Skeletal Evidence

Unlike many historical facial reconstructions, this study does not claim to work from remains. And for good reason: Anne Boleyn's skeletal record is essentially nonexistent as a forensic resource.

After her execution on May 19, 1536, Anne's head and body were placed in an elm chest and buried beneath the chancel pavement of the Chapel of St Peter ad Vincula at the Tower of London [5]. During Victorian restoration work in 1876-1877, remains were exhumed. A Dr. Mouat examined one set of bones and concluded they belonged to "a female of between twenty-five and thirty years of age, of a delicate frame of body," noting a small atlas vertebra that he linked to accounts of the queen's "little neck" [6].

But that identification was based on circumstantial physical characteristics, not scientific analysis. No DNA testing has ever been performed on the remains [5] [6]. No skull cast was taken before reinterment [7]. The bones were reburied — not in their original location, but to the left of the altar, where they remain today [6]. Without verified remains, no forensic facial reconstruction from Anne Boleyn's skull has been or can be conducted. Claims circulating on social media that a waxwork by artist Emily Pooley represents a "skull-based reconstruction" are incorrect; Pooley based her model on Holbein sketches and a living friend's face [7].

How Forensic Facial Reconstruction Actually Works

Standard forensic facial reconstruction requires a skull. Practitioners use tissue-depth markers at anatomical landmarks, applying population-specific data about average soft tissue thickness for a given age, sex, and ancestry [8]. Muscles and skin are then built up in clay or digitally. In criminal investigations, the goal is typically to produce a likeness recognizable enough for someone who knew the deceased to make an identification — not to produce a photographic duplicate [9].

Even under optimal conditions, accuracy is variable. The field lacks a single recognized standardized methodology [8]. Available tissue-depth datasets remain limited across age ranges, sexes, and body types [8]. Practitioners acknowledge the process involves substantial artistic interpretation: two reconstructions from the same skull can look markedly different depending on the practitioner.

The Bradford study sidesteps these issues entirely by working not from a skull but from existing artworks — which introduces a wholly different set of problems.

Source: OpenAlex

Data as of Jan 1, 2026CSV

Research publication data from OpenAlex shows the broader field of facial reconstruction research has expanded significantly, with over 110,000 papers published to date and a peak of 14,251 papers in 2024 alone [10]. The growth reflects increasing interest in the technology, but also the wide spectrum of methodological rigor within the field.

What 16th-Century Sources Actually Said

Contemporary descriptions of Anne Boleyn are sparse and often contradictory, shaped by the political allegiances of the writers. The Venetian ambassador described her as "not one of the handsomest women in the world," with a "swarthy complexion, long neck, wide mouth," and striking black eyes [11]. Other accounts noted dark hair, an oval face, and olive skin [12]. She was considered rather tall by 16th-century standards.

However, Anne's central role in England's break from the Catholic Church made her a deeply polarizing figure. Descriptions written after her fall and execution were frequently hostile. The persistent myth that she had six fingers on one hand, for example, originated decades after her death and has no contemporary support [11].

The Bradford team argues that RCIN 912190 — showing a dark-haired woman — aligns better with these descriptions than the traditionally accepted Holbein sketch, which depicts a lighter-haired subject in informal dress with an 18th-century inscription [4]. This is one of the study's stronger points: whatever the AI analysis adds, the documentary case for reconsidering the traditional identification has some independent basis.

A Long History of Proposed Identifications

The latest study joins a long queue of attempts to pin down Anne Boleyn's face. The only undisputed lifetime likeness is a small lead medal from 1534, held by the British Museum, which shows a long oval face with high cheekbones [12]. Everything else is contested.

Source: Historical records compilation

Data as of May 3, 2026CSV

Around 30 portraits or drawings have been proposed as depicting Anne Boleyn at various points, and no two look exactly alike [12]. The most recognizable image — the National Portrait Gallery's posthumous portrait showing a woman in a "B" pendant necklace — may not depict Anne at all. Recent scholarship has noted its striking resemblance to three royal portraits painted in the late 16th and early 17th centuries, and some researchers have questioned whether it actually shows Elizabeth I or another figure [11] [13].

A lost painting called the Lumley Portrait, possibly painted from life by Joos van Cleve around 1532, was sold to a private collector in 1773 and photographed in 1926 before vanishing again [14]. The Lumley-based analysis suggests Anne had dark red or auburn hair and a fuller mouth than derivative portraits indicate [14].

In 2015, Artnet reported on a separate facial recognition study comparing the 1534 medal to various portraits, with mixed results [15]. Digital reconstructions by Royalty Now Studios and other content creators have generated millions of views [14]. Each new attempt generates headlines. None has settled the question.

Who Is Behind This Research?

The study's lead author, Karen L. Davies, is an independent historian and novelist who, by her own account, self-funded her research while working as a cleaner [4] [16]. She connected with Professor Ugail's visual computing team after mentioning her project to a client. Davies describes herself as pursuing "research projects and mysteries that institutions have left unsolved" [16].

Professor Ugail is a recognized expert in visual computing at the University of Bradford, though his primary expertise is in computational methods rather than Tudor history or art history [4]. The third author, D.G. Stork, is also listed on the paper [3].

The study was published in npj Heritage Science, a peer-reviewed open-access journal in the Nature portfolio, which lends it more credibility than a self-published claim [3]. No specific external funding source or associated book deal has been publicly identified at the time of publication. Davies does maintain a website and social media presence focused on her Anne Boleyn research [16], but this is consistent with standard academic self-promotion rather than evidence of a commercial scheme.

The Art Historical Pushback

The criticism has been sharp. Art historian Dr. Bendor Grosvenor, known for his work on portrait identification and his BBC television appearances, called the study "a load of old phooey" and "a load of rubbish" [1] [4].

Grosvenor's objection is methodological: "They're trying to take a 21st century technology, which is facial recognition and AI, and assume that it can just be put immediately back onto drawings and artworks made 500 years earlier in the 16th century. The methodology here is so profoundly misplaced that it almost is impossible for it to get off the ground in the first place" [1].

The core of this critique is that Holbein's drawings are not photographs. They are artistic interpretations governed by the conventions, tools, and intentions of a 16th-century court painter. Proportions may be stylized. Features may be flattered or de-emphasized. Angles vary. Materials — chalk on paper — impose their own constraints. Facial recognition algorithms trained on photographic data may detect patterns in these drawings, but whether those patterns correspond to actual biological facial geometry is an open and unresolved question [17].

Grosvenor has also raised broader concerns about AI in art attribution, arguing that while such technology may become a useful supplementary tool, it "lacks human researchers' ability to engage in primary study of techniques and the capacity to contextualize an artwork within history" [17].

Other skeptics note that the study's logic is partly circular: without a confirmed portrait to anchor the analysis, the researchers must assume that family resemblance to Elizabeth I and cousins is sufficient to identify Anne — but family resemblance in 16th-century chalk drawings is exactly the kind of signal most vulnerable to artistic convention rather than biological reality.

Whose Face Gets Reconstructed — and Why

The Anne Boleyn identification debate sits within a broader pattern that scholars of gender and history have documented: the disproportionate focus on female historical figures' physical appearance.

Anne Boleyn's political accomplishments — her role in triggering England's break with Rome, her patronage of religious reform, her influence on the English court — receive consistently less popular attention than the question of what she looked like [18]. Susan Bordo's The Creation of Anne Boleyn traces how successive generations have reimagined her "as a whore, martyr, cautionary tale, proto 'mean girl,' feminist icon, and everything in between," with her appearance serving as a canvas for each era's anxieties about women [19].

Henry VIII, by contrast, has also been the subject of facial reconstructions and digital restorations [20]. But the framing differs. Reconstructions of Henry tend to focus on his physical transformation — the athletic young king versus the obese older monarch — as an index of power, excess, and political change. Reconstructions of Anne focus on beauty, desirability, and the mystery of attraction. The question "what did she look like?" implicitly asks "what did he see in her?" — centering her story on his gaze rather than her agency.

This pattern is not unique to Anne Boleyn. From Cleopatra to Nefertiti to the "Ice Maiden" of Peru, female historical figures are routinely subjected to appearance-focused analysis that frames physical beauty as their most historically significant attribute [18]. Male counterparts — warriors, kings, political leaders — are more often reconstructed to illustrate character, power, or historical change.

What This Latest Study Actually Tells Us

The Bradford study makes a reasonable documentary case for reconsidering which Holbein sketch might depict Anne Boleyn. The observation that the traditionally accepted sketch conflicts with contemporary written descriptions is not new, but it is worth revisiting systematically. The peer-reviewed publication in a Nature portfolio journal distinguishes it from social media speculation.

But the AI facial recognition component — the element that generated the headlines — rests on assumptions that art historians find deeply problematic. Applying biometric algorithms designed for photographic images to 16th-century chalk drawings involves a category error that peer review in a heritage science journal may not have been equipped to fully evaluate, given that the relevant expertise sits in art history rather than computer science.

The study does not tell us what Anne Boleyn looked like. It tells us that one Holbein sketch shares certain measurable facial proportions with portraits of Anne's relatives — according to an algorithm trained on photographs, applied to drawings. Whether that finding is historically meaningful or an artifact of the method remains genuinely uncertain.

Nearly five centuries after her execution, Anne Boleyn's face continues to be reconstructed, reinterpreted, and debated. Each new attempt reflects the technology and preoccupations of its own era more than it recovers a lost reality. The 2026 Bradford study, whatever its merits, is unlikely to be the last word. The real Anne Boleyn — the diplomat, the reformer, the political actor — remains obscured, as she has always been, behind the question of her face.