Richard Wilson's Figurative Constraint: A Critical Review

Iconic affordance, anatomical convergence, stacked constraints, pareidolia and the statistical questions left unanswered by the 2026 Fontmaure presentation.

In January 2026, Richard Wilson wrote to me:

I have spent some time reading through your site and blog material

He proceeded to explain that much of what I had presented as pioneering research had already been discussed for more than a century.

Interesting.

On 6 June 2026, Wilson delivered Bear-like Forms at Fontmaure? Testing Figurative Constraint in Neanderthal Lithics at the Prehistoric Society's York conference.

His presentation introduced a six-component assessment of what he calls "iconic affordance", incorporating anatomical convergence, modification, orientation, recurring patterns and material properties.

These are closely related to the methodological questions I have been investigating and publishing for years.

Wilson's familiarity with my work is not speculation. He acknowledged reading it.

So what exactly is new about his latest methodology, what was already established, and where did the individual components originate?

1. Figurative Constraint versus Image Logic and Stacked Constraints

My research distinguishes two related evidential problems.

Tool Logic: Does the physical evidence demonstrate human modification, including coherent flake removals, fracture characteristics and technological organisation?

Image Logic: Do physically grounded features form anatomically coherent, recurring and recognisable representations?

My Stacked Constraints approach examines the combined significance of these observations, rather than relying on any single resemblance or modification.

Now compare this with Wilson's six iconic affordance criteria on slide 8 of his 2026 presentation.

Wilson's 2026 terminology	Closely corresponding elements of my research
Consistency of perceived form	Coherent Image Logic and anatomical consistency
Anatomical convergence	Anatomical topology, proportions and related visual features
Modification–affordance alignment	Technical grounding; flaking and modifications corresponding to figurative features
Orientation stability	Natural presentation, hand-fit and orientation
Patterned/composite complexity	Recurring motifs, combinations of features and assemblage repetition
Material salience	Patina, surface properties, inclusions and other material evidence

These are not identical frameworks, and the criteria do not appear in precisely the same order.

Nevertheless, the correspondence in their analytical functions is substantial.

My May 2017 publication, Eoliths, flint tools and figure stones, examined the combined probability of anatomically related features, flake removal, repeated topology and perceived imagery.

I specifically considered the significance of eye-like features occurring alongside correctly positioned mouths, ears, noses and other details.

My 2021 portable rock art article explored visual triggers, partial recognition, multiple imagery and orientation.

My subsequent Flint Flake Mechanics and Stacked Constraints work formalised the physical agency argument, while my Eoliths.org identification methodology brings the mechanical and visual assessments together.

The important similarity is not any single criterion. It is the combination of criteria into an assessment intended to distinguish accidental resemblance from physically supported figuration.

2. The Statistical Problem: Counting Bears Is Not Testing Their Anatomy

Wilson's 2018 Fontmaure paper, Eschewing Bear Tracks, examined 501 lithic specimens (he claims, as there are not 501 different 'bear' items in the paper).

His principal statistical tables classified the perceived imagery and recorded 467 specimens as apparently iconographic.

But where was the systematic statistical examination of the actual anatomical features?

How many specimens possessed recognisable eyes, ears, noses and muzzles?

How frequently did these occur in appropriate anatomical relationships?

How many features coincided with demonstrable flake removals or retouch?

How many specimens displayed corresponding imagery on opposite faces or through rotation?

And how did these combinations compare with a properly selected natural control sample?

These are the questions that test whether combinations of physical and anatomical features provide evidence beyond subjective resemblance.

Wilson was already observing such features in 2018. Indeed, his published reply reported that 86 of 483 modified specimens, approximately 18%, exhibited related bear interpretations following rotation.

Yet his main statistical tables did not systematically present joint frequencies for the anatomical and mechanical features supporting those interpretations.

The comparison also encountered a serious sampling problem.

His Table 1 recorded perceived iconography in 93.2% of modified specimens and 94.4% of the unmodified specimens.

Wilson subsequently acknowledged that the unmodified stones had been selected for their apparent imagery and were therefore not directly comparable.

He reported a further comparison using 100 unmodified stones, but that did not supply the missing feature-by-feature model.

If anatomical convergence and modification alignment are central to the 2026 framework, why were these relationships not comprehensively quantified in the 2018 study?

3. From Cumulative Observations to Probability: Where Is the Actual Test?

Wilson's 2010 paper, Cultural Cobbles or a Load of Cobblers?, already discussed multiple visual cues, possible modification and the cumulative significance of features suggestive of intentionality.

That is an important precedent.

But the paper did not publish a calibrated probability model for combined anatomical and mechanical constraints. Wilson stated that the detailed results of one visual-cue assessment were omitted.

In my 2017 article, I explicitly explored a hypothetical probability problem: a spot and curved feature combining to form an apparent eye, followed by the additional conditions required for anatomically appropriate ears, mouth, nose and other features.

The point was that several coordinated physical conditions present a different chance problem from an isolated resemblance.

That was an illustrative probability argument, not a completed statistical model. A rigorous test would also need to account for dependence between features.

Interestingly, in his 2018 published response, Wilson expressed the ambition to develop an attribute-based system producing realistic probability ratings.

Now, eight years later, his 2026 presentation offers six descriptive criteria.

But where are their empirically established weights, combined probabilities, control distributions or demonstrated likelihood ratios?

Slide 18 still identifies independent observer validation as a future step.

In 2018 he proposed probability ratings. In 2026 he presents a more formal vocabulary. Where are the probabilities?

A graded rubric may be useful, but assigning criteria and calculating their evidential significance are not the same achievement.

4. If It Was Established a Century Ago, Identify the Method

Wilson's January 2026 email dismissed claims of pioneering work by invoking historical archaeological research, including Boucher de Perthes, Rutot, Commont, Boule, Breuil and Bednarik.

In our later exchange, he pointed to Raymond, Hervé and Dharvent.

Their historical relevance to the study of eoliths, lithic modification and figure stones is not disputed.

But a historical discussion of suggestive stones does not, by itself, establish the origin of a particular modern analytical framework.

Which specific publications established the combined tests now described as anatomical convergence, modification–affordance alignment, orientation stability and patterned complexity?

Where were their operational definitions and statistical relationships demonstrated?

Wilson has earlier published work of his own, and his 2018 paper includes relevant methodological contributions and commentary from other researchers. Those are legitimate sources that need to be considered.

But citations should explain the development of the actual method, not simply establish that the general subject is old.

There is another relevant fact.

Wilson acknowledged reading my published research in January 2026, before delivering his June presentation.

That establishes familiarity, although it does not establish when he first developed the six criteria or prove that he adopted them from my work.

Nevertheless, the methodological parallels and that documented familiarity make the question of acknowledgement reasonable.

If the framework was developed independently, its intellectual history should be traceable.

If my research influenced its later formulation, that contribution deserves recognition.

5. What Does "Iconic Affordance" Actually Add?

The test of a new scientific method is not the sophistication of its terminology.

It is whether the method produces better, more reproducible results.

Wilson's 2026 presentation proposes an analytical sequence beginning with pareidolia, proceeding through lithic organisation and iconic affordance, and ultimately requiring observer-level testing.

That may be a useful development.

But does it distinguish intentionally modified figurative stones from convincing natural mimics more successfully than the earlier approaches?

Does its scoring reliably separate stronger from weaker examples?

Can independent observers reproduce the assessments?

And does it demonstrate a meaningful cumulative effect when mechanical and anatomical criteria converge?

These questions require data.

The presentation does not provide completed validation establishing those outcomes.

Until then, iconic affordance remains a proposed evaluative framework, not a demonstrated statistical solution to the figure-stone problem.

Conclusion: New Terminology Is Not New Evidence

I have no objection to Richard Wilson improving his methods. I welcome serious attempts to investigate possible prehistoric imagery.

But the correspondence between his latest framework and the questions and methodologies I have publicly developed is considerable.

My own publications establish that I independently developed specific anatomical, topological, physical and probabilistic arguments before his 2018 Fontmaure study.

And his January 2026 email confirms that he had read my work.

The important question is how these lines of research relate to one another.

What is genuinely new in the 2026 framework? Which earlier publications supply its methods? What has been statistically validated? And what acknowledgement, if any, is due to contemporary research Wilson knew about?

I raised the issue directly. The explanation I received relied on historical authorities rather than a specific comparison of the modern methodologies.

That leaves the question unresolved.

There is an irony in demanding rigorous evidence for the origins of physical features while leaving the intellectual development of the interpretative method insufficiently explained.

So, let me get this straight. Richard tells me my research offers nothing new because it was all done over a century ago, yet months later presents a strikingly similar methodology under the shiny new labels of “figurative constraint” and “iconic affordance”. Apparently, these ideas are century-old irrelevancies when I publish them, but worthy of a conference presentation when Richard does. Perhaps he could spare us the historical name-dropping and simply show us where this supposedly century-old methodology actually exists? Not the general ideas, Richard. The method.

---

This is a critical review of published methodologies, statistical arguments and documented correspondence. Similarity between approaches does not, by itself, prove plagiarism or intentional misattribution.

Sources and methodological comparisons

My published research

• Eoliths: Flint Tools and Figure Stones (2017)

• Portable Rock Art and Figure Stones (2021)

• Flint Flake Mechanics and Stacked Constraints (2025)

• Image Logic and Portable Rock Art Identification

Publications examined in this review

• Wilson, R. (2010). Cultural Cobbles or a Load of Cobblers?

• Wilson, R. (2018). Eschewing Bear Tracks: Fallacies, Figure-Stones and Fontmaure.

• Wilson, R. (2026). Bear-like Forms at Fontmaure? Testing Figurative Constraint in Neanderthal Lithics. Prehistoric Society conference, University of York.

References to Wilson's publications are included for scholarly identification and verification, not as endorsements of their conclusions.

Stunning Ape Face in Flint: Fossil or Prehistoric Art?

Every so often, something turns up among my South Downs flints that really makes me look twice.

This is one of those finds.

I've recovered a remarkable piece of flint displaying an extraordinarily convincing primate face. My first impression is something rather baboon-like, although other primates also come to mind.

It is the sort of face that seems to stare straight back at you.

An ape face in flint: but what exactly are we looking at?

What interests me is the combination of features, rather than one accidental resemblance. The facial structure appears remarkably coherent, while parts of the surface have an almost fossil-like appearance.

And that introduces another question.

Could there be a genuine fossil structure within the flint? Could it conceivably be part of a fossilised primate? Or are we looking at a naturally suggestive stone that was recognised, selected and deliberately modified by prehistoric people?

I've found numerous ape-like faces across my assemblage, with what is clearly repeated evidence of deliberate modification. At some point, simply saying pareidolia stops being a satisfactory explanation for the entire collection, and will produce mockery of the entire field of archaeology for the ignorance.

If these are all chance resemblances, the South Downs must be home to a particularly industrious magic pareidolia fairy, working overtime to produce the best pareidolia anywhere in the world, because million to one chances happen 9 times out of 10 here.....

A fossil that shouldn't exist?

Flint can contain genuine fossils, including the remains and traces of ancient marine organisms. But what if this object really did contain part of a primate skull?

The conventional account places the evolution of baboons and other recognisable modern primates far more recently than the Cretaceous Period, approximately 145 to 66 million years ago.

But there is an important distinction here: the fossil record is not a complete record of life on Earth. It is a fragmentary collection of what happened to survive, what has been discovered and what researchers have been able to identify.

New discoveries can, and do, change our understanding of when organisms first appeared.

If a genuine primate fossil were ever convincingly demonstrated within securely dated Cretaceous flint, we would have to ask some difficult questions. Is the evolutionary history represented by the existing fossil record incomplete in a far more significant way than currently recognised? Is the geological context or dating correct? Or would aspects of both interpretations require reconsideration?

Those are legitimate questions to investigate, although a convincing facial resemblance on its own cannot establish such an extraordinary fossil identification.

My own position on geological dating is quite deliberate.

I tentatively accept the established geological timescale as a working framework because I need a consistent measuring stick against which to examine my finds.

I cannot sensibly attempt to correct archaeology, the conventional account of human evolution and geological chronology all at the same time. If I abandon the geological framework as well, I lose the very reference against which some of my archaeological arguments can be tested.

That doesn't mean geology is beyond question. It means I am choosing to work with its established chronology unless compelling evidence demonstrates that something needs revising.

There is also an important difference between the age of the Cretaceous chalk and the age of an artefact subsequently found on its surface. One does not automatically date the other.

So, no, I'm not announcing a 66-million-year-old baboon. That would give palaeontology a very difficult Monday morning.

But I am asking what this striking flint actually represents. Is there an ancient fossil structure contributing to the appearance? Was an unusual natural form recognised and enhanced by prehistoric people? Or is there another explanation we haven't yet identified?

Those possibilities deserve examination.

This find certainly has had an interesting history, with probable tar and ochre deposits all over it.

What about the unusual markings?

Another interesting feature of my assemblage is the presence of iron-rich areas, dark patches and occasional sooty or tar-like surface deposits.

Comparable features occur on other finds from the site, including objects discussed on my Prehistoric Manuports page.

I've also commissioned FTIR spectroscopy on other specimens, and those results are 100% indicative  of substance manufacture and application, to question whether all these deposits can simply be dismissed as ordinary mineral staining.

Of course, the composition of any particular marking needs to be established through examination and testing.

But the question becomes particularly interesting when contrasting patches coincide with recognisable anatomical features, as they often do in many of my other finds.

I am interested not merely in whether a stone resembles an animal or face, but whether its form, markings and identifiable modifications work together to produce a coherent image.

Another example of prehistoric visual communication?

This find does not stand alone.

For years, I have been documenting recurring animal heads, facial profiles, hand shapes and other figurative devices in my South Downs assemblage.

My interpretation is that some of these stones preserve a very early system of visual communication, involving the selection of naturally suggestive forms and, in many cases, deliberate enhancement.

The important evidence is the recurrence of comparable features across different specimens, especially where the details appear to have been constructed using recognisable stone-working techniques.

One convincing resemblance is interesting. A collection of recurring, deliberately constructed figures is something much more significant.

That is why I am documenting this ape-like face alongside the wider assemblage rather than treating it as an isolated curiosity.

See the face for yourself

I've included photographs and video so you can examine the object from different angles.

Look closely at the facial structure and the unusual surface details. Which features belong to the original shape of the flint? Which might be fossil-related? And which appear to have been deliberately enhanced?

Whatever its eventual explanation, I consider this one of the most striking primate-like forms recovered from my site.

And if you think it resembles a different primate, you're welcome to make the case. The stone has so far declined to identify itself.

For related examples, visit my Prehistoric Manuports and Portable Rock Art Identification pages.

Brett Martin - Eoliths.org

A Hematite Handaxe from the South Downs: Tool, Figure Stone, or Both?

A large iron-rich object with embedded flint fragments and a possible ape or early-hominin profile from the South Downs Palaeolithic assemblage

The Iron Axe next to one of the most modern looking handaxes from the site circa 600kbp. The iron axe has a fairly reasonable face profile, facing right (common at the site). 

Figure 1. Principal face of the iron-rich axe-form. This view is intended to document overall morphology, surface texture and any visible flint inclusions.

Among the many unusual objects recovered from my South Downs find site, close to the wider Boxgrove Palaeolithic landscape, one recent find stands apart simply because of its material, size and form.

It is a large, exceptionally dense hematite or hematite-rich ironstone formation, one of the largest examples of this material I have recovered from the site. Iron-rich nodules themselves are not unusual here. On the contrary, they form a recurring component of the surface assemblage alongside heavily patinated flints, cores, handaxe, blades, scrapers, hammerstones, figure stones, fossils and other selected stones.

This specimen, however, is different.

Its overall morphology resembles a substantial axe or bifacial tool form, while one aspect of the object presents what I interpret as a remarkably coherent ape-like or archaic hominid head in profile. The profile is not merely an isolated marking on an otherwise shapeless stone. It appears to interact with the overall form of the object, raising the possibility that natural morphology was deliberately selected, enhanced or incorporated into a functional artefact. Both the shape and size of this axe-like form recur among other flint artefacts from the site, suggesting a recurring typology.

There is another unusual feature. Within the iron-rich mass are small angular pieces and flakes of flint. Determining exactly how those fragments entered the matrix may prove crucial to understanding the object. Logically this must have been mined and worked, because that is what we have, something out of its natural location and natural form.

First question: what exactly is the stone?

I have provisionally described the material as hematite because of its appearance, density and its relationship to the many iron-rich nodules recovered from the locality. But “hematite” should eventually be treated as a mineralogical diagnosis rather than simply a visual description.

Hematite is iron oxide, Fe₂O₃. Archaeological “ochre”, ironstone and ferruginous rocks can contain hematite without being composed principally of it. A rock can therefore look intensely iron-rich while actually being a composite of iron oxides, quartz, clay minerals and other constituents.

That becomes especially important because of the tiny pieces of flint embedded within this specimen.

The wider geology of the South Downs includes naturally occurring ferruginous rocks and flint conglomerates. Geological descriptions from Sussex record iron-rich deposits in which sub-angular flint and quartz clasts occur within a matrix cemented by iron oxides. Some deposits form tough sandy ironstones associated with ferruginous flint conglomerate. South Downs National Park Building Stone Atlas.

No surprise to see an ape face in my materials, Pareidolia? try growing a brain!

A face stares out at you, looking almost identical to palaeoanthropological skull finds, fine eye, eyeball, pupil, brow, nose and even teeth details can be interpreted here.

Figure 2. Close-up of small angular flint fragments within the iron-rich matrix. Their size, angularity, orientation and relationship to the surrounding material are relevant to distinguishing a natural conglomerate.

Their size, angularity, orientation and relationship to the iron matrix can be examined microscopically. If they are simply natural components of ferruginous sediment, petrography should reveal that.

This makes the internal structure of the object potentially as interesting as its external form.

Ironstone really was used for Palaeolithic tools in Britain

One reason this find deserves proper investigation is that iron-rich stone is by no means unknown as a prehistoric tool material.

The clearest British comparison comes from Robin Hood Cave at Creswell Crags.

Excavations there produced implements manufactured from both quartzite and ironstone. Nineteenth-century descriptions record no fewer than 86 quartzite and ironstone implements and associated pieces, including axe-like forms. Quarterly Journal of the Geological Society, Vol. 32 (1876).

Figure 3 (comparison). Clay-ironstone handaxe from Robin Hood Cave, Creswell Crags. This is a comparison of raw material, not a claim that the South Downs specimen has the same age or manufacturing history. British Museum collection reference. Photograph reproduction is subject to the rights of the image owner.

That comparison matters because it demonstrates something very simple: prehistoric people in Britain did not restrict themselves to flint. Where suitable ironstone fractured effectively, they could deliberately reduce it into bifacial implements.

My South Downs find is not being claimed to be equivalent in date or technology to the Robin Hood Cave handaxe. The importance of the comparison is material: an iron-rich rock can unquestionably be an intentional stone tool raw material.

Hematite could also become sculpture

There is an even more remarkable European comparison.

At Petřkovice in the Czech Republic, a Gravettian settlement dated to approximately 21,000–23,000 years ago produced the celebrated Petřkovice or Landek Venus. Unlike the better-known ivory and fired-clay Palaeolithic figurines, this tiny human sculpture was actually carved from hematite. Archaeologists have also reported a fragment of another figurine from the same material. Archeologický ústav AV ČR, v.v.i..

Figure 4 (comparison). The Petřkovice (Landek) Venus, a Gravettian figurine carved from hematite. Its relevance here is the use of hematite for figurative carving, not a direct chronological comparison with the South Downs object. Creative commons wiki link Attribution

So we know that hematite was not exclusively a pigment source. At least occasionally, prehistoric people valued the stone itself sufficiently to carve figurative objects from it.

The Petřkovice figurine is enormously younger than the chronology I am investigating on the South Downs, and it would be wrong to use it as evidence for the age of this find. Its significance lies elsewhere: it removes the objection that hematite is somehow an inherently implausible material for prehistoric sculpture.

Worked hematite elsewhere in Palaeolithic Europe

Other sites add to the picture.

At Hohle Fels Cave in Germany, researchers studied hundreds of pieces of ochre and iron oxide extending through the Upper Palaeolithic. Among them were pieces of specular hematite that had been ground into pointed forms, as well as circular hematite objects intentionally rounded and centrally perforated. Twenty-seven pieces in the assemblage showed anthropogenic modification. PLOS ONE (2019).

Far earlier, at Maastricht-Belvédère in the Netherlands, archaeological sediments dating to at least about 200,000–250,000 years ago contained hematite brought into the site from a non-local source. The material occurs alongside Middle Palaeolithic flint-working evidence. PubMed Central / PNAS (2012).

None of these parallels authenticates my South Downs object. What they establish is the broader behavioural background: iron oxides and hematite were selected, transported, ground, modified and sometimes sculpted by Palaeolithic populations.

That makes an unusual hematite object found within a lithic-rich Palaeolithic locality worthy of examination on its own merits.

The figurative element: an ape or hominid profile

The most provocative aspect of the object is its profile and the central forward facing ape like head image shown above.

Viewed from one principal orientation, the outline presents what I interpret as an ape-like or archaic hominid head. Brow, cranial outline, facial projection and lower facial region appear to resolve into a coherent silhouette, although fairly weak here, it's expected from the site. The face frontal is much harder evidence to ignore.

On the reverse, If we were to rotate the find about 45 degrees CW, a fairly plausible ape face profile can be seen. There are some strange linear striations pointing middle wards, central in the artifact, a tree shape can almost be interpreted?

This is precisely where analysis has to go beyond resemblance.

A naturally formed stone can resemble almost anything when viewed from enough orientations. The archaeological question is therefore not simply:

“Can I see a face?”

The better question is:

“Which parts of the apparent image were created or modified by technically identifiable actions?”

The South Downs context

The find did not appear in isolation.

The locality has produced a large surface assemblage that includes heavily patinated flint implements and cores, Oldowan-like chopper forms, Acheulean-like bifaces and ovates, hammerstones, plate-like or tabular flints, fossils and substantial quantities of iron-rich nodules. Start at the https://eoliths.org/science.html Science page.

My working hypothesis is that parts of this landscape were repeatedly exploited for raw material - particularly flint, fossils and iron-rich stone - and that some of the concentrations may relate to deliberate extraction or sorting.

That remains a hypothesis requiring archaeological testing.

There is nevertheless a very important established Palaeolithic landscape nearby. Boxgrove and related deposits along the West Sussex coastal plain preserve one of Britain’s most extraordinary Lower Palaeolithic records. The Slindon Formation has been mapped over approximately 26 kilometres, with substantial surviving areas capable of preserving archaeology comparable with Boxgrove. UCL Boxgrove Wider Area Project.

At Boxgrove itself, handaxes, knapping scatters, butchered animal remains and hominin fossils occur in deposits dated to approximately 478,000–524,000 years ago.

That established archaeology makes the region exceptionally important. It does not automatically date surface material found elsewhere on the Downs, but it demonstrates beyond dispute that early hominins occupied and exploited this broader Sussex landscape.

What about an age measured in millions of years?

The possibility I have been investigating at this locality is considerably older than Boxgrove.

Some of the site’s lithic material possesses extremely developed patination and forms that I compare technologically with very early core-and-flake traditions. Together with geological observations and the wider eolith literature, this has led me to investigate whether parts of the assemblage could extend into a far earlier chronology.

At present, however, a multi-million-year age for British artefacts would be an extraordinary archaeological claim, no matter how logically evidenced by myself, it's not yet 100% proof.

The oldest securely contextualised evidence presently accepted in Britain (ignoring eoliths) comes from Happisburgh in Norfolk, where artefacts and footprints indicate hominin occupation roughly 850,000–950,000 years ago. British Museum — First Pioneers: The Earliest Human Occupation of Britain.

Recent work elsewhere in western Europe is beginning to push occupation towards and beyond the million-year threshold: the French site of Lunery-Rosières has been reported at approximately 1.1 million years. The authors nevertheless note that evidence in north-western Europe before one million years remains sparse. Scientific Reports / Nature (2024).

Consequently, neither the hematite axe-form nor the South Downs assemblage should simply be labelled “millions of years old” because their appearance seems archaic. That is the hypothesis to be tested.

A genuinely multi-million-year British artefact would not merely add another specimen to the Lower Palaeolithic record. It would fundamentally alter the currently understood chronology of hominin dispersal into north-western Europe.

That is exactly why the evidence needs to be exceptionally good.

How this object can be tested

The next stage should begin with the material itself.

• Mineralogical analysis: Non-destructive XRF, Raman spectroscopy or X-ray diffraction could establish whether the object is predominantly hematite, goethite, another ironstone, ferricrete, or a mixture.
• Internal structure: Micro-CT or high-resolution microscopy could investigate the embedded flint fragments and determine whether they form part of a natural sedimentary fabric.
• Surface recording: Controlled macro photography, 3D photogrammetry and microscopic examination. Areas interpreted as flaking, pecking, grinding or use-wear can be mapped independently of the figurative interpretation.
• Technological indicators: Particularly important would be any negative flake scars with identifiable initiation points, systematic edge regularisation, opposed removals, crushing associated with percussion, grinding striations or differential polish.
• Blind mapping of the proposed face: Instead of outlining the ape profile first and then looking for supporting features, the modifications should ideally be mapped blind. If the technical modifications independently reconstruct the same profile, that would be far more compelling.
• Separating context from age: A hematite nodule itself could be millions of years old geologically while human modification of it could be vastly younger. Dating the formation of the iron mineral therefore would not automatically date the manufacture of a tool. For a surface find, establishing age will ultimately depend upon tying comparable objects to a surviving stratigraphic context.

Why this find matters

Whatever the eventual interpretation, this object deserves more than a passing glance.

It is unusually large for the iron-rich material recovered from the site. It possesses an axe-like overall morphology. It apparently contains numerous small pieces of flint within its iron-rich matrix. And one orientation produces an unusually coherent primate or hominid profile.

Around it lies an assemblage in which hematite and other iron-rich nodules occur repeatedly alongside apparently worked lithics.

There are therefore several testable questions contained within a single stone.

1. Is it natural ferruginous conglomerate or another form of ironstone?
2. Was its exterior intentionally shaped?
3. Was it employed as a tool?
4. Were parts of its natural form deliberately retained?
5. And, most unusually, was that form exploited to produce a recognisable hominid image?

Those questions do not require us to decide the answer in advance.

The object can be photographed, scanned, analysed mineralogically and examined technologically. Its flint inclusions can be studied. Its suspected worked surfaces can be compared with both natural ironstone fracture and securely identified prehistoric ironstone artefacts.

That is where this find becomes genuinely interesting.

If it is natural, the geology should explain it.

If it is a tool, the fracture mechanics should demonstrate it.

If it is figurative, the modifications themselves should construct the image.

The stone now needs to be allowed to answer those questions.

References and further reading

• British Museum. “First Pioneers: The Earliest Human Occupation of Britain.”
• British Museum Collection. “Clay-ironstone handaxe, Robin Hood Cave, Creswell Crags.”
• UCL Boxgrove Wider Area Project. “The Slindon Formation and Lower Palaeolithic archaeology.”
• UCL Discovery. “Boxgrove: dating and stratigraphy.”
• South Downs National Park. “East Sussex Building Stone Atlas: Ferruginous deposits.”
• Hovers, E. et al. (2019). “Ochre use and modification at Hohle Fels Cave.” PLOS ONE.
• Roebroeks, W. et al. (2012). “Hematite at Maastricht-Belvédère.” PNAS.
• Archeologický ústav AV ČR. “Petřkovice Venus / Landek Venus.”
• Nature / Scientific Reports (2024). “Lunery-Rosières: 1.1 million years in western France.”
• Quarterly Journal of the Geological Society (1876). “Robin Cave, Creswell Crags: implements and fauna.”
• Eoliths.org. “South Downs Palaeolithic assemblage documentation.”
• Eoliths.org. “Iron-rich nodules across the South Downs locality.”

Comparing Prehistoric Art in 3D: A New Archaeology Tool

I've spent years examining flint tools and other prehistoric stones for signs of intentional imagery. One problem has always stood out: photographs can only show so much.

There is so much to see in this palaeolithic handaxe, but I thought I would just show you the elephant front half, a glyph I've been banging on about for almost 15 years.

You might see an animal's head in one photograph, a facial profile in another, or a particularly interesting arrangement of flake scars. But how do you compare those features properly when the objects are in different collections, photographed from different angles, or located hundreds of miles apart?

And what if you want to examine the other side of the stone?

This is something I've wanted to address for some time. I've now developed a 3D archaeological artefact comparison workspace on Eoliths.org, bringing several interactive models together on one screen.

It's a fairly simple idea, but I think it has considerable potential for research into prehistoric figuration, stone-working techniques, portable rock art and the possible origins of visual communication.

Try the 3D Archaeological Comparison Workspace on Eoliths.org

Six 3D models, side by side

The workspace opens with six archaeological objects displayed in a grid of three across and two down on a desktop computer.

These aren't just photographs. They're interactive 3D scans that you can turn, tilt and zoom in on.

Each window has its own model selector, allowing you to replace the object with another from the available collection. You can compare two different handaxes or, just as usefully, display the same handaxe in two windows and examine different faces or viewing angles at the same time.

I've also included a small control to fold away the descriptive text beneath each model. That leaves a much cleaner screen when you want to concentrate on the actual stones.

The starting collection consists of Lower Palaeolithic flint artefacts digitised by James M. Hicks and published by 3Djimmetry on Sketchfab. I've retained the links and attribution to the original scans.

But the important development isn't simply having six models on the page. It's being able to choose which objects to compare and then examine them independently.

Why this could matter for prehistoric art research

Imagine finding what appears to be an elephant's head worked into a flint tool.

You can see a trunk, an eye and perhaps a distinctive bump above the forehead. The boundaries of flake removals appear to form parts of the anatomy.

You might be convinced that the image was intentional. Someone else might dismiss it as coincidence.

A photograph alone isn't necessarily going to settle that disagreement.

Now imagine finding several more artefacts with similar arrangements. Perhaps the head shape recurs. Perhaps the eyes occupy corresponding positions. Perhaps the same anatomical divisions are formed by intersecting flake scars, with small pecked details appearing in similar places.

At that point, we're no longer talking about one stone that happens to resemble an animal. We're investigating a recurring combination of visual and technological features across an assemblage.

That is potentially much stronger evidence.

My research has led me to propose that some prehistoric people incorporated recognisable subjects and repeated visual devices into their stone-working. These include animal profiles, human and ape-like faces, hands, fingers and geometric marks.

I have described this possibility as a form of prehistoric visual communication.

The challenge has always been demonstrating that the patterns are genuinely repeated and physically constructed, rather than simply recognised by an observer.

Being able to put several artefacts together and examine their actual three-dimensional surfaces offers a useful way forward.

Looking beyond the outline

One of the things I find particularly important is how an image has been made.

If a proposed animal figure is present, what actually defines it?

Is the line of the back formed by a deliberate flake removal? Does an intersecting scar form the jaw or neck? Is a small indentation positioned where an eye would be expected? Is there a relationship between the working sequence and the final appearance?

These are questions that deserve more attention than they usually receive.

A recognisable outline is a starting point. But when an outline is supported by consistent anatomical details, repeated technical interventions and comparable examples from other stones, the argument becomes much more substantial.

This is closely related to the Stacked Constraints approach I've described elsewhere on Eoliths.org. Rather than depending on one striking resemblance, I look for several independent lines of evidence that support the same interpretation.

The new workspace makes it easier to study those relationships directly.

It can also expose weaknesses in an interpretation. A feature that looks persuasive from the front might turn out to be a normal fracture boundary when viewed from the side. Another feature might become clearer once the model is rotated.

Either result is useful.

Try comparing the same stone from two angles

This is one of my favourite uses of the workspace.

Choose the same artefact in two adjacent windows. Rotate one model to show its broad face, then turn the other to reveal the opposite face or an edge-on view.

You can examine how the flake scars relate to one another, whether a proposed figure has any depth, and whether a particular feature belongs to the original stone surface or a later removal.

Then try a second artefact displaying a similar feature.

Do the anatomical relationships correspond? Are the same kinds of marks involved? Does the image depend on a very particular angle, or is its structure still understandable as the stone is turned?

These are the sorts of comparisons I want to encourage.

There are perfectly ordinary technological questions to investigate as well: edge preparation, bifacial working, cortex retention, symmetry, tool thickness and possible differences between manufacturing traditions.

The workspace isn't restricted to figurative interpretations.

You can bring your own artefacts into the comparison

This is where I think the project could become particularly interesting.

Suppose you've found an unusual worked stone or you're studying a museum object that appears to contain deliberate imagery.

If you can obtain a suitable 3D scan, you may be able to compare it directly with archaeological models already available online.

There are several ways of recording an object in three dimensions. Photogrammetry uses overlapping photographs to reconstruct its surface, while structured-light scanners can capture detailed shape and texture information.

Once your model is prepared and uploaded to Sketchfab with embedding enabled, its model link can be added to the Eoliths.org comparison workspace.

You could then place your own find alongside another artefact and examine both independently.

Of course, the quality of the scan matters. Poor lighting, missing surface detail or inaccurate textures can be misleading, particularly when you're investigating fine pecking, possible pigment or subtle flake boundaries. A recorded scale is also important for meaningful measurements.

And naturally, anyone scanning or publishing an archaeological object needs the appropriate permission.

But the basic possibility is there: researchers and collectors could compare their own documented artefacts with material held elsewhere, without needing the physical objects together in one place.

The workspace does not identify motifs automatically. That's not its purpose. It gives people the means to inspect and compare the evidence themselves.

The Goodwood handaxe and a possible elephant figure

One example I've included on the Eoliths.org page is a Goodwood handaxe with unusually thick white patination.

I see what appears to be an elephant front half figure worked into this object. The interpretation involves the arrangement of flake scars, the lines where those scars meet, and small details that appear to help define the anatomy, no surprises here with this glyph, it's on my chart, just like other things visible here.

I've drawn around the proposed figure in an illustration so visitors can see exactly what I'm referring to.

A feature I find especially interesting is the distinctive bump on the animal's head. I've observed comparable shapes in other stones discussed in my research on prehistoric manuports.

Is this a recurring convention? Is the similarity significant? Are the apparent anatomical features deliberately constructed?

These are precisely the questions that comparative 3D examination can help us investigate.

Rather than asking readers simply to accept my interpretation, I want them to look at the surfaces and decide which features they can independently recognise and verify.

Not just handaxes

Although the first models come from a collection of ancient flint tools, I see no reason why the same approach couldn't be used much more widely.

Possible figure stones, selected natural objects, portable sculptures, engraved artefacts, worked pebbles and other archaeological materials could all be examined in this way, provided suitable scans are available.

For researchers interested in prehistoric cognition, that opens an intriguing possibility.

Could certain subjects or visual conventions have appeared repeatedly across different assemblages? Could some of the same methods of constructing imagery be recognised in objects made in different places or periods?

Those are big questions, and I wouldn't expect a single comparison to answer them.

But a growing body of accessible, well-documented 3D material could make them much easier to investigate systematically.

It would also make it easier for researchers to disagree constructively. Instead of arguing over a single photograph, they could identify the exact physical feature under discussion, rotate the same specimen and compare it with other examples.

An invitation to look for yourself

I've been documenting what I consider to be recurring imagery and visual conventions in prehistoric stone artefacts for many years.

This new comparison workspace is another step in that research. It gives me a better way to examine the material, but just as importantly, it allows other people to look for themselves.

Perhaps you'll recognise an animal figure I haven't noticed. Perhaps you'll identify a repeated arrangement of flake scars that deserves closer attention. Or perhaps you'll find a perfectly ordinary technological explanation for something that initially appeared figurative.

All of those observations could contribute to a better understanding of the material.

What matters is whether we can identify repeated, recognisable features and demonstrate how they were formed.

If you're interested in prehistoric art, ancient technology, portable rock art or the beginnings of human visual communication, I hope you'll give it a try.

Open the Eoliths.org 3D Archaeological Artefact Comparison Workspace

You can also read more about my approach to identifying intentional imagery on the Portable Rock Art Identification page, or explore the wider evidence on Science and Methods.

Brett Martin - Eoliths.org

The comparison workspace and associated interpretations are part of my independent research. Original 3D archaeological scans remain credited to their respective creators and publishers.

Beyond the Goose Egg: Do Elongated Handaxes Echo Dinosaur Eggs?

Could some Acheulean handaxe forms correspond to the geometry of dinosaur eggs?

In my earlier post on ovate handaxes and the symbolism of the egg, I pushed the argument past simple resemblance. A goose egg and a well-made Acheulean Boxgrove ovate do not merely share a similar length-to-breadth ratio: when their outlines are normalised and compared, the likeness is measurable across the form.

UPDATE: EOLITHS.ORG BOXGROVE HANDAXE OVATES

A new page under construction with some deeper insights and a very cool gadget.

The match goes beyond proportion. Using the geometry formalised in the 2021 Universal Formula for Egg Shape, the egg profile can be described from its length, maximum breadth, the displacement of maximum breadth from the half-length, and the diameter at one-quarter of the egg's length from the pointed end. Applied alongside direct outline comparison, this captures asymmetry, taper and changing perimeter curvature. When the goose egg and Boxgrove handaxe are plotted at the same scale, the two contours converge across almost the entire outline; the overlap is so close that the plotted goose-egg line is virtually hidden beneath the handaxe. This is not simply something that "looks egg-shaped". It is a measurable morphometric correspondence in precise geometric form. That raises the harder question: what are we to make of the much longer, narrower Acheulean bifaces?

Do they simply represent another functional solution in stone, or could some occupy the geometry of a completely different kind of egg?

Acheulean handaxes do not stop at the proportions of a goose egg. Across Britain and the wider Acheulean record, some assemblages contain many forms around 2:1, while individual pieces reach 2.5:1 and, exceptionally, approach or exceed 3:1. These include elongated ovates but also lanceolate, ficron-like and other pointed bifaces. If the Boxgrove ovate can be tested mathematically against a goose egg, then these elongated forms deserve the same treatment: what natural egg geometries occupy the 2:1–3:1 range, and do their complete outlines — not merely their proportions — match too?

The comparison becomes particularly interesting when we move beyond ordinary bird eggs. Pterosaurs and some non-avian dinosaurs produced strikingly elongated eggs, in some cases two, two-and-a-half, or more than three times as long as they are wide. These long forms occupy much the same proportional territory as many elongated Acheulean bifaces, making them obvious candidates for direct morphometric comparison. The pterosaur connection is particularly intriguing to me because among the worked lithics and apparently altered flint nodules from my South Downs find site I have identified numerous figurative forms that appear to resemble pterosaurs, some with remarkably specific likenesses. Those interpretations remain observational rather than proven, but they give an added reason to test whether the geometry of certain bifaces also corresponds to the geometry of pterosaur and elongated dinosaur eggs.

The Egg and Axe Ratio Test

The simplest first measurement is the same one I used before:

Elongation Index (E) = Length ÷ Maximum Breadth

A circular outline has a value of 1.00. As a form becomes longer and narrower, the number rises. Elongation alone does not establish an egg match — the decisive comparison must also include curvature, asymmetry, taper and maximum-breadth position — but it tells us which handaxes and eggs occupy the same broad region of geometric shape-space.

Elongation index comparison between Acheulean handaxe forms and fossil egg types

Form	Approx. L ÷ B	Geometric implication
Large modern bird eggs / goose-like zone	1.3–1.5	Strong overlap with broad ovate handaxes
Acheulean ovate handaxes	about 1.4–1.6	Broad, rounded egg geometry
Elongated Acheulean handaxes	commonly around 1.7–2.0 in some assemblages	Moves decisively beyond the ordinary goose-egg zone
Hamipterus tianshanensis pterosaur eggs	1.70–2.63	Broad overlap with elongated handaxe proportions
Fordwich handaxes, Kent	mean dimensions ≈ 1.92; narrowest recorded form ≈ 3.2	Shows that highly elongated British handaxes are part of the archaeological record
Jalindri handaxes, India	mean 1.85; range 1.20–2.50	The longest specimens enter pterosaur and theropod egg proportional space
Strongly elongated Acheulean handaxes	about 2.0–2.5+	Direct proportional overlap with elongated pterosaur and non-avian theropod eggs
Oviraptorosaur and troodontid dinosaur eggs	about 2.0 to >3.0	Moderately to extremely elongated fossil egg geometry
Pterodaustro guinazui pterosaur egg	60 × 22 mm ≈ 2.73	Extremely elongated pterosaur egg geometry within the range of very long bifaces
Macroelongatoolithus dinosaur eggs	about 2.6–3.2	Extreme elongation comparable only with the narrowest handaxe forms
Titanosaur / megaloolithid dinosaur eggs	near 1.0	Generally spherical to sub-spherical; a poor comparison for elongated handaxes

The Overlap Is Larger Than It First Appears

It is too restrictive to treat 1.55–1.65 as the normal limit of an elongated handaxe. Acheulean handaxes extend far beyond that. At Jalindri, handaxes have a published mean elongation of 1.85 and range to 2.50. At Fordwich, the mean dimensions of the published sample produce a ratio of about 1.92, while the narrowest recorded width-to-length value corresponds to roughly 3.2:1 when expressed as length divided by breadth. These are not imaginary forms created to fit the hypothesis; they are already present in archaeological datasets.

Adding pterosaurs makes the comparison more continuous. Published Hamipterus eggs range from 1.70 to 2.63, while the known Pterodaustro egg measures approximately 60 × 22 mm, an elongation of about 2.73. Oviraptorosaur and troodontid eggs extend broadly from around 2:1 to more than 3:1. The proportional ranges therefore overlap substantially.

But ratio is only the first filter. The Boxgrove ovate matched the goose egg not merely in length and breadth, but in its complete mathematical profile: maximum-breadth position, asymmetry, taper and perimeter curvature. When plotted together, the overlap was so close that the goose-egg outline was virtually hidden beneath the handaxe.

The same standard must therefore be applied to elongated forms. If a handaxe with an elongation of 2.1, 2.4 or 2.7 also reproduces the corresponding fossil egg's breadth position, asymmetry, taper and curvature, then the comparison becomes far more exacting than two objects simply being long.

The prediction is simple: broad ovates should correspond with broad egg geometries, while progressively elongated handaxes should correspond with progressively elongated egg geometries. The first test is L ÷ B. The decisive test is whether the entire mathematical outline also overlaps.

Note: Pterosaurs were flying archosaurs, but they were not dinosaurs. They are included here as a separate fossil egg group because their measurable egg proportions overlap the elongated handaxe range.

Addendum: The Match Is Now in Scale as Well as Shape

Since writing this post, the comparison has produced something I was not originally looking for.

The argument began with shape. Acheulean handaxes and eggs overlap in proportional shape-space. At Boxgrove the comparison goes much further: the ovate and egg profiles correspond in elongation, taper, maximum-breadth position and changing perimeter curvature. Boxgrove is also one of the clearest examples of deliberate handaxe standardisation, with hundreds of finished tools repeatedly converging on a closely controlled oval form.

Now there is another layer.

When published mean handaxe dimensions are compared with actual egg dimensions, three remarkable size correspondences appear:

• Boxgrove handaxes: mean 117.8 × 76.2 mm. Whooper swan eggs occupy almost exactly this physical scale — around 112 × 73 mm as a typical modern measurement, while the morphometric bird dataset used in the comparison produces a Whooper record of approximately 116.7 × 76.4 mm.
• Fordwich handaxes: mean 142.4 × 74.3 mm. A published oviraptorid egg measures 145 × 75 mm.
• Jalindri handaxes: mean 171.77 × 91.84 mm. A published Macroolithus egg measures 175.3 × 92.1 mm.

That is important because normalisation removes size. Nothing in the original egg-shape hypothesis required handaxes and eggs to be physically the same size. Yet some independently measured Acheulean assemblage means appear to match not only egg geometry, but egg scale.

The Boxgrove result is particularly difficult for me to ignore.

Long before finding this dimensional correspondence, I had repeatedly identified the swan as an important motif in the South Downs assemblage close to Boxgrove. The swan interpretation was therefore not invented to explain these measurements afterwards. The proposed motif came first. The numerical comparison came later.

So we now have several independent observations converging on the same idea: egg-like shape-space, close outline curvature, strong archaeological standardisation, correspondence in absolute physical dimensions, and—independently—a previously identified swan motif from the same wider landscape.

Unfortunately, none of this constitutes proof that an Acheulean knapper deliberately copied an egg. Geometry cannot recover intention from half a million years ago.

But there comes a point where simply saying “coincidence” stops being an explanation and becomes something that itself needs explaining.

One resemblance can be coincidence. A ratio can be coincidence. Even an outline might be coincidence. But when proportion, curvature, standardised form and absolute scale repeatedly converge, while an independently proposed swan motif is already present in the associated interpretive record, the egg comparison becomes considerably harder to dismiss as an accidental likeness.

The full dataset, Boxgrove standardisation evidence, interactive shape modeller and extended egg/handaxe comparisons are now collected on the Eoliths.org Acheulean Handaxe Egg Geometry page. 

Frequently Asked Questions

Do these measurements prove that Acheulean handaxes were modelled on eggs?

No. Geometric and morphometric overlap can show that a handaxe and an egg share closely corresponding proportions and outlines, but it cannot by itself prove that the maker intended to copy an egg. Intent requires independent archaeological evidence.

Is the Boxgrove ovate and goose egg comparison compelling?

Yes, as a geometric comparison. The correspondence is not limited to length divided by breadth. It also concerns maximum-breadth position, asymmetry, taper and perimeter curvature. When normalised and plotted, the two outlines overlap extremely closely. That makes the resemblance quantitatively interesting, although it still does not prove prehistoric intent.

Are all elongated Acheulean handaxes ovates?

No. Acheulean handaxes include ovates, elongated ovates, cordiforms, lanceolates, ficrons and other forms. The comparison here concerns continuous geometry and elongation across handaxe forms, not the archaeological label "ovate" alone.

Do elongated handaxes overlap with pterosaur and non-avian dinosaur egg proportions?

Yes, at the level of elongation ratio. Published handaxes extend into the same broad 2:1–3:1 proportional space occupied by some pterosaur and non-avian theropod eggs. A stronger test must compare the complete outline rather than length-to-breadth ratio alone.

Sources and Further Reading

• Narushin, V. G., Romanov, M. N. & Griffin, D. K. (2021). Egg and math: introducing a universal formula for egg shape. Annals of the New York Academy of Sciences 1505(1): 169–177.
• Iovita, R. & McPherron, S. P. (2011). The handaxe reloaded: A morphometric reassessment of Acheulian and Middle Paleolithic handaxes. Journal of Human Evolution 61(1): 61–74.
• Key, A. et al. (2022). On the earliest Acheulean in Britain: first dates and in-situ artefacts from the MIS 15 site of Fordwich (Kent, UK). Royal Society Open Science 9: 211904.
• Vyas, V., Pottentavida, A. & Jha, G. (2026). Assessing Acheulean biface variability at Jalindri. Frontiers in Earth Science 14: 1669759.
• Unwin, D. M. & Deeming, D. C. (2019). Prenatal development in pterosaurs and its implications for their postnatal locomotory ability. Proceedings of the Royal Society B 286: 20190409.
• Grellet-Tinner, G. et al. (2014). The first pterosaur 3-D egg: implications for Pterodaustro guinazui nesting strategies. Geoscience Frontiers 5: 759–765.
• Varricchio, D. J. & Jackson, F. D. (2016). Reproduction in Mesozoic birds and evolution of the modern avian reproductive mode. The Auk 133(4): 654–684.
• Wang, S., Zhang, S., Sullivan, C. & Xu, X. (2016). Elongatoolithid eggs containing oviraptorid embryos from the Upper Cretaceous of Southern China. BMC Evolutionary Biology 16: 67.

Interpretive caution: the numerical overlap discussed here establishes geometric comparability. It does not demonstrate intentional copying, fossil recognition or symbolic meaning. Those interpretations require independent contextual evidence.