Atlas Essay — 2026-07-23

Identical Forms

Dordogne · Cantabria · the Pyrenees · the Sepik · the Upper Xingu

In 2025 two music archaeologists wrote a sentence about the Ice Age bone plaques that is either the most interesting claim in the field or an unexamined habit of speech: the forms are identical with bullroarers all over the world. It is an unusual claim in that it can actually be checked. Both corpora are measurable, both have been measured, and so far as we can establish, nobody had put the two sets of numbers beside each other. Doing so gives a split verdict — and the half that survives is the more remarkable one.

In brief

Median thickness-to-width is 0.164 across the Magdalenian plaques and 0.182 across the living instrument. That is the same slenderness — the proportion bullroarer-makers hold to at every size, from a boy’s toy to a metre-long Sepik board.
They are not the same size. Median length runs 140 mm in the Ice Age against 270 mm in the ethnographic record; median width, 16 mm against 45 mm.
So the claim is right about shape and wrong about scale — a distinction that matters, because a bullroarer is defined by its proportions and scaled freely in its dimensions everywhere it is found.
The comparison also vindicates a typology built by eye: Ignacio Barandiarán’s three-way sorting of the Spanish and French material separates cleanly on arithmetic he never ran.
The Aurignacian “batons” of the Swabian Jura, sometimes offered as the earliest bullroarers of all, are a stubbier form — landing at 0.323, with Barandiarán’s elipses rather than his blades.

A sentence worth checking

Arnd Adje Both and Michael Praxmarer, writing on Stone Age music in 2025, set out the European inventory: at least fourteen findspots yielding possible bullroarers, from Spain, France, Belgium, Germany, Austria, Bohemia and elsewhere, the oldest of them — the Vogelherd and the Tischoferhöhle — reaching back to 38,000 BC. Then they say what the objects are.

Die Formen sind identisch mit Schwirrhölzern auf der ganzen Welt und auch die Verzierungen der Objekte auf einer der Seiten … sind vergleichbar.Both & Praxmarer, “Musik in der Steinzeit” (2025), p. 138

The forms are identical with bullroarers all over the world, and the decorations on one of their faces are comparable. Both convenes the international study group for music archaeology; this is the settled view of the field speaking in its own voice, in an open-access chapter, without hedging.

It is also a claim of a rare kind: falsifiable by tape measure. Every serious Palaeolithic candidate has been published with its length, width and thickness in millimetres. So has a large fraction of the ethnographic corpus, because museums measure what they accession. If the forms are identical, the numbers will say so. If they are not, the numbers will say that instead, and will say where the difference lies — which is the part that turns a slogan into knowledge.

The trap in the obvious method

Archaeological entries in a bullroarer catalogue are there because somebody looked at a small pierced blade and thought of a bullroarer. Measure them and you will find that they are small pierced blades — the selection criterion handed back as a result. Any comparison built on such a list is answering its own question.

The escape is to let somebody else choose the objects — somebody with a different purpose, working before the question was asked. Ignacio Barandiarán Maestu’s 2012 survey in Veleia is exactly that. He set out to sort the Franco-Cantabrian bone industry into types, and along the way described every object anyone had ever called a bramadera, giving each its measurements and his own verdict. He also, and this is what makes the survey usable as a control, kept two other classes separate: a set he calls elipses, which he insists are a distinct thing, and the ordinary perforated pendants, which he treats as the background population that the bullroarers must be distinguished from.

That gives three groups sorted by a specialist’s eye, for his own reasons, decades before this atlas existed — and a built-in control group chosen by the same eye. Against them stands the ethnographic side: the ninety-seven catalogued bullroarers here that carry dimensions, from the Sepik to the Kalahari to the Great Basin, every one of them an instrument collected in use, not inferred from its shape.

What the two corpora measure

The medians, in millimetres:

Franco-Cantabrian Magdalenian material after Barandiarán 2012, against the measured ethnographic corpus.
Set	n	Length	Width	Thickness	Thickness ÷ width
Magdalenian bramaderas	16–26	140	16	2	0.164
Barandiarán’s elipses	3–5	156	22	7	0.318
Perforated pendants (control)	3–7	91	18	3	0.214
Swabian Aurignacian batons	5	147	31	10	0.323
Ethnographic bullroarers	49–97	270	45	9	0.182

Read the last column first. Then read the two before it.

The proportion that carries

Why should proportion be the thing that survives forty thousand years, rather than size? Because proportion is what bullroarer-makers converge on everywhere, at every scale. The instrument is documented from hand-sized boards a boy swings at play to monumental Sepik blades well over a metre long, and across that whole enormous range it stays a slender thing — long for its width, and thin for its length. Size is the local variable: it answers to the wood on hand, the room there is to swing, the distance the sound has to carry. Slenderness is the constant that travels.

So a match in proportion between two corpora separated by the width of the Ice Age is the match that means something, and a difference in size is the kind of difference bullroarers show between one village and the next.

Median thickness-to-width across the Magdalenian plaques is 0.164. Across the living instrument, gathered from every inhabited continent over the last century and a half, it is 0.182. The two figures differ by less than a millimetre of bone. On the measurement that a maker holds constant while everything else is free to change, Both and Praxmarer are simply correct, and correct with a precision their sentence does not claim for itself.

And it has been put to the test, at least once. Michel Dauvois cut a replica of the Lalinde plaque and swung it, and it sounded like a bullroarer — the one Magdalenian object anyone has actually set spinning, behaving exactly as the shape predicts.

This is a positive result about the deep past, and worth stating plainly. A person at Lortet or Altamira or Bolinkoba, thinning a strip of bone with a flint, stopped thinning it at the same proportion that a Wauja man in the Upper Xingu stops at when he shapes a matapu from hardwood. Neither of them could have told you the number. Both of them were answering to the same air.

Half the length, a third of the width

The other half of the sentence does not survive. The Magdalenian median is 140 millimetres long and 16 wide; the ethnographic median is 270 by 45. Roughly half the length and roughly a third of the width. The distributions do overlap at the tails — there are small ethnographic roarers and there are longer Palaeolithic plaques — but their centres are nowhere near each other.

The obvious objection is that the Palaeolithic numbers are inflated downward by breakage: most of these objects are fragments, and a fragment is short. Barandiarán handles this by reconstructing, writing things like mediría —completo— unos 140 mm. Nine of his sixteen lengths are reconstructions of that kind, and reconstruction is precisely where a typologist’s expectations would leak into the data.

They do leak — in the wrong direction for the objection. The nine reconstructed lengths have a median of 150 millimetres; the seven that are simply measured have a median of 85. His reconstructions push the corpus longer. Throw them out and keep only what a caliper touched, and the distance to 270 gets wider, not narrower.

One honest weakness remains, and it is on the modern side. The ninety-seven ethnographic objects with numbers are the ones that passed an accession desk, and accession desks like big things. A collector carrying a metre of painted Sepik hardwood out of a men’s house has brought back a museum piece; a boy’s hand-sized noisemaker is more likely to be left where it lies. How much that inflates 270 millimetres is unknown, and it is the strongest argument anyone can make against this half of the finding.

A typologist checked by his own numbers

Barandiarán sorted this material by looking at it. He never calculated a thickness-to-width ratio for anything. Run the arithmetic across the classes he built and they come apart exactly where he said they would.

His control group of ordinary pendants is markedly shorter than his bullroarers — 91 millimetres against 140 — so the distinction he drew is carrying real signal and not merely a label. His elipses, which he insisted throughout were a separate class, sit at a thickness-to-width ratio of 0.318: nearly twice the bramadera figure, and in a band of their own. He set them apart by eye, and the arithmetic keeps them apart.

The same arithmetic settles a question from an entirely different quarter. The Aurignacian ivory and bone “batons” of the Swabian Jura — from Geissenklösterle, the Vogelherd and Hohle Fels — are periodically offered as the oldest bullroarers in existence, on the strength of being flat, perforated and very old. Measured, the five of them land at a median 0.323 — nowhere near the bramaderas, and sitting instead on top of Barandiarán’s elipses. They are a different, stubbier form, and they land there by the arithmetic of a corpus assembled in another country for unrelated reasons. That is a reason to keep them out of the blade series, whatever else they may turn out to be.

What a 140-millimetre bullroarer sounds like

If the proportions match and the sizes do not, then the interesting question is no longer whether the Ice Age objects are bullroarers. It is what kind of bullroarer they are.

Length drives the fundamental. A blade twice as long, swung at a comparable rate, turns more slowly about its own axis and beats the air at a lower frequency; it also sweeps far more air, which is what lets the great Sepik boards be heard across a river at night. A 140-millimetre plaque at the same proportions is a different instrument in performance if not in kind: higher in pitch, quieter, shorter in range. Something to be heard inside a cave or within a small circle of people, rather than a sound sent out over a valley to warn women and children away.

That is a claim about what Magdalenian ritual sounded like, and it follows from measurements alone. It also fits the settings. These objects come out of La Vache, Isturitz, El Pendo, El Valle — enclosed spaces where a small, high, insistent sound is entirely sufficient, and where a Sepik board could not be swung at all.

And they are decorated, which is the part of the 2025 sentence still standing open. The Magdalenian engraving that survives is bovids and abstract geometry: a wolverine running the length of the Les Eyzies blade, engraved bison on both faces at Lortet and Abri du Morin, a bison reduced to head and spine at Raymonden. Game animals and pattern. Not, on a single European bramadera, a serpent — and the serpent is among the most insistent creatures carved on bullroarers in the ethnographic world, from the Sepik to the Vaupés. The exception sharpens the point rather than blunting it: a serpent does run the length of one Magdalenian plaque, and it is an object from the Lortet excavation that Barandiarán classes as an elipse — the stubbier neighbour of the blades, not one of the blades themselves. The one European serpent sits on the fringe of the type, not at its centre. The forms are identical; the decorations, so far, are not, and that is the sharpest question the comparison leaves behind.

Key sources

Both, Arnd Adje & Michael Praxmarer, “Musik in der Steinzeit,” in Fontaine et al. (eds), Interdisziplinarität in der Disziplin (2025), pp. 119–153, doi:10.25366/2024.12. The quoted sentence is at p. 138; the fourteen-findspot inventory at p. 137. Open access.
Barandiarán Maestu, Ignacio, “Bramaderas en el Paleolítico superior peninsular,” Veleia 29 (2012), pp. 307–334 — the source of every Magdalenian measurement used here, including the elipses and the pendant control group. His 2015 study in Kobie BAI 6 carries the Bolinkoba and El Pendo material.
Dauvois, Michel, on the replica experiments in which a copy of the Lalinde plaque was swung and sounded.
Peyrony, Denis, “Sur quelques pièces intéressantes de la grotte de La Roche, près de Lalinde (Dordogne),” L’Anthropologie 40 (1930), p. 22, fig. 3 — the founding side-by-side of Lalinde and an Australian churinga.
Dimensions on the modern side are drawn from the museum records behind the atlas’s own entries; the Palimbei board is Ethnologisches Museum Berlin VI 48045.
As far as we know, the two corpora have not previously been measured against each other, and the thickness-to-width agreement reported here is original to this page.

Browse every dated archaeological record in the atlas, see them on the map, or read the essay on the Raymonden bison.