Thoughts on ‘proto-writing’ in the Upper Palaeolithic

There’s a lot of buzz about the recent paper by Bacon et al ‘An Upper Palaeolithic Proto-writing System and Phenological Calendar’.

In short, the authors analyse repetitions of dots and and lines that are sometimes found superimposed on paintings of animals in European Upper Palaeolithic cave art. They argue that the signs are numerical and represent calendrical information pertaining to the life cycles of these animals. Each dot or line in a linear sequence encodes one lunar month from the new year (assumed to be late Spring) and a Y-shaped symbol indicates the time that the depicted animal gives birth.

Upper Palaeolithic horse in Lascaux cave, photographed by Emma Groeneveld. Note this image does not appear in the paper but is included here because it is public domain.

Many of my favourite scholars generate interesting visions of the world that shake up my thinking in useful ways, even if their models aren’t particularly well substantiated. In the case of this paper, I really do think the hypothesis is provocative and well-articulated and it’s quite possible that it’s statistically robust. But I remain unexcited by the vision and very cautious about the assumptions. I want to start by confessing that I’m automatically prejudiced against anything that presents itself as a ‘decipherment’. The world is already awash with folk interpretations of rongorongo, Linear A, and the Voynich manuscript to name a few, and I’m frankly jaded.

Ethnographic analogy

With that aside, my real objection is that the authors make assumptions about graphic codes and how they work that are far from universal in human societies. I’m aware that ethnographic analogy has its vigorous critics, but the approach is intended to be heuristic rather than predictive. It’s all about revealing the scope of what is possible. This makes it a much more effective safeguard against ethnocentrism than any thought-up-in-the-bath hypothesis that cannot see itself in the mirror.

A primary assumption of the authors is that a sequence of dots or lines is a numerical tally. Here’s the problem: examples can be found in contemporary societies of tally-like marks being merely decorative, or when they do carry information it is not necessarily numerical information. A ‘tally’ may indicating imprecise quantities (‘a large amount’, ‘all the women’), unnumbered stages in a journey, or core elements in a story or explanation (see figure below). Certainly, they are doing the work of structuring information but it is by no means always numerical information.

A non-numerical ‘tally’. Figure 2.13 from page 61 of Green (2014). Drawn from the ground: Sound, sign and inscription in Central Australian sand stories. Cambridge: Cambridge University Press. Reproduced here with the author’s permission.

Related to this issue is the invisible role of orality.

A common expectation of literate minds is that sequences of standardised graphic symbols must represent self-contained strings of information—like written inscriptions—that are relatively context-independent. Yet in traditionally non-literate societies, graphic symbols may only become activated and meaningful within an oral interaction such as a sung performance, a diplomatic exchange or a narration. In these scenarios, the oral channel is doing almost all of the heavy lifting in terms of the information load. The graphic symbols may help to structure, reinforce or authenticate the oral message but they never encapsulate it in such a way that it can be independently reconstructed (or ‘deciphered’ like glottographic signs). To use a term applied by Pierre Déléage (2013), the graphic code is bound to the oral channel.

From the 19th century onwards, many anthropologists assumed that such codes were mnemonic, but again this is a gross simplification. Members of oral-dominant societies can remember things quite effectively without writing them down, though prompts can certainly help to structure and condense the information, or give it social legitimacy. Consider that some oral societies have transmitted observations of sea-level rises or volcanic eruptions without any ‘mnemonic’ whatsoever except the landscape itself. Some of these observations may have remained accurate for over 13,000 years (update: for a critique see Davidson et al. 2021). Writing, meanwhile, has been with us for a mere 5000 years and widespread global literacy only since the 1970s. Pure orality should not be underestimated. Indeed, on such measures it has yet to be outmatched by writing.

This makes me question the presumed function of these Upper Palaeolithic cave marks as recording and transmitting information about animal birth cycles in lunar months. For a hunter-gatherer community—whose members might potentially recall hours-long song cycles, genealogical lists, complex kinship arrangements, not to mention vast quantities of environmental knowledge—this is an odd thing to want to make a permanent record of. What’s more, animal life-cycles in any environment are integrated with countless other seasonal indicators that make predictions easier. Hunter gatherers will know, for example, that the arrival of a certain migratory bird announces the maturation of a certain frog. The drying out of a lagoon predictably coincides with the prevalence of a grub that attracts the interest of raptors and so on. In other words, I suspect that there would be many better ways of predicting when wild animals are going to give birth than numerical counting from the onset of a seasonal change. The subsequent need to actually record such counts for future generations is hard to fathom, though I could certainly imagine ceremonial possibilities, as per rock art in Australia.

With all this in mind, the hypothesis may still be correct, and I do love a bold vision! But I would want to see a more thorough exploration of alternative explanations.


The least inspiring part of the paper is the section arguing that that Upper Palaeolithic striations and dots represent a form of proto-writing. The suggestion that abstract cave signs are precursors to writing is an old chestnut that was entertained by Edward Clodd (1900), then Henri Breuil (1905) and then taken for a decent spin by Arthur Evans (1909) who imagined a Phoenecian alef in the carving on a reindeer horn fragment. The idea also turns up in Hans Jensen (1958) among other later writers who I won’t defame here.

But ‘proto-‘ anything is a suspect hold-over from progressivist thinking whereby a given phenomenon is characterised as an undeveloped stage of a higher-status phenomenon. As Graeber and Wengrow (2021) put it: “terms like ‘Proto-palatial Crete’, ‘Predynastic Egypt’ or ‘Formative Peru’ convey a sense of impatience, as if Minoans, Egyptians or Andean peoples spent centuries doing little but laying the groundwork for such a Golden Age.” Likewise, early graphic codes are not trying and failing to be writing. They are performing a fixed task that simply does not require the modelling of any linguistic structure. Standardised non-linguistic graphic codes are everywhere and do very specific jobs. They predate writing, are used alongside writing, and emerge after writing. To suggest that they are embryonic of ‘full’ multi-purpose writing is to ignore what they already do, or to assume that what they’re doing is deficient. Contemporary musical scores, for example, are a great example of a narrow yet potent non-linguistic code. It would be absurd to suggest that the system will eventually be abandoned in favour of a more effective written description of the music.

The authors maintain that the position of ‘Y’ in a tally sequence is ordinal and thus logographic, per contemporary Hindu-Arabic numerals. This dispensed with the need for orality to explain meaning and increased the reliability of the signal, just like writing. They speculate that ‘Y’ encapsulated morphological information in the verb form ‘give birth’. On these grounds, they assume that the signs could be articulated in language and thus represent writing. Again, this is easily refuted by contemporary ethnographic evidence where visual codes may prompt or reinforce elements of speech—especially formulaic speech such as ritual recitations—without modelling linguistic structure.

To further bolster their ‘proto-writing’ theory, the authors appeal to the Schmandt-Besserat (1989) hypothesis that the earliest invented writing in Mesopotamia was an outgrowth of numeric markers on clay tokens. This hypothesis is another bold idea that I enjoy but it has to be admitted that it has taken a battering in recent years on the grounds of archeological evidence (for one excellent critique see Bennison-Chapman 2018, cited by the authors). It doesn’t stand up to comparative generalisation either. At the three other sites in which writing was invented from scratch—Egypt, China and Mesoamerica—there is no good evidence that glottographic writing emerged from counting.

An aside: If abstract symbols in cave art are not proto-writing, who is to say that they not ‘writing’ proper? What is ‘writing’ anyway? This is a controversial question but minimally writing involves that modelling of linguistic structure such that a linguistic message can be reconstructed. Even by the most generous interpretation, these sequences of abstract cave symbols are not writing. For those interested, my own attempt at clarifying the question of ‘What is writing?’ is here.

Anyway, I don’t want any of this to come across as gripey or grumpy. Bennett Bacon and colleagues know a great deal about Upper Palaeolithic visual culture than I do, and I love seeing outsider scholars making profound contributions.

These are just my personal reactions and I’m very happy to be told where I’ve got it wrong!

Bennison-Chapman, Lucy E. 2018. “Reconsidering ‘tokens’: The Neolithic origins of accounting or multifunctional, utilitarian tools?” Cambridge Archaeological Journal:1-27.

Breuil, Henri. 1905. “La dégénérescence des figures d’animaux en motifs ornementaux à l’époque du renne.” Comptes rendus des séances de l’Académie des Inscriptions et Belles-Lettres 49:105-120.

Clodd, Edward. 1900. The story of the alphabet. London: George Newnes.

Davidson, Iain, Heather Burke, Pearl Connelly, Stephen Porter, Hazel Sullivan, Lance Sullivan, Isabel Tarragó, and Lynley A Wallis. 2021. “Oral tradition, history, and archaeohistory of Indigenous Australia.” In The Oxford Handbook of the archaeology of Indigenous Australia and New Guinea, edited by Ian J McNiven and Bruno David. Oxford: Oxford University Press

Déléage, Pierre. 2013. Inventer l’écriture: Rituels prophétiques et chamaniques des Indiens d’Amérique du Nord, XVIIe-XIXe siècles. Paris: Les Belles Lettres.

Evans, Arthur J. 1909. Scripta Minoa: The written documents of Minoan Crete with special reference to the archives of Knossos. 2 vols. Vol. 1. Oxford: The Clarendon Press.

Graeber, David, and David Wengrow. 2021. The dawn of everything: A new history of humanity. New York: Farrar, Straus and Giroux.

Green, Jennifer. 2014. Drawn from the ground: Sound, sign and inscription in Central Australian sand stories. Cambridge: Cambridge University Press.

Jensen, Hans. [1958] 1970. Sign, symbol, and script: An account of man’s efforts to write. Translated by George Unwin. London: George Allen and Unwin.

Schmandt-Besserat, Denise. 1989. “The precursors of writing: Plain and complex tokens.” In The origins of writing, edited by Wayne M. Senner, 27-41. Lincoln: University of Nebraska Press.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s