Stone Age Carvings from 40,000 Years Ago Match the Information Density of Early Writing

The oldest known writing system - proto-cuneiform, developed in Mesopotamia around 3,000 BCE - is roughly 5,000 years old. It emerged at a particular moment in the history of early civilization, when trade, administration, and the need to record transactions created pressure for a standardized notation system. That origin story is well documented and widely accepted.

But a study published in PNAS complicates the picture by asking a different question: not when writing appeared, but when humans began using sign sequences to encode and transmit information with a statistical structure equivalent to writing. The answer, according to computational analysis by a linguist and an archaeologist working across Palaeolithic collections in European museums, may be 40,000 years ago or more.

Three thousand signs on 260 objects

The research team - Christian Bentz, a linguist at Saarland University, and Ewa Dutkiewicz, an archaeologist and curator at the Museum fur Vor- und Fruhgeschichte in Berlin - examined more than 3,000 geometric signs found on approximately 260 Palaeolithic objects. The artefacts date to between 34,000 and 45,000 years ago, placing them in the period when Homo sapiens first arrived in Europe and encountered Neanderthals.

Many of these objects come from sites in the Swabian Jura region of southwestern Germany - caves including Vogelherd, Geissenklosterle, and Hohlenstein-Stadel, where exceptional preservation has kept ivory and bone artefacts intact for tens of millennia. A small mammoth figurine carved from mammoth tusk bears rows of crosses and dots. An ivory plate depicting a lion-human hybrid is adorned with rows of notches. The Lion Human sculpture from Hohlenstein-Stadel shows notches placed at regular intervals along its arm.

The signs themselves - lines, notches, dots, crosses, and other geometric marks - have been debated by archaeologists for decades. Are they decorative? Tallies of hunted animals or lunar cycles? Territorial markers? Ritual symbols? The current study does not attempt to decode their specific meaning. Instead, it asks a statistical question: do these sequences have the kind of information structure that would be expected if they were being used to convey distinct messages?

What the statistical fingerprint reveals

The computational approach draws on methods from information theory - mathematical tools developed to measure the structure of communication systems. One key measure is information density: how much information is conveyed per unit of the sequence, and how consistently that density holds across different parts of the sequence. Another is the degree of redundancy - the extent to which some signs are more common than others in a patterned way, rather than appearing randomly.

Applied to the Palaeolithic sign sequences, the analysis produced a statistical fingerprint. The team then compared that fingerprint to the same measures calculated from proto-cuneiform, the earliest known formal writing system. The comparison was, according to Bentz, clear enough to surprise even the researchers.

The Palaeolithic signs and proto-cuneiform showed the same level of complexity and information density. The statistical structure - the patterns of repetition, variation, and sequencing - was equivalent between sign systems separated by 35,000 to 40,000 years of human history. This finding does not mean that the Stone Age signs are writing in the conventional sense. It means they exhibit the same formal properties that make writing effective as an information-encoding system.

The cognitive implication

If the signs of 40,000 years ago carry the same statistical information structure as the first writing, then the cognitive capacity underlying both is also 40,000 years old. The ability to create sequences of distinct symbols with controlled variation, to maintain those sequences with consistent structure across a collection of objects, and to produce sequences whose information density matches that of a formal communication system - all of this was present in the minds of Palaeolithic humans long before any civilization developed formal writing.

Dutkiewicz, who travels through European museums and archaeological sites with Bentz to catalog signs, emphasizes both the scale of what has been found and what remains to be examined: "There are many sign sequences to be found on artefacts. We've only just scratched the surface." The project is funded by the European Research Council and continues to build a larger database of Palaeolithic sign systems across the continent.

What the study does not claim

The researchers are careful about what their statistical analysis can and cannot establish. It demonstrates equivalent information structure; it does not demonstrate equivalent meaning or function. Proto-cuneiform encoded specific commodities, numbers, and transactions in a way that could be reliably decoded by trained readers. Whether Palaeolithic signs functioned similarly - with shared conventions that allowed consistent decoding by others in the same community - is a question the statistical analysis cannot answer.

The sample, while large for this type of study, is also skewed by preservation. Only objects made from durable materials - ivory, bone, stone - survive 40,000 years. Signs carved into wood, drawn on perishable surfaces, or made in other media are lost entirely. The 260 objects in this analysis represent the surviving minimum, not the original population of sign-bearing artefacts.

Those caveats aside, the findings push back significantly on the timeline of human symbolic behavior - not as a speculative argument from isolated finds, but as a statistical pattern measured across hundreds of objects.