📜 Shattered record: these artifacts may show writing from 40,000 years ago

Published by Adrien,
Source: Proceedings of the National Academy of Sciences
Other Languages: FR, DE, ES, PT

Engravings decorating Stone Age tools and figurines could represent the very first records of information, and thus the dawn of writing.

40,000 years ago, our ancestors were already adorning ivory objects with repetitive signs, such as dots and crosses. Recent research shows that these marks are not simple ornaments. Their elaborate structure shows similarities with the earliest known writing systems, pushing back the origins of symbolic communication.


The mammoth figurine from Vogelherd Cave, dating to approximately 40,000 years ago, displays sequences of crosses and dots on its surface.
Credit: Universität Tübingen / Hildegard Jensen, CC-BY-SA 4.0

Led by Christian Bentz from Saarland University and Ewa Dutkiewicz from the Museum for Prehistory and Protohistory in Berlin, a team examined over 3,000 signs across 260 Paleolithic artifacts. Their study, published in PNAS, uses computational methods to analyze the information density of these sequences. The results indicate that these intentional engravings likely served to store or exchange data, long before the emergence of formal writing.

The marked objects have often been found in caves of the Swabian Jura in Germany and date from a period ranging from 34,000 to 45,000 years ago. Notable pieces include a mammoth figurine from Vogelherd Cave, adorned with crosses and dots, and the sculpture of the Lion Man of Hohlenstein-Stadel, marked with regular notches. Carefully crafted, these artifacts were palm-sized, suggesting they were carried and handled regularly by communities of the time.

After digitizing the sign sequences, the scientists used quantitative linguistics tools. By analyzing the frequency and predictability of the marks, they calculated their entropy, a measure of information density. This method allowed for a comparison with proto-cuneiform, a Mesopotamian system that appeared around 3,000 BCE. The statistical resemblance between the two sets surprised the researchers.


The Adorant figurine from Geißenklösterle Cave, approximately 38,000 years old, features an ivory plaque with an anthropomorphic figure and sequences of dots and notches. These marks suggest a possible notational system.
Credit: Landesmuseum Württemberg / Hendrik Zwietasch, CC BY 4.0

According to the authors, these repetitive signs, of the 'cross, cross, cross' type, differ from modern writing that transcribes speech. Nevertheless, their ability to encode information is statistically equivalent to that of proto-cuneiform. This observation suggests that humans developed symbolic systems long before the creation of writing proper. Figurines, in particular, display a higher information density than tools, which could signal a more specialized or ceremonial use.

This ability to record data may have helped social groups coordinate or strengthen their chances of survival during the Ice Age. Although the precise meaning of the engravings remains unknown, this work allows for the interpretation of the cognitive behaviors of early Homo sapiens.


Proto-cuneiform tablet from the Uruk V period, dating from approximately 3500 to 3350 BCE. It contains numerical signs and an ideogram.
Credit: Staatliche Museen zu Berlin, Vorderasiatisches Museum / Olaf M. Tesmer, CC-BY-SA 4.0


Statistical analysis methods for ancient signs


To understand prehistoric engravings, researchers use quantitative linguistics methods. They convert sequences of signs into numerical data, then apply statistical models to measure their regularity and predictability. This approach allows for estimating information density without having to speculate on the meaning of the marks.

The concept of entropy is central to this analysis. It quantifies the average amount of information contained in a sign, considering its frequency and context. Entropy values close between Paleolithic engravings and proto-cuneiform indicate a similar ability to encode data, even though the systems are separated by millennia.

Machine learning algorithms help classify signs and identify patterns. Comparison with other symbolic systems, such as modern writing, allows for tracing the evolution of visual encoding. This work demonstrates that an elaborate structure is not the exclusive domain of spoken languages, but also exists in older forms of communication.

These techniques pave the way for studying other ancient artifacts worldwide. By avoiding subjective interpretations, they provide an objective basis for exploring the beginnings of human symbolic thought, long before the appearance of alphabets.
Page generated in 0.218 second(s) - hosted by Contabo
About - Legal Notice - Contact
French version | German version | Spanish version | Portuguese version