Lost and Found Knowledge

We're used to mocking the remedies of our ancestors. Bloodletting, horoscopes, concoctions of emeralds and obscure animal parts. The trajectory of medical improvement seems obvious and linear — hygiene, anaesthetics, germ theory, antibiotics. But this linear narrative turns out to be wrong in ways that matter, because our ancestors were often surprisingly good at finding things that actually worked, even when they had no idea why they worked — and those discoveries were then prematurely discarded by the very march of science that was supposed to improve on them.¹

Terra Sigillata and Accidental Antibiotics

The case of terra sigillata — stamped earth — is remarkable. A clay exported from the Aegean island of Lemnos had been popular since ancient times as a cure-all for various diseases and poisons. In ancient times it was stamped with a head of Artemis; by the seventeenth century, with the Ottoman sultan's seal, who controlled its export as a revenue source. Various imitation stamped earths soon appeared from England, Ireland, France, Malta, and elsewhere. By the eighteenth century people became skeptical. Nineteenth-century chemical analysis put an end to the whole tradition: Lemnian clay, disappointingly, turned out to be just clay.¹

But the nineteenth-century scientists got it wrong. In 2020, teams of scientists and archaeologists analyzed terra sigillata specimens at the Pharmacy Museum at the University of Basel and discovered that some of them actually had antibacterial properties. The key wasn't the clay itself but its ritual treatment. In ancient times, priestesses of Artemis mixed the mud with water and left it to stagnate before drying and stamping it. In the fifteenth and sixteenth centuries, it was similarly covered with spring water and left, or dug only at certain times of year, only to a shallow depth, near water sources. Thanks to DNA sequencing, we now know why: the ritual treatment introduced a fungus closely related to Penicillium, called Talaromyces, which produced an antibacterial and antimalarial compound called bioxanthracene B.¹

The Armenian boles — another type of terra sigillata — are effective against even more kinds of bacteria, though the mechanism isn't yet understood. There are plenty of other terra sigillata yet to be tested. What nineteenth-century chemical analysis too hastily took away, twenty-first century DNA sequencing is giving back.

The Pattern of Premature Dismissal

This isn't an isolated story. Anton Howes documents a recurring pattern in the history of innovation: practical knowledge that works for reasons nobody understands is discarded when the theoretical framework catches up and judges the explanation inadequate, even though the practice was sound.¹

Mercury for syphilis sounds barbaric. But mercury, while poisonous, appears to have worked along the same lines as chemotherapy — hopefully killing the disease before the cure killed the patient. It was effective under certain conditions. The replacement — bismuth salts in the 1880s — worked similarly but with less toxicity. Even the early twentieth-century wonder drugs Salvarsan and Neosalvarsan were arsenic compounds. The whole history of treating syphilis before penicillin was a matter of picking the right poison, and the pre-scientific practitioners who used mercury weren't being irrational — they were doing empirical medicine without the theory to explain it.¹

Sixteenth-century alchemy sometimes got real results. The mechanical ventilation of confined spaces was invented by following the wrong theory (that "noxious airs" caused disease) but saved lives anyway by reducing airborne transmission of actual pathogens. Mediterranean cities had quarantine islands — Lazarettos — centuries before germ theory. These were sophisticated public health measures based on a model of disease that was technically wrong but practically effective.

This pattern should make us nervous about our own era. How much practical knowledge are we currently dismissing because the theoretical framework says it shouldn't work? The cultural-evolution literature documents how the Polar Inuit lost kayak-building technology when their population shrank below the threshold needed to maintain the full toolkit. But knowledge can also be lost by an expanding, theoretically sophisticated civilisation that dismisses what it can't explain.

The Nazi Famine and Celiac Disease

Sometimes catastrophe is the key that unlocks knowledge that was sitting right in front of us. In the winter of 1944, the Nazis cut off food supplies to the western Netherlands in retaliation for a railway workers' strike. The resulting "hunger winter" starved the civilian population. But in the Juliana Children's Hospital in the Hague, pediatrician Willem Karel Dicke noticed something extraordinary: his celiac patients were improving. While other children suffered from the famine, children with celiac disease — previously among the sickest in the hospital, with a 35% mortality rate — were thriving.²

Dicke had suspected wheat since the early 1930s. He'd observed that a child's growth patterns correlated with periods of wheat-free eating at the hospital versus a normal diet at home. But it took a famine — the forced removal of bread from everyone's diet — to produce the natural experiment that proved it. When the Allies liberated the Netherlands and bread returned, celiac children got sick again. Dicke spent five more years meticulously documenting the effect, eventually pinpointing gluten as the culprit.

The bitter irony: the banana diet that preceded Dicke's discovery had been working for twenty years — ripe bananas contain no gluten — but for the wrong stated reason (that complex carbohydrates caused celiac). Its inventor, Sydney Haas, refused to accept Dicke's findings and delayed adoption of gluten-free treatment in the United States for years. Knowledge can be lost not only through disaster but through a correct treatment being yoked to a wrong theory, such that defending the theory means attacking the cure.

What We Can't Read

The Indus script may be the most humbling reminder that knowledge loss isn't always dramatic. More than 4,000 inscriptions survive from the Harappan civilization (c. 3200-1800 BCE), one of the three earliest literate cultures alongside Egypt and Mesopotamia. We can read Egyptian hieroglyphics and Sumerian cuneiform. We cannot read a single word of the Indus script.³

The obstacles are stacked. There is no bilingual text — no Rosetta Stone equivalent. The underlying language is probably Dravidian, based on the survival of the Brahui language in western Pakistan and Dravidian loanwords in nearby Elamite texts, but this remains uncertain. The script has 250-400 distinct symbols, suggesting a logo-phonetic system like Egyptian, but the longest known inscription is only 17 symbols long. Even if we crack the script, most of what we'd find are probably seal impressions — the ancient equivalent of "government property" or "tax paid."

Some scholars have questioned whether the Indus signs represent true writing at all, comparing them to medieval European heraldry: conventional symbols that serve as identification marks without encoding spoken language. The texts show less internal repetition than expected for language, and several non-linguistic pictographic systems from around the world look similar. Against this: the symbols are linear (written in lines), and they crowd at line endings — a distinctive feature of true writing, where the writer avoids breaking up words.

Whether the Indus script is ultimately deciphered or not, it represents a class of knowledge loss that the cultural evolution framework doesn't handle well. This isn't a practice that was slowly eroded by population decline or prematurely dismissed by scientists. It's an entire writing system — the accumulated intellectual infrastructure of a major civilization — that was simply abandoned, probably when the Indo-Europeans arrived around 1800 BCE. The knowledge didn't decay; the civilization that carried it was replaced.

Medieval Multiverse: Knowledge Hiding in Plain Sight

Not all "lost" knowledge was actually lost — some was merely suppressed long enough to be forgotten. In 1277, Etienne Tempier, Bishop of Paris, condemned 219 propositions as heretical, forbidding the Sorbonne's arts faculty from teaching them. One of the condemned ideas was Aristotle's principle that God could not have made more than one world. Tempier's reasoning was theological: an all-powerful God couldn't be constrained by Aristotle's physics. But the effect was to license a remarkable burst of cosmological speculation.⁴

Richard of Middleton responded by arguing that God could create multiple universes, each with its own center of gravity. William of Ware pushed further: if the universe by definition contains all created things, multiple universes would have to be entirely separate, with no way of interacting — what we would now call causally disconnected spaces, or parallel universes. By the 15th century, Nicholas of Cusa was arguing for multiple luminous bodies existing alongside our world — far-off stars, planets, and moons — some possibly inhabited. This was a century before Galileo.

The standard narrative says the medieval period was intellectually dark, with the scientific revolution arriving as a sudden dawn in the Renaissance. The historian Pierre Duhem argued the opposite: there was no "scientific revolution," only a continuation of work already happening in medieval universities. Whether Duhem is right in general, the 1277 case is specific and traceable — a religious authority accidentally opened up the space for cosmological speculation by shutting down Aristotelian dogma. Knowledge sometimes advances not by addition but by the removal of a constraint that nobody realized was a constraint.

The Antikythera Problem

The loss of knowledge isn't always gradual. The Antikythera mechanism — a clockwork computer from ancient Greece, capable of predicting eclipses and tracking the motions of the known planets — represents a level of mechanical sophistication that wasn't matched until the fourteenth century. It's not that the Greeks were secretly more advanced than we thought. It's that one specific line of mechanical knowledge reached an extraordinary level and then vanished, probably because the tradition of craftsmanship that produced it was disrupted by conquest or social collapse.⁵

This connects to the broader argument in the civilisational collapse literature: what collapses is almost always political structure, not knowledge or culture — except when the political collapse specifically targets knowledge-bearing institutions or reduces the population below the threshold needed to maintain complex skills. The Mycenaeans lost their palaces but kept their gods. The Maya lost their divine kings but kept their agricultural techniques. What destroyed knowledge comprehensively was usually deliberate cultural destruction by conquerors — the Spanish burning thousands of Maya books, for instance.

The implication for The Material Basis Of Civilisation is that technological progress isn't monotonic. There isn't a ratchet that prevents regression. The history of medicine suggests we may be sitting on top of useful knowledge that was discarded for bad reasons — not because it didn't work, but because the mechanism wasn't understood, and a culture committed to mechanistic explanation threw the baby out with the bathwater. The question is whether we can be humble enough to go back and check.