The Dying Words
THE GEARS OF LOST KNOWLEDGE — Post 3 of 10 by pazooter
In Post 2, we traced the network that made the Antikythera Mechanism possible—Babylonian temple scribes, Greek mathematicians, Levantine metalworkers, all flowing together at a commercial workshop on the island of Rhodes. This post is about what happened to the language that held all of that together.
The Problem with Dead Words
Every serious field of knowledge develops its own vocabulary. This is not jargon for the sake of jargon. It is precision. A technical term earns its place by naming a distinction that the field has found—through hard experience—to be real and useful. It doesn’t just label something, it names a distinction that practice has proven is real. It’s not just a label, it’s a working instrument.
When the practice dies, the word loses its footing. It either disappears entirely, or it survives in a hollowed-out form—carrying the sound of the original without the operational content. It looks like knowledge. It no longer functions as one.
The Antikythera Mechanism sits at the center of this problem. Its inscriptions include technical astronomical terms in Greek that we can partially read. But the Babylonian vocabulary underneath those Greek words—the specific language used in the cuneiform records that gave the mechanism its predictive power—exists now only in fragmentary clay tablets that took most of the twentieth century to decode. The words survived in clay. The practitioners who knew what those words did, who could use them as tools rather than read them as museum pieces, were gone within a few centuries of the mechanism sinking.
The Word That Wasn’t Translated—It Was Replaced
Here is the clearest example of how this happens.
In the Babylonian astronomical tradition, the term AN.MI referred to an eclipse. But it didn’t mean simply “the sun went dark.” In the Babylonian predictive system, an AN.MI was a specific configuration within the Saros cycle—an 18-year repeating pattern of eclipses. The term encoded a predictive procedure: a specific class of eclipse was expected, at a calculable time, with specific political and geographic consequences attached to it. The word carried all of that. It was an operational term, not a descriptive one.
When Greek translators absorbed Babylonian astronomical knowledge, they translated AN.MI as ekleipsis—which means, roughly, “disappearance.” That word captured the visual phenomenon—the sun goes away—and lost everything else. The predictive procedure. The position in the Saros cycle. The consequence framework. Gone.
The Antikythera Mechanism encodes the Saros cycle’s eclipse predictions using the Babylonian arithmetic procedures. The Greek word on the device’s inscriptions describes something shallower than what the gears are actually doing. The surface language and the operational substrate are from different traditions, and the gap between them is invisible unless you know both.
The Word That Survived by Becoming Something Else
Not all vocabulary loss is this clean. Sometimes the more damaging pattern is a word that survives—but changes what it means without anyone noticing.
The Greek word planetes—wanderer—described stars that moved against the fixed background of the sky. That word carried an observational procedure: a specific way of watching the sky, tracking motion, distinguishing the movers from the fixed points. It named a doing, not just a thing.
It became the Latin planeta, then the Arabic kawkab sayyār, then the medieval European planet, and eventually the English word you use today. At every step of that journey, the word kept some of its original content and shed some. By the time it reached medieval European astronomy, it had become a classification label—a category that planets belong to—rather than a description of an observational act. The procedure was gone. The word remained.
A gap is visible. You can notice when a word is missing and go looking for what filled its place. A misunderstood term is invisible. It looks like knowledge. It passes unquestioned through generations of use. It only reveals itself as a problem when someone tries to build something on it and it fails to hold.
The Compounding Effect
Lost vocabulary doesn’t just vanish once. It degrades in stages, and each stage filters out a little more of the operational content.
Consider what happened to Greek astronomical knowledge on its way to medieval Europe. It traveled through Syriac translation, then Arabic, then Latin. At each step, the receiving tradition absorbed what it had existing categories for and quietly dropped or distorted what it didn’t. Over four or five transmission cycles, a vocabulary precise enough to support the arithmetic prediction procedures in the mechanism’s gears could be reduced to a set of approximate descriptive terms that sounded like knowledge while being operationally insufficient to reproduce what they originally described.
By the time Copernicus was working in the early 16th century, the European astronomical vocabulary was a four-layer palimpsest—Greek terms absorbed into Arabic nominal forms absorbed into Latin case endings absorbed into medieval European categories. Copernicus went back to Greek originals that the Latin-mediated tradition had obscured. He learned Greek specifically to access Ptolemy’s Almagest without the distortions introduced by its Arabic and Latin translations. His complaint was not just mathematical—he wrote explicitly that the inherited Latin astronomical vocabulary had absorbed compromises and approximations that obscured what the original Greek sources actually said. He was, among other things, reading against the grain of his own tradition’s received language.
This is the compounding effect in action: each transmission event filters the vocabulary through the receiving tradition’s existing categories, and what doesn’t fit a category either gets dropped or absorbed into the nearest available approximation. A hundred years of this is manageable. Five hundred years is catastrophic.
The Antikythera Mechanism predates Copernicus by fifteen centuries. The vocabulary it would take to fully read it—to understand not just what the gears do but what the maker knew in order to make them do it—exists now only in fragmentary cuneiform tablets, partially decoded, partially misread, and partially gone for good.
The Living Lexicon and the Written One
This is where something important needs to be said carefully, because it is easy to get wrong.
A living lexicon and a written lexicon are not the same thing. The difference between them is not just a matter of medium. It is a matter of what each one can and cannot preserve.
A living lexicon—a vocabulary maintained through active practice—encodes operational precision. A term means what it means because the person using it has done the thing the term describes, has been corrected by someone who knew the standard, and has calibrated their use of the word against a real referent. The word and the practice are in continuous contact. The practice corrects the word when it drifts. This is how the tupšar Enūma Anu Enlil—the Babylonian priestly scribes who kept the astronomical records—maintained their vocabulary for five centuries. The terms stayed precise because the practitioners using them were still doing the work the terms described.
A written lexicon—a glossary, a codex, a technical manual—encodes the word and its definition at the moment of writing. That definition is already one step removed from the practice. It describes what the term means to someone who already knows the practice. A reader who doesn’t know the practice reads the definition and acquires something: a starting point, a framework, a rough approximation. What they don’t acquire is calibration—the felt sense of what the term points at in the actual doing.
The written lexicon is invaluable when the living chain breaks. Without it, the vocabulary dies entirely when the last practitioner dies. With it, something survives—a pointer toward the original meaning, enough to orient future recovery efforts when physical evidence is found. The decipherment of Babylonian cuneiform in the 19th century was possible partly because written records preserved enough of the structure to crack.
But the written lexicon is also insufficient in a specific way. It presents its definitions as complete. It becomes the authority. The practice recedes behind it. And when that happens—when the codex is the whole of what remains—the definitions begin their slow drift away from the standards they were developed to name. They still look like knowledge. They no longer fully function as one.
The most honest written lexicon would say explicitly: this term names a distinction that can only be fully understood through practice, and this definition is a pointer toward that understanding, not a substitute for it. Very few do.
The ideal is both at once—codex and living practice in continuous contact, each keeping the other honest. The codex updated by the practice. The practice corrected by the codex. When that contact breaks, the written record is what survives, and what survives is a representation of the lexicon—valuable, recoverable, but not the thing itself.
Why This Matters More Than It Sounds
The nomenclature problem is easy to dismiss as a linguistics puzzle. It is not. It is an engineering problem.
What gets built on degraded vocabulary is policy, regulation, public understanding, and institutional response. If the people building those things are working with words that have lost their operational precision—if the term means one thing in the technical paper and something different in the public debate—then the responses being built will address something adjacent to the actual problem while leaving the actual problem untouched.
The Greek word ekleipsis sat above Babylonian predictive procedures on the mechanism’s inscriptions. The word described something shallower than what the gears were actually doing. No one who read only the inscriptions—and didn’t know the Babylonian arithmetic underneath—would understand what the device was actually capable of, or why it worked.
This is the nomenclature problem in physical form. And it did not end with the ancient world.
Next: What Lost Looks Like—Greek fire, Roman concrete, Damascus steel, and Stradivari. Four cases of technology that vanished, and the structural pattern they all share.
References
Freeth, T. et al. “Calendars with Olympiad display and eclipse prediction on the Antikythera mechanism.” Nature 454, 614–617 (2008). — Documents the mechanism’s eclipse inscription terminology and its relationship to Babylonian predictive procedures
Rochberg, F. The Heavenly Writing: Divination, Horoscopy, and Astronomy in Mesopotamian Culture. Cambridge University Press, 2004. — Foundational study of Babylonian astronomical vocabulary and the operational precision of cuneiform eclipse terminology
Steele, J.M. “Eclipse Prediction in Mesopotamia.” Archive for History of Exact Sciences 54 (2000), 421–454. — Primary scholarly source on the Saros cycle predictive system and its arithmetic procedures
Neugebauer, O. A History of Ancient Mathematical Astronomy. Springer, 1975. — Standard reference for the transmission of astronomical knowledge from Babylonian through Greek to medieval European traditions
Pingree, D. “The Greek Influence on Early Islamic Mathematical Astronomy.” Journal of the American Oriental Society 93 (1973), 32–43. — On the Syriac and Arabic transmission of Greek astronomical vocabulary
Fascinating read. I'm reminded of so many structures in our historic town, that are still standing, still usable, built by those who had the knowledge. Unfortunately, the knowledge was not passed on. Thus we now have crumbling infrastructure, crumbling sidewalks, and more.
Sounds a lot like 'USA government' today.