The Babylonians Invented Prediction, Not the Greeks

The first people to predict the future by arithmetic, and to be right about it, were not Greek. They were the temple scribes of Babylon, and they got there by doing something almost nobody credits as a discovery: they wrote down what the sky did, every night, for roughly seven hundred years, in a ruled ledger whose format barely changed. The Greek "invention of science" story — Thales, then Pythagoras, then Plato's geometry and Aristotle's spheres — treats that ledger as raw material at best and superstition at worst. My reading is the reverse. The Babylonians built a working predictive science without a single geometrical model of the heavens, and the two Greeks we most credit with mathematical astronomy, Hipparchus and Ptolemy, could not have done what they did without the Babylonian numbers.

That is not a fringe position any more. It is close to the consensus of the people who actually read the tablets — Otto Neugebauer, Abraham Sachs, Hermann Hunger, Gerald Toomer, Francesca Rochberg, John Steele, Mathieu Ossendrijver. But it has not made it into the textbook, and I think the reason is worth spelling out, because it says something about what we are willing to call "science."

The ledger

Start with the object itself. The Astronomical Diaries are clay tablets, each covering half a year or so, on which a scribe recorded, night by night, the positions of the Moon and planets relative to a fixed set of reference stars (the "Normal Stars," in modern jargon), the times between moonrise and sunset and similar intervals around new and full Moon, eclipses, halos, weather, the level of the Euphrates, the market prices of barley, dates, sesame, wool and cress, and whatever news reached the city — a king's illness, a battle, a plague. The earliest surviving fragment dates to 652 BCE; the latest to the first century BCE.¹ The bulk were excavated at Babylon in the nineteenth century, sit in the British Museum, and were edited by Sachs and Hunger in the 1980s and 1990s.

Two things about the format matter more than any single observation. First, it is standardised. The same phrases, the same units, the same order of items, whether the tablet is from the reign of an Assyrian overlord, a Persian satrap, Alexander, or a Parthian governor. A scribe in 200 BCE could read a Diary from 500 BCE without translation. Second, when the sky was clouded, the scribe did not leave a blank. He wrote what the position should have been, marked as not observed, or copied the value from a prediction. Prediction and observation live in the same columns from very early on.²

Nobody keeps a record like that for its own sake. The Diaries existed because Babylonian scholars believed the sky carried signs about the state, and because their institution — the temple of Marduk, Esagil, and its astronomers, whose employment contracts we actually have — paid them to watch.³ The motive was divinatory. The product was a database. That combination is the whole story, and it is where the argument about "science" begins.

Omens are not the opposite of prediction

Behind the Diaries stands an older text, Enūma Anu Enlil, the great omen series: some seventy tablets and several thousand entries, each of which runs, to give the form in my own words rather than a translator's, if the Moon is eclipsed in a given month and the shadow comes in from a given quarter, then such-and-such will befall the king of Akkad. It was compiled by the early first millennium from material that in places reaches back to the Old Babylonian period.⁴ The most famous single piece is Tablet 63, the "Venus Tablet of Ammiṣaduqa," which preserves what appear to be genuine observations of Venus's disappearances and reappearances from the seventeenth century BCE, wrapped in omens.

The textbook line treats this as astrology, therefore not astronomy, therefore not science, and treats the moment the Babylonians "graduated" to arithmetic tables as the moment they stopped being astrologers. Rochberg has spent a career dismantling that division, and I find her persuasive.⁵ Her point is not that omens are secretly rational; it is that the omen literature already assumes the sky is regular. An omen conditioned on the Moon being eclipsed on the fourteenth is useless unless you can tell the fourteenth from the fifteenth, which requires a calendar tied to lunar phases; an omen about a planet's first visibility requires knowing what "first visibility" is and roughly when to expect it. The letters from scholars to the Assyrian kings Esarhaddon and Ashurbanipal in the seventh century show the mechanism working: the astronomer writes that an eclipse is due, or is not due, on such a day, and advises the king accordingly.⁶ Sometimes he is wrong and says so. That is a predictive practice under institutional pressure to be right, with a written record of hits and misses. Call it what you like; it is not the opposite of science.

The motive was divinatory. The product was a database. That combination is the whole story.

I should flag the disagreement honestly. Neugebauer, who did more than anyone to recover the mathematics, drew a firm line between the "scientific" ephemerides and the omen material, which he treated as a separate and lesser thing; and David Brown has argued that a real conceptual break happened in the eighth or seventh century, when the emphasis shifted from reading signs toward predicting them, driven precisely by the demand of the Assyrian court.⁷ Rochberg's counter is that the actors themselves — the same men, with the same title, ṭupšar Enūma Anu Enlil, scribe of the series Enūma Anu Enlil — never drew that line, and that imposing our category of "science" on them tells us more about us than about Babylon. You can hold both views at once, I think: something did change in the seventh century, and it changed inside a divinatory institution rather than in defiance of one.

What they could actually predict

Here is the part that gets skipped. By the Persian and Seleucid periods (roughly the fifth to first centuries BCE), the Babylonians had at least three layers of predictive tool, and each is documented in surviving tablets rather than reconstructed from Greek hearsay.

Periods

The simplest layer is the recognition that planetary phenomena repeat after fixed intervals: Venus returns to the same phase against the same stars after 8 years, Mercury after 46, Saturn after 59, Jupiter after 71 (or 83), Mars after 79 (or 47), and lunar eclipses recur after 18 years — the cycle we now call the Saros, 223 synodic months.² The "Goal-Year Texts" are exactly what the name says: for a target year, the scribe pulls from the Diaries the observations from one Venus-period, one Jupiter-period, and so on earlier, adjusts by small known corrections, and hands the result to whoever will watch that year. It is prediction by lookup. It requires no theory. It requires a ledger long enough and consistent enough to look things up in, and no one else in the ancient world had one.

Eclipse prediction is the showpiece. Steele's study of the surviving eclipse texts shows the Babylonians tracking eclipse possibilities in Saros-based schemes from at least the sixth century, arranging them in tables that identify which months could carry an eclipse and predicting the time of night to within a plausible margin.⁸ The Diaries duly record predicted eclipses that failed to appear — the standard formula, in my rendering of it, notes that the eclipse passed the city by — alongside those that did. A record of your own misses is the signature of a science, not a superstition.

Arithmetic ephemerides

The second layer is the one Neugebauer edited in his Astronomical Cuneiform Texts and which everyone means by "Babylonian mathematical astronomy": tables, month by month, of the Moon's or a planet's position and the dates of its characteristic phenomena, generated not from observation but by rule.⁹ Two families of rules exist, labelled System A and System B by modern scholars. System A models a variable quantity (say, the Sun's monthly progress along the ecliptic) as a step function: one constant speed in one arc of the zodiac, another in the rest. System B models it as a zigzag: the quantity rises by a fixed increment each month to a maximum, then falls by the same increment to a minimum, and back. Combine several such columns — solar velocity, lunar velocity, latitude, the length of daylight — and you can compute the moment of new Moon, the visibility of the first crescent, the possibility of an eclipse, for years ahead.

Notice what is missing. There is no sphere, no orbit, no epicycle, no cosmos at all. There is nothing that would answer the Greek question, "what is the shape of the heavens?" The Babylonian question was different: "given the numbers I have, what number comes next?" And the numbers were remarkable. The value the System B lunar theory uses for the mean synodic month, in sexagesimal 29;31,50,8,20 days, is about 29.530594 days. The modern figure is 29.530589. They were within half a second per month, from a table.⁹

Geometry after all — but not the Greek kind

The third layer is recent news. In 2016 Mathieu Ossendrijver published a group of tablets, datable to between roughly 350 and 50 BCE, in which Babylonian scribes compute the distance Jupiter travels over sixty days by treating its daily velocity as the height of a trapezoid and calculating the area — and then, more strikingly, splitting the trapezoid into two of equal area to find the time at which Jupiter has covered half the distance.¹⁰ That is a graph of velocity against time, and integration under it, of a kind that historians had assigned to fourteenth-century Oxford and Paris. It is still not a model of the cosmos. It is geometry applied to numbers, not to space. But it kills the tidy line that says the Babylonians did arithmetic and the Greeks did geometry.

What the Greeks took

None of this would matter for the "who invented science" question if Greek astronomy had grown up independently. It did not. Toomer, whose translation of Ptolemy's Almagest is the standard one, argued in a series of papers that Hipparchus's lunar theory in the second century BCE was built on Babylonian parameters and Babylonian eclipse records — that his mean motions for the Moon are the System B values, and that his method of testing them against eclipses separated by long intervals depends on having a run of dated Babylonian eclipses to test against.¹¹ Ptolemy, three centuries later, says outright that he has observations going back to the era of Nabonassar, that is, to 747 BCE, and the earliest eclipses he actually uses in the Almagest are Babylonian ones from 721 and 720 BCE.¹² The Greek geometry — the eccentrics and epicycles — is Greek. The numbers that were fitted into it, and the record against which the fit was checked, were carried west from Babylon, probably in the wake of Alexander, and possibly through named intermediaries; Strabo lists Babylonian astronomers by name, including Kidinnu, whom later tradition credited with a lunar theory.¹¹

A record of your own misses is the signature of a science, not a superstition.

Neugebauer's point, put bluntly, is that the sexagesimal place-value system in which Ptolemy writes his tables, and in which we still write minutes and seconds of arc and time, is Babylonian, and so is the habit of tabulating. That habit is not a footnote to Greek astronomy. It is the technology of prediction. The Greek philosophical tradition of the fifth and fourth centuries had cosmologies in abundance and could not tell you when the next eclipse would fall. Eudoxus's nested spheres are beautiful and predict nothing usable. Something happened between Aristotle and Hipparchus that turned Greek astronomy quantitative, and the something has cuneiform on it.

I want to be careful about how far this goes. The Greeks added things the Babylonians never had: the demand that a model be physically true, the trigonometry (Hipparchus's chord table is Greek), the ambition to unify the whole sky in one system. Ptolemy's Almagest is a different kind of book from any tablet. If "science" means a geometrical model of nature that explains as well as predicts, the Greeks have a claim. But if "science" means a disciplined institution that observes, records in a standard format, extracts regularities, predicts, checks the prediction against the next observation, and keeps the record of failure — then the temple astronomers of Babylon were doing it a good three centuries before Hipparchus, and the Greeks inherited the practice along with the numbers.

Why the story got told the other way

Part of it is simple access. Greek astronomy survived in Greek manuscripts that Europe could read; Babylonian astronomy was buried until the 1870s and undeciphered as astronomy until the Jesuit scholars Epping, Strassmaier and Kugler worked out the tables between the 1880s and the 1920s, and Neugebauer's editions were not complete until 1955.⁹ The Greek story had a two-thousand-year head start.

Part of it, though, is a prejudice about what counts. The Babylonians left no treatise on method, no argument for why the zigzag function should be trusted, no named genius with a biography. They left procedures and tables, unsigned, in an institution that also produced horoscopes and read the livers of sheep. We are comfortable calling Kepler a scientist despite his astrology because he wrote books explaining himself. The Babylonian scribe wrote nothing of the kind, so his omens are held against him and his tables are treated as a curiosity. This is the same reflex that reads the Code of Hammurabi as a legal system because it looks like one, when the evidence for it ever being applied is thin, and that remembers Nebuchadnezzar as a destroyer from a single Biblical angle when his own records show a builder. We read the ancient world through the genre we expect, and Babylon keeps turning out to have been working in a genre we did not have a name for.

There is a comparison with Greece I find sharper than the usual one. Delphi, the great Greek engine of prediction, worked by aggregating human information through a sanctuary and dispensing it in ambiguous verse. Babylon's engine worked by aggregating the sky through a temple and dispensing it in numbers that could be checked. Both were religious institutions. Both were consulted by kings. One of them left us the synodic month to six sexagesimal places.

The Babylonians did not "anticipate" science, as if science were waiting in Athens for someone to arrive. They built one particular kind of it — the kind that says the future is a column you have not filled in yet, and that if the ledger is long enough and the format strict enough, you can fill it in early. That is not the whole of what we mean by science, and it is not what Aristotle meant. But it is what the weather forecast means, and the tide table, and every model that predicts from data without pretending to explain, and it was invented in Babylon by men whose job title made them scribes of the omen series, and who would not have understood why we needed to decide whether they were scientists before we were willing to admit that they were right.