The Septuagint is not Christian

Nearly every complete ancient copy of the Greek translation of the Hebrew Bible was made by Christian scribes, and the large bound volumes from the fourth and fifth centuries that carry it are Christian books. Only about fifteen fragmentary Greek scriptural manuscripts can be dated by their handwriting to before the destruction of the Jerusalem temple in 70 CE, and apart from the Greek scrolls recovered in the Judean desert, almost nothing that escaped Christian copying survived at all.^[1] A text preserved almost entirely by one community comes to look like that community’s own property, which is why the Septuagint is so often taken for a Christian production, assembled to supply Christian arguments and to stand at the front of the Christian Bible.

The translation belongs to a much earlier world. After Alexander the Great died, the empire he had built was divided among his generals, whose rival kingdoms then passed to their descendants, and Egypt went to the Ptolemies, who governed from the new city of Alexandria. “Hellenistic Jewish” names the Greek-speaking Jewish communities that lived under those successor kingdoms, and the largest and best documented of them was in Egypt, where tens of thousands of Jews arrived as soldiers, captives, and economic migrants, settled in the cities and the countryside, organized their own districts, and within a generation or two were living daily life in ordinary spoken Greek rather than in Hebrew or Aramaic.^[2] The five books of Moses were rendered into Greek in Alexandria in the third century BCE, and the remaining books followed over the next two centuries, some in Alexandria and some in the Jewish homeland. Those dates place the whole enterprise centuries before there were any Christians, and equally before the rabbinic movement, whose teachers and traditions took shape in the land of Israel only after the temple fell.

Everything known about who read the translation confirms that setting. Its core readership was the Greek-speaking Jews of the eastern Mediterranean cities, with Alexandria as the community that most fully represents Hellenistic Judaism, and the Greek scriptures served as the shared inheritance of the Jewish diaspora, the communities living outside the land of Israel, for at least five hundred years.^[3] The language of the translation keeps pointing back to that setting as well. Festival names carry their Aramaic forms straight into Greek, separate words distinguish Israelite altars from those of the surrounding cults, and a number of Greek renderings agree with later rabbinic interpretation more closely than the later Jewish revisions of the Greek do, while literary evidence shows the Greek text being read aloud in Greek-speaking Jewish gatherings from the first century BCE onward.^[4]

Even the familiar name comes from the readers who inherited the translation rather than from the people who made it. The story of seventy-two elders translating for King Ptolemy is itself Jewish, told first in a Hellenistic Jewish text, and the earliest Jewish sources apply it only to the five books of Moses, whereas Christian writers from the second century CE onward stretched the achievement of the Seventy to cover any and every book of the Greek scriptures they accepted.^[5] The word “Septuagint” for that whole collection first appears in a Christian author and turns up in no surviving pre-Christian Jewish source, which leaves a historical question in place of a theological one. The question is how a body of Jewish writings in Greek, read in synagogue communities and larger than the set of books preserved in Hebrew, came to function as the authoritative scriptures of Christian congregations by the end of the second century.^[6]

Setting aside the question of who later used the translation brings the translators back into view, and their most discussed departures from the Hebrew have reasons that fit their own time and place. A command against reviling God becomes a command against reviling the gods of other peoples, which suits a minority living under Gentile kings among neighbors who had cults of their own. A scene in which Israel’s elders see God becomes a scene in which they see the place where God stood, which resolves a difficulty the Hebrew text of Exodus raises against itself. Balaam’s oracles about a coming ruler are recast in language shaped for an age of Gentile kingship rather than a restored Israelite throne. Each of those choices is an act of interpretation, and each one answers a Hellenistic Jewish question rather than anticipating a Christian one.

Several renderings gave Christian readers trouble rather than support. Greek Isaiah lacks the royal titles that Christian writers most wanted in the famous verse about the child who is born, offering a single descriptive name instead, and Greek Isaiah inserts the names Jacob and Israel into the passage about the servant, tying that figure to the people rather than to an individual. Jewish revisers were already pulling the Greek back toward the Hebrew in the first century BCE, before Christianity spread, and rabbinic teachers centuries afterward were still discussing the translation as something Jewish elders had made for King Ptolemy. Christian scribes are the reason so many good manuscripts of the Greek scriptures survive,^[7] and what they copied was a translation Jewish communities had already been reading for centuries.

The Letter of Aristeas spends most of its length inside a Ptolemaic palace, where a king commissions a Greek translation of the Jewish Torah for his collection, pays for the project, and entertains its translators through a week of banquets. Once the seventy-two elders have finished their work, the closing scene carries the new translation out of the palace to the Jewish population of Alexandria, where it is read aloud to the assembled community, approved by its priests, elders and communal officers, and then guarded by a curse laid on anyone who would tamper with it.

Everything up to this point had been royal business, an acquisition for the library at Alexandria commissioned by Ptolemy II and managed by his librarian, and in these final paragraphs that motive almost disappears. All that survives of it is the request that a copy be handed on to the community’s own leaders, while the translators, the Jewish community and its leaders step forward as the people the project was actually for.^[1]

The shape of the ceremony draws on Israel’s own scriptures, since reading a document aloud to an assembly and receiving the people’s spoken consent is the standard way a text becomes binding in the Hebrew Bible, the procedure followed at Sinai in Exodus 24:3-7, at Josiah’s hearing of the scroll found in the temple in 2 Kings 23, and at Ezra’s reading of the Torah in Nehemiah 8. The gathering is described as the Jewish people as a whole rather than as one congregation among several, and the seventy-two elders standing among them are meant to recall the seventy elders who went up the mountain with Moses, an echo already prepared earlier in the book, where the translators are called elders and the twelve tribes are brought back into the story so that each can send delegates.^[2] What the scene produces is a second giving of the Torah, in which the Greek carries the same standing as the Hebrew and can serve the Alexandrian community independently of it.^[3]

The prohibition attached at the end draws on Deuteronomy 4:2, the instruction against adding anything to the commandments or taking anything away, though Deuteronomy itself curses those who break the covenant rather than those who edit a scroll. Whoever issues that prohibition settles the question of ownership, and the library at Alexandria is absent from the whole scene, with the demand that nothing be changed coming from the Jewish leaders on their own authority rather than from a Ptolemaic official protecting an expensive purchase.^[4] Understood as a ceremony of authorization rather than as an early copyright notice, the passage does something larger than protect a manuscript, because the community’s approval covers both the obvious change of language and every less obvious change that turning Hebrew into Greek required, ratifying all of it as genuine. The Greek text is not treated as a second-best stand-in for the real thing, and the people who will actually live by it are the ones who declare it accurate.^[5]

The book that preserves this story is itself Alexandrian Jewish writing. Its narrator presents himself as Aristeas, a Greek courtier serving Ptolemy II in the early third century BCE, a persona that does not hold up, since his knowledge of Jewish practice is far too accurate for an outside observer, the court titles and official formulas he uses belong to the second century rather than the third, and the librarian he casts as supervisor of the project had in fact lost the king’s favor early in that reign. The author was most likely an anonymous Alexandrian Jew of the second century BCE writing under a borrowed Greek name, a common ancient habit of composing in someone else’s voice.^[6]

Because the account explains where a community’s central possession came from, it is often described as a “charter myth,” meaning a story a society keeps about its own beginnings that justifies its present arrangements more than it records events.^[7] Alexandrian Jews had a Greek Torah they treated as authoritative in its own right, and the closing ceremony explains how it came by that standing. The label needs one qualification, since the Letter of Aristeas is first of all a piece of literature shaped by one author with a message of his own, which makes it a literary reflection of the community’s founding story rather than that story itself.^[8]

Whatever the Letter of Aristeas can and cannot report about how the Greek Torah actually came into being, the people it names as authorizing the translation are Alexandrian Jews acting in their own city, from the assembly that receives the scrolls to the priests and elders who rise to approve them and the leaders who pronounce the prohibition. Their approval follows the procedure that made the Torah binding at Sinai, and the curse they attach is drawn from the Torah’s own warning against changing what was commanded, which puts the Greek translation under the control of the community that had it made.

Philo of Alexandria, a Jewish philosopher who wrote in Greek during the first half of the first century CE, told the story of the Greek Torah roughly two and a half centuries after the translation was made, and he told it inside a biography of Moses rather than as a chapter of Egyptian history. That setting governs the whole account, since the argument running through the second book of On the Life of Moses is that Moses surpasses every lawgiver among the Greeks and among other peoples, that his laws alone have come down unaltered, and that they are honored even by nations who never received them. The translation into Greek enters as the proof of those claims, the moment when laws readable by one people became available to everyone else.^[1]

What Philo describes is not the working committee of the Letter of Aristeas. Aristeas has seventy-two elders compare their results and settle on one agreed version, while Philo never says how many translators there were and has them work apart from one another, each alone with his own thoughts, arriving independently at Greek that agreed word for word. He does not argue for the inspiration he claims so much as assert it, treating the exact agreement of separate renderings as a fact reported by others, and he calls the men who did the work prophets rather than interpreters.^[2]

The claim about the two languages reaches well beyond praise for the translators. Anyone who learned both Greek and Hebrew, the language Philo usually calls Chaldean, would on his account find the two versions the same throughout, so closely matched that Philo treats them as one text in two languages. A Greek Torah fully equal to the Hebrew is not a concession to people who have lost the original language but a form of the Torah in its own right. That position suited a community whose reading of scripture, whose schooling, and whose public argument all happened in Greek.

The festival that closes the account appears in no other version of the story. Every year, Philo reports, a gathering was held on Pharos, the small island off Alexandria where the translation was said to have been carried out, and those who came paid their respects at the site and gave thanks before spreading along the beach to eat in the open with relatives and friends. Nothing in the description is turned into allegory, and the details read like those of a person who had been present or who knew people who had, which makes the festival a real communal event rather than a literary device, an annual re-enactment of what Alexandrian Jewry remembered about its own beginnings.^[3] Egyptian Jews were still keeping the day in the first century CE, some two hundred and fifty years after the Torah had first been put into Greek.^[4]

Aristeas has an annual commemoration too, though of a different kind, since there the king tells the translators that the anniversary of their arrival will be observed each year during his own lifetime, which ties the observance to the court and to the reign of a single monarch. Philo’s festival belongs to the community rather than to the palace, it is held at the place where the work was done, and it outlasted the dynasty that Aristeas flatters.^[5] The gathering also reversed the conditions the translators had wanted, since they had crossed to Pharos for quiet away from the crowds and sickness of the city, and the anniversary filled the same stretch of shore with a crowd.^[6]

Non-Jews sailed across for the festival as well, and Philo reports their presence with evident satisfaction, closing his account with the hope that other nations would one day set aside their own customs and take up the laws of Moses instead.^[7] That openness was genuine, and it did not make the celebration any less a Jewish one, because the Greek Torah being honored on that beach was the possession of the Alexandrian Jewish community, the text its members heard in their assemblies and taught to their children, and the festival was theirs to keep or to let lapse.

Philo’s version of events is not a reliable record of how the Greek Torah was actually produced, since the individual books show the hands of different translators working at different times. It is excellent evidence for what Greek-speaking Jews in first-century Alexandria believed about the book at the center of their common life. Talk of divine prompting, of translators kept apart who still arrived at identical Greek, and of a translation that loses nothing of the original was already in circulation long before any Christian writer took it over and pushed it further. The highest claims ever made for the inspiration of the Septuagint were made first by a Jewish philosopher, about a Jewish translation, for a Jewish community that kept an annual holiday in its honor.

The Alexandrian community that kept an annual festival for its translation also had to live with what the translation said, and its rendering of the laws covering everyday disputes shows the pressures of life under Ptolemaic rule more plainly than its stories about origins do. The Hebrew text sets two prohibitions side by side, one guarding speech about God and one guarding speech about the man who leads the community. The word behind the first is elohim, a Hebrew noun that carries a plural ending yet ordinarily names the God of Israel alone, and the laws about judges and courts surrounding it give it that sense here. Jewish interpreters in the land of Israel later read the word in this verse as a reference to judges, which keeps the whole sentence focused on the community’s own courts and its own leadership.

The translators working in Alexandria in the third century BCE took the plural ending at face value and produced a sentence about gods rather than about God, then reshaped the second clause to match, so that one leader became a whole class of officials. A single word ending accounts for the difference, and it changes what the law is about, since a rule governing how Israelites speak of their own God and their own leader becomes a rule governing how a minority speaks of everyone else’s gods and everyone else’s magistrates.

Nothing in the Hebrew required this. The Hebrew sentence was not understood to be speaking about other deities at all, and the translator exploited a plural ending the Hebrew never intended as a plural, holding onto a surface resemblance to the source while producing a Greek sentence that says something the source does not say.^[1] The fullest account ties the choice to the position of Greek-speaking Jews under the Ptolemies, where the Torah had become a text that non-Jewish neighbors could read for themselves, and where a law forbidding insults to locally worshipped gods answered a standing suspicion that Jewish separateness meant contempt for everyone else’s worship.^[2] A second reading treats the plural less as outreach than as self-defense, anticipating Gentile criticism rather than stating a principle about other peoples’ gods.^[3] A third possibility leaves room for accident, since an automatic rendering of a plural ending could produce the same Greek without any intention behind it, and what a translator meant is hard to prove.^[4] None of the three explanations needs Christianity in order to work.

The Hebrew protects a single office inside Israel, while the Greek protects officials in the plural, language that suits the layered administration of Ptolemaic Egypt far better than it suits the tribal leadership of the Hebrew text. That fit is not isolated, since the Greek legal vocabulary running through this whole block of Exodus law repeatedly reaches for terms at home in Egyptian courts and contracts, which suggests a translator with one eye on how the sentence would sound to the officials who governed him.^[5]

The earliest readers of the Greek verse took it in the accommodating sense without hesitation. Philo, writing in Alexandria in the first century, understood the command as forbidding abuse of the gods that other peoples acknowledge, on the practical ground that such abuse invites the same treatment in return against the God of Israel. Josephus, addressing a Greek-speaking audience a generation later, reported that the Jewish lawgiver had expressly forbidden deriding or blaspheming the gods recognized by others, out of respect for the word for god itself.^[6] Neither writer needed the plural explained to him, and neither handled it as an embarrassment to be managed.

Three centuries later the same words turned up in an argument aimed at Christians, when the emperor Julian, raised as a Christian and later a public opponent of the movement, quoted the Greek command by name to show that the Jewish lawgiver himself had required restraint toward the gods of other peoples, restraint Julian said Christians had thrown away.^[7] Cyril of Alexandria, answering him, read the plural as a reference to human leaders, who can be called gods because they bear God’s image.

The plural belongs to a community that lived among cults it had no part in, under kings it had no say in choosing, and the rendering reads as one attempt to make the community’s position in the city easier to defend. Putting the Torah into Greek set it in front of neighbors and officials who could now read it without an interpreter, so the Torah itself became part of how the community presented itself in public, and the choices made at this verse present it as a text that asks restraint of its own people rather than one that licenses attacks on the worship of everyone around them. Philo and Josephus both reached for the verse when they wanted to tell a Greek-speaking audience what the Torah required, which shows the sentence still doing that work generations after it was made. A minority with no king of its own and no say in who governed Alexandria had few ways of shaping how it was seen, and its own scripture, now readable by the people it lived among, was one of them.

Two chapters after the command about the gods of other peoples, the same translators met a difficulty of another kind, since Exodus 24 brings the covenant ceremony at Sinai to its height, with Moses, Aaron, Nadab, Abihu, and seventy elders of Israel climbing the mountain together for a meal in the divine presence. The Hebrew account of what happened on the mountain is direct, stating the sight as a plain fact and then noting that the group ate and drank afterward without coming to harm. That directness creates a difficulty inside the book of Exodus itself, since the same book later has God refuse Moses’ request to be shown the divine glory with the flat statement that no one shall see him and live.

The Greek translation made in Alexandria in the third century BCE reports something else at those two points. Where the Hebrew has the elders looking at God, the Greek supplies a noun meaning “place,” so that what the group sees is a location associated with God rather than God himself, and at the close of the scene the leaders stand in that location rather than looking at God.

The Hebrew here contains nothing that could have been misread as the word for a place, so the usual innocent explanation, that the translator was working from a Hebrew text different from the one that survives, is not available. One noun has been added in each verse, reducing the direct sight of God in the first and eliminating it from the second. Changes of this kind fold interpretation into the act of translating itself rather than adding it afterward as a comment, and the motive is visible in the surrounding narrative rather than in anything outside the book. Moses is ordered earlier at Sinai to warn the people against trying to look at God, since many of them would otherwise die, the people themselves then ask Moses to mediate so that they will not die, and the refusal in chapter 33 is repeated a few verses after it is first given. Reading the elders as seeing a place rather than a person removes the contradiction and brings the covenant meal into line with everything else the book says about looking at God.^[1]

Even the vocabulary of the solution came from the Torah. Immediately after the refusal, God tells Moses that there is a place nearby where he may stand, which supplies both the noun the Greek uses and the logic behind using it, and that noun already occurs nine times across the book. A comparable restraint shows up in two further divergences in the Greek Exodus, where an angel meets Moses on the road in place of God meeting him, and where Moses climbs the mountain of God rather than climbing up to God, each of them the addition of a single familiar word. The translator was not rewriting the scene so much as inserting the smallest possible buffer, using terms the Hebrew text had already made available.^[2]

Reluctance to describe God in human physical terms, with a face, hands, or a body that can be looked at, was never confined to Greek. The Aramaic translations of the Torah arrived independently at their own solutions for this verse, having the elders see the glory of the God of Israel or the glory of the divine presence, and Symmachus, revising the Greek in the second century CE, wrote that Moses and the elders saw God in a vision. Three separate lines of Jewish translation, working in different languages across several centuries, all found the same sentence uncomfortable and all reached for a way of putting distance between the elders and the sight.^[3]

The most frequently repeated example of this kind of caution is the Song of the Sea in Exodus 15, where the Hebrew calls God a man of war and the Greek instead describes God as the one who shatters wars, and that example is considerably less secure than its popularity suggests. The Hebrew copy in front of the translator may already have read differently, since the Greek answers closely to a Hebrew phrase used elsewhere for God breaking war, in which case the rendering is word for word rather than squeamish. The Greek may also carry no squeamishness at all, since in a book that reports the plagues, the crossing of the sea, and the battles that follow, it states that God always wins rather than that fighting is beneath God. No consistent avoidance of human descriptions of God appears anywhere in the Greek Torah, which means every case has to be argued on its own evidence.^[4][5][6]

Giving up the Song of the Sea costs nothing, since the elders on the mountain do not depend on it, and the disputed case teaches something on its own. Each explanation on offer makes a claim about what Greek-speaking Jewish translators in Alexandria were doing with their own Torah, whether it harmonizes the passage with the rest of Exodus, traces the difference to a different Hebrew copy, or hears a victory song. What Exodus 24 shows is a translator who read the book as a whole, noticed that one passage sat badly against another, and settled the difficulty with language the book had already given him, so that the reverence at work in the Greek came from tensions inside the Torah rather than from anything outside it.

Resolving a tension inside Exodus took nothing more than vocabulary the Torah had already supplied, whereas other passages put the translators in front of a difficulty that came from outside the text, from the political world their readers lived in. Balaam enters the book of Numbers as a foreign diviner hired by the king of Moab to curse Israel, and each time he opens his mouth a blessing comes out instead. His speeches are oracles, pronouncements delivered under compulsion rather than by choice, and the third of them turns from the fertility of Israel to the standing of its future ruler. In the Hebrew, the promise ends in a comparison with Agag, the Amalekite king whom Saul captured and spared and whom Samuel then executed at Gilgal.

The Septuagint reorganizes the verse around a single human figure and replaces the name of the enemy.

Agag barely appears in the Hebrew Bible, losing one battle in 1 Samuel 15 and dropping out of the story afterward apart from a passing description of Haman as an Agagite. He carries almost no weight in later Jewish tradition, so a pledge that Israel’s king would outrank him would have read to Jewish readers of the last centuries BCE as an obscure comparison. Gog carries far more weight, since in Ezekiel 38 and 39 he is the foreign ruler whom God draws against a resettled Israel in order to destroy him and display his power before the nations, and the name became shorthand for the final assault on God’s people in texts written long after Ezekiel. Swapping one name for the other shifts the verse out of the era of Saul and into a future conflict the Hebrew never mentions.^[1]

That change may not have originated with the translator. The Samaritan Pentateuch, an independent Hebrew edition of the Torah preserved by the Samaritan community, also reads Gog here, and agreement between the Greek and the Samaritan text against the standard Hebrew usually indicates that the Hebrew manuscript behind the Greek already had Gog in it.^[2] Whoever introduced it, the readers who encountered the oracle in Greek met a text about a coming conflict rather than a settled score with Amalek.

The fourth oracle receives similar handling, since the Hebrew pairs a star coming out of Jacob with the emblem of royal authority rising out of Israel, two images of kingship set in poetic balance.

The Greek keeps the star and puts an ordinary word for a human being where the Hebrew has the ruler’s staff. The choice cannot be explained as mechanical habit, since the Hebrew term normally means tribe in Numbers and the translator renders it that way everywhere else in the book, and the obvious Greek word for a staff was available and was not used. One long-standing explanation of the substitution is that it takes the royal character away from the promised figure, and the translator of Numbers does decline to call this person a king, a restraint that fits a wider pattern in the Greek Torah of avoiding the title for Israelite rulers, apparently because the translators held that it belonged to God alone.^[3]

Messianism, the hope that God would raise up an anointed deliverer to rule Israel, sharpened in Judea during the upheavals of rule by the Seleucid kings of Syria and the Jewish fight for independence that produced the Hasmonean dynasty, while the Greek-speaking Jewish communities of Egypt lived as a minority under Ptolemaic kings whose favor they depended on, and their hopes were tempered by that political climate.^[4] A verse openly promising a native king who would rise above the ruler of the nations was awkward in such a setting, and the softening of its royal language reads as a politically sensitive reworking of the passage for readers living under Gentile kings.

That reading is contested, since the figure in the third oracle is still given a kingdom of his own, which means royal character has hardly been erased, and pairing an individual man with the name Gog arguably turns the passage into a prediction about a final deliverance rather than a memory of old wars. On that account the Greek strengthens rather than weakens the expectation, though only in a qualified way, since the translator still refuses the words king and anointed while describing a future rescuer who holds a kingdom. No independent evidence, however, places expectation of an anointed deliverer in Ptolemaic Egypt at the time the Torah was translated, and the Greek Jewish literature surviving from Egypt tends to pin its hopes on a well-disposed Gentile ruler instead.^[5]

Neither explanation changes the dating or the direction of influence, since the Greek of the oracles in the Torah belongs to the third century BCE, and the earliest readers who found a coming deliverer in these Greek lines were themselves Jewish. Philo quotes the Greek of the third oracle in his treatise On Rewards and Punishments and applies it to a figure who will lead an army and subdue nations, and the Aramaic translations read in the synagogue understand the star and the staff of the fourth oracle as a king and an anointed one arising out of Israel. When Justin Martyr and Irenaeus eventually reached the verse, they did not even reproduce the Greek word under discussion, using a term meaning leader instead, and they built their case on the star rather than on the man. On either account of these renderings, what the Greek carries is Israel’s hope for a ruler of its own, reworked by Jews for the political moment they were living in. Even the language about a coming deliverer, the part of the verses later readers found easiest to claim, took the shape it has inside the Jewish community that produced it.^[6]

Isaiah promises King Ahaz a sign during a military crisis in the eighth century BCE, the birth of a child whose name will announce that God has not abandoned Judah. The Hebrew describes the mother with the word almah, which identifies a girl who has reached marriageable age and makes no direct statement about whether she has had sexual relations. Whoever put Isaiah into Greek chose parthenos, a word that in ordinary usage means “virgin,” and that single word choice became the best known of all the differences between the Hebrew and Greek forms of the prophetic books.

No Christian setting existed when the choice was made. Greek Isaiah is generally understood as the work of a single translator active in the Hellenistic period, most likely during the second century BCE, and the case for that dating rests on his vocabulary, his sentence patterns, his use of terms tied specifically to Hellenistic Egypt, and his occasional allusions to events under the Ptolemies.^[1] He was working for the Jewish community of Alexandria, to which he himself belonged, at a point when Hebrew had become unfamiliar to many Jews in Egypt and Greek was the language in which they read their scriptures. The rendering that later Christian tradition would treat as a prophecy of Jesus was therefore settled roughly two centuries before he was born, by a Jewish translator serving Jewish readers.

What that translator meant is a narrower question than the later argument assumed. Where the Hebrew leaves it unclear whether the woman is already pregnant or will become so, the Greek makes the conception plainly future, which allows the sentence to describe someone who is a virgin at the moment the sign is given and who will conceive by ordinary means once she is married.^[2] The translator may have judged that a sign worth announcing required more than an already pregnant woman giving birth, and that God’s care for Judah would show more clearly if a girl still unmarried when Isaiah spoke became a mother afterward. Nothing in the Greek claims a conception without a human father, and nothing in the surviving literature of Alexandrian Jewry suggests that this verse produced any expectation of such a birth.

Matthew takes the verse over in its Greek form rather than translating the Hebrew afresh. The gospel writer very probably knew a tradition about the birth of Jesus before he went looking for a text, and what the Septuagint supplied was a prophetic passage on which that tradition could rest, which an argument built on the Hebrew could not have provided.^[3] The Greek word was one of several defensible ways of rendering the Hebrew, and Matthew’s use of it determined which of them Christian readers would afterward treat as the meaning of the prophecy. A Jewish translation became a Christian proof text and not the reverse.

The Jewish response came later, and it shows which of the two readings arose first. During the second century CE several Jewish revisers, among them Aquila, Symmachus, and Theodotion, reworked the Greek scriptures to track the Hebrew far more closely than the Septuagint had, correcting an existing translation rather than beginning one from nothing. All three replaced parthenos at Isaiah 7:14 with neanis, a neutral term for a young woman that answers the Hebrew word much more exactly.^[4] Jewish readers of Greek scripture in the second century and after had access to renderings that removed the ambiguity Christian writers were exploiting, and the neutral rendering is the later of the two.^[5]

By the time Justin Martyr staged his long argument with the Jewish figure Trypho around the middle of the second century, the two Greek readings were already opposed to each other, with Justin insisting on the phrasing of the Alexandrian translation and Trypho holding to the revised one, though the fictional Trypho is never permitted to discuss the rival translations or the Hebrew word behind them.^[6] What prompted the original choice remains uncertain, since the Alexandrian translator may have inherited an existing Jewish reading of the verse or may simply have wanted to indicate that the girl was a virgin at the time she conceived. Either way the choice belongs to Hellenistic Judaism, and the accusation of Christian tampering has the history backwards.

The council described in Acts 15 met to settle whether non-Jews joining the movement around Jesus had to be circumcised and keep the whole Torah, and James ends the debate by appealing to the prophet Amos. The version he cites turns an oracle about the restored kingdom of David into a promise that the nations will turn toward God, which is the question under discussion, and that version exists only in Greek.

Greek and Hebrew Amos agree that the fallen shelter of David will be raised again, and then they part company entirely.

In the Hebrew, the restored kingdom takes over what remains of Edom along with the other nations under God’s rule, so the promise concerns Israel recovering land and standing. The Greek turns the same clause into a statement that the people who remain, together with the nations, will look for God. An oracle of conquest has become an oracle of response, and the argument being made in Jerusalem about admitting non-Jews gains support from prophecy that the Hebrew does not supply.^[1]

Three small decisions produced the difference, all of them possible because Hebrew was written without vowels and often without clear breaks between words. The verb meaning to take possession differs by one letter from the verb meaning to seek, and a short Hebrew word that signals which noun is being acted on was left untranslated, which allowed the survivors rather than Edom to become the subject of the sentence, while the name Edom was read as the ordinary word for a human being, from which it differs by a single vowel.^[2] Whether the translator misread his text or read it the way he thought it ought to be understood remains an open question, since the Hebrew is not particularly hard and the result suits the outlook of a Greek-speaking Jewish community living among non-Jews, for whom the nations were neighbors rather than territory.^[3] The date, however, is not in question, since the rendering belongs to the Greek Minor Prophets, twelve shorter prophetic books translated by Jewish hands well before the first century, and Luke received it along with everything else he read as scripture.

A second New Testament argument leans on the Greek just as heavily, since the writer of Hebrews needs a voice from scripture that sets aside animal sacrifice in favor of an offered human body, and the Greek form of the psalm numbered 40 in Hebrew supplies exactly that while the Hebrew says something else. The Greek Psalms number this psalm 39 because they divide two earlier psalms differently, so their numbering runs one behind the Hebrew from that point on.

The Hebrew line has God opening the speaker’s ears, an image for being made able to hear and obey. The Greek has a body fashioned instead, and that one noun carries the entire argument of Hebrews 10, which contrasts the bodies of sacrificial animals with a single body handed over in obedience.^[4] How the Greek came to read that way is unsettled. Every major Greek manuscript of the Psalms carries the reading about a body, and the only support for an earlier reading about ears comes from Latin translations made from the Greek, which is why the standard scholarly edition of the Greek Psalms, reconstructed from the manuscripts, prints ears while nearly every interpreter of Hebrews assumes a body.^[5] One reconstruction has the writer of Hebrews making the change himself, after which it spread through the manuscripts, while the other has the Greek Psalms reading a body from the beginning, with the wording pulled back toward the Hebrew afterward.^[6] The manuscript evidence favors the second, and the change back is visible in the Jewish revisions of the Greek made by Aquila, Symmachus and Theodotion, each of which restores the ears in order to match the Hebrew.^[7]

Christian alteration of the Greek did happen, and it looks nothing like either of these cases. In Ezekiel 17 the Hebrew has God planting a shoot on a high and lofty mountain, and the Greek translator, faced with a Hebrew word that occurs nowhere else, connected it to the root meaning to hang and called the mountain suspended. An early papyrus copy of Greek Ezekiel preserves that reading, with the mountain still described by an adjective. Most later manuscripts convert the adjective into a verb and add a person as its object, so that the line announces the hanging of someone on the height of Israel, and early Christian writers took it as a reference to the crucifixion.^[8] Intervention by Christian scribes leaves a recognizable trail, with the altered reading confined to part of the manuscript tradition, the earlier form still recoverable from a surviving copy, and the direction of the change running toward a Christian claim rather than away from one.

The Ezekiel pattern does not repeat at Amos 9:12 or at the psalm. The Greek Amos reading is carried by the whole manuscript tradition of the Greek Minor Prophets, and the sentence it produces was available to any Greek-speaking Jewish reader who had never heard of Jesus, so the warrant James finds in Jerusalem comes from a rendering Jewish translators had already put into circulation for their own communities. Where the Greek of the psalm came from is harder to settle, and two points hold whichever reconstruction is right, since the argument of Hebrews 10 works only in Greek, and the correction made against the Greek wording about a body came from Jewish revisers pulling the Greek back toward the Hebrew rather than from Christian ones pushing it further forward. Two arguments that shaped the earliest Christian communities were built on Greek renderings the Hebrew does not supply, and in the Amos case on one that Greek-speaking Jewish translators had made generations before there were Christians to read it.

A translation produced to serve Christian claims would read most favorably at the points where Christian readers pressed hardest. The Greek command about the gods of other peoples already showed what happens when a rendering offers no such help, since Christian writers had to explain that line away once it was turned against them. Isaiah 9 puts the difficulty in a sharper form, since the Hebrew gives the newborn royal child four royal titles, two of which describe a human ruler in language otherwise reserved for God, and those two carried much of the weight in later Christian argument about the identity of Jesus. The Greek translation keeps none of the four.

In place of the whole sequence the Greek supplies a single title built around God’s great plan and the figure who announces it, so that the passage turns away from what the child is called and toward what his reign achieves, a peaceful kingdom with Jerusalem at its center.^[1] Greek Isaiah uses the language of God’s planning again and again, naming his intentions for his people and against foreign powers, and the child is presented as the one who knows that plan and makes it known, a description that suits a teaching priest better than a divine figure.

That rendering, useful to Christian readers at first, turned into a liability within three centuries. Writers of the first four centuries regularly described Jesus in angelic terms and sometimes called him an angel outright, drawing on Jewish traditions about heavenly intermediaries, and the Greek Isaiah passage supplied one of the readiest titles for doing so. The fourth-century dispute known as the Arian controversy, which turned on whether the Son was eternal or a created being, changed what such language could safely mean, and describing Jesus as an angel fell out of use as the discussion moved toward questions about his nature.^[2] A phrase that had once been an asset became something to qualify, and writers who still wanted the verse had to explain that the title described a job rather than what the Son was.

The first of the servant passages in Isaiah raised a different difficulty, since there the Greek translator added material the Hebrew does not have, and what was added settles the servant’s identity in a way that ran against the reading Christian tradition wanted.

The Hebrew opens with nothing more than the word “behold” and leaves the servant unnamed, which is what made centuries of argument about his identity possible in the first place. The Greek puts two names in the slot that word occupied, Jacob and Israel, settling the question before it can be raised and identifying the servant with the people rather than with any individual.^[3] Greek-speaking Jewish communities read the passage in exactly that way and took their lead from the translation, while the Aramaic translation read in the synagogue moved in the opposite direction and set the title Messiah, the anointed deliverer awaited by Israel, alongside the servant. Matthew quotes this passage at greater length than any other text from Isaiah, and the text he quotes is not the Greek translation. Only three of the gospel’s Isaiah quotations differ from Greek Isaiah, and this is the longest of the three, mixing wording that agrees with the Greek, wording closer to the Hebrew, and wording found in neither.^[4] The two inserted names are precisely what the quotation lacks, the feature that makes the Greek hardest to read as a description of a single person, and at that point the gospel agrees with the Hebrew and with a Jewish reworking of the Greek carried out in Palestine around the turn of the era, revision work that later circulated under the name of Theodotion.

Proverbs 8 created a larger problem of the same kind. The Greek translator rendered the verb of Proverbs 8:22 with the ordinary Greek word for creating, so that the figure of Wisdom says God created her at the beginning of his ways. Christian writers took that figure to be the Son before his birth, which left the verse reading as a statement that the Son had been made, and it was fought over by writers throughout the fourth century. Athanasius gave it one of the longest treatments in the second of his orations against the Arians, arguing that the verse could not be allowed to say anything about the Son’s status until it had been set inside a framework built from other texts.^[5] The elaborate defense was necessary because the rendering had been inherited rather than chosen.

The clearest case involves an entire book. Jerome reported that the Greek Daniel was not read at all, Theodotion’s version being used in its place, and he admitted that he did not know how the substitution had come about, though he judged the Greek translation to have strayed far from the truth of the text. The rejection was thorough enough that only one Greek copy survived, since Theodotion’s version had taken its place in every other Greek manuscript of the book.^[6] The corrections, the arguments, and the replacement of an entire book were all necessary because the collection had not been assembled with Christian arguments in view.

Justin Martyr, teaching in Rome around the middle of the second century, addressed his First Apology to the emperor as a defense of Christian belief before an audience that knew nothing of Jewish scripture. Needing to explain how books composed in Hebrew came to be readable in Greek, and why anyone should trust that their predictions had not been manufactured after the events they describe, he told the story of the Alexandrian translation in a form no Jewish writer would have recognized.

The Jewish accounts run differently. In the Letter of Aristeas, an Alexandrian Jewish text of the second century BCE, and in the retelling by Josephus, a royal librarian wants the Jewish Torah for the collection at Alexandria, and elders sent from Jerusalem produce a Greek translation of the five books of Moses. Justin keeps the Egyptian king and the library, drops the elders entirely, and makes the object of the request the prophetic books rather than the Torah, which is exactly the material his argument requires. He also has the king write to Herod, who ruled more than two centuries after Ptolemy, an error suggesting he worked from a secondhand summary rather than from the Jewish account itself.^[1] The retelling does real work for him, since if the prophecies were sitting in an Egyptian library long before Jesus was born, no Christian could be accused of writing them in afterward, and the whole argument from prophecy stands on a document a Roman official could in principle go and check.^[2]

The Dialogue with Trypho, a much longer text cast as a debate with a learned Jewish opponent, turns the same legend against that opponent. Justin says he trusts the translation made by the seventy elders at the court of Ptolemy and not the renderings Jewish teachers were producing for themselves, and he lists passages he claims have been cut out of scripture so that predictions about Jesus would be harder to see. Whenever two versions of a verse are in play, the appeal to the Alexandrian translation settles the matter for him, with the wording he quotes counted as authentic and the competing version treated as deliberate Jewish alteration.^[3] The best known of these disputes concerns the woman of Isaiah 7:14, where he defends “virgin” against “young woman.”

The accusation reverses the history, since the Greek wording Justin defends belongs to the Hellenistic Jewish translation as it already stood, and the renderings he attacks came from Jewish revisers pulling the Greek back toward the Hebrew, so what he describes as corruption is a correction, and the text he holds up as uncorrupted is the Jewish one. The specific charges do not survive checking either, since none of the supposedly deleted passages can be found in the main manuscripts of the Greek scriptures, and verses he credits to Jeremiah and to Ezra appear in no known version at all. The line he valued most is the phrase “from the wood,” attached to the announcement that God has reigned in Psalm 95 in the Greek numbering and 96 in the Hebrew, and it is far more likely a Christian addition than a Jewish deletion, since it is absent from the Hebrew and from the Greek manuscript tradition, and Justin himself later quotes the psalm without it.^[4]

Justin also stands near the beginning of a process in which the origin story grew steadily more miraculous in Christian hands. The Letter of Aristeas has the translators consult one another and compare drafts before settling on a single agreed text, while Philo already has each of them working alone and arriving at identical Greek under a divine prompting, without saying how the agreement was ever checked. A third-century text circulated falsely under Justin’s name makes the separation physical, shutting the seventy in individual cells so that the king could set seventy complete translations beside one another and find them agreeing word for word, and its author claims the foundations of the cells could still be seen on the island of Pharos.^[5] Irenaeus made the isolation itself the proof, arguing that seventy men who never spoke to each other and still wrote the same sentences must have been guided by God, which allowed the Greek to be treated as the equal of the Hebrew and in some places as better.^[6] Jerome, two centuries later, dismissed the cells as an invention, noting that neither Aristeas nor Josephus reported anything of the kind.

The claim Justin made about the prophetic books was carried further after him, until second-century and later Christian writers credited the seventy with the whole of the Greek scriptures.^[7] Augustine carried the development to its furthest point in response to Jerome’s argument that new Latin translations should be made from the Hebrew. Unable to deny that the Hebrew and the Greek differ, and unwilling to see the Greek demoted, Augustine proposed that the same Spirit had spoken twice, once through the prophets in Hebrew and once through the translators in Greek, so that a difference between the two versions was not an error to be repaired but a second revelation to be received.^[8]

Every stage of the argument was needed because the Greek scriptures had come from outside Christian tradition, and by the end of the second century Christian writers had largely taken them over. The Jewish origin of the translation had to be argued away before the Greek text could serve as Christian scripture.

Roughly five centuries after Greek-speaking Jews in Alexandria turned the Torah into Greek, rabbis living in the Hebrew- and Aramaic-speaking communities of Palestine and Babylonia were still telling the story of how it had been done. Their fullest version sits in the Babylonian Talmud, in the tractate Megillah, which is cited by leaf rather than by chapter, so that 9a names the front side of the ninth leaf. The passage is introduced as a baraita, an early rabbinic teaching that was left out of the Mishnah, the collection of legal traditions compiled around 200 CE, yet is quoted as carrying the same age and weight. It is brought in to support a ruling of Rabbi Yehuda, who held that when Greek was permitted for written copies of the biblical books, it was permitted for a Torah scroll and for nothing else, and the reason offered for that narrow permission is the incident involving King Ptolemy.

The shape of the account differs sharply from the Alexandrian legend behind it, where the translators work in consultation and the finished text is ratified by the assembled Jewish community. In the Babylonian telling the king isolates the elders because he does not trust them, setting up a situation in which their separate copies can be checked against one another for signs of collusion, and the agreement God grants them works to hide the changes rather than to prove the translation perfect.^[1] A tradition that reads at first as praise for an inspired translation turns out, on its own terms, to be a story about elders who quietly withheld something from a Gentile king.

The list that follows the miracle names around fifteen places where the elders departed from the Hebrew. The first two fall in the opening lines of Genesis. In the first, the word order of the opening clause is rearranged so that the divine name stands ahead of everything else, which blocks any attempt to take the first word as the name of a prior power who made God, and in the second, the plural used for the creation of humanity is turned into a singular so that no reader could infer several creators or a God shaped like a person. Those are questions rabbinic commentary was already working through on its own, without reference to any Greek text, since the same rearrangement of the opening clause answers difficulties raised and resolved in the later rabbinic commentaries Genesis Rabbah 1:14 and Tanhuma.^[2]

Further down the list the elders are said to have avoided the ordinary Greek word for the hare among the forbidden animals of Leviticus, and the rabbis explain the avoidance as a courtesy to the king, whose mother or wife they believed carried that name. No Ptolemaic queen was called anything of the kind, though the connection they sensed belongs to a different member of the royal family, since the Greek word for hare sounded exactly like Lagos, the name of the founder of the dynasty that ruled Egypt.^[3] It is often said as well that the Greek translators themselves set the word aside out of political caution, and that has not held up, since the term the Septuagint uses is an equally good rendering of the Hebrew and appears interchangeably with the alternative.^[4]

The list fits the surviving Greek Torah only loosely, with about five of the items identical to the known Greek text and one more coming close, and it circulated as a document of its own, gathering real differences between Hebrew and Greek together with variations among Hebrew copies that had nothing to do with the translation at all. The actual Septuagint diverges from the Hebrew in far more than ten or fifteen places, and several entries look like misunderstandings of accurate Greek renderings rather than records of Greek departures, which suggests that those who preserved this tradition knew the Greek Torah only in part and had little direct access to a copy.^[5] What was inherited was the story and a short list, not the translation itself.

Ptolemy is neither the sponsor of the project nor the person who conceived it. The account casts him as the one the Torah is written for, and rabbinic tradition already knew what a Torah written for a king meant, from the instruction of Deuteronomy 17:18 that a king write out a copy for himself under priestly supervision. A translation prepared under those terms is a royal copy for the king to learn from rather than a text for communal reading, which places the decision of what to hand over squarely with the elders.^[6]

The passage holds admiration and criticism together in one continuous story, which makes it hard to read as either praise or complaint, since respect for a divinely guided translation sits directly against an account of translators who altered what they had been given. The Septuagint was a Jewish enterprise, and the translation was taken to be Christian literature only later, once its language shaped the New Testament and the tradition that grew from it.^[7] Christian writers of the same centuries were building the same motif of the isolated translators into an argument of their own, yet nothing in the rabbinic memory imagines the Greek Torah as anyone else’s book. Whatever these rabbis got wrong about the text in front of Ptolemy, they remembered its makers as Jewish sages weighing what a Gentile king could safely be given, and they preserved that memory in Hebrew and Aramaic, in communities that had no stake in Alexandria at all.

The Jewish revisers who pulled the Greek back toward the Hebrew have already come into view in the dispute over the young woman of Isaiah 7:14 and in the corrections made to Greek verses that New Testament writers had quoted. The earliest hard evidence for what Greek-speaking Jews did with the Septuagint after they made it comes out of a cave near the Dead Sea rather than out of any later argument about who owned the translation. A Greek scroll of the Minor Prophets recovered at Nahal Hever, copied no later than the first century of the common era and most likely in the second half of the first century BCE, preserves twenty-four fragmentary columns in which the twelve prophets stand in the Hebrew order rather than the order familiar from the great Greek manuscripts, and its language pulls away from the main Septuagint tradition again and again. What the scroll holds is not a fresh translation but a reworking of Greek already in circulation, adjusted so that it follows the Hebrew text then current in Judea more closely.^[1]

Adjustments of the same kind turn up in the Greek of Judges, of Samuel and Kings, and of several smaller books, and the whole pattern is named after one of its own habits, the reviser’s practice of rendering a common Hebrew word with the same fixed two-word Greek phrase, kaige, wherever it appeared, even where ordinary Greek offered something smoother. No ancient writer names a school or a teacher behind the activity, and the evidence comes from the manuscripts alone, which show Jewish scribes across several generations reshaping the earliest Greek translations so that they would line up with the Hebrew tradition.^[2] The changes were slight and sporadic rather than wholesale, which is part of why they went unnoticed for so long, and the same trend runs forward into the second-century versions attributed to Aquila, Symmachus, and Theodotion, all of which emerged inside Greek-speaking Judaism as alternatives to the Septuagint.^[3]

Dating this activity settles a common misunderstanding about why Jewish communities eventually moved away from the Septuagint. Correction of the Greek toward the Hebrew was already under way before Christianity existed as a movement with any claim on the text, so the revising cannot have started as a reaction against Christian use of it. The likelier pressure was internal, since the translation carried from Egypt into Judea was found to differ considerably from the Hebrew text in use there, and tying scripture reading firmly to that Hebrew text meant producing Greek translations that reflected it. Heavy Christian use of the Septuagint did later push Jewish communities away from the translation, though that development came afterward and explains nothing about revision that had already happened.^[4]

Aquila, a convert to Judaism working in the second century, carried the method to its limit, coining new Greek words and accepting word orders no Greek writer would tolerate so that a reader could see the shape of the Hebrew through the Greek. Rabbinic memory of him is warm rather than defensive, and the Jerusalem Talmud reports that he translated the Torah in the presence of Rabbi Eliezer and Rabbi Joshua, who praised him with a line from Psalm 45 about being fairer than the children of men, a compliment that plays on his name in Hebrew and on the name of Japheth, the ancestor of the Greeks. A second tradition places him before Rabbi Akiva, and Jerome later made him a product of Akiva’s school, though that connection rests on a single line in the Jerusalem Talmud and on a writer whose grasp of rabbinic Judaism was thin enough that he took Hillel and Shammai for the founders of the scribes and the Pharisees.^[5]

Greek scripture also kept a formal place in rabbinic law. The Mishnah permits scripture to be written in any language, and Rabban Shimon ben Gamliel narrows that permission to Greek alone, a minority opinion in the Mishnah that the Jerusalem Talmud treats as the settled position and that the Babylonian Talmud endorses through Rabbi Yoḥanan, who anchors it in the blessing of Japheth in Genesis 9.^[6]

Rather than a grudging concession, the ruling singles out Greek as the only language other than Hebrew in which a scroll may be written, and it grounds the choice in a verse of Torah rather than in convenience, which places Greek scripture inside the rabbinic system instead of at its edge.

Three centuries later an emperor still found it necessary to legislate on the matter. Justinian’s Novella 146, a piece of imperial legislation issued at Constantinople in 553 CE, ruled on which languages Jewish communities might use for synagogue readings, put the Septuagint forward as the most accurate text, and allowed Aquila’s translation as well, with the odd aside that Aquila was a Gentile. A ruling of that kind makes sense only if Greek scripture was still being read in synagogues, since an emperor set on drawing Jewish communities toward Christianity would have gained nothing by pressing on them a text they had long since dropped. The insistence that Aquila stood outside the Jewish community concedes, without meaning to, that the Septuagint did not.^[7]

The line comparing the day of the translation to the day the golden calf was made comes from neither the Mishnah nor either Talmud. It belongs to a pair of late minor tractates on scribal practice, Soferim and Sefer Torah, generally printed now as appendixes to the Talmud, and the outlook those tractates express is likely to belong to the seventh or eighth century.^[8] Within Soferim the bleak version sits directly beside a second telling in which God grants wisdom to each translator so that all of them independently arrive at the same Greek, and the pairing has long been recognized as a joining of Palestinian and Babylonian material.^[9] What the rabbinic sources preserve is a divided attitude rather than a rejection, and it took centuries for that attitude to turn negative. Jews commissioned the translation, ratified it, corrected it, and kept reading it into the Byzantine centuries, and rabbinic law made room for it the whole time, so the Septuagint is a Jewish text and the long Christian use of it is one chapter in its later history rather than a change in what it is.