Begin by taking away the object everyone is arguing about. There was, in the century the New Testament was written, no volume called the Septuagint. No bound codex, no agreed table of contents, no edition that could have been handed to St. Paul at Tarsus or consulted by St. Matthew. The name is a convenience invented long afterwards for a library, and treating the library as a book is the mistake that generates the entire dispute.

What existed was this. Sometime in the third century before Christ, in Alexandria, the five books of Moses were translated into Greek for a Jewish community that had largely lost Hebrew. The Letter of Aristeas tells the story of seventy-two elders sent from Jerusalem to do the work, the legend that gave us the number and the name, and whatever one makes of the legend, it describes the translation of the Torah alone. The prophets, the psalms, and the wisdom books were rendered later, by other hands, across the two centuries that followed, with wildly varying skill and philosophy — and the revising of that Greek ran on for two centuries more. The Greek of Ecclesiastes is stiff and mechanical; the Greek of Job is free to the point of paraphrase and shorter than the Hebrew by a sixth. These are not chapters of one book. They are the output of a translation industry that ran for generations.

By the first century that output was already being revised. Jewish scholars, uneasy that the Greek had drifted from the Hebrew as they knew it, produced corrections; the process continued into the second Christian century in the versions associated with Aquila, Symmachus, and Theodotion, and traces of earlier revising work have turned up in manuscripts from the Judaean desert. So when an apostle quoted the scriptures in Greek, he was quoting a Greek text, one form among several already in circulation, some closer to the Hebrew, some further.

And the great codices that people actually mean when they say “the Septuagint” (Vaticanus, Sinaiticus, Alexandrinus) are Christian manuscripts of the fourth and fifth centuries. They post-date the New Testament by three hundred years. They are, in the most literal sense, evidence about what fourth-century Christians were copying, not about what first-century Jews were reading.

Set the canon question aside for a moment: it is a separate question, and confusing the two is how this debate goes wrong. Ask only this: when the New Testament writers quoted the Old, did they quote a text that matches the Hebrew Bible we print today? Sometimes yes. And sometimes, unmistakably, no. You can verify the most important instance yourself, in a Catholic Bible, in under two minutes.

Turn to the prophet Amos, at the end of his book, where he foretells the restoration of the fallen tabernacle of David. The purpose of that restoration, he says, is “that they may possess the remnant of Edom, and all nations, because my name is invoked upon them: saith the Lord that doth these things” (Amos 9:12). A prophecy of territorial recovery: Israel restored will possess her old neighbour and rival.

Now turn to Acts 15, and the Council of Jerusalem. The apostles are settling the gravest question the infant Church had yet faced, whether Gentile converts must be circumcised and keep the Mosaic law. After Peter speaks and Barnabas and Paul report their work, St. James rises to give the decision, and he grounds it in Amos. He quotes the rebuilding of David’s tabernacle exactly as we should expect. Then comes the purpose clause: “That the residue of men may seek after the Lord, and all nations upon whom my name is invoked, saith the Lord, who doth these things” (Acts 15:17).

That is not a loose citation. It is a different sentence. Where the Hebrew tradition has Israel possessing Edom, the text in James’s mouth has the residue of men seeking the Lord. And the mechanism behind the difference is well understood and perfectly innocent: ancient Hebrew was written without vowels, and the bare consonants that spell they shall possess are the same consonants, with one letter’s difference, that spell they shall seek; and the consonants that spell Edom are, unpointed, the consonants that spell man. One tradition read and pointed those letters one way; the Greek translators, working long before Christ and with no Christian conclusion to protect, had read them the other.

Here is why it matters more than any statistic. St. James’s argument does not work on the Hebrew reading. A prophecy that restored Israel will possess Edom gives no warrant whatever for admitting uncircumcised Gentiles into the people of God. A prophecy that the rest of mankind will seek the Lord, and all the nations over whom his name is called gives exactly that warrant, and it is the warrant James uses. The first doctrinal decree in the history of the Church, the decision that made Christianity a faith for the nations instead of a movement inside Judaism, is argued from a Greek Old Testament, against the reading the Hebrew text would later carry.

If the Masoretic Text is the sole authentic Old Testament, then the apostles reasoned from a corrupt one and arrived at the right answer by accident. That is a great deal to swallow in order to keep a tidy theory of the canon.

A Catholic reader who has got this far may feel the argument running downhill in his favour. It does not, and the honest thing is to apply the brakes ourselves rather than wait for an opponent to do it. Four difficulties stand against the case just made, and two of them are serious.

First: the apostles were not committed to the Greek. St. Matthew, recounting the flight into Egypt, writes: “Out of Egypt have I called my son” (Matthew 2:15). He is quoting Osee 11:1, and the Hebrew there has the singular, my son, which is precisely what Matthew needs, because he is applying it to Christ. The Greek tradition renders it in the plural, of Israel’s children, which would be useless to him. So here the evangelist follows the Hebrew against the Greek. A theory that the New Testament simply is the Septuagint cannot survive its own first chapter. The apostles used whichever text-form served the truth they were expounding, and they moved between them without embarrassment, because the notion of a single fixed textual authority is ours, not theirs.

Second: our prettiest example may be circular. Hebrews 10:5 reads the Psalm as “a body thou hast fitted to me” where the Hebrew has the ear-piercing idiom, and it is a marvellous verse for the Incarnation. But the Greek Psalter manuscripts that carry that reading are Christian codices copied after Hebrews was in circulation, and a substantial body of scholarship holds that the Psalter was conformed to Hebrews rather than the reverse — with some arguing the author of Hebrews made the change himself. If so, the verse proves nothing about what the apostles found in their Bibles. We include it because it is famous and because you will meet it; we do not build on it. Amos carries the argument.

Third: the statistics are worthless. You will see it asserted that the New Testament follows the Greek in some three hundred and forty quotations against thirty-three where it follows the Hebrew. The figure circulates everywhere and is almost certainly counting every place where the Greek and the Hebrew agree, which is the vast majority of them, and which proves precisely nothing about preference. The related claim that two-thirds of New Testament citations are Septuagintal has no computation behind it that anyone can produce. Where the Greek and the Hebrew agree, a quotation is evidence of nothing. Only divergences count, and divergences are rare enough to be listed by name, which is why this article names them instead of counting them.

Fourth: the argument from allusion has a price. Catholics rightly point out that the New Testament breathes the atmosphere of the disputed books. The taunt flung at the Cross — “He trusted in God; let him now deliver him if he will have him; for he said: I am the Son of God” (Matthew 27:43) — is regularly produced as an echo of Wisdom 2, and we ought to be exact about it, because half the credit belongs elsewhere. The first two-thirds of that taunt is Psalm 21:9 almost verbatim: “He hoped in the Lord, let him deliver him: let him save him, seeing he delighteth in him”, and the verse before it supplies the wagging heads. What the Psalm does not supply is the closing clause. That is Wisdom’s, where the wicked say of the just man that he “calleth himself the son of God” and resolve to “condemn him to a most shameful death” to see whether God will deliver him (Wisdom 2:13, 18, 20). The mockers begin in the words of the Psalm and finish in the words of Wisdom. The roll of martyrs in Hebrews 11 remembers those who “were racked, not accepting deliverance, that they might find a better resurrection” (Hebrews 11:35), and that is the martyrdoms of 2 Maccabees 6–7 and nothing else: the verb rendered “racked” occurs in the Greek Old Testament only of Eleazar’s torture (2 Maccabees 6:19, 28), and the better resurrection is the hope of the seven sons (2 Maccabees 7). These are real echoes and they tell us what the apostolic Church was reading.

But if literary echo establishes canonicity, St. Jude ends the argument against us. He writes that “Enoch also, the seventh from Adam, prophesied” (Jude 14) and quotes him — using the language of prophecy, which is stronger than anything said of Wisdom or the Maccabees — from a book no Church of the West or the Byzantine East has ever received as scripture, and which is canonical today only in the Ethiopian Tewahedo tradition. Allusion is evidence of reading. It is not evidence of canon. Whoever presses it must take Enoch with it.

For nineteen centuries the argument had a fixed shape. On one side stood the Hebrew, ancient and authentic; on the other the Greek, a translation, and where the two differed the translators were assumed to have slipped. Then in 1947 a boy went after a stray goat near the Dead Sea, and the shape of the argument changed permanently.

The scrolls from the caves at Qumran are Hebrew manuscripts a thousand years older than the oldest complete Hebrew Bible previously known. And among them are manuscripts that agree with the Greek against the received Hebrew, a text of Samuel that repeatedly supports Greek readings long dismissed as translators’ errors, a short Hebrew Jeremias matching the shorter Greek Jeremias against the longer received text. The Deuteronomy scroll bears on the very verse quoted in Hebrews 1:6: it has the summons the received Hebrew lacks.

The implication is precise, and it should be stated precisely, because it is regularly overstated in our favour. It does not mean the Greek scriptures were dominant in Palestine — they were not, and the Hebrew manuscripts at Qumran that align with the Greek are a distinct minority of the collection. What it means is that when the Greek translators produced a reading that later looked like a mistake, they were often faithfully translating a real Hebrew text that afterwards fell out of use. Their integrity is vindicated, not their supremacy.

And that carries a second implication, which is the one that dissolves the original question. In the first century the Hebrew text was not fixed either. Different forms of the Hebrew scriptures were being copied side by side in the same community. The standardisation that produced the text behind every modern Old Testament came later, and the Bible Protestants call original is the outcome of that process, not its starting point. “The Hebrew Bible Jesus used” cannot be produced, because in the sense the phrase intends it did not yet exist.

There is a version of the Catholic case that runs like this: the Jews of Alexandria had a wider canon than the Jews of Palestine; that wider Greek canon included the deuterocanonical books; the apostles used the Greek Bible; therefore the apostolic canon was the wider one, and Luther narrowed it. It is clean, it is memorable, and it is taught in a great many parish classes. It is also, in its decisive premise, not defensible.

The scholarly demolition came from a Protestant. In 1964 Albert Sundberg argued that the “Alexandrian canon” was a modern construction with no ancient evidence behind it, that there is no sign of a distinct, closed, wider canon held by Greek-speaking Jewry, and that in the first century no Jewish community had a closed canon of the writings at all. The category itself was invented to explain a difference that had a simpler explanation: the boundaries were still open, everywhere.

The manuscripts confirm it in the most awkward possible way. If the Greek Bible carried the Catholic canon, the great codices should show it. They do not, and they do not even agree with each other. Vaticanus has no book of Maccabees at all. Sinaiticus has the first and fourth books of Maccabees but not the second or third, and then continues past the New Testament into the Epistle of Barnabas and the Shepherd of Hermas. Alexandrinus carries all four Maccabees and two epistles of Clement, and its own index lists a further appendix, the Psalms of Solomon, on leaves since torn away and lost — its scribes thought the volume’s edges ran wider still. Not one of these manuscripts matches the canon of Trent, and no two of them contain the same books.

Which means the honest Catholic must give up the argument from the codices, and lose nothing worth keeping. The Church has never taught that she found her canon in a Greek manuscript. She teaches that she received the scriptures and discerned them, under the guidance of the Spirit promised to her, and that the discernment took centuries and involved genuine dispute among the saints. That is a stronger position, because it is the one the historical record actually supports.

One further caution, since the claim is common. It is often said that a Roman council of 382 under Pope Damasus fixed the canon. The list associated with that date comes down inside a document of contested date and authorship, and it is not a safe foundation. The African councils of the 390s are firm ground. Stand on those.

Gather it up. The apostles quoted the Jewish scriptures in Greek, and the Greek they quoted sometimes differed from the Hebrew text now printed as the Old Testament: occasionally in ways that carry doctrine, once in a way that carries the Church’s first solemn decision. Those Greek scriptures were older and more textually respectable than the Reformation supposed, and the Dead Sea has vindicated their translators repeatedly. But they were not a book, they did not come with a table of contents, and they never delivered the seven disputed books into apostolic hands by way of a formal citation.

So the objection, pressed to the end, turns out to prove something neither side quite intended. It proves that there was no Bible to inherit. There were scriptures: plural, various, in more than one language and more than one text-form, with edges that had not yet been drawn. Every Christian who has ever opened a Bible has opened one whose contents somebody determined. The Protestant who receives sixty-six books has received a judgment as surely as the Catholic who receives seventy-three; the only difference is which body he trusts to have made it, and when.

And that is the question a manuscript can never settle, for the reason that ought to have been obvious from the start. No book contains its own table of contents. The index is always written from outside. The dispute over the Septuagint is, at bottom, an attempt to find in a library an authority that only a living Church can exercise — and the library, honestly consulted, says so itself.