A history of the Textus Receptus, the printed Greek that the Reformation was translated out of

Set in five months from a handful of late manuscripts, with the last verses of the Bible written by its own editor because a leaf was missing — and given its name, a century afterward, by a line of advertising copy.

A history in 14 parts · about 83 minutes · 1437 – today

Basel · A Gospel book with the printer’s marks still on it

The Red Chalk

In the University Library at Basel there is a small parchment Gospel book, two hundred and forty-eight leaves, about nineteen centimeters by fifteen, twenty lines to the page. It is a perfectly ordinary book. It was written in the twelfth century by somebody whose name nobody knows, in the ordinary Greek of the Byzantine church, and there are several thousand like it. On the last leaf a fifteenth-century hand has recorded what it cost: two Rhenish florins, bought in Basel. What makes it worth crossing a city to look at is the marks on top of the writing.

Down the margins are corrections in the hand of Desiderius Erasmus. Ruled across the Greek, at intervals, are strokes of red chalk. They are casting-off marks: they tell a compositor where one printed page will end and the next begin. There are ink smudges on the parchment of the kind a book gets when it stands for months in a printing house. F. H. A. Scrivener, who examined it in the nineteenth century, said it had been “barbarously scored with red chalk to suit his pages,” and the phrase is fair, though it is his judgment and not a fact. On one opening, at John 20 and 21, you can see Erasmus’s press notes down the right-hand margin and, down the left, a Greek sentence added by Martin Crusius, who borrowed the book in Tübingen and wrote that he had read it through in 1577.

That book is Gregory-Aland minuscule 2, and it is the copy from which the Gospels of the first printed Greek New Testament were set in type. What was printed from it went into Tyndale’s English, and out of Tyndale into the Geneva Bible, and out of Geneva into the King James, and out of the King James into the ear of every English-speaking congregation for four hundred years. The clean rectangle in the dust beside it is not a figure of speech. Erasmus had a second manuscript on that table, older and better and lent to him by a famous man, and he took it away from his own proof-readers because he did not trust it.

The thing that book helped to make is called the Textus Receptus, the Received Text. It is worth saying at the outset what sort of object that is, because a great deal of confusion comes from expecting it to be something else. It is not a manuscript. It is not a text-type. It is not a reconstruction of anything. It is a family of printed books, running from Basel in 1516 through Alcalá, Paris, Geneva and Leiden to about 1633, which were largely copied from one another and which differ from one another in ways nobody has ever fully counted. The name was not given to it by a scholar. It was lifted out of a sentence of advertising copy in the preface of a small Dutch pocket edition, and applied backward, more than a century after the first of these books was printed, to men who never saw the phrase.

This page is about how that happened: who made these books, out of what, in how long, and under what pressure; where the printed Greek of the Reformation actually came from, sentence by sentence, in the places where it can be traced; and which of the confident things everybody says about it survive being checked. A good many do not. The story has attracted, on every side, a body of figures and anecdotes that turn out on inspection to be somebody’s estimate, somebody’s misreading, or somebody’s footnote-free gloss that the next four generations copied. Saying which is which is most of what a page like this can usefully do.

1435–1559 · How the books got to Basel, and who would not give one back

A Dominican Who Went Shopping in Constantinople

The manuscripts Erasmus printed from were not gathered for him, and they were not gathered for a printing press, which had not been invented when they were collected. They were, without exception, part of one library assembled by one man.

John Stojković of Ragusa — Dubrovnik — was a Dominican who went east on the business of the Council of Basel and came back in 1437. Between September 1435 and November 1437 he bought something like fifty-nine Greek codices in Constantinople, in the last years in which a westerner could do that at leisure; the city fell sixteen years after he left it. Two of his purchase notes survive at the backs of the books, in the ordinary way a careful man records what he paid. One reads that it cost three hyperpera. The other, on the Gospel book with the chalk on it, records that this one was bought not in Constantinople at all but in Basel, for two Rhenish florins, most likely off the Byzantine delegation while the council was sitting. He died in 1443 and left the books to the Dominican convent at Basel.

They were still there seventy years later. Around 1511 another Dominican, John Cuno, copied out an inventory of them; it survives at Sélestat, and it is how a modern scholar matches a surviving codex to a fifteenth-century shelf. Cuno’s list has a quiet problem in it that turns out to be evidence: he lists a book that had not been in the house since 1488, which means he was copying an older inventory rather than walking along the shelf. The convent was dissolved in 1525 and the books reached the University Library in 1559, several of them still in the bindings they had come west in.

The book that was missing from the cupboard had gone to Johann Reuchlin, and the way he got it is one of the better letters in the story. In 1488 Reuchlin wrote to the prior of the Basel Charterhouse begging for a small Greek volume in the Dominicans’ library. His argument is entirely about size. The book is written in the tiniest hand, so that it is more fruitful and extremely portable; there are other, better-written copies of the same matter in the same library, not double but many times as many; but this one is so slender and so hand-sized that you would not wrongly call it an Enchiridion, a hand-book. And therefore, he writes, unless I have it, I shall breathe out my life — which you could easily give me back even when I am dead, if only you would agree to let me have on good terms a thing that is not yours. He got it. The Dominican who smoothed the loan through was Jakob Sprenger, provincial of the German Dominicans and the man conventionally named as co-author of the Malleus Maleficarum, who is reported to have said that the house would rather lose the book than Reuchlin’s friendship. The loan ran for life. When Reuchlin died in 1522 the last Dominican librarian wrote in the front of the book that it had been with him more than thirty years and had at last, after his death, been given back to the convent.

That book is minuscule 1, and it matters more than its neighbors because of what happened to it in 1515. Erasmus wrote to Reuchlin in August 1514: they say you have a most correct exemplar, and if you make it available to Johann Froben you will oblige not only me and him but all scholars; the codex will come back to you whole and unspoiled. Reuchlin sent it, and sent a second book with it, and what happened to the pair of them is the difference between a manuscript that has printer’s chalk on it and a manuscript that does not. Nobody wrote on Reuchlin’s books. They came back clean, because a promise had been made about them. The friars’ books were another matter, and one of them came back without its boards, its pastedowns or its flyleaves, its margins trimmed by the binder, and did not get a binding again until 1980.

How many manuscripts, then, and how old? The honest answer has moved recently, and it has moved in an awkward direction. The codicologist Patrick Andrist went through the surviving books quire by quire in 2016 and reached eight codices — five carrying continuous biblical text, three commentaries with the text embedded in them. He also redated three of them against the standard handbooks, and the redatings go the wrong way for the popular account: the manuscript everyone calls twelfth-century he assigns to the thirteenth or fourteenth, and the one everyone calls fifteenth-century to the thirteenth or fourteenth as well. Against that he found one genuinely old thing. The first half of a commentary manuscript of the Pauline Epistles, which the handbooks put in the eleventh or twelfth century, he assigns with Paul Canart to the second half of the tenth. So the flat statement that Erasmus had nothing older than the twelfth century is not quite right; there was tenth-century writing on his table, and it was half of one book, and that book was a commentary. What is true, and is the point, is that not one of his manuscripts was an uncial, none was a papyrus, none was a plain continuous copy of the whole New Testament, and the two he actually set type from were the ordinary work of ordinary provincial scriptoria three or four hundred years before he was born.

There is one more thing in the cupboard worth knowing about, and it is the sharpest irony available. There was an eighth-century uncial Gospel codex in Basel — Codex Basilensis, part of the same Ragusa bequest, four centuries older than anything Erasmus used. In 1943 C. C. Tarelli tried to show that Erasmus had used it. The attempt failed and the field now agrees that neither he nor his team touched it. Andrist offers a reason nobody had thought of: a good many of the Ragusa books were not in the Dominican library in 1515, because they were out on loan, several of them to Reuchlin. The best Greek book in the city may simply not have been on the shelf.

1514–1515 · A printer with a proposition, and an origin story that does not check out

Froben Asks

The whole commercial approach survives as one sentence in somebody else’s letter. Beatus Rhenanus wrote to Erasmus from Basel on 17 April 1515, and between a long discussion of emendations to Seneca and some gossip about a man who had ridden a horse into a hall full of sophists, he passes on a message: Froben asks to have the New Testament from you, for which he promises to give as much as anybody else would. Two weeks later he says it again, more briefly. Froben wants to have the New Testament from you.

Read that first letter to the end and the motive stops needing any speculation. A few lines above the offer, Beatus reports the state of the shop’s stock: of the eighteen hundred copies of the Praise of Folly, only six hundred are left. Twelve hundred copies of an Erasmus book had gone in about three weeks. Johann Froben made his proposal in the week he could watch that happen.

What he was proposing to publish is not quite what the popular account says. Erasmus told the story in print, more than once, that the idea of putting out his own Latin translation of the New Testament had been urged on him at Basel by Froben and his friends, and that he had had no such plan before. P. S. Allen, who edited the letters, demolished that with objects. A copy of Erasmus’s Latin version, written out beside the Vulgate in parallel columns by the scribe Pieter Meghen at John Colet’s direction, survives in three parts, and the colophons carry dates: Matthew and Mark, 8 May 1509, with a corroborating note that Henry VII had died in April of that year; Luke and John, 7 September 1509; the whole of the Epistles, October 1506. A second copy of the entire New Testament, interlined with the Vulgate, is at Corpus Christi College, Oxford. The Latin translation existed in England, in fair copy, six to nine years before the Basel conversation. Allen’s way of handling this is worth copying: he does not call Erasmus a liar. He observes that the statements “admit equally well of the interpretation that the idea of publishing it had not occurred to him until he went to Basle.” That is a real historian declining a cheap shot, and it is also, on the evidence, the likeliest reading.

It matters because it settles what the book was for. The Greek New Testament of 1516 is not, in its own conception, a Greek New Testament. Its title is Novum Instrumentum omne, and the Greek is not named first. Henk Jan de Jonge argued the case in 1984 and it is now the standard view: Erasmus’s aim was a new Latin New Testament, in better and more classical Latin than the Vulgate’s, aimed at theologians as a study book rather than at churches as a lectern Bible, and the Greek was printed alongside so that a reader could check the translator’s work. The Greek column is the evidence. The Latin is the case. Four hundred years of readers have taken the exhibit for the argument.

Erasmus’s own account of how the job grew is one of the better letters he ever wrote about work. Writing to Guillaume Budé in June 1516, with the thing finally off his hands: I had decided to go at it with a light arm, to deal with a few small points and just to point at passages with a fingertip. Then, when the work was already going to press, certain people urged me to change the Vulgate edition by my own correction or translation. I reckoned the addition of that little bundle would be very light indeed; I found it in fact by far the heaviest. Then they pressed me to add rather fuller annotations. And at once, as you know, everything had to be unwoven. The word he uses for unwoven is the one used of Penelope’s web.

There was opposition before a sheet was printed, and it came from inside the church rather than from any reformer. Martin Dorp of Louvain wrote in 1514 urging him not to publish a Greek New Testament at all: nothing is wrong with the Vulgate, and Greek copies are not to be trusted because the Greeks seceded from Rome. Erasmus’s reply is savage and courteous at once. He answers with Aristotle, a pagan whom everybody cites; and with the Psalms and the Prophets, written by a nation that on the same logic seceded from Christ; and he lists the actual points of controversy between the Greek and Latin churches — the procession of the Spirit, the ceremonies of consecration, the power of the Roman pontiff — and observes that not one of them turns on a falsified New Testament text. He opens by saying he would rather have Dorp’s approval than a thousand other men’s, and that of all the flattering letters that reach him daily calling him the sun and moon of Germany, he would die if any had pleased him as much as Dorp’s scolding one.

One last thing about the arrangement, because it is the beginning of the race that shaped the book. The title page of 1516 carries an imperial privilege forbidding anyone to print a Greek Testament within the Holy Roman Empire for four years, and for the second edition Leo X granted Froben a papal privilege as well. Auction catalogs and exhibition captions like to say that this embargoed the rival edition being printed in Spain. It may have contributed; but the best-evidenced account of why that book was late has nothing to do with Erasmus, and is told in its own place below.

1515–1516 · Five months, one gathering a day, and the sentence everybody half-quotes

A Ternion a Day

The popular timeline of the 1516 New Testament has three fixed points, and each of them has a document behind it, though not always the one people think.

Printing had begun by the end of August 1515: Erasmus tells Thomas Wolsey on 30 August that the New Testament is being printed, the Greek as it was written by the apostles, the Latin as translated by me, together with my annotations. Then it stopped. Writing to Pieter Gillis in September: the New Testament had been begun, but was interrupted again for want of a corrector. The correctors arrived on Saint Matthew’s day, and the printer said so in his own hand in a letter that survives in the Basel archives: on Matthew’s day came Oecolampadius from Heidelberg, and also Nicolaus Gerbell from Strasbourg, and they want to come to my help with the New Testament, and they are in my house. Had we had such men from the start — and there Allen’s quotation of the letter breaks off, with the printer in September already wishing he had hired them earlier.

The famous date, 2 October 1515, comes from a letter to Andrea Ammonio, and what it actually reports is a resumption. It is also the best room in the story. All has been pleasant enough for me so far, Erasmus writes, apart from the stink of the stoves, which they have lately begun to light. The Jerome is in progress. They have now set about the New Testament. I can neither stay, because of the intolerable reek of the heating-flues, nor go, because of the work begun, which cannot possibly be finished without me. A printing house in Basel in the first cold week of October, the flues newly lit and unbearable, and the editor unable to leave the city because the book will not finish itself.

The pace is Erasmus’s own, in the same letter to Budé. The work was being edited and printed at the same time; a ternion was struck off every single day, for so they call it nowadays; and even so I was not free to give the whole of my time to this business. A ternion is three sheets folded once: six leaves, twelve pages. Twelve pages a day, of a book neither of them had measured. Immediately afterward comes the sentence that gives the scale of the misjudgment: the size of the volume deceived us; the printer declared it would run to about thirty ternions, and it exceeded eighty-three, unless I am mistaken. Thirty ternions is three hundred and sixty pages. Eighty-three is nine hundred and ninety-six, and the finished folio runs to about a thousand and twenty-seven. Froben underestimated his own book by a factor of nearly three, and Erasmus’s offhand “unless I am mistaken” lands within thirty pages of the truth.

There was a deadline, and it was not the one usually given. Everyone says Froben drove the book to be ready for the Frankfurt spring fair, and it is a reasonable guess, since Reuchlin received his copy through that fair. What is in a document is Erasmus’s own resolve, stated in the same sentence as the ternions: either to die at the work, or to extricate myself from that treadmill before Easter. Easter 1516 fell on 23 March. The colophons run February and 1 March 1516; the dedication is dated 1 February; on 7 March he wrote that the New Testament is published. Budé’s copy reached Paris on 26 April and the book was on sale there a week later. So the defensible statement about the speed is Allen’s, not the popular one: a folio of more than a thousand pages was printed, and partly written, in the space of five to six months. The line about six years’ work in eight months is real Erasmus, and it is about his entire Basel workload — the Jerome, the Seneca, the Institutio principis and the rest — written from Antwerp on 1 June 1516 after he had got out. It is not a statement about how long the New Testament took, and quoting it as one is the commonest slip in the popular accounts.

Now the sentence. In April 1516, writing to Nikolaus Ellenbog, a Benedictine of Ottobeuren whom nobody would otherwise have heard of, Erasmus produced the line that every account of this book quotes: the New Testament was rushed out rather than brought out. What almost nobody quotes is the rest of the sentence, which runs straight on: and yet brought out in such a way that in this kind of thing we have beaten all our predecessors. The famous confession is not a confession. It is a boast with a hedge on the front of it, in a private letter to an obscure monk, and it changes character completely when you read it whole.

He was more honest with scholars. Writing to William Latimer on 5 June 1516, three months after publication: it was published accurately enough, considering how short the time was; but contrary to expectation a good part of my time had to be spent correcting the copy beforehand, and finally correcting the formes — even though two learned men had been hired at great expense for that duty. And then, in the next breath: for I shall soon be preparing a second edition. But take care no one gets wind of this, for the printer’s sake, whose volumes would stay at home if the buyers found out. He asked Budé the same favor, in Greek, in a letter written the same month. Three months after the book appeared he was collecting corrections for its replacement and swearing his friends to secrecy so the first edition would go on selling.

The most honest thing he ever wrote about the book is four words in that letter to Budé. Some things I deliberately passed over; at many I knowingly winked; and soon after publication I disagreed with myself about them. And the reason he had to read the formes himself is in the same place, and explains the state of the printed page: I had thought corrected copies were to be had at Basel. Since that hope failed me, I was forced to pre-correct the codices the printers were to use. The two hired men, a jurist and a theologian who also knew Hebrew, were raw at the work and could not deliver what they had undertaken, so the final reading fell to him. The jurist is Gerbell; the theologian is Oecolampadius, who checked the Hebrew citations, because Erasmus knew no Hebrew.

While they were at it, the two of them started improving his Greek from Reuchlin’s manuscript, and he stopped them. Erasmus tells the story himself, years later, in the course of an argument with Edward Lee: two men were put in charge of correcting the formes, Johannes Oecolampadius, theologian, and Nicolaus Gerbelius, doctor of both laws, and they had Capnio’s copy — Reuchlin’s — elegantly illuminated indeed; but an elegant book is not straight away the same thing as a corrected one. He took it away from them and gave them a plainer one he thought more accurate. His reason for setting aside the better manuscript was not its age, or its text, or its legibility. It was that being handsome is not the same as being right. The joke that history has played on him is that minuscule 1 is the head of a family of manuscripts that modern editors do cite, in every critical apparatus, while the plainer book he preferred is cited by nobody.

1 February 1516 · What the book claimed on its first leaf, and who signed for it

The Oldest and the Most Correct

On the second leaf of the Novum Instrumentum stands a letter to Pope Leo X, dated 1 February 1516, and in the middle of it is the claim on which four centuries of readers took the book at its word. We have revised the whole of what they call the New Testament against the authority of the Greek original, and that not rashly nor with light labor, but taking into counsel many manuscripts in both languages — and not just any manuscripts, but the oldest and the most correct.

Set that beside the list. Eight books, five of them carrying continuous text; the two used as setting copy twelfth-century and thirteenth or fourteenth; the only Apocalypse a commentary manuscript with the scripture buried inside the exposition; no uncial, no papyrus, nothing that any sixteenth-century scholar would have called very old if he had been told the truth about it. It is the single most misleading sentence Erasmus wrote about his own materials. Scrivener saw the gap in 1894 and put it plainly: it is almost painful, he wrote, to be obliged to remember that a portion of ten months at the utmost could have been devoted to his task, and that the working manuscripts were neither ancient nor particularly valuable. Set that against a letter written eight weeks after the dedication, in which the same man says the book was rushed out rather than brought out, and the essay has its spine.

The dedication has a second half that is charming and, in the circumstances, faintly absurd. Erasmus imagines great men sending Solomon marble and ivory and gold and gems for the building of the temple, and adds that we thin little kinglets, or rather manikins, are anxious to send materials of whatever sort, or at any rate goatskins, so as not to contribute nothing. Goatskins for the tabernacle, from the man who had just claimed the oldest and most correct manuscripts in Christendom.

He was dedicating the book into silence. Leo X had in fact written to him, a friendly letter of 10 July 1515 composed by Jacopo Sadoleto, but it went astray for a year: it reached England in August, was held by Ammonio, who was not sure where Erasmus was, was duplicated in February 1516 and sent after Richard Pace, followed Pace into Italy, and reached Basel after Erasmus had left. He first saw these proofs of papal favor in July 1516 at Antwerp, five months after the dedication was written. Compare the anxious postscript in the letter of 2 October 1515: I have had my letter to Leo published along with some others; if any reply has come that matters much to me, let me know by letter, but carefully.

The endorsement, when it came, is the best small piece of theater in the whole book’s history. Printed at the head of the second edition of 1519, and in every Froben edition after it, is a papal brief dated Rome, 10 September 1518, in which Leo says that your labors on the New Testament published some while ago greatly delighted us, and that he forms his expectation of the new edition from that first one, which seemed most perfect. Absolutissima. The pope calls the first edition most perfect; its editor had called it rushed out rather than brought out. Put the two sentences next to each other and stop.

And then read the next letter in the file. On 1 October 1518 Paolo Bombasio wrote to Erasmus from Rome to explain how the brief had been got, and Erasmus published the letter himself, in his own collection, alongside the endorsement it describes. Bombasio drafted the brief. Cardinal Lorenzo Pucci approved his draft. It was transcribed on parchment and sent after the pope, who had left Rome two days before the date it bears. It was entrusted to a feverish young Frenchman named Silvius, who was waylaid on the road and sent the document ahead; the pope read it, signed it, sealed it, called for Silvius, and Silvius could not be found anywhere. Bombasio wrote to the papal secretary complaining that while he sends back countless briefs to the City almost every hour, he keeps Erasmus’s so long. The sealed brief then vanished with an unnamed courier and a duplicate had to be expedited. And then the sentence that gives the game away: and do not be surprised at the date of issue, which I wanted to be Rome rather than anywhere else, as being much more auspicious. Allen’s note on the brief says what follows from that in five words — the dates are fictitious — and corroborates the pope’s absence from a Venetian diarist. The endorsement at the head of the Greek New Testament of 1519 was drafted by the dedicatee’s friend, approved by a cardinal, and postdated to a city the signer was not in, and the man it flattered printed the confession of all three in his own next book.

What Leo called most perfect, Scrivener called the most faulty book he knew, and it is worth quoting him exactly because the sentence is comparative and the comparison is the point. He has just given the typographical errors of the rival Spanish edition, and goes on: Erasmus’s first edition is in that respect the most faulty book I know. He immediately explains why, without malice: Oecolampadius had undertaken that department and was glad enough to serve under such a chief, but Froben’s hot haste gave him little leisure to do his part. He then counts. No fewer than five hundred and one itacisms — the iota-type vowel confusions of Byzantine spelling — were carried over from the manuscripts into the printed text. That is a count of spellings, not of readings, and it is Scrivener’s own, published in 1894; anyone who quotes the number ought to say both those things.

The errata leaf at the back of the 1516 is the longest of any of the four folio editions. And two years later the second printed Greek New Testament in the world, the Aldine Bible of Venice, followed Erasmus’s first edition so closely that it reproduced his press errors — including, Scrivener notes, the ones Oecolampadius had already corrected on that errata leaf. The Textus Receptus is beginning to breed, and on its second generation it is copying mistakes that the first generation had printed the corrections for.

1502–1522 · The book that was printed first and read last

Printed First, Read Last

Two hundred and fifty miles from the nearest press that would have competed with him, a Spanish cardinal had been building the same book for fourteen years.

Francisco Jiménez de Cisneros — Archbishop of Toledo, Primate of Spain, Regent of Castile, and a Franciscan who wore a hair shirt under the vestments — conceived a polyglot Bible in 1502 and founded a university at Alcalá de Henares to make it. The New Testament volume was finished first. Its colophon, at the end of the Apocalypse, is dated 10 January 1514, and it prints the Greek and the Vulgate in facing columns in a Greek type cut in imitation of manuscript hands that is still admired by people who admire nothing else about the book. The Old Testament volumes followed in 1517, the last of them dated 10 July. Cisneros saw the final sheets and died on 8 November of that year. Papal license was not granted until 22 March 1520, and the set did not circulate until 1522, which is when Erasmus first saw it.

So the Complutensian Polyglot was in print two years before its rival and in readers’ hands six years after it. Scrivener’s formulation of that is the cleanest ever written and is worth borrowing whole: Erasmus’s New Testament was by six years the earlier published, though it was printed two years later than the Complutensian.

Why it waited is where the accounts part company. Auction catalogs say Erasmus’s imperial privilege embargoed it. Scrivener, who had actually followed the dates, attributes the delay to the cardinal’s death and the license: this event must have retarded the publication of the whole, since Pope Leo’s license was not granted until March 1520. The privilege may have helped; the documented cause is a dead patron and a slow curia.

The money is the figure everybody quotes and nobody sources. Fifty thousand gold ducats for the project, four thousand of them for manuscripts. Both numbers come from Álvar Gómez de Castro’s life of the cardinal, printed at Alcalá in 1569 — fifty-two years after the last volume, commissioned by the university, begun by a man who had actually worked on the Polyglot and finished by somebody who had not. It is a Toledo humanist’s report, not an account book, and when the internet gives the figure it is quoting him without knowing it. The other number is Scrivener’s, asserted without a source in the passage where he gives it: not more than six hundred copies were printed, which is why the book was scarce and dear from the first. The two figures sit oddly together. A set sold for six and a half ducats. Six hundred sets at six and a half ducats is under four thousand, against an outlay of fifty thousand: the cardinal spent something like twelve times what the edition could ever return, and appears to have understood that perfectly well.

What manuscripts did the Spanish editors use? Nobody knows, and the failure to know is old and well documented. Their own preface claims very ancient and very correct copies sent from the Apostolic library by Leo X. Leo was not elected until 11 March 1513, and the New Testament volume was finished on 10 January 1514, so the claim can hardly cover the New Testament; Vercellone found that only two manuscripts were sent from the Vatican in Leo’s first year, and neither contains any part of the New Testament. The one codex their editor Stunica names by name has disappeared. Delitzsch went looking in 1871 and the question is still open; Metzger and Ehrman recorded it as unresolved in 2005.

Which produces a symmetry the essay cannot improve on. Two Greek New Testaments were made within two years of each other by rival Catholic scholars, and both of them assert on their opening leaves that they were made from the oldest and most correct manuscripts obtainable. In one case the claim is demonstrably an exaggeration; in the other it is unverifiable and, as stated, chronologically impossible. Neither claim was tested for three hundred years, because nobody could test it. Both were believed.

There is one story attached to those vanished Spanish manuscripts that deserves following to the end, because it is the best-documented myth in the whole subject and the chain can be printed entire. The story is that in about 1749 a librarian at Alcalá, wanting shelf room, sold the old vellum manuscripts to a fireworks-maker named Toryo, to be cut up for rockets, and entered the sale in his accounts as useless parchments.

Here is how it travels. In 1784 the Danish professor D. G. Moldenhawer went to Alcalá specifically to find the Complutensian codices, found none, suspected they were being hidden from him, and was eventually told the rocket story. In 1793 it reached print in English, in Herbert Marsh’s translation of Michaelis, with a sentence that ought to be quoted whenever anybody repeats it: this prodigy of barbarism I would not venture to relate, till Professor Tychsen, who accompanied Moldenhawer, had given me fresh assurances of its truth. Dibdin repeated it in 1804 and Horne carried it on Michaelis’s authority. In 1819 John Bowring visited Alcalá and reported in the Monthly Repository that he had inquired for the manuscripts of Ximenes Cisneros, and they had been cut up for sky-rockets, to celebrate the arrival of some worthless grandee. The grandee is Bowring’s own contribution, and it is the version that survives on the internet.

Then, in 1821, Bowring retracted it, in full, in a letter dated from Hackney on 29 March. Having been instrumental in the circulation of a misstatement, he begins, originally but certainly unintentionally made by Michaelis, I beg you will allow me to correct it. His grounds are five, and every one of them is checkable. The oldest catalog of the Alcalá library is dated 1745, complains in its prologue about damage to other and lesser manuscripts, and says nothing about the loss of any scriptural document. The fireworks-maker was named Torija, not Toryo, lived at Alcalá in the middle of the century, and was a man of letters with whom the most eminent of the professors were accustomed to associate. Gómez de Castro says the university had seven Hebrew manuscripts, and seven remain. The library at the date in question was under a careful librarian, and its hundred and sixty manuscripts were all handsomely rebound at that time. And there are no Greek manuscripts of the whole Bible at Alcalá, because Gómez says Leo X lent Cisneros what he needed from the Vatican and got them back.

Tregelles, gathering all of this in 1854, arrived at a verdict subtler than a flat denial, and it is the reason the story is worth telling. A sale to a rocket-maker did take place at the time mentioned, he concludes; but it cannot have been of manuscripts belonging to the library, so that the useless parchments disposed of were almost certainly the old covers of the books, compacted of vellum and folded paper — stripped off at exactly the moment the whole collection was being rebound. And then he adds the twist that makes it a parable rather than an anecdote. The Greek New Testament is in none of the Alcalá manuscripts. Moldenhawer was inquiring at Alcalá for something that library had never possessed; the librarians, wanting to allay a foreign professor’s suspicion that they were concealing something, reached for a real sale of waste parchment and offered it as the answer. Neither the Danish professor nor yet the Spanish librarians, Tregelles writes, seem to have thought of the previous question: were any such manuscripts ever in the library at Alcalá? The story has outlived its own refutation by two centuries because it is a better story, and because the man who first printed it in English said in the same breath that he did not believe it.

1516 · The last leaf of the Bible, and the Greek that was written to replace it

The Verses He Wrote Himself

For the book of Revelation, Erasmus had one manuscript. He says so himself, and the reason he gives is worth keeping: the copy that was then my only one, for that book is rarely to be found among the Greeks.

It is the second book Reuchlin lent him, and it is now at Augsburg. It is a mixed volume: the first eight gatherings are parchment and the rest is cotton paper, with the final leaf made good on linen paper by a later hand. Delitzsch, who found it, dated it to the twelfth century and carried it to Leipzig to show Tischendorf, who confirmed the date on the spot and corrected him on one point. The hand is a minuscule still carrying traces of uncial: frequent scriptio continua, word running into word with no separation; punctuation limited to comma and colon with no distinction from a full stop; no iota subscript, no final sigma; older and younger letterforms mixed; and the vowels constantly confused. Delitzsch’s judgment on the man who wrote it was that he was a badly educated and none too conscientious hired copyist who, especially toward the end, gave an astonishing proof of his thoughtlessness: the leaves of the older book he was copying from had got out of order, and he copied them one after another anyway, without letting the abrupt disconnection stop him.

It is not a Bible. It is Andreas of Caesarea’s commentary on Revelation, with the scripture set in blocks between blocks of exposition. The manuscript does mark which is which — the word for text is written in red in the margin where scripture begins and the word for interpretation where the commentary does, and the first word of each stretch of scripture is picked out in red — and that matters, because the textbook line that Erasmus could not tell the commentary from the text is an inference made by Tregelles in 1844 from a book he had never seen. Delitzsch, holding the object, allows one case in the whole of Revelation where Erasmus mistook the one for the other, and in that case the scribe himself had tangled them.

The front of the book has a picture of the wrong author. On the verso of the first leaf is a bold pen drawing of a bishop enthroned, holding tablets and writing, and a later and less literate hand has squeezed a title in above the opening line attributing the commentary to Hippolytus. The margins further in belong to somebody else again. The glosses are in medieval vernacular Greek and they read Revelation as a prophecy of Islam and the Ottomans: the dragon of chapter twelve is Muhammad; its seven heads are the seven readings of the Qur’an; the beast from the sea is Tamerlane; the number six hundred and sixty-six is worked out as the letters of Mecca. From the rulers named, Delitzsch dated the ownership to the fifteenth century, and concluded that the codex was then in the hands of a Christian groaning under Turkish rule. On one page the note about the number of the beast is clipped along its outer edge, because somebody trimmed the page after it was written. The only Greek Apocalypse behind the printed New Testament of the Reformation is a book that already had the Turks in its margins and a binder’s knife through them.

And it has a hole in it. Correcting a slip that every popular retelling makes: it is not the last leaf. The commentary runs on past the gap for two more leaves. What is gone is one leaf out of the middle of the last half-gathering, carrying the end of the scripture and a stretch of exposition, lost before the codex was bound. The text breaks off in the middle of Revelation 22:16 — I am the root and the offspring of David — and Andreas’s comment begins, and stops four words in. The next surviving leaf picks up in the middle of the epilogue. What the leaf carried away with it was five verses and part of a sixth, about a hundred and thirty Greek words, beginning at “the bright and morning star.”

What he did is not in dispute and never has been, because he said so in print. In the annotations of 1516, in the last note on the book, he wrote: however, at the end of this book I found some words in our copies which were lacking in the Greek copies; we added them nevertheless from the Latin. Delitzsch, who admired him, indicts that one sentence on two counts. Erasmus writes of Greek copies in the plural when only one was accessible to him, as he admits elsewhere; and he presents the deficiency as if it followed from the Greeks’ neglect of the Apocalypse, when it had its ground in an accidental defect in his single manuscript. Both of these, Delitzsch says, are not in accordance with the truth. The sentence was dropped from the second edition onward.

The descriptions shrink as he is pressed. In 1516 it is some words. Answering Edward Lee in 1520 it is one or two lines. In the fuller reply it is very few. And in the same reply, stung, he asks: shameless face, where do I correct Greek that is sound? Is it because I once added three words, which were certainly absent, that I therefore correct the Greek codices? Three words, for a hundred and thirty. Delitzsch, generously, proposes that we take the three words as a figure of speech not to be pressed.

He also produced a diagnosis for the gap, and the diagnosis is the most interesting thing in the episode. Answering Lee, he explains that at the end, through the carelessness of the scribes, these words were missing, and gives them in Latin; and then: we perceived that the scribe had erred on this occasion, because, since “in this book” is written twice, he let his eyes slip down to the second one, leaving out what lies between. For there is no stone on which copyists stumble more often. That is a specific, expert, entirely plausible piece of textual pathology — a scribe’s eye skipping between two identical phrases — offered as the explanation for what was in fact a whole leaf torn out of a book. Delitzsch, who could see the gathering, calls it a veiling of the true state of affairs, excusable only if Erasmus never looked at the manuscript again with that question in mind. And notice what he is quoting when he gives the missing words in Latin. They already contain the phrase “the book of life.” He is reading his Vulgate, and his Vulgate says book.

His fullest defense is the best quotation in the topic, and it is not a defense a modern reader of any party expects. There was no doubt that the words had been omitted, and they were only a few. To avoid leaving a gap in my text, I supplied the Greek out of our Latin version. I did not want to conceal this from the reader. And then: and yet I would not have dared to do in the Gospels, or even in the apostolic Epistles, what I have done here. The language of this book is very simple, and the content has mostly a historical sense, not to mention that the authorship was once uncertain. Finally, this passage is merely the conclusion of the work. He knew exactly how large a liberty he was taking, and he graded the liberty by the standing of the book he was taking it in.

How much of the result is his own invention has been counted twice, two hundred and eighty-one years apart, and the two counts agree. Wettstein, working in 1730 from Erasmus’s own statements and without ever seeing the manuscript, correctly identified the stretch as 22:16 to 21 and wrote that in so short a passage his Greek departs from the Greek of the manuscripts at least thirty times. Jan Krans, collating the 1516 text against the modern critical text in 2011, counts thirty-seven differences, and breaks them down: ten come straight from the Vulgate; one comes from Lorenzo Valla; and twenty-six are simply Erasmus’s Greek failing to be the transmitted Greek. Twelve of those twenty-six involve the definite article, omitted eleven times and wrongly added once, which is exactly the fingerprint of a man composing in a language that has articles while thinking in one that does not. Krans then does the control that makes the number mean something: the six verses immediately before, which are in the manuscript, show nineteen differences, and another six-verse stretch shows nineteen. The seam is visible in the arithmetic.

The fingerprint has a second impression. In the Corsendonk manuscript, which he borrowed for the second edition, Erasmus wrote a note in his own Greek hand recording that he had used the book — and wrote it without the article, exactly as in the printed Revelation. The handwriting sample matches the tic in the type.

Which brings us to Revelation 22:19, the one line of this whole business that a reader of the King James Bible meets without knowing it. The King James threatens the man who takes away from the words of the prophecy with the loss of his part out of the book of life. Every Greek manuscript of Revelation that is not itself copied from a printed edition says the tree of life. The corruption happened in Latin, centuries earlier, and it took two letters: a scribe wrote libro for ligno, book for tree, in a passage where the words for book already stand twice within two verses. The great early Vulgate codices split on it. The Gutenberg Bible says book of life. So do the Jenson Vulgate of 1479, the Sixtine and Clementine Vulgates, both Wycliffe versions, and the first printed German Bible. And the Complutensian Polyglot, in the same opening, prints the tree of life in its Greek column and the book of life in its Latin, so that the fault line runs down the middle of a single page of a book printed two years before Erasmus’s existed. Tyndale has “boke of lyfe” in 1526, ten years after the retranslation and long before the King James men were born.

Three Greek witnesses have been produced over the years for Erasmus’s reading, and all three fail in ways that can be checked in an afternoon. One is a sixteenth-century manuscript copied from a printed edition, probably Erasmus’s own fourth; Josef Schmid struck it from the list of witnesses in 1955 on exactly that ground. One is, as Hoskier himself noted, a copy of Colinaeus’s edition of 1534. And the third has, in its margin, a different word from the one it is cited for. The defense that has circulated most widely on the other side — that there are only eighteen differences in those six verses, so no retranslation can have happened — comes from a self-published book of 2000, and Krans shows how the figure was reached: by comparing the critical text against a later Textus Receptus rather than against Erasmus’s own, which hides eight readings that Erasmus or Stephanus had quietly repaired in the meantime, and by counting several differences as one. One of the eighteen is not even inside the retranslated stretch.

Two of the smaller consequences are worth having, because they are the sort of thing this history is made of. At Revelation 22:11 the King James reads “he that is filthy, let him be filthy still” — six verses before the famous six. The manuscript does not have it; the words are Erasmus’s Greek, made out of the Latin, and Bengel pointed out long ago that one of them is not Greek at all. And at Revelation 17:8, where the King James has the beast that was, and is not, and yet is, the manuscript reads two words that mean “and is present.” Delitzsch, with the page under his eye, saw what had happened: the copyist had written the two words with a slight gap in the middle of the second, and somebody read across the gap and got a different phrase entirely. He lithographed that very line as the frontispiece of his book so his readers could see the gap for themselves. His sentence about it is the one to keep: to this day Erasmus has here deceived New Testament textual criticism, after having perhaps first been deceived himself by his copyist. Luther’s German has it, because Luther translated from the second edition. So does the King James.

And then there is the circle, which is the strangest documented fact in this whole history. When Erasmus sent the revised copy to Basel for the second edition, he wrote to his friends to restore the passage out of the Aldine edition, because, he says, I had not yet bought that work; and they did as I told them. The Aldine Greek Bible of 1518 printed its New Testament from Erasmus’s own 1516. He instructed his friends to correct his invented Greek from a book that was reprinting it. And it went further than a wasted instruction: having corrected one of his own errors to the true reading in the second edition of 1519, he put his invention back in the third edition of 1522, encouraged, Delitzsch says, by the Aldine — which had copied it from him. He kept it in 1527 and 1535, and it is in the Received Text still.

1516–1522 · The most argued-over sentence in the Bible, and a promise nobody made

The Sentence That Was Not There

At 1 John 5:7 the King James Bible reads: For there are three that bear record in heaven, the Father, the Word, and the Holy Ghost: and these three are one. The words from “in heaven” to “in earth” in the following verse are called the Comma Johanneum, and the word comma there is not a punctuation mark. It is the Greek for a clause, a piece cut off.

Nothing else in the New Testament says quite what it says. A number of passages name Father, Son and Spirit together; this is the only sentence in the Bible that says the three are one. Western theologians for centuries cited it as the most explicit statement of the Trinity in scripture, which is the whole reason the fight about it was a fight.

Erasmus’s three Greek manuscripts of 1 John did not have it, and in 1516 he said so, in about forty words, without heat. In the Greek codex I find only this about the threefold testimony, he writes, and gives the short reading; and then two philological remarks, one of them that the Greek does not say “are one” but “are into one.” That second remark is the loaded one, and he never withdrew it: he took the unity in the verse to be a unity of witness rather than of essence, and he was still saying so in 1523, after he had printed the words.

The reply came from Edward Lee, later Archbishop of York, in a book of 1520 in which the Comma is the last of twenty-five complaints. Lee argued in three moves: that a prologue then believed to be Jerome’s says the clause was cut out by unfaithful translators, so the manuscripts prove nothing; that Erasmus had followed a manuscript corrupted by a heretic and had been too idle to check others; and — the move that did the work — that dropping the verse would revive Arianism. Diego López de Zúñiga came at him from Spain with the same prologue and a professional grievance, having been one of the editors of the book Erasmus had beaten into print. It went further than pamphlets. In the summer of 1527 the Spanish Inquisition convened theologians at Valladolid on articles drawn from Erasmus’s writings, this one among them, and the range of opinion recorded there is worth knowing: some delegates wanted Bibles lacking the clause banned from sale, and some proposed that denying its genuineness be made a capital offense, while at the other end one delegate said Erasmus would have been within his rights as an editor to go on excluding it, and another drew the distinction the whole quarrel needed — that what the manuscripts attest and what the church has canonized are two different questions.

He printed it in the third edition of 1522, after a Greek manuscript containing the words was produced in England. The manuscript is Codex Montfortianus, now Trinity College Dublin MS 30, and the first thing to say about it is what it physically is. It is a duodecimo, about six and a half inches by five, on paper, not parchment: a pocket book, and not a fine one. The most argued-over manuscript in the printed New Testament is a small, plain, hand-sized volume that a man could carry in a coat.

The dating evidence is physical rather than editorial. One parent manuscript was copied in England in the late fifteenth century; the watermark in the paper indicates manufacture in the decades around 1500; and the exemplars behind it were only gathered in one place after 1502. The usual range is about 1500 to 1520. And there is a harder fact than any of that. In the Catholic Epistles this manuscript is a copy of another surviving manuscript, and that one does not have the Comma. We possess the book it was copied from, and the sentence is not in it. Its scribe also had Erasmus’s 1516 edition open beside him at some points, adding readings from it in the margins in a different ink.

Whether it was made to order is a live disagreement, and it should not be tidied. De Jonge thought the case as good as certain. Grantley McDonald, who has done more work on this manuscript than anybody, leaned that way in his thesis and pulled back in his book: until it can be dated more precisely, he wrote, it is difficult to know for certain whether the scribe meant to influence Erasmus’s editorial choices — but that a recent Greek manuscript containing the clause, one of only two in the world, should have appeared in the homeland of Erasmus’s critic and been presented at the moment when it might make a difference, can certainly be described as a remarkable coincidence. Others find it implausible that anyone would copy a whole New Testament to win one verse. McDonald’s own argument against the plot is neat: if it were made to order, we should expect it to support Lee’s other criticisms too, and it does not.

And now the promise. Every account of this episode you are likely to have read says that Erasmus rashly promised to insert the clause if a single Greek manuscript could be produced containing it; that such a copy was found, or made to order; and that he kept his word. Metzger printed it in that form in 1968, complete with an Oxford friar named Froy or Roy. Henk Jan de Jonge went looking for the promise in Erasmus’s works in 1980 and reported that it has no foundation in them.

His argument has four legs and they are all checkable. First, the silence of the men who would have used it: Richard Simon gives the Comma an entire chapter in 1689 and does not mention it; nor does Mill’s prolegomena, in the very section discussing the clause’s entry into the third edition; nor Le Clerc, nor Wettstein, nor Michaelis. Second, the earliest telling he could find is from 1818. Third, the retellings cannot agree who received the promise — some say Zúñiga, some say Lee, some say both. And fourth, the chronology forbids the Zúñiga version outright: Erasmus did not address him at all until an apology of late 1521, and in that very work he announces that he has already received the clause from an English codex and has already put it into the forthcoming edition. There was no window in which to make the promise.

What Erasmus actually wrote is two passages, both misread. The first, in his reply to Lee: if a single manuscript had come into my hands in which stood what we read in the Latin, then I would certainly have used it to fill in what was missing in the others; because that did not happen, I have taken the only course permissible, and indicated what was missing from the Greek. That is a past unfulfilled conditional. It is a report of what he did in 1516 and 1519, not an undertaking about 1522. The entire legend rests on reading a Latin pluperfect subjunctive as a future. The second passage, answering the charge of idleness: let Lee produce a Greek manuscript in which is written what my edition lacks, and let him prove that I had access to that manuscript, and then let him accuse me of indolence. The clause at the end is not “and I will print it.” It is “and then you may call me lazy.” And there are two conditions in it rather than one.

De Jonge’s list of the people who passed the story on without ever asking for the passage is the part worth sitting with, because it is not a list of careless men. Tregelles, Hort, Scrivener, Westcott, Kenyon, Nestle, Turner, Darlow and Moule, Bainton, and P. S. Allen, who edited Erasmus’s letters and states the promise as a fact in his introduction to one of them. De Jonge adds, in a footnote about himself, that the writer of these lines cannot plead innocence in this respect. Of the whole company, exactly one — Roland Bainton — ever went looking for the text, and what he found was not a promise.

There is one more link, and it takes the story back another twenty-three years. De Jonge left the trail cold at Thomas Hartwell Horne in 1818, writing that it remains unclear from which source Horne derived his information. But Horne prints the passage inside quotation marks and gives his source in a footnote, and the source is Herbert Marsh’s Letters to Mr Archdeacon Travis, printed at Leipzig in 1795. There, in a preface note, is the earliest telling anyone has yet found. Look at how it is built. Marsh quotes a real sentence of Erasmus in Latin, correctly, with the reference — I had not taken on the job of adding out of my own head what the Greek lacks. And then, in the very next sentence, with no citation at all, he supplies the promise: that in order to give as little offense as possible to the members of the Church of Rome, he promised to insert the passage in his third edition if a Greek manuscript could be discovered which contained it. A genuine quotation, immediately followed by an uncited gloss, so that the reader’s eye carries the authority of the first across into the second. Then Horne quotes Marsh inside quotation marks, and everybody quotes Horne, and the footnote that would have shown there was nothing behind it never existed at any stage.

The last thing to notice about Marsh’s sentence is its temperature. He is defending Erasmus, not attacking him. The clause immediately before the promise calls the omission “which honest conduct,” and the promise is offered as a courtesy extended to Rome by a scholar under pressure. The story was born in 1795 as a compliment. It became an indictment only much later, in a quarrel neither Marsh nor Erasmus lived to see.

What Erasmus said his own reason was, he said twice, in the same four words: so that no one should have a handle for slander. He also said what he thought of the manuscript. He called it recent, and added that he suspected this one too to have been corrected against the manuscripts of the Latins. The word that matters there is “too.” He believed, wrongly, that Greek manuscripts generally had been conformed to the Vulgate by a decision of the Council of Ferrara-Florence, and cited a papal bull for it; in 1534 he admitted he had never read the bull and knew its contents by hearsay. The bull says nothing of the kind. His remark about Montfortianus is his standing theory applied to one more book, not a charge of forgery aimed at it.

As for the Greek evidence, it is small enough to print. Ten manuscripts carry the clause. Only three of the ten match the form printed in the Received Text. Only one was copied before Erasmus was born: a fourteenth-century Latin-Greek diglot whose Greek has been extensively altered toward the Latin beside it, and whose wording differs from the Received Text in eight places in twenty-odd words. Of the rest, several are marginal additions in later hands, and two of those can be dated. In one, a tenth-century manuscript, the note in the margin was written after 1854, because the Bodleian catalog printed in that year records that the book lacks the verse. The oldest manuscript on the list carries the youngest Comma. In another, the marginal note reproduces Erasmus’s third edition exactly, missing definite articles and all, so it was copied after 1522 and before he added those articles in 1527 — and elsewhere in the same book the annotator writes Erasmus’s name. Against those ten stand something like five hundred Greek manuscripts of 1 John that do not have it.

The clause is old, but it is old in Latin. The earliest datable citation of it in full is in the Liber apologeticus of Priscillian of Ávila, who was executed about 385, the first Christian put to death by a Christian state for heresy. His version is not the one in the King James. The earthly witnesses come first; they are water, flesh and blood rather than spirit, water and blood; and the sentence ends “in Christ Jesus.” The Latin witnesses after him keep shuffling the order in the same way, which is the strongest internal argument that the clause grew out of somebody reading the short verse allegorically and writing the allegory in the margin. And the best object in the Latin half of the story is Codex Fuldensis, the oldest surviving Vulgate New Testament, finished under Victor of Capua’s correction on 2 May 546. Its text of 1 John does not contain the Comma. Its prologue to the Catholic Epistles — the prologue Lee and Zúñiga were leaning on — accuses unfaithful translators of leaving it out. One sixth-century book, in which the preface charges the text that follows it with heresy.

1546–1551 · The first apparatus, and the numbers in your Bible

The Printer Who Numbered the Verses

Robert Estienne’s shop stood in the rue Saint-Jean-de-Beauvais, a stone’s throw from the Sorbonne. He printed and sold over his counter the annotated Bibles the Faculty of Theology was condemning, from an address a few minutes’ walk from the men condemning them. His printer’s mark was an olive tree with several branches broken off and the motto NOLI ALTUM SAPERE, which is Romans 11:20 — do not be high-minded, but fear. The wild-olive passage, made into a trademark.

He was the king’s printer, and the quarrel with the Sorbonne ran for years before it broke him. In October 1547 the Faculty formally condemned his Bibles, on a charge that is worth reading closely: they contain, it says, several things added beyond the true reading which has hitherto been received in the Church, which it is not lawful for any private person to do, nor to publish. The vocabulary of a received reading is in the air in Paris a lifetime before a Dutch printer put it on a title page. Francis I backed his printer and approved his suggested compromise — that the Faculty draw up a list of its censures and Estienne print the list inside the Bibles themselves, so readers could beware of the suspect passages. The Faculty never produced the list. Estienne spent years demanding it, and the book he eventually wrote from safety is built on the grievance; his motto in it is that the readers should be judges by the sight of the things themselves. In November 1548 the new king approved the condemnation and pointedly did not require the list. On 13 November 1550 Estienne applied for permission to live in Geneva.

It is worth knowing what he was leaving. The Chambre ardente, the special heresy court, had been set up in October 1547 and issued something like sixty death sentences in two years. Étienne Dolet, a printer, had been burned on the place Maubert in 1546. The Sorbonne was lobbying to have Estienne’s case sent to that court. From Geneva he called them a Pharisaic synagogue, leeches on people who must depend on them for salvation, and devilish enchanters who cloud the understanding of the faithful; and what he says wounded him most is not the stake but the damage to his trade, since scarcely one man in ten, he writes, did not form an odious judgment of him.

The book he had made in the middle of all this is the one that mattered. The Editio Regia of 1550 is a folio in the grecs du roi, the Greek type Claude Garamond cut on a royal contract, its letterforms following the hand of the king’s Cretan calligrapher; the largest of the three sizes was finished in the year of the folio. It has been called, by people who disliked everything else about it, the most beautiful Greek book ever printed. Bound in with the New Testament are Chrysostom’s first homily on Matthew, printed there for the first time anywhere, the Eusebian canons, an index of Old Testament passages quoted in the New, and seventy-two lines of Greek hexameter signed by Henri Estienne, the printer’s son, to every God-loving reader.

What makes it a landmark is the inner margin. Down it runs a set of variant readings, each tagged with a Greek numeral naming the copy it came from: the first is the Complutensian, the second a most ancient copy collated in Italy by friends, several more are from the king’s library, and the rest, in Estienne’s own phrase, are whatever could be scraped together from anywhere. This is the first printed Greek New Testament to carry an apparatus of manuscript variants, and the qualification matters — Estienne had done the same thing twenty-two years earlier for the Latin Vulgate, collating manuscripts from Saint-Denis and Saint-Germain. Two of the fifteen manuscripts he cites have never been identified, from that day to this. The identifications of several others are still being argued about.

And then people checked. Estienne says in the preface that he collated the manuscripts himself, as carefully as possible. The work was in fact done by his son Henri, who was eighteen. Michaelis’s verdict on Henri is that he was too young, too impatient, and too little experienced in criticism for an undertaking of that nature, and the numbers bear him out. Scrivener audited the citations of the second manuscript, the one collated in Italy, against the actual book, and reported: three hundred and eighty-nine citations in all, of which three hundred and nine are quite correct, forty-seven a little loose, eight give a corrector’s reading as the first hand’s, and twenty-five are left over — nearly all of them readings that belong to a different manuscript in the same margin, so that the wrong numeral was set. That is about four fifths right. The other errors are the kind that come from a compositor reaching into the wrong box of type, and Michaelis says exactly that. One manuscript that contains only the Gospels is cited five times in Acts; Mill, trusting the margin, concluded the book must therefore contain Acts, and it does not. Another that has no Apocalypse is cited in the Apocalypse.

Scrivener then made the comparison that settles the question of how good the collating was. The Complutensian was a printed book, in clear type, widely circulated, and Estienne’s most valued source; and setting aside errata and spellings, it differs from his text in more than two thousand three hundred places, of which it is cited correctly five hundred and fifty-four times and falsely fifty-six. If a printed book in clear type was handled that way, what were the manuscripts like? Tregelles had made the same argument in 1854 in one sentence: it would be remarkable if manuscripts were examined with greater accuracy than a printed book.

The hardest single fact about the 1550 is this. Its text stands against all of its own cited authorities — every manuscript and the Complutensian together — in a hundred and nineteen places. Eighty-one of them are in the Gospels. In the Apocalypse, where the apparatus rests on three authorities only and one of them breaks off partway through chapter seventeen, their united reading is rejected fifty-four times. Tregelles’s summary is unkind and difficult to answer: the various readings in the margin seem rather to be appended as an ornament to the text than as giving it any real utility. Scrivener, gentler, gets to the same place: the text is perpetually at variance with the majority of them.

One quiet thing came out of the collation and did not get printed. Beza afterward published, out of Henri’s working papers, forty readings from one manuscript that Estienne had left out — and among them was the whole of the Shorter Ending of Mark. One of the most consequential variants in the New Testament sat in a teenager’s notebook while the folio went to press without it.

The following year Estienne printed a small two-volume edition at Geneva, without naming the city on the title page, with Erasmus’s Latin down one side, the Greek in the middle and the Vulgate down the other. Its text is the 1550’s, near enough; it carries no apparatus; some copies misprint their own year. What it carries instead are the verse numbers. They are the ones in your Bible.

They were not made for readers. Estienne was preparing a Greek concordance and needed a finer division than the chapter to refer to; his son finally published the concordance in 1594. The preface gives a second and purely typographic reason, which is that with three columns of text to align, numbered verses let a reader see at a glance which words of each Latin version answer to which words of the Greek. The idea was not new either: Estienne’s own father had numbered verses in a Psalter in 1509, and Sante Pagnini had numbered a whole Latin Bible at Lyon in 1527. What Estienne invented was the particular cut, markedly shorter than Pagnini’s, and it is his cut that survived. As for the story that he made the divisions on horseback and that this is why they fall in silly places — the phrase is his son’s, written forty-three years later, and it will not carry the weight put on it. The Geneva Bible’s own history takes that story apart at length, and this page defers to it.

1565–1598 · The editor the King James men followed, and the two great books he gave away

The Two Codices He Gave Away

Theodore Beza inherited Geneva when Calvin died in 1564, and published his last folio New Testament at seventy-nine. The Greek text the King James translators followed more often than any other is his. It was not what he thought he was doing.

His project was the Latin. The folios print three columns — his Greek, his own new Latin translation, and Jerome’s Vulgate — so that a reader watches the Vulgate being corrected in the next column across; and the Greek is there, as Jan Krans puts it, to show the sound basis of the new translation in the original. He reviewed the Latin for every edition. He never described how he edited the Greek at all. His base was Estienne’s text, and he changed it only occasionally: the fullest count anybody has made lists a hundred and eighteen differences between Estienne and his second edition, of which twenty-seven concern only accents and punctuation, ten are corrections of Estienne’s errors, eight are new errors of his own printers, and seventy-one are real changes. The changes cluster in Acts and Revelation, and in Revelation he is generally moving away from Erasmus toward the better Spanish text.

He said he had manuscripts, and the number he gives will not sit still. Twenty-five in some prefaces, seventeen in others, nineteen in the last. Wettstein thought the twenty-five plain hyperbole, through which the editor makes his Greek text seem to rest on a firmer base than it actually did, and listed seven separate charges against his handling of Estienne’s material, of which the mildest is that he presents readings as if he had consulted the manuscripts himself. He called the whole performance most disgraceful.

The apparatus was thrown away, and the printer explained why in print. A note in the 1556 edition says that Beza at first cited the manuscripts by number — first, second, and so on — but later on refrained from naming them thus because of the tedium that it involved for him together with the fact that it is of no use to you, and limited himself to giving their number. Krans can locate the moment it happens: the precise enumerations survive through Luke 22 and 23, and from John onward the references stop being exact. The apparatus of the text that would carry the Reformation into English was abandoned partway through because the editor got bored somewhere around the end of Luke. What follows from that is Krans’s and it is severe: as far as textual criticism is concerned, the edition can no longer be called critical; it becomes a commentary which can at best be used as a companion volume to Estienne’s.

And then there are the two books. Beza owned Codex Bezae, a Greek-Latin copy of the Gospels and Acts from about the year 400, and Codex Claromontanus, a Greek-Latin Paul from about 550. Between them they are among the oldest New Testament manuscripts in Europe, and he had them on his desk. He gave the first to the University of Cambridge in 1581 and the second went to Paris. What he did with them in between is very little.

The structural joke sits underneath the whole arrangement. Estienne’s second manuscript, the one collated in Italy by friends and cited three hundred and eighty-nine times in the 1550 margin, is Codex Bezae. Beza had been reading its readings in print for a decade before he owned it, and he never worked out that they were the same book.

The letter he sent to Cambridge with the gift is the center of the story, and it is not the letter of a man handing over a treasure. He tells them the manuscript is somewhat incomplete, and not very correctly copied from beginning to end, nor kept with sufficient care; that in Luke especially he found such a great discrepancy between this manuscript and any others however old that he would think it better to store it than to publish it, in order not to raise offense; and that he can see inserted pages in different characters and strange remarks added occasionally by some ignorant elder Greek monk. He hands over the oldest thing he owns with a note saying, in effect, you had better lock this away.

The readings of Codex Bezae that got into his Greek text can be listed on one hand — six places from the 1582 edition, and one more that reached the Greek only in 1598. He gave his reason himself, at Acts 18:27: at this place my very old manuscript has many things which I would not want to be put in the text, as they do not occur anywhere else as far as I know, but I wanted to insert them here. Uniqueness was a disqualification. It is the exact reverse of the instinct that has governed the discipline since Lachmann, and Beza states it without embarrassment because in 1582 there was nothing embarrassing about it. He thought Claromontanus the more accurate of the two, and said so in a preface, and used that one even less.

He also made conjectures — a great many of them. The paradox is Krans’s: though he professes time and again not to change the text lightly or out of mere conjecture, he offers at the same time an astonishingly high number of conjectures. Nobody has published a total, including the man who wrote the book about it. Almost all of them stayed in the annotations. Three did not, and two of those are in your Bible.

The famous one is Revelation 16:5. The angel of the waters says that God is the one who is and who was, and then, in every Greek manuscript that has ever been collated, the Holy One. Beza printed instead a participle meaning “the one who shall be,” and the King James reads: Thou art righteous, O Lord, which art, and wast, and shalt be, because thou hast judged thus. His reasons are given at length and they are not stupid. The article in the transmitted phrase, he says, is contrary to all rules of language and shows that the reading is corrupt; and John uses a three-part formula for God everywhere else in the book — past, present, future — so a third member is demanded here. He chose “shall be” rather than John’s usual “is coming” because in this vision God is already seated in judgment.

He also said where it came from. And so I cannot doubt, he wrote, that the genuine reading is the one I have restored from an old manuscript codex of good faith. No such manuscript has ever been found. For a long time that was where the matter rested, and Jan Krans, writing the standard book on Beza’s conjectures in 2006, left Revelation 16:5 out of it precisely because Beza said he had a manuscript and Krans believed him. In 2024 he changed his mind, and the reason is an object rather than an argument. Beza’s own working copy of his 1565 edition survives in Geneva, carrying the notes he made preparing the third edition. At Revelation 16:5 he has underlined the transmitted reading and written his own in the margin, and named no manuscript. Twelve lines up the same page, at Revelation 1:4, he wrote a marginal note that does name one: in an old codex it reads thus. Krans’s reconstruction — which he labels a kind of conjecture itself — is that Beza jotted the word down early, not from a manuscript and perhaps not even as a conjecture but as the reading he would have expected; and that years later, preparing the 1582 edition, he could not remember where it had come from and, following the example twelve lines above it, assumed it was the same old and reliable codex. The strongest defenders of the Received Text on this verse do not dispute the Latin; they dispute what follows from it, and argue that a man should be taken at his word about his own sources. Neither of Beza’s own two great codices could have supplied it. Neither of them contains Revelation.

The second conjecture in your Bible is quieter and better. At Romans 7:6 Beza changed one letter of one word, on the strength of what he took to be Chrysostom’s reading — he had in fact found the word in Chrysostom’s commentary rather than in the biblical text quoted at its head, and defended the mistake by supposing that later copyists had corrected Chrysostom toward the common reading. The King James reads “that being dead wherein we were held,” and Krans says outright that it depends on Beza’s Greek here. The reading is absent from the surviving Greek manuscripts.

What makes Beza worth the space is not the conjectures but what he said about conjecture while making them. Early on, at Revelation 18:14, he had written that it was a matter of scrupulousness with him not to change anything out of mere conjecture. He re-edited that sentence in 1589 to say that it had always been a matter of the utmost scrupulousness not to change even a tittle in these holy books out of mere conjecture. In 1598, at Hebrews 12:6, he wrote that may the crime be far from me to change even so small a thing in these sacred records against all manuscripts and interpreters. He was introducing fresh conjectures in every one of those editions. Krans’s formulation of the predicament is the fairest thing anybody has written about it, and it applies to more men in this history than Beza: in order to be perfect, the text of Scripture stands in need of emendation; whether that paradox is a matter of irony or tragedy will depend on the beholder.

1624–1633 · Where the name came from, and what it was selling

A Sentence on the Back of a Book

In 1620 Isaac Elzevier was appointed sworn printer to the University of Leiden, and the following year, for eighteen florins a year, he obtained leave to build in the courtyard of the university itself, between the entrance gate and his father’s house, a lodge to hold his workshops and his stock. He put six presses in it, at a time when the three largest printing houses in the town had four each. His device was an elm with a vine climbing it, heavy with grapes, and a hermit standing by, under the motto Non solus.

Out of that shed in 1624 came a Greek New Testament thirteen centimeters tall, and it is the beginning of the phrase this whole page is named for — though not in the way it is usually told, and almost every detail of the standard account is wrong.

It was not printed by Bonaventure and Abraham Elzevier, who are named for it everywhere. It was printed by Isaac, who did not sell them the press until a notarial act of 24 December 1625, in which he made over the type, five presses and one intaglio press, the punches, matrices and cases, for nine thousand florins, and sold them the premises the next day for two thousand more. Nor were Bonaventure and Abraham brothers. Bonaventure was Abraham’s uncle. And Isaac left the trade fourteen months after printing the book, telling the university curators on 8 February 1626 that by reason of the wars and troubles in Germany the profession of printer had fallen so low that he saw himself constrained to give it up. He went to Rotterdam, joined the navy, made captain in 1632, and ended as a partner in a brewery at Delft.

The 1624 has no preface at all. This is checkable and it is decisive, and it is confirmed three separate ways: the title page is followed straight by an index of Old Testament testimonies and then by Matthew. The famous sentence is in the second edition, of 1633: Textum ergo habes, nunc ab omnibus receptum: in quo nihil immutatum aut corruptum damus. You have therefore the text now received by all, in which we give nothing altered or corrupted.

Notice what the name kept and what it threw away. The Latin boast is that the text is received by everybody, and now. The label that survives keeps the participle and drops the two words that made it a claim about the market. Notice as well that in the sentence both words are accusatives, the object of “you have,” so that somebody had to put them back into the nominative to make a title out of them. The name of the Greek text behind the Reformation is a fragment of a sales sentence, re-declined.

And the evidence the preface offers, inside the advertisement, for the text being received by all is that the first printing had sold out. The copies being dispersed and sold off, it says, we submitted the edition, acceptable to all, once more to the eyes of the learned. The rest of the preface is the work of somebody with a good Latin education — it contains a joke about Pandora, whose name means all-gifts, and a comparison of the edition to the tabernacle built out of freewill offerings in Exodus. It is unsigned. The 1633 title page names nobody but the shop. The attribution of the preface to Daniel Heinsius, and of the Greek text to Jeremias Hoelzlin, rests on two specialist studies by Henk Jan de Jonge in 1971 and 1978, and before those the honest answer was that nobody knew, and three of the best men in the field said so in print. Griesbach called the text patched together out of Estienne and Beza by an unknown man. Tregelles: the editor, if any, is wholly unknown. Scrivener: of the person entrusted with its superintendence we know nothing. Reuss went further and said the name was not worth chasing, since by no unjust fate the man had gone without his praise among posterity.

The important thing about the blurb is that it was not unusual. Eduard Reuss, cataloguing the printed Greek New Testament family by family in 1872, records an edition printed at Hanau in 1629 which claims on its title page to be faithfully printed from Estienne and the other best editions, and a Hamburg reissue of the very same sheets which claims instead to be taken chiefly from Erasmus’s edition. Both are the Elzevir text unchanged. Reuss’s verdict on the pair is four words long: an impudent lie, both of them, which was the custom of the booksellers of that age. The 1633 sentence is not an outlier. It is house style. Reuss also noticed that the Elzevirs sold the same book into Catholic countries under a title page giving the place of printing in a form that could be taken for Lyon, with the red-and-black title reserved for those copies.

Then the phrase went travelling, and it travelled backward. The earliest technical use I can date is Griesbach’s, in the prolegomena to his edition of 1796, where the received text stands as a settled term of art needing no gloss and is to be printed word for word and syllable for syllable in one typeface with every departure religiously marked. In English, Herbert Marsh in 1802 still has to quote the Latin rather than use the name. By 1818 Thomas Hartwell Horne teaches the term on one page of his Introduction and uses it flat on another, which is a word crossing over into ordinary use and caught mid-crossing. By 1854 Tregelles has to explain to his readers where the phrase came from, which means that by then it was a name whose origin had been forgotten. And both Marsh and Horne attach it to the wrong book — to the 1624, which made no such claim, because it has no preface to make one in. From there it went backward again over Beza, over Estienne and finally over Erasmus, none of whom ever saw the words.

The last thing to say about the Elzevirs is what Reuss said when he opened his chapter on them. He had by then cataloged a hundred and eighty-eight printed editions that in no way depend on the 1624, forty of them printed after it; and he observed that those editions differ from one another so greatly that scarcely two or three of them agree. His verdict on the whole afterlife of the phrase is worth having in full, because it was written in 1872 and has not dated: how anyone today can hold the Leiden edition of 1624 as the norm of a received and common text never varied thereafter, I frankly cannot grasp; nor does it seem explicable except by an ignorance of literary history.

How far apart are the two books everyone means by the Received Text? Estienne’s 1550 and the Elzevir text have been collated four times, and the number goes up every time somebody looks harder. Mill in 1707 observed twelve differences. Tischendorf in 1841 gave a hundred and fifteen, most of which, he notes, had escaped Mill’s eyes. By his seventh edition he listed a hundred and fifty. For the first edition of his own handbook Scrivener drew up a list of two hundred and eighty-seven. Nobody has ever said the earlier counts were wrong. They simply had not looked.

1707–1883 · Thirty thousand variants, and the man who said they were harmless

The Scarecrow of Thirty Thousand

John Mill spent thirty years on an edition of the Greek New Testament and lived a fortnight after it appeared. It came out at Oxford on 9 June 1707 and he died on 23 June. What he printed was Estienne’s 1550, with its errata corrected and no other intentional change; what he added was an apparatus of every variant reading he and his correspondents could gather, printed under a text he did not defend. Bishop Fell had paid for the printing and died when fifteen sheets were done, whereupon Mill refunded everything Fell had laid out and carried on alone. Isaac Newton helped with the collating. The reissue of 1710 did not sell, and the unsold stock was put out again with a new title page in 1723, and again in 1746.

Mill never gave the famous number. Richard Bentley, writing in 1713 as a man who had known him, says the readings amount in the whole, by a late author’s computation, to above thirty thousand — and attributes the arithmetic to somebody else. Which somebody is itself disputed, between a reprint of 1710 and an edition of 1711, and both are after Mill is dead. His name is on the figure because his apparatus was what got counted.

It frightened the devout first. Daniel Whitby published an examination of Mill’s variants arguing that the readings were unfit to overturn the received text and that Mill had repeatedly contradicted himself, and gave his motive plainly: I have found much in Mill’s Prolegomena which seems quite plainly to render the standard of faith insecure, or at best to give others too good a handle for doubting. Six years later Anthony Collins took the handle and used it against revelation. Tregelles saw the sequence for what it was: if Mill had not been blamed for his endeavors to state existing facts, this line of argument could not have been put into Collins’s hands. The alarm made the weapon.

Bentley’s anonymous reply of 1713 is short, and every one of its moves has been made again since by somebody who had not read it. Dr Mill did not make and coin them, he only exhibited them to our view; if religion therefore was true before, though such various readings were in being, it will be as true and consequently as safe still, though everybody sees them. More copies will always mean more readings. And if there had been but one manuscript at the restoration of learning, we had had no various readings at all — and would the text be in a better condition then? So far from that, that in the best single copy extant we should have had hundreds of faults, and some omissions irreparable. He had collated manuscripts of Terence, an author not near so big as the whole New Testament, and had seen twenty thousand various readings in him; scripture’s tally looks worse than any classical author’s only because its variations have been noted with a religious, not to say superstitious exactness. So: when you hear more of this scarecrow of thirty thousand, be neither astonished at the sum, nor in any pain for the text.

And then, in the same pamphlet, the paragraph that is the reason Bentley belongs at the center of this page rather than at the end of it. The present text, he writes, was first settled almost two hundred years ago out of several manuscripts by Robert Stephens, a printer and bookseller at Paris, whose beautiful and generally speaking accurate edition has been ever since counted the standard; and if the conceit is but spread and propagated, within a few years that printer’s infallibility will be as zealously maintained as an evangelist’s or apostle’s. What Mill would say, he goes on, is that this text fixed by a printer is sometimes by the various readings rendered uncertain, and is proved certainly wrong; but that the real text of the sacred writers does not now, since the originals have been so long lost, lie in any single manuscript or edition, but is dispersed in them all.

That was 1713, and the discipline spent the next century and a half catching up with it. The thirty thousand kept rising, every figure in the sequence an estimate rather than a count, until Peter Gurry produced the first one with a stated method in 2016 and arrived at about half a million. His gloss on it is the part to keep: Mill had fewer than a hundred Greek manuscripts and the register now lists more than five thousand six hundred, so the number reflects the frequency with which scribes copied more than their infidelity in doing so.

For three hundred years every printed Greek New Testament had a printed Greek New Testament underneath it. Editors corrected, annotated and occasionally departed from the received text, but they set their type from it. In 1831 a Berlin professor of classical philology named Karl Lachmann — not a theologian — printed a small Greek Testament built from documents, with, in his own phrase, the so-called Textus Receptus being allowed no place.

Two things about that are routinely got wrong. He did not claim to have recovered the original text, and said so: his object was the oldest attainable text, namely that of the fourth century, as a basis for further inquiries which might lead nearer to the primitive text. And the word first needs its qualifiers. Edward Harwood had built his Gospels and Acts on Codex Bezae in 1776, and Reuss, testing a thousand passages, found him deserting the received text in more than seventy percent of them and agreeing with Lachmann in six hundred and forty-three. The formulation that survives scrutiny is Tregelles’s, qualifiers intact: the first Greek Testament since the invention of printing edited wholly on ancient authority, irrespective of modern traditions. Lachmann’s evidence was also very thin — through almost a quarter of the New Testament he had at most two manuscripts, sometimes one, so that an error in a single copy is promoted to a widely spread reading of the fourth century.

What followed came fast. Tregelles went to Rome in 1845 and spent five months failing to get permission to read Codex Vaticanus properly. Tischendorf took forty-three leaves of Codex Sinaiticus from Saint Catherine’s in 1844 and the bulk of it in 1859, under circumstances the monastery still disputes; the story that the leaves were in a basket destined for the fire is his own, published twenty-two years afterward in a popular book, and his own telling gets its leaf count wrong. Then in 1881 Westcott and Hort published a text and a theory, the Revised Version appeared built on the same principles, and John William Burgon wrote three long articles calling the Revision the most astonishing as well as the most calamitous literary blunder of the age, and the theory behind it as flimsy and as worthless as any spider’s web.

Burgon is the most misreported figure in this history, and the correction is in his own footnote on page twenty-one. Once for all, he writes, we request it may be clearly understood that we do not, by any means, claim perfection for the Received Text. We entertain no extravagant notions on this subject. Again and again we shall have occasion to point out that the Textus Receptus needs correction. What he insists on is that it is better than Lachmann’s or Tischendorf’s and that improving it will take entirely different principles from those in fashion. He was not a King-James-only advocate and was not opposed to revision; he wanted a different one. His own term is not the Received Text but the Traditional Text, which he distinguishes from it and dates to the year 375 — so he and Hort are both reaching for the fourth century and disagreeing about which fourth-century text is older. Two smaller corrections run in his favor. The sentence most often quoted as his — that the worst corruptions the New Testament ever suffered originated within a hundred years of its composition — is Scrivener’s, and Burgon is quoting him. And Burgon counted: testing Codex Bezae against his Traditional Text through the surviving Gospels he tallies more than thirteen thousand deflections, and he says how he arrived at the figure, which a good deal of the rhetoric against him never did.

1881 · A Greek text made backward out of an English one

Which Textus Receptus?

The Greek New Testament sold today under the name Textus Receptus, printed by the Trinitarian Bible Society and used by most people who defend the Received Text, was made in 1880 by a man working backward from an English translation, and it had never existed before.

It exists because a margin was too narrow. Convocation had instructed the Revisers of 1881 that wherever their Greek differed from the Greek behind the Authorized Version, the alteration should be indicated in the margin. They found that obeying the instruction literally would crowd and obscure the margin of every page, so they handed the two university presses a corrected list of their departures and asked them to publish it independently in some shape or other. Oxford printed the Revisers’ Greek with the old readings at the foot. Cambridge, through F. H. A. Scrivener, printed the mirror image: the King James’s presumed Greek in the text, with the Revisers’ readings underneath. It is the Cambridge arrangement that a later century could lift out and use on its own, and nobody in 1881 intended that.

Scrivener says in his own preface what he is doing, and the sentence is the load-bearing one for everything that has been built on his book since. The Authorised Version, he writes, was not a translation of any one Greek text then in existence, and no Greek text intended to reproduce in any way the original of the Authorised Version has ever been printed. The man making the text says plainly that he is making something new.

His method was not to translate English into Greek. He took Beza’s last edition of 1598 as his base, because it was more likely than any other to have been in the translators’ hands and agrees more closely with their English than any other printed text; and then, wherever their rendering matched some other Greek reading that could naturally have reached them through a printed edition, he displaced Beza’s reading in favor of the more truly representative one and marked the substitution with an asterisk. He set himself a limit and stated it: it was manifestly necessary to accept only Greek authority, though in some places the version corresponds but loosely with any form of the Greek original while exactly following the Latin Vulgate. Which means that even Scrivener’s text does not represent the Greek behind the Authorized Version at those points, because he would not invent Greek to make it. It is the closest anybody has printed, and he says so, and it is not the thing itself.

He also tidied, and said so: manifest errors of the press, which often occur in Beza’s New Testament of 1598, have been silently corrected. He standardized accents and breathings where his base wavered, and chose one spelling and kept it. The jots and tittles of the Received Text as it is sold today are, in those places, the editorial housekeeping of an English clergyman in 1880. And the edition the Trinitarian Bible Society prints is his re-typeset without his notes, so the apparatus in which he recorded exactly where the English departed from Beza has been dropped from the book that carries his text.

How many departures? About a hundred and ninety, Scrivener says, and the word is his: all variations from Beza’s text of 1598, in number about 190, are set down in an appendix. He gives no exact figure for his own work, and the appendix is the only place they are itemized. Three published counts circulate today — a hundred and ninety, a hundred and ninety-one, and a hundred and twenty-six — which is a fair sign that nobody has re-tallied the appendix in a century and a half.

The more interesting number is what the King James translators actually did, and Scrivener published that twice, with different results, and the version in circulation matches neither. In 1873 he examined a hundred and thirty-one places where the printed editions differ enough to affect the English, and reported that the translators go with Beza against Estienne in eighty-one, with Estienne against Beza in twenty-one, and with the Complutensian or Erasmus or the Vulgate against both in twenty-nine. In 1884, having enlarged the list, he examined two hundred and fifty-two places and reported a hundred and thirteen, fifty-nine and eighty. Both sets sum correctly. The triplet that circulates on the internet and in a good deal of the secondary literature — a hundred and eleven, fifty-nine, sixty-seven — matches neither of his published tallies; the fifty-nine is from 1884, the hundred and eleven is near the 1884 figure, and the sixty-seven corresponds to nothing at all. Somebody transcribed the 1884 numbers imperfectly, attached the 1873 citation, and it has been copied ever since. There is a parable in that which does not need drawing out: an argument about how faithfully a text is transmitted, whose central statistic has been corrupted in transmission.

Take the 1884 figures at face value and they say something plain. In eighty places out of two hundred and fifty-two, roughly a third, the King James follows neither Estienne nor Beza but the Complutensian, or Erasmus, or the Latin Vulgate. The translators of 1611 were not following the Received Text, because there was no such thing to follow. They had four or five printed Greek Testaments on the table and they chose between them, verse by verse, on their own judgment, exactly as they said in their preface they had done with everything else. Scrivener’s own headnote adds the last necessary qualifier: the list began as two pages of his in 1845, was checked against a footnote of Westcott’s, and was enlarged twice, each generation correcting the errors of the last. The famous hundred and thirteen is a fourth-generation hand-collation, not a machine count.

Now · Three things that get confused, and the numbers people quote

What Is Argued About Today

Three different objects are routinely called by one another’s names, and almost every unproductive argument about this subject begins there. The Textus Receptus is a family of printed books made between 1516 and 1633, plus one retrospective reconstruction of 1881. The Majority Text is a statistical construct: the reading found in more surviving Greek manuscripts than any other, corresponding exactly to no manuscript that exists. The Byzantine Textform as an edited text is a third thing again, an attempt to print the text that actually predominated in transmission, using judgment where the Byzantine witnesses are themselves divided.

The person best placed to separate them has done it in one sentence, and it deserves to be better known on both sides of the argument. Maurice Robinson, who edits the Byzantine Textform, wrote in 2001: certainly the Textus Receptus had its problems, not the least of which was its failure to reflect the Byzantine Textform in an accurate manner; but the Byzantine Textform is not the Textus Receptus, nor need it be associated with the Textus Receptus or those defending such in any manner. The preface to his 2018 edition is blunter, calling the printed Received Text a partially deficient form of the Byzantine Textform, and adding in a footnote that its editions print many readings which do not represent the dominant Byzantine consensus nor even come close to it, some of them with no actual Greek manuscript support at all, having been rendered into Greek from Latin sources. The leading Byzantine editor alive says the Received Text is a bad copy of the thing he is trying to print.

Then there are the numbers, which is where the topic rots fastest. The sentence you will meet most often is that the Textus Receptus differs from the Majority Text of Robinson and Pierpont in 1,838 Greek readings, of which 1,005 are translatable. It is in the standard reference articles and has been copied outward from them for twenty years, and it welds together two counts made by two different people against two different Majority Texts and two different Received Texts. The 1,838 is Daniel Wallace’s, published in 1989, and it is a count against Hodges and Farstad — Robinson and Pierpont’s edition did not exist in 1989 — using the Oxford Received Text of 1825. The 1,005 is Michael Marlowe’s, from 2001, comparing Hodges and Farstad with Scrivener’s text of 1881. Both counts are real and both are honestly reported by the men who made them. The sentence that fuses them describes a comparison nobody has run.

Marlowe’s own footnote undoes the idea that the Received Text is a single object. In a hundred and fifty-six of those thousand-odd places — about one in six — the Majority Text agrees with some other representative of the tradition, whether Estienne, Beza or Elzevir, so that it becomes necessary, he writes, to enter into fine distinctions and definitions of the term. And six hundred and seventy-eight of the thousand are places where the Majority Text departs from the Received Text toward the critical text, not away from it.

The most reprinted number of all is in the preface of a Bible on a great many shelves. The New King James Version, whose New Testament rests on the Received Text, states that fully eighty-five percent of the New Testament text is the same in the Received Text, the Alexandrian text and the Majority Text. There is no source for it in the preface, no method, and no definition of the unit being counted. That same preface is worth reading rather than citing, though, because it concedes the point at issue in one line — the Majority Text is similar to the Textus Receptus, but it corrects those readings which have little or no support in the Greek manuscript tradition — and because of a human hinge behind it. Arthur Farstad was executive editor of the New King James Version and co-editor of that Majority Text: the same man edited the translation that keeps the Received Text in the body and the Greek edition that corrects it.

The case for the Received Text is worth stating as its serious defenders make it, because that is not the case that usually gets refuted. The movement calling itself Confessional Bibliology rejects being described as a preference for old Bibles, as a label for the Textus Receptus, or as a cautious form of King-James-onlyism; its own definitional statement says the confessional text is not identical to the infallibility of every decision made by Erasmus, Estienne, Beza or the Elzevirs, that no printer received a new act of inspiration, and that the early editions differ among themselves. The line it actually draws is theological, and it is about where inerrancy is located: it aims at the formulation associated with B. B. Warfield, which locates formal inerrancy in the original autographs while criticism reconstructs their wording, and which they say relocates the errorless object to documents nobody can inspect. That is an argument about the doctrine of scripture rather than about counting manuscripts, and refuting it by counting manuscripts answers a different question. To its own side it says something worth quoting as well: that a person cannot defend the purity of God’s Word by employing impure methods.

Robinson, from the other side of the same room, disowns the theological argument outright, granting that many critiques of the majority-text position are valid, particularly the refutation of extreme claims and questionable appeals to providential preservation. The two positions that get lumped together are not one position, and neither is King-James-onlyism, which is younger than both.

The last thing to say is the thing this page has been circling, and it cuts against the tidy version of the story that both sides prefer. The men who set these books in type were working from late manuscripts, in a hurry, with tools that were not adequate, and they made mistakes that are still in print. And they were not always wrong.

At 1 Peter 5:8 the Received Text carries a small word that is in neither the Byzantine Textform nor the modern critical editions. It is in the third-century papyrus that is our earliest Greek witness to that part of the letter. It is in a corrector of Sinaiticus, in two of the best minuscules, in the Syriac and the Coptic. Nobody put it there to prove a point; it is in the Received Text because it was in a late Greek manuscript in a Basel cupboard, and it is in that manuscript because somebody copied it from something older. James Snapp’s conclusion about it is the right note to end the argument on, and it belongs to no party: in the field of New Testament textual criticism, there is no such thing as an unimportant Greek manuscript.

1435 – today

The whole thing, in order

1. 1435–1437John of Ragusa buys Greek books in Constantinople. One Gospel book he buys later, in Basel, for two florins.
2. 1443He dies, and leaves the library to the Dominican convent at Basel.
3. 1488Reuchlin writes that unless he is given a small Greek codex from that library he will breathe out his life. He is given it.
4. 1502Cardinal Cisneros conceives a polyglot Bible and founds a university at Alcalá to make it.
5. c. 1511A Basel Dominican copies an inventory of the collection, including a book that had not been on the shelf for twenty-three years.
6. 10 January 1514The Complutensian New Testament is finished at Alcalá. It will not reach a reader for eight years.
7. 17 April 1515Froben asks to have the New Testament from Erasmus, in the same week twelve hundred copies of another book of his sell in three weeks.
8. 30 August 1515Printing has begun at Basel.
9. September 1515It stops, for want of a corrector.
10. 21 September 1515The correctors arrive and lodge in the printer’s house. The printer wishes he had hired them earlier.
11. 2 October 1515The stoves are lit. Erasmus can neither stay for the smell nor leave for the work.
12. 1 February 1516The dedication to Leo X claims manuscripts at once very ancient and very correct.
13. 1 March 1516The Novum Instrumentum is finished: about a thousand pages, five months, one gathering a day, and the last verses of Revelation translated out of Latin because a leaf was missing.
14. April 1516Erasmus, privately: rushed out rather than brought out — and yet in this kind of thing we have beaten all our predecessors.
15. June 1516He asks friends for corrections and swears them to secrecy so the first edition will keep selling.
16. 1518The Aldine Bible reprints his text, press errors and all, including ones his own errata leaf had corrected.
17. 10 September 1518A papal brief calls the first edition most perfect. Its drafter admits, in a letter Erasmus prints, that he chose the date and the city himself.
18. 1519The second edition. Luther translates the September Testament from it.
19. 1520Edward Lee argues that omitting the heavenly witnesses will revive Arianism.
20. 1522The third edition prints them, from an English pocket manuscript copied out of a book that does not contain them. The Complutensian finally circulates.
21. 1526Tyndale’s English New Testament carries the book of life at Revelation 22:19, ten years after the retranslation.
22. 1527At Valladolid, some theologians propose making it a capital offense to deny the heavenly witnesses.
23. 1546Estienne’s first Greek Testament, in a new royal type, claims manuscripts almost to be adored for the look of their age.
24. 19 October 1547The Sorbonne condemns his Bibles for adding to the true reading hitherto received in the Church.
25. 13 November 1550He applies for leave to live in Geneva.
26. 1550The Editio Regia: the first Greek New Testament with a margin of manuscript variants, collated by the printer’s son at eighteen, and standing against all its own authorities in a hundred and nineteen places.
27. 1551The verse numbers, made for a concordance, on the road to Lyon.
28. 1565Beza’s first Greek edition. He stops naming his manuscripts somewhere around the end of Luke.
29. 1581He gives Codex Bezae to Cambridge with a note saying it would be better stored than published.
30. 1582He prints a conjecture at Revelation 16:5 and, years later, remembers a manuscript for it.
31. 1598His last folio: the Greek the King James men follow more often than any other, though in a third of the disputed places they follow neither him nor Estienne.
32. 1624Isaac Elzevier prints a pocket Greek Testament at Leiden. It has no preface.
33. 1626He gives up printing because of the wars in Germany, and goes to sea.
34. 1633The second Elzevir edition advertises the text now received by all. The evidence offered is that the first printing sold out.
35. 1707Mill prints Estienne’s text with an apparatus of variant readings, and dies a fortnight later.
36. 1710–1711Somebody — Küster or von Maestricht — computes the readings at thirty thousand. Mill never gave a number.
37. 1713Bentley: that printer’s infallibility will be as zealously maintained as an evangelist’s.
38. 1795Herbert Marsh writes, without a citation, that Erasmus promised to print the heavenly witnesses if a Greek manuscript could be found.
39. 1818Horne quotes Marsh inside quotation marks, and the story is loose.
40. 1831Lachmann prints a text with no printed edition under it, and says he is recovering the fourth century, not the apostles.
41. 1861Delitzsch finds the lost Reuchlin Apocalypse and lithographs the line Erasmus misread.
42. 1881Westcott and Hort, the Revised Version, and Scrivener’s reverse-engineered Greek, all in one year.
43. 1883Burgon, in a footnote: we do not, by any means, claim perfection for the Received Text.
44. 1980De Jonge reports that the promise has no foundation in Erasmus’s works.
45. 2011Krans counts thirty-seven divergences in the six retranslated verses. Wettstein had said at least thirty, in 1730.
46. 2016Andrist re-examines the Basel manuscripts and makes two of them younger than the handbooks say.
47. 2024Krans, on the evidence of Beza’s own annotated copy, moves Revelation 16:5 into the conjectures.

Who says so

A note on sources

This is a secondhand history, assembled from the works named below, and where they disagree the disagreement has been left standing in the prose rather than tidied out of it. On this subject that is a matter of accuracy rather than of manners. A great many of the figures in circulation turn out to be somebody’s estimate, somebody’s misreading or somebody’s uncited gloss, and separating those from the counts that were actually made is most of the work.

Several things here rest on a single witness, and the sentences say so where they do. Everything about the state of Erasmus’s desk in 1515 comes from his own letters, and he is an interested party in every one of them; the print runs of 1516 and 1519 are two figures he gave in the middle of a polemic, where he had an interest in the numbers looking large. How the papal brief of 1518 was obtained rests on one letter from the man who drafted it. That Estienne divided the verses on a journey rests on one sentence written by his son forty-three years afterward. What Beza thought he was doing at Revelation 16:5 rests on one annotation and one marginal note in his own copy.

Where a number is somebody’s reckoning rather than a count, the page tries to say whose. The fifty thousand ducats of the Complutensian is a Toledo humanist writing fifty-two years later. The five hundred and one itacisms and the two hundred and eighty-seven differences are Scrivener’s own counts. The thirty thousand variants are not Mill’s. The eighty-five percent in the New King James preface has no source anywhere in that preface. The 1,838 and the 1,005 that are quoted together as one statistic were made by two people against four different books. And two figures are given as agreeing precisely because they were reached independently: Wettstein’s at least thirty, in 1730, and Krans’s thirty-seven, in 2011, for the same six verses.

Two claims are contested in a way the page does not resolve, and both are marked where they occur: whether Codex Montfortianus was copied in order to force Erasmus’s hand, and whether Beza had a manuscript at Revelation 16:5. On the first, specialists who have handled the book disagree and one of them has changed his mind in print. On the second, the evidence is recent and the disagreement about what follows from it is live.

One thing here is offered as an addition rather than a report. H. J. de Jonge, whose 1980 article took the promise story apart, traced its earliest appearance to Horne in 1818 and wrote that it remained unclear where Horne got it. Horne names his source in a footnote, and it leads to Herbert Marsh in 1795. Whether the specialist literature has already noticed that, this page cannot say; it is given as the earliest telling reachable by following Horne’s own reference, not as a discovery.

Further reading

• Jan Krans, Beyond What Is Written: Erasmus and Beza as Conjectural Critics of the New Testament (2006) — The book behind two whole sections of this page. Krans counted what other people asserted, and he is scrupulous about saying when he is reconstructing rather than reporting.
• Grantley McDonald, Biblical Criticism in Early Modern Europe: Erasmus, the Johannine Comma and Trinitarian Debate (2016) — The full history of one verse across five centuries, by the person who has done most of the archival work on it. Long, and the length is the point.
• Martin Wallraff, Silvana Seidel Menchi and Kaspar von Greyerz, editors, Basel 1516: Erasmus’ Edition of the New Testament (2016) — Where the current state of every question about the making of the first edition is argued by the people arguing it. Patrick Andrist’s chapter on the manuscripts is the one that changed this page.
• P. S. Allen, editor, Opus epistolarum Des. Erasmi Roterodami (1906–1958) — The letters, with Allen’s notes, which are frequently better than the letters. Volumes two and three carry the whole of 1515 and 1516. Free online.
• F. H. A. Scrivener, A Plain Introduction to the Criticism of the New Testament (1894) — The fourth edition, and still the place where most of the counts everybody quotes were actually made. Scrivener was on the conservative side of the argument of his own day, which is what makes his figures worth having.
• S. P. Tregelles, An Account of the Printed Text of the Greek New Testament (1854) — The best narrative account written in the nineteenth century, and the one that traces the Alcalá rocket story to the end. Free online.
• Eduard Reuss, Bibliotheca Novi Testamenti Graeci (1872) — Every printed Greek New Testament to 1869, sorted by textual family. This is the book that shows the Received Text is a shelf.
• Bruce M. Metzger and Bart D. Ehrman, The Text of the New Testament: Its Transmission, Corruption, and Restoration (2005) — The standard handbook in its fourth edition. Useful, and worth reading alongside the specialists who have corrected several of its stories.
• Jerry H. Bentley, Humanists and Holy Writ: New Testament Scholarship in the Renaissance (1983) — What the sixteenth-century editors thought they were doing, on their own terms rather than ours.
• Franz Delitzsch, Handschriftliche Funde (1861) — The rediscovery of Erasmus’s Apocalypse manuscript, with a verse-by-verse collation and a lithograph of the line he misread. In German, free online, and unmatched on its subject.
• Maurice A. Robinson, The Case for Byzantine Priority (2001) — The Byzantine-priority argument made by the man who edits the text, free online, and notable for how much of the popular case for it he declines to make.
• Peter J. Gurry, The Number of Variants in the Greek New Testament: A Proposed Estimate (2016) — Where the three-centuries-long chain of guesses about the number of variants is finally laid out with its sources, and replaced with a method.