Archetypes and Autographs

Contents: Introduction *The Autograph *The Archetype *Footnotes


It is customary to say, in performing textual criticism, that we seekthe "original text." But what is the "original text"?Take, say, Shakespeare. Is the original text the manuscript he wrote? Oris it what the actors actually spoke when the plays were firstperformed?[1]

Or take Charles Dickens's novel Great Expectations. Hegave it a rather downbeat ending. Edward Bulwer-Lytton read it andconvinced Dickens that the ending was too melancholy. Dickens wroteanother ending, somewhat ambiguous; it could be read to meanthat Pip and Estella will never meet again -- which is what I assumedwhen I first read it, knowing how Dickens meant the book to end --but it could also beread to mean that they would never part again. The secondending is, in its way, a masterpiece of ambiguity, allowing readersto take away whatever message they wanted -- but it wasn't whatDickens originally intended. Which is the "proper"ending of the book? Dickens's original intent, or his finalversion that was the standard for many years?

At least, in this case, we can print both endings, and many moderneditions of Great Expectations do just that. Other recent writingsproduce more complex problems. HenryJames, when he wrote his book The Ambassadors, published amagazine version, then an American version, then an English version;the English version was the most accurately produced. Yet James,six years later, produced another American edition, and seems tohave worked from the inaccurate first American edition rather thanthe accurate English edition. So which is the "correct"James text? The 1903 English edition, which represented his bestview of the book at the time? Or the 1909 American edition, whichrepresents his revised thinking but contains all the numerouserrors which derived from the 1903 American edition which Jamesnever corrected? Or is it a combination of the 1903 English and 1909American editions?

Or take Langland's Piers Plowman. Almost all scholars agreethat Langland produced (at least) three editions, called A, B, and C. A is theearliest, so it could be called the original. C is the last, so it could becalled Langland's last word. And yet, the largest share of scholarly attentionhas gone to B, which is neither the first nor the last edition. (Why? Becauseit's generally considered the best.)

Or, again, Martial wrote his Epigrams while Domitian was RomanEmperor, then revised Book X after that Emperor was assassinated. Which versionshould we print, the original version (which is a unified whole) or the revisedversion (which probably is closer to what he actually thought.)

In a more recent example, consider Shakespeare's King Lear. There aretwo basic printed editions, a quarto and the First Folio. Although it is commonto find quarto and folio editions of Shakespeare plays that differ substantially,this is usually explained on the grounds that the quarto text is a "bad"quarto reconstructed by an actor rather than from Shakespeare's manuscript. But,in King Lear, neither edition shows signs of this; each has significantmaterial not found in the other, and in each case the material appears to beShakespearean and not corrupt. So which is the original of King Lear? Manynow think that the two versions are different stage presentations. Was there anoriginal Shakespearean manuscript in this case? Very possibly not; rather, the lastcommon ancestor of the two was a collection of scenes and materials that was notyet fully worked up into an actable play. Shakespeare, the hypothesis goes,twice worked these materials up into a play, but it wasn't the same play! So whichis the original -- the manuscript from which the folio text descended, the manuscriptfrom which the quarto is derived, the raw materials from which both descend? Somethingelse? We actually see a number of editions print both texts, sometimes inparallel to allow comparison.

Such problems occur throughout the field of textual criticism -- James Thorpehas collected dozens of examples just from Shakespeare's time until now. We shouldalways keep in mind what we are trying to reconstruct. Although we strive torecreate the autograph, the author's original writing,what we actually are working on is the archetype,the earliest common ancestor of all surviving copies.

The distinction is very important. As Erick Kelemen notes,

[T]he archetype is still -- and this is key -- nothing more than the last manuscript from which all the surviving manuscripts are ultimately derived, not an ideal version of the work, but a manuscript always at some incalculable remove from that ideal.[2]

Nelson Goodman made an interesting point here: the original of a written work,or of a piece of music, is conceived differently than the original of a paintingor work of art. A reproduction of a painting by Holbein is explicitly a copy, buta printing of Beethoven's Fifth Symphony is still Beethoven's Fifth Symphony. Ifproperly printed, it's the same as every other version of Beethoven's Fifth.Goodman's term for works where there is a completely duplicable original is"allographic," and that original is an "allograph."A performance may be an interpretation, but the interpretation is, so to speak,post-textual. But, for Beethoven's Fifth, we have the manuscript -- or someone did.Not so, obviously, with the New Testament as a whole. We have to keep in mindwhat original we mean.

The Autograph

"Autograph" is the accepted term for the original edition of aparticular work, written or dictated by the author. It is the earliest copyfrom which all later copies are ultimately descended (note that it may not be the latestcopy from which the manuscripts descend). Thus in most instancesit is what the textual critic would like to reconstruct (there are exceptions -- as, e.g.,when an author later edits his work).This is not always possible, however; in many cases, all we can reconstruct isthe archetype.

It should be noted that not all documents have an autograph.As noted above; Shakespeare's plays probably don't, in a pure sense; there was nodocument that represented Shakespeare's "final draft."In the case of Chaucer's Troilus and Criseyde it is widely(though not universally) believed that Chaucer continued to makemodifications to his manuscript even after the first copies hadbeen made. Thus the autograph in that case was a moving target.There can also be "autographless" documents as a resultof compilation. We see this with some commentaries, for instance.A church father might write a commentary, leaving out the longer Biblicalquotations, and hand it to a scribe to finish off. The scribe copiesthe text and inserts the Biblical quotations. So: The autograph of thecommentary is the Father's original text, but the autographof the quotations is Bible itself (or, in another way, the manuscriptthe copyist used to supply the quotations), and there is no actualautograph of the combined text. Nor is this complex process confinedto commentaries; ancient histories often quoted sources verbatim atgreat length -- as Livy took over Polybius, or Josephus used theassorted sources at his disposal. Nor was it only ancient authorswho did this; Holinshed and Shakespeare, e.g., both took large textsverbatim out of Hall.

By contrast, every extant manuscript -- of every writing ever made! --traces back to an archetype. (Technically, this is true even of theoriginal manuscript: It is its own archetype, and would be sotreated in mathematical discussions of generations of copying.)

The Archetype

Although the term "archetype" is well established in classical textual criticism, there is now a tendency in New Testament circles to call it the "initial text," that is, the ancestral text to the existing copies. Obviously making up a new term tends only to introduce confusion, so in this article I will continue to use the term "archetype."

The archetype is the direct ancestor from which a particular group of copies is derived.For example, Dabs1 and Dabs2 are both copied fromD/06 (Claromontanus), so D/06 is the archetype of the group D/06, Dabs1,Dabs2.

In most cases, of course, the archetype of a particular group is lost.We do not, e.g., have the archetype of Family 1 or Family 13, let alonesuch a vague thing as the Alexandrian Text (which may not even have anarchetype; text-types are loose enough collections of readings that notall copies containing readings of the type may go back to a singleoriginal). For classical works, however, it is often possible to identifythe archetype of some or all surviving copies. Arrian's Alexander, forinstance, exists in about 40 copies. Every one of these has an obviouslacuna at the same point (in Book 8, the Indike).It so happens, however, that the manuscript Vienna hist. gr. 4 chances tobe missing a leaf which corresponds exactly with the lacuna. Thus it isapparent that this manuscript is the archetype of all surviving copies.(There are even a few exceptional cases where it is possible to determinethe archetype in cases where it is lost. All copies of Suetonius'sLives of the Twelve Caesars, for instance, lack the beginning ofthe life of Julius. From this and other evidence, including colophonsand excerpts and cataloguing data, it is apparently possible to provethat all these copies go back to the lost Codex Fuldensis.)

There are instances where we can demonstrate the difference between autographand archetype. An example is Chaucer's "Boece," derived from Boethius'sConsolation of Philosophy. We have good knowledge of the Latin source, andalso of French versions Chaucer consulted. Knowing that Chaucer rendered theLatin quite literally in most places, we can reconstruct his actual autograph withfair exactness. It can be shown that the archetype of the extant copies wassimplified at many points.

It is possible to speak of an archetype for the New Testamenttext. It does not absolutely follow that this archetype is theAutograph. Consider the followingstemma:

          A          |          B          |          C          | ------------------- |     |     |     | D     E     F     G

with all surviving copies being descendants of D, E, F, and G. In thiscase, the autograph is A, but the archetype is C. All survivingmanuscripts are derived directly from C, with A several removes furtherback. It is worth noting that all textual criticism can directly reconstructis the archetype C; A is beyond our direct reach, and any difference betweenA and C can only be reconstructed by means of emendation. (For furtherbackground on this process, see the article on ClassicalTextual Criticism).

Now it should be noted that we cannot construct the ancestry of any part ofthe New Testament in detail. But we can approximate it. Westcott and Hort, forinstance, proposed the following sketch-stemma:

              Autograph                  |       ---------------------       |                   |       D                   E       |\                 /|       | \               / |       |  \             /  |       |   \           /   |       |    \         /    |       |     \       /     |       |      \     /      |       |       \   /       |       |        \ /        |       |         |         |Alexandrian  Byzantine  Western   Text        Text      Text

We should keep in mind, however, that we cannot by any means tell this stemma from thefollowing:

              Autograph                  |                  B                  |                  C                  |       ---------------------       |                   |       D                   E       |\                 /|       | \               / |       |  \             /  |       |   \           /   |       |    \         /    |       |     \       /     |       |      \     /      |       |       \   /       |       |        \ /        |       |         |         |Alexandrian  Byzantine  Western   Text        Text      Text

Indeed, Westcott and Hort suspected the existence of some copies before priorto the earliest recoverable text, as they marked a handful of primitive errorsin their text.

An additional complication is that the archetype of a particular New Testamentwork may differ recensionally from the autograph. This is perhaps best illustratedfrom the Pauline Epistles. At some very early point -- assuredly before the time ofour earliest papyri -- most of Paul's letters were assembled into a collection.(Hebrews, of course, is an exception, and perhaps a few others such as the Pastorals.But most of the letters must have been collected by the mid-second century at thelatest.) It is therefore perfectly possible -- perhaps even likely -- that thiscollection is the archetype, and that the individual letters are not even the sourceof the textual stream. So, e.g., Zuntz; on page 14 of The Text of the Epistles,he points out that Ignatius and Polycarp apparently knew a Pauline corpus, but theauthor of I Clement seemingly did not, and so concludes, "Thus A. D.±100 is a probable date for the collection and publication of theCorpus Paulinum; that is, forty or fifty years after the Epistles werewritten. Here then, as in the tradition of all ancient authors, 'archetype'and 'original' are not identical."

Even if the archetype of the Pauline collection is pristine, and the letterhave individual archetypes, thisdoes not mean that the archetype is a pure descendent of the autograph. Severaldocuments are thought by at least some form critics to be composite. This ismost evident in the case of 2 Corinthians, where many authorities believe thatat least two letters have been used to produce the present document. Therefore, theearliest document entitled to the name "2 Corinthians" is not an autograph;it is the conflation we now have. Properly speaking, even if we could recover thecomplete texts of the component letters of 2 Corinthians, the portions not foundin 2 Corinthians cannot be considered canonical.

We see another clear, and even more complicated, case in the Hebrew Bible,in 1 Samuel 17-18, the story ofDavid and Goliath and its aftermath. It will be obvious that two stories havebeen combined here: One in which David, Saul's courtier, volunteers to slayGoliath, the other in which David is unknown to Saul and comes out of nowhereto slay the giant and be taken on by Saul. The former story is a clear part ofthe continuous history of Saul; the latter is a folktale about David.

This is just literary analysis, but it has strong textual support: The HebrewBible has both stories -- but the Old Greek, as represented by Codex Vaticanusand others, has only the court history. Nor can this be credited simply to editorialwork to eliminate doublets; the separation is too clean and clear. (Sorry, folks,but I study folklore, and it is.) Somehow, the Old Greek was taken from acopy of 1 Samuel into which the Hebrew folktale had never been incorporated.

So what is the true autograph? If we consider the Hebrew version canonical,then we're reconstructing a version redacted after the initial draft found in LXX.

We should note that it is not the task of the textual critic todisentangle the strands of 2 Corinthians or any other such work. The task of thetextual critic is to reconstruct the archetype. If we are fortunate, this will proveto be identical with the autograph -- or, at least, so close as makes no difference.But it does not matter in practice whether the autograph and archetype are nearlyidentical or wildly different. We reconstruct the earliest available text.To go beyond that is the task of a different sort of critic. The textual criticshould simply be aware that the archetype may not be the autograph -- and also toconsider how the existence, e.g., of a Pauline collection, might affect the readingsof a particular letter. It is quite possible that certain letters were altered tofit an anthology, just as certain passages were adapted to fit the lectionary.

Chances are that, in the New Testament, only Paul suffers from problem.The Acts and the Apocalypse, of course, were standalone documents, neverincorporated into a corpus. The Catholic Epistles cannot have been assembledas a collection until quite late (this follows from their canonical history: 1 Peter and1 John were universally acknowledged, but the other five were slow to achieve recognition,and became canonical in different areas at different times; note, for instance, thatP72 contains 1 and 2 Peter and Jude, with non-canonical materials,but not 1 John, even though that book was certainly regarded as canonicalby the time P72 was compiled, and Jude was still questionable). Thegospels probably came together much earlier than the Catholic Epistles (clearlythey were accepted as a collection by thethird century, when P45 was written), but they also circulated widelyas separate volumes. Thus, while a four-gospel collection may have exercised someinfluence, it was not the archetype.

On the other hand, every part of the New Testament may have suffered from the"which copy" problem. This is most obvious in Paul: He dictated atleast some, and probably most, of his letters. It's also widely believed thathe kept copies of these letters. Note what happened here, because it's asituation actually analogous to the situation in Shakespeare outlined in thefootnote: Paul dictated a rough draft. Unless two scribes took it downsimultaneously (in which case those two scribes would doubtless produceslightly different transcriptions), someone would then have to produce acopy of that dictated text, either for circulation or for Paul's filecopy. This second copy would doubtless be neater, and might well includesome corrections of Paul's errors. So which one did Paul send out? Wedon't know, though we'd suspect it was the "fair copy" ratherthan the original "foul" edition. But which is the autograph?And which formed the basis for the later canonical edition? There is noway to answer this.

Other New Testament authors weren't sending out letters, but they werepresenting copies to patrons. Would Luke really write a gospel, and giveit away without keeping a copy? It seems most unlikely. But which of thosefirst two copies became The Gospel? And could the different traditions havecross-contaminated? The answer again is not obvious. But it likely isimportant.

Amanuensis copies have another interesting problem: The influence of theamanuensis on the text. Suppose the amanuensis says to the author, "Areyou sure you want to say that?" And, sometimes, he might convince theauthor. So the version produced by the amanuensismight not be what the author would have produced on his own. A ratherhilarious version of this happened to James Joyce. According to Kelemen, p. 169,he was dictating to Samuel Beckett when someone knocked on the door. Joycesaid, "Come in," and Beckett mechanically put it into the text.When Beckett read back the section, Joyce queried is, was told that he hadsaid it, thought about it (and, supposedly, realized what happened) -- butsaid, "Let it stand." To be sure, the book was Finnegans Wake,where nothing means anything anyway, so if Beckett had copied down,"The aliens have elected one of the own as president of the UnitedStates," Joyce might have accepted even that on the grounds that itmade as much sense as anything else in the book....

We note incidentally that classical scholars actually have a notationfor distinguishing archetype and autograph. The autograph is denoted bysome symbol (e.g. the autograph of Chaucer's Canterbury Tales is sometimesgiven the symbol O), and the archetype by that symbol followed by a' (so the Canterbury Tales archetype was O', read -- at least in mycircles -- as "O prime."). We also note that at least somescholars, both classical and NT, have not tried to go beyond thearchetype (though they didn't really express it this way). Thus Lachmanntried to reconstruct "the text of the fourth century," and, as noted above,Westcott and Hort marked "primitive errors" -- readings wherethe original had been lost before the ancestors of all the main types.


1. In the case of Shakespeare and other Elizabethandramatists, the question is even more complicated than these choices mightmake things appear. The relationship between original writing and originalstage presentation could be extremely complex. The likely process ofcomposition was as follows: Shakespeare would prepare a rough draft (the"foul papers"). This would certainly be full of corrections andrevisions, and quite unusuable for production purposes. So someone --perhaps Shakespeare himself, but perhaps not -- would produce a fair copy.The foul copy would go in some archive somewhere, in all its disorder. Butthe foul copy might be the last and only copy from Shakespeare's pen. (Thisis even more true of Shakespeare than of other Elizabethan dramatists,because there is evidence that his hand was hard to read.)

And the fair copy, even if written by Shakespeare, probably wouldn't beuseful for dramatic purposes. There is reason to think that Shakespeare'swork was sorely lacking in stage directions, for instance. He also usedsome rather peculiar and confusing spellings. So someone wouldhave to convert the fair copy to an official prompt book. This, in additionto adding stage directions and such, might involve levelling of dialect,cleaning up of unacceptable language -- and, in at least some instances,clarification of errors. This stage of the production would not be underShakespeare's direct control; the producer of the play would be incharge. But Shakespeare would be available for consultation, and might wellbe responsible for the revised language of any changes.

And it's thought that Shakespeare acted in at least some of his ownplays, so he himself might have been involved in the give-and-take.

And this is before the play has even been put into production! Aftercreation of the prompt book, additional changes might be made --and, if the changes were cuts, the alterations might not appearin the prompt book. In addition, Shakespeare might not have much controlover these; if the producer said, "we need to cut twenty minutes,"he might be allowed to choose what was cut, but if a part called for anactor to do something he physically couldn't do (e.g., perhaps, jump awall), then tough luck to the script.

So the question of what we should reconstruct is very real. Thefoul papers, the only copy known to have been entirely byShakespeare? (We should note that this copy often contains errorswhich the author clearly did not intend -- e.g., characters whosestage directions are identified by the wrong name, as the infamoususe of "Oldcastle" for "Falstaff" on occasionin Henry IV Part I.) The fair copy, which -- if by Shakespeare -- wouldundoubtedly have contained some additional corrections by the author?The prompt book, which is not in Shakespeare's hand and may containcorrections he did not make -- but which also contains material hedid suggest, and which will have the full stage directions and properidentifications of the speakers? Or the production version?(W. W. Greg, The Editorial Problem in Shakespeare, third edition,pp. xxxii-xxxiii, is certain we should go for the last version entirelyby Shakespeare. I can only say that this is not obvious to me.)

And once we decide which to manuscript to target, we still have to sort throughthe materials. Some Shakespeare plays exist only in the printing of theso-called "First Folio" and editions taken from it. The playsin the folio are believed to derive from all sorts of sources, ranging fromShakespeare's foul papers to the prompt book to editions produced byother printers.

Other plays exist also in individual quarto volumes. Some of theseare "good" quartos, taken from sources similar to the folio.Others as "bad" quartos, taken from the memories of authorswho had performed the plays, often misremembered and often cut by theproducers. Yet they are the only line of evidence outside the folioedition, and may represent a more advanced state of the script.

Many other writings have suffered similar complications, and thereis no reason to think Shakespeare, or the New Testament, is any wayunique in this. The problem of what to reconstruct is very real. Indeed, in recent years there has been a movement of sorts that seems almost to want to scraptextual criticism because of this problem. This strikes me as going toofar. The fact that we have to decide what to reconstruct does not mean thatwe should give up on reconstruction! [back]

2. Erick Kelemen, Textual Editing and Criticism: An Introduction, W. W. Norton, 2009, p. 82. [back]