How do we represent the critical apparatus when we make our critical edition?

Posted on September 2, 2025 by Roger Pearse

My current project, the Latin text of the “Life of St Botolph”, composed around 1100, has reached us in a number of manuscripts, together with abbreviated forms, also in manuscripts, and a couple of early editions. These are the “witnesses” to the text.

I have now compared the lot, and I have a word document with all the variations. There are indeed differences in words, mostly accidental. Only one of these affects the sense, where some idiot copyist in a single late manuscript decided that the people living nearby were Scottish, and introduced the word into the text. It’s usually fairly obvious what the correct reading is. It’s fairly straightforward to produce a stemma, a tree-diagram showing what is copied from what, based on those errors and omissions. This I have done.

But what to put in the notes? There are about a dozen witnesses. This means that it is possible to put the whole list of variants into the notes. That’s not going to be very readable, tho. The jargon for this is a “positive apparatus.”

Or I could just ignore every manuscript except where it differs from what I have decided the text is. That will be much shorter. But it will also be a lot less usable. The jargon for this is a “negative apparatus.” One manuscript came to hand late, and I can tell you that it helped a *lot* that I knew what every manuscript said, rather just what was different.

Indeed a critical apparatus is not a very readable thing anyway. So how best to do it?

There are papers on these kinds of questions. Addressed these issues very well indeed was Sebastien Moureau, “The apparatus criticus” (2015), looking at how to do the apparatus for an Ethiopian text.¹ The article is online and recommended. This assesses the advantages and disadvantages of various approaches.

In particular Moureau points out that the long-established format of classical editions – numbering the lines of the page, with variants by line number at the page foot – is completely impossible to reproduce with any widely-used software. Here’s a random page from a recent edition of the letters of Isidore of Pelusium, vol. 3:

How on earth do we do that in Word?

Now there is an online course with textbook, M Burghart, Digital Editing of Medieval Texts: A Textbook, online here. Chapter 4 discusses how to encode the variants in something called TEI. This is a computer file format. It’s horrible.

TEI is no doubt possible to learn, but working with tagged text files is horrible. It imposes a huge initial cost on anyone wishing to adopt this approach. It’s very error-prone.

However this does introduce an interesting general issue.

Basically in order to do a critical edition, you must need to go through two stages.

Compile a database, in a structured format. which includes the text and all its variants and any notes upon it, from all the witnesses.
Present a subset of this information in some other format in your publication. Possibly even in more than one format for different purposes.

Anyone familiar with modern methods of computer program design will immediately recognise a common computing pattern – pull the presentation layer out of the code, and do it as a separate layer. Don’t mix it in with the main logic. Let the logic get the data in whatever way is useful and reasonable; and then pass that to the front end to display in whatever way is useful. The two sections of the program can be developed separately. and if you need to change the output format, or the on-screen appearance, you don’t have to change every part of the program.

The Burghart paper does indicate that TEI (the database, essentially) can be used in this way, to generate different outputs.

But creating a file/database encoded in TEI means doing things in text files. That’s very error-prone at the best of times. TEI is just a format. What is needed is some “casual casual easy thing,” some bit of software that stores what you put in in TEI, and redisplays it in an easy way. I know people have attempted this, but the results that I have seen look awful.

Is there a tech billionaire in the house?

TEI is a very tech-heavy approach, that imposes a lot of pain upfront for no gain. It gets in the way. I can’t face doing it, and I’m a former programmer! In Word on the other hand I can just start, and get on with what I want to do. But I’m then stuck. There’s no easy way to transform the output into some other format, say from positive to negative or back again.

I’ve no idea what the answer is. In the meantime we’re stuck with Word.

For the New Testament, where there are thousands of witnesses, and stemmatics does not work, the apparatus is always presented negatively. Indeed in the Nestle-Aland edition, special symbols are introduced, known as “critical sigla”, indicating insertion and omission etc. These are supported by special Greek fonts, such as Gentium. I found an article online by Brent Nongbri here, describing them:

Um… yuk. I’ve not seen any classical text use these, however. This again is a presentation-layer thing, designed to reduce the amount of gunk at the bottom of the page, at the expense of intelligibility. Here’s a page of Jude, using these techniques.

That’s pretty awful, and only gives a tiny amount of data.

So there’s a world of pain here, for something that ought to be simple and an activity that is carried out in universities around the world.

In the meantime, I will stick with Word, and, since it is possible for me to give a positive apparatus, I will!

Sébastien Moureau, “The Apparatus Criticus,” In: Edd.: Alessandro Bausi, &c., Comparative Oriental Manuscript Studies: An Introduction, Tredition : Hamburg (2015), p. 348-352 http://hdl.handle.net/2078.1/155652[↩]

What is a critical edition, and how do I find one?

Posted on April 14, 2021December 27, 2022 by Roger Pearse

I have just been asked this basic question, on this post on the manuscripts of Pliny the Elder, and to my surprise a quick google does not give a satisfactory answer. So … here goes!

Ancient literary texts were dictated or written by their authors more than 15 centuries ago. They were then hand-copied for many centuries, initially in papyrus rolls, and then into the modern book format, the parchment codex. During this time most ancient texts were lost, forever. Only 1% of ancient literature is estimated to survive.

Those that do survive do so in medieval hand-written copies. These are known as the “manuscripts” of ancient authors. (For modern authors, we use the word “manuscript” differently, to mean the handwritten copy sent to the publisher by the author, but these almost never survive from antiquity).¹ These copies are few. Most ancient texts survive in copies no older than about A. D. 800, many of which descend from a single manuscript that had survived from Antiquity. The only exceptions are texts that were used a lot during antiquity and after, such as the bible, and the works of the major church fathers.

Other losses of text happened. Some texts survive in an incomplete form. Sentences are missing. Chapters are missing. Whole books are missing.

For instance, the start of Suetonius’ Lives of the Twelve Caesars is not known to us, although it still existed in the 6th century AD, as a stray quote tells us. Many ancient histories were written in tens of “books”, each originally a single scroll. Livy’s Roman History is an example, written in at least 140 books. The groups of 10 books travelled down the centuries separately. Of Livy, all that survives is books 1-10 – one unit; books 21-30; books 31-40; and a single damaged manuscript which originally contained books 41-50 but the back is lost, and today it includes only books 41-45 and the first page of book 46.

Also, in the process of copying, scribes made mistakes. Sometimes they went back and fixed them. Sometimes a later copyist fixed it. Sometimes a later copyist guessed wrong! Sometimes there were odd abbreviations. Also there were changes to the type of handwriting used for books – “book hand” is the jargon phrase – which means that later scribes could get confused.

Printing arrives in 1450. The first printed editions of ancient texts arrive then.

But these were not “critical editions”. Instead the publisher found a manuscript – often a late manuscript -, and simply printed whatever text was in it. He might include some corrections, or not. If it was a Greek text, he would often supply a translation into Latin.

These “pre-critical editions” were printed, and reprinted, for centuries. Sometimes a work would be printed; and then a later publisher would find another manuscript, which contained parts of the work that the first edition did not contain. But often it was just a case of different punctuation, typeface, and notes. (The text that “everybody” knew is sometimes called the “textus receptus”).

Imagine that you are an editor. You have more than one manuscript. They differ, in small ways. What text do you print? Well, the early editors bodged along, guessing at the correct text.

But in the early 19th century, scholars in Germany began to evolve some rules to decide how to handle this problem. The rules are not scientific; they merely make common sense explicit. The creation of these rules marked the creation of “textual criticism” as a discipline, dedicated to making it possible to restore a text to something like what the author wrote, and remove scribal errors.

The editions that arise from this process are known as “critical editions”. A critical edition is one where the text is as far as possible what the author wrote, but with the process of creating that text documented, and a “critical apparatus” of footnotes that shows where there might be uncertainty.

The scholars creating a critical edition try to assemble all the surviving manuscripts, where possible. They try to compare them all. They apply the rules of textual criticism to decide which versions of the text are original, and which are derived from the process of copying the work down the centuries.

Now it has been found by experience that various traps lie in wait for scholars doing this. Sometimes the “obvious” text is not right. Greek texts in the 4th century BC were written in Attic Greek. Later Hellenistic texts from the 3rd century BC to the 1st century AD were in a later form of Greek. But from the 2nd century AD onwards, there was a revival of the custom of writing in Attic Greek, which persisted as late as 1453. Consequently Hellenistic Greek texts could be “corrected” by an Atticising copyist. The same process happens in Latin, where a difficult or unorthodox author can be “corrected”. Some early editors certainly did the same, falsely correcting the author, rather than the manuscript.

Modern academic editions of ancient texts, in the original language, are always critical editions. So to find a critical edition of any ancient author, you can use a library catalogue like the Library of Congress, or COPAC, and sort by date, most recent first. Any academic edition of an ancient text published after about 1850 will probably be a critical edition.

Sometimes one critical edition may become the “standard” edition of that text. That is usually the one you want. The only way to find out which edition this might be is to read around the subject, read reviews of the editions, and see which edition is referenced. Other critical editions of the same author will normally indicate if one edition is widely used.

Some critical editions are still not very good. Furthermore, most ancient literary texts do not even have a critical edition at all; the only editions are pre-critical. The vast majority of ancient texts are of the church fathers, and modern scholars have preferred to edit classical texts instead.

You use what you have.

My thanks to MDR for pointing out that such an original has survived, of the epigrams of the 6th century author Dioscorus of Aphrodito has survived, plus some other corrections – thank you.[↩]