Saturday, 21 May 2022

How to read an app crit

All ancient and mediaeval texts have textual variants. That’s because they’re transmitted in manuscripts, and all manuscripts contain scribal errors. No scribe is immune. Even if a text exists in only one manuscript, it’s often possible to infer that it contains errors.

This is what the app crit is for: the apparatus criticus (‘critical apparatus’). The app crit is the chunk of gibberish at the bottom of every page in a modern critical edition. The app crit is a meticulous record of textual variants, scholarly suspicions, and other annotations to the text.

This is an introduction to deciphering the app crit. We’ll look at examples from four ancient and mediaeval texts, in four languages. They’re arranged in ascending order of difficulty.

1. Old High German: the Hildebrandslied

The Hildebrandslied is a fragmentary 8th/9th century poem in which the hero Hildebrand unwittingly encounters his son Hadubrant in combat. Here are the opening lines, with text and app crit as they appear in the edition by Wilhelm Braune, revised by Karl Helm in 1952.

The app crit at the bottom is in German: fair enough, for a German poem. The entries are separated by a large horizontal gap. Each entry gives line number; text as printed; and further information, depending on context. Where a word from the text is quoted, it’s followed by the source for that variant: in this example, the source is always the manuscript, because there’s only one manuscript of the Hildebrandslied.

Text in roman type is used for the readings. Italics are used for the editor’s own annotations. This practice is fairly standard and widespread: we’ll see it again in examples 2 and 3, below.

Because the Hildebrandslied survives in only one manuscript, the variations in the app crit are modern editors’ alterations, not different readings in different manuscripts. The first entry, for line 3 ‘Hiltibrant’, reads:

Statt des n hat die Hs. h (Hiltibraht). So auch 7. 14. 30. 36. 45.

Instead of n the MS has h (Hiltibraht). Similarly in (lines) 7, 14, 30, 36, 45.

Hs. is short for Handschrift (‘manuscript’). In English, Latin, and French the standard abbreviation is ‘MS’, for manuscript, manu scriptum, or manuscrit.

This entry indicates that the editor has normalised the manuscript spelling Hiltibraht to Hiltibrant every time it appears. But the manuscript is very consistent: which spelling is more authentic? Hard to say, without expertise in Old High German linguistics. Maybe you disagree with the alteration, maybe you agree: the point of the app crit is to give you enough information to consider it for yourself.

The next three entries are also spelling normalisations. The manuscript has ringa (line 6), wer (9), and welihhes (11). The editor’s knowledge of Old High German leads them to add an extra h to the start of each word. The last entry, line 13 mir, is a more radical alteration by the editor: that’s a grammar thing, not a spelling thing.

One thing this app crit does not comment on — and it ought to — is the string of dots in lines 10–11. By convention, a string of dots indicates a lacuna, a gap in the text. The manuscript doesn’t have a gap, but a modern editor has judged that there’s a chunk missing. Here it’s because of the syntax. Hildebrand is making a speech: the first part is in indirect speech (‘he asked in a few words who his father was among the men in the people’), but from eddo hwelihhes onwards it switches to direct speech, quoting his exact words (‘Which lineage do you come from? Tell me the one thing and I’ll know the other ...’).

Every edition of the Hildebrandslied does the same thing here, and since we’re talking about it, I’ll say that I don’t agree with it. Slipping from indirect to direct speech may be a solecism in 19th century prescriptive grammar, but that doesn’t mean it’s wrong for 8th/9th century poetry. Moreover, there’s alliteration in the half-lines on either side of the supposed lacuna, exactly as there should be in German alliterative verse: fireo – folche – hwelihhes is a legitimate alliteration. (For the record, I disagree with the lacunas marked later on in the poem too.)

2. Old Norse: the Vǫluspá 

The Vǫluspá, ‘prophecy of the seeress’, is the most famous poem in the elder Edda, mainly because of its account of Ragnarok. Here are the first two stanzas, with text, translation, and app crit from Ursula Dronke’s 1997 edition.

This app crit is in English, but it’s also denser: there are multiple manuscripts, and that means textual variants. As in the Hildebrandslied, we get line number, printed text, then a variant followed by the source for that variant; and any other relevant information, as needed.

The first entry, ‘1/2 helgar] so H; om. R’, unpacks as follows:

  • ‘1/2': stanza 1, line 2.
  • ‘helgar]’: text as printed.
  • so H’: short for ‘this is what manuscript H has here’.
  • om. R’: ‘manuscript R omits this word’

The second entry, for stanza 1 line 4, indicates that the editor has chosen the spelling Heimdallar from manuscript H, rather than Heimdalar from manuscript R. And so on. The last two entries, for 2/3 þá and 2/6 iviðiur, have an extra comment: ‘See MS. R corrections 1’ and ‘2’. These are cross-references to an explanatory section, following the poem at page 88. They’re places where the scribe wrote the wrong word and then corrected it. (There are tidier ways to report a scribal correction.)

What about ‘H’ and ‘R’: what are they, how are you meant to know what they are, or how to track them down? Dronke’s edition isn’t generous in explaining these. You’d be excused for trying to find them in her list of abbreviations on page xiv — but that list doesn’t include manuscripts.

Instead, Dronke gives a paragraph-by-paragraph description of the surviving manuscripts on page xi, in the book’s general introduction, with the abbreviations buried in the middle of each paragraph. She gives a separate, longer, account following the poem, at pages 61–62. She doesn’t tell us which libraries the manuscripts are in or their callmarks, and she doesn’t give the folio numbers for which pages of the manuscript contain the poem.

Dronke hasn’t made things easy. She could have made this information more convenient, useful, and complete. I have to say, I’m not impressed.

The best practice is to give a list of manuscript abbreviations at the start of the text, with the manuscript’s location, library, callmark, page numbers in the manuscript, and the date when the manuscript was made. If I were doing that for this edition, it would look like this:


Codex Regius: Reykjavik, Árni Magnússon Institute GKS 2365 4to, ff. 1r–5r (ca. 1270)
Hauksbók: Reykjavik, Árni Magnússon Institute AM 544 4to, ff. 20r–21r (early 14th cent.)
portions quoted in Snorri Sturluson, Gylfaginning
text of Snorri in Codex Regius: Reykjavik, Árni Magnússon Institute GKS 2367 4to (ca. 1325)
text of Snorri in Codex Trajectinus: Utrecht, Universiteitsbibliotheek Traj 1374ˣ (Tˣ) (ca. 1595; derived from a copy dating to ca. 1250–1300)
text of Snorri in Codex Upsaliensis: Uppsala, Universitetsbibliotek DG 11 (U) (ca. 1300–1325)
text of Snorri in Codex Wormianus: Copenhagen, Arnamagnæan Institute AM 242 fol. (ca. 1350)

(The excerpt shown above only cites R and H. The other manuscripts pop up in later parts of the poem.)

3. Latin: Statius, Thebaid 

Statius’ Thebaid is a twelve-book epic about the mythical war of the Seven against Thebes, written in the 80s–90s CE. Here’s an excerpt from the 1972 edition of book 10 by R. D. Williams.

Once again the app crit gives line number, the text printed and its source, then variants and their sources. Other annotations are in italics. Williams additionally uses bold type for the manuscript abbreviations, which I like. He gives a simple and easily accessible list of manuscript abbreviations on the page facing the first page of the text, headed sigla (‘abbreviations’), including the callmark and approximate date of each manuscript.

Things are getting more complicated, though. First, the manuscript list includes 15 manuscripts. Second, Williams’ app crit is in Latin. That makes sense, for a Latin poem, but it does have the knock-on effect that the library names in the manuscript list are given in Latin too. Even for an experienced reader, they can be impenetrable. One of them is actually wrong.

If a manuscript is called Codex Parisinus with a catalogue number, you may be able to guess that means it’s in the Bibliothèque Nationale in Paris. But others, like Codex Matritensis, Codex Guelferbytanus, and Codex Mediceus, are harder. Matritensis doesn’t look much like ‘Madrid’ (which is what it means). Guelferbytanus looks even less like Wolfenbüttel, and it’s extra hard because Wolfenbüttel isn’t exactly a famous name. And Mediceus doesn’t even refer to a place: it refers to a specific collection of manuscripts at the Biblioteca Laurenziana in Florence.

The fact is, these designations are intended for specialists who already know what the manuscripts collections are and where they are. Things would be easier if we were given city names in a modern language, and library names, even if Latin is used for the callmark.

Fortunately, the American scholar Ryan Baumann has made a list of Latin library names used in manuscript lists, and has released it for free here.

Another problem is that Williams’ list of manuscripts is mostly just copied from Klotz’s 1908 edition, with a couple of additions. As a result he records one manuscript, N, as being in the Phillipps Collection in Cheltenham (Codex Cheltoniensis). But that collection hasn’t existed since the early 1900s. The manucript was purchased in 1925, and is now in Dublin in the Chester Beatty Library, manuscript number 76.

The app crit itself is also more complex than the previous ones:

One abbreviation for lots of manuscripts. The manuscript list includes a special abbreviation, ω. This means ‘a reading found in manuscripts other than P’ (consensus codicum praeter P). It’s a group of manuscripts, not a single manuscript.

Lower-case letters, Greek letters. Two manuscript abbreviations are lower-case (t, r), and two are designated by Greek letters (δ, Θ). Williams didn’t invent these abbreviations, but he should explain inconsistencies like that. The normal convention is that a lower-case letter represents a derivative manuscript, where we also have the manuscript it was copied from; and a Greek letter represents either a group of manuscripts (like ω, above), or a lost manuscript that was the ancestor of some surviving manuscripts (a ‘hyparchetype’).

The obelus symbol. The symbol † appears in the app crit, but not in the main text. Normally in a modern edition † indicates a corruption, a place where the editor knows the text is wrong, but hasn’t been able to work out what the correct text was. Sometimes it’s jokingly called a ‘dagger of despair’. But that isn’t what it means here. Here it indicates that the scribe wrote a word, then wrote † above the line, along with a variant. It’d be nice if the editor explained this.

4. Ancient Greek: the ‘Homeric’ Hymn to Aphrodite 

The Hymn to Aphrodite is a 7th century BCE narrative poem about Aphrodite’s love affair with Anchises. Here’s the opening, in Filippo Càssola’s standard edition of the Hymns.

Càssola’s manuscript list is at the start of the collection of Hymns. But it’s pretty threadbare: he gives the library names (in Latin) and a catalogue number, but no locations, and no dates. He doesn’t give a detailed description of the manuscripts in his introduction to the book. By 21st century standards this is really slack.

Out of the app crits we’ve looked at, this one is by far the most esoteric. That’s partly because it’s in Latin. There’s no good reason to use Latin, other than to mark Greek literature as ‘belonging’ with Latin literature, ring-fencing them both and creating a gulf between them and ancient Asian literature. For the record, there is no gulf. For the Hymns, Hittite and Levantine poetry are much more important parallels than anything in Latin. (Editions of Akkadian, Ugaritic, Hebrew, and Sanskrit poetry regularly use English for the app crit; for Hittite, either German or English.)

The fact that Càasola’s introduction is in Italian rubs salt in the wound. This edition requires you to know three languages — two of them totally unrelated to the language of the poem itself. The only value of this combo is as a barrier to entry. It’s a great big sign saying ‘Hittitologists keep out’. It’s certainly not for anyone’s convenience.

Many Greek app crits use Latin, but mostly they stick to conventional expressions. That means you can decode the app crit even if you’re not a latinist. Another American scholar, Karl Maurer, has provided a list of common Latin terminology here. That would suffice for the Statius in example 3, above. It partly covers Càssola’s Latin. But not much of it.

Càssola’s basic format is familiar — line numbers, text as printed and source, variants and their sources — but he also cites suggestions made by modern editors (in line 4, διειπετέας comes from Wilhelm Schulze’s 1892 book Quaestiones epicae; the full title is on page lxxvi). One of them (Abel, cited for line 6) doesn’t appear in Càssola’s bibliography. And two sources, ‘Aldina¹’ and ‘Chalcondyles’ (cited for line 10), aren’t modern editors, but a 15th century edition and a mediaeval Byzantine commentator. You need to know all of this off the top of your head to use this app crit. Definitely not user-friendly.

Worse still, Càssola’s Latin is much harder than it needs to be. There’s no way past it: to read Càssola’s annotations, you need to know Latin, and you need to know it well.

Lines 1–151: ‘om. Π foliis avulsis’. (Notice that Càssola doesn’t put his annotations in italics, as in examples 1 to 3.) Maurer’s cheatsheet covers om.: it means all these lines are omitted in manuscript Π (cod. Parisinus suppl. 1095). But for foliis avulsis you’ll have to brush up on your Latin ablative absolutes. This bit means ‘because the pages have been torn out’.

Line 6: πᾶσι δὲ ἔργα is a correction suggested by two modern scholars, Hoffmann and Flach. After that we hit a wall of abbreviations, with no punctuation to help the reader follow the structure of the annotation. What it means is this: πᾶσι δ’ ἔργα is the variant given in six manuscripts (M, At, D, B, Γ, and V), while πᾶσιν δ’ ἔργα, with an extra letter, appears in seven others (A, Q, P, and x, where x represents the group E, T, L, and Π).

Lines 10–11 (written ‘10–1’): ‘usque ad πόλεμοί τε et ab καὶ ἀγλαὰ in unum conflarunt ET’. This is Càssola being a mean-spirited bastard. First, conflarunt is a high-faluting version of the verb form conflaverunt, so you don’t just need to know Latin, you need to be familiar with literary Latin. The note as a whole means: ‘Manuscripts E and T have joined together the text up to πόλεμοί τε, and from καὶ ἀγλαὰ onwards.’ Or, to put it another way, ‘the words ἅδον ... μάχαι τε are missing in E and T.‘ But why use a simple om., when you could use usque ad ... et ab ... in unum conflarunt? Mustn’t make it too easy for the proles, I guess.

Final thoughts

It’s easy to criticise. But it’s also pretty easy to provide clear information and to use punctuation marks. I have the utmost respect for the critical skills of all these editors, but where they’ve been thoughtless, I resent that.

The point of scholarship, the point of an app crit, is to make information easily accessible. Examples 2 to 4 do an uneven job of that. Càssola, in particular, has gone out of his way to make things accessible only for an inner circle of experienced scholars. It’s the very definition of ‘esoteric’.

There are harder app crits out there. But the very hardest ones — the Nestle-Aland edition of the New Testament, and M. L. West’s editions of Homer — are actually much better than these ones. They look extremely forbidding, because they take up a huge amount of space on the page. It takes a lot of work to read them. But they’re much better thought out, and much more clearly explained.

Their difficulty lies in the fact that there are hundreds of manuscripts for those texts, and that means lots of citations when it comes to variants. But their manuscript lists, and the way they write in the app crit, are crystal clear. West does use Latin, which is very regrettable; but at least he doesn’t mix it up with extra languages. And in his case, Maurer’s cheatsheet will see you through almost every note.


  • Baumann, R. 2020. ‘Latin names of libraries commonly used in the naming of manuscripts.’ [github link]
  • Braune, W.; Helm, K. 1952. Althochdeutsches Lesebuch. 12th edition. Tübingen.
  • Càssola, F. 1975. Inni omerici. Milan.
  • Dronke, U. 1997. The poetic Edda. Volume II. Mythological poems. Oxford.
  • Klotz, A. 1908. P. Papini Stati Thebais. Leipzig. [Internet Archive link]
  • Maurer, K. 2018. ‘Commonest abbreviations, signs, etc. used in the apparatus to a classical text.’ [University of Dallas link]
  • Williams, R. D. 1972. P. Papini Stati Thebaidos liber decimus (Mnemosyne supplement 21). Leiden.