2014-11-11

Why Greek diagrams are rude

How come it took until Latin Late Antiquity for a great infographic like the Great Stemma to explode on the scene? How come that the earlier Hellenistic culture of the West never evolved a graphic technology like this, with its flows of data that drag the eye down and onwards, with its callouts that divert your attention in the fashion of hyperlinks, and with its blending of multiple data classes - in this case chronography and genealogy?

Last year, the Israeli historian of mathematics Reviel Netz, who is professor of classics at Stanford University, spoke at the British Academy in London and offered an interesting new synthesis of his ideas on diagrams. As far as I know Netz has not paid any attention to the Great Stemma, but his thoughts on the Hellenistic intellectual world have a strong bearing on what followed on from it.

Netz is not only famous as the author of The Archimedes Codex, but also as the scholar who introduced in 1999 the idea that the geometrical diagrams in the works of Euclid, Archimedes and others should not be understood as after-the-fact explanations of geometrical discoveries or as mere ornaments to the text, but as integral to the logical proof, as snapshots of the discoveries themselves.

The geometers thought spatially, or diagrammatically, then honed the explanations in words afterwards.

Ideas evolve, and Netz seems to have moved onwards to a view that, in a sense, restores some of the primacy of text in that polarity. That would be my somewhat exaggerated take on what he was saying last year during Leaping out of the Page, a lecture in London that you can see on YouTube.

The question he explored in his 2013 March 14 presentation at the British Academy was: why are those Greek mathematical diagrams so bare, so sketchy, one might almost say, so rude?

Netz seeks to explain the fact that there is no evidence of anyone debating a better way to draw such diagrams (at 51.10 by elapsed time in the video) and he comes up with the following answer.  Literary texts on papyrus in the classical period lack any word-breaks, paragraphs or illustrations. The readers had to inject all of that. The text they held in their hands was what Netz calls "schematic": rude cues for the educated reader to unlock and "perform" the text in his mind.

The diagrams conform with this: they are schematic too (52.18). Nothing more than that was expected of them. This seems to come out of ideas in Netz's book, Ludic Proof, in which he explored the startling similarities between Hellenistic poetry and mathematical texts from the same era.

So what could this reveal to us about Late Antique diagrams?

Firstly, it highlights the way in which Greek mathematical figures are so very different from Late Antique flow diagrams like the Great Stemma, the arbor porphyriana and the 37 stemmata of Cassiodorus. We may use the English term "diagram" for both types, but they do not belong to the same genera of things at all. The Late Antique graphics are startlingly new and creative, with no identifiable roots in mathematical techniques or any existing literary practices.

The idea of devoting an entire papyrus roll to a graphic without any accompanying text must have seemed strange and new-fangled to people when they first saw it.

Secondly, these observations reinforce our understanding that flow diagrams are part of the haptic world of things that we manipulate. Netz argues (24.57) that geometrical figures are cerebral and differ utterly from touchy-feely arithmetic: "When Greeks do counting, what they do is to operate on an abacus. They have counters that are being moved around on an abacus... You have a flat surface, and on this flat surface you are moving stuff."

Something parallel would have happened when the designer of a flow diagram was laying it out. He had to use counters or ostraca to plan it. Later on, the reader of the flow diagram will inevitably handle it too, putting fingertips on its "icons" and tracing its flows. The flow diagram is something everybody has an urge to manipulate, a Late Antique precursor to the iPad: it's a physical thing, designed to be not only looked at, but touched, and aimed at the "manual" user, even the semi-literate, not the cerebral reader showing off his paideia.

Netz stresses how, in Euclid or Archimedes, text and geometrical figures marry together. Papyri are one of the main media of Antiquity and a "happy" literary papyrus, as Netz calls it, always contains the same thing: column after column of text to be read left to right. The figures are subsumed into these slabs of text (which is why I say, not entirely seriously, that Netz now sees the text having primacy).

So my third reflection is on the thorough-going difference between the unidirectional content of a papryus book and the jump-in-anywhere nature of the first great infographic. The Great Stemma does not have any accompanying text: all its words are build directly into the drawing. It has no mandatory start or finish: you can read it from the right, or the left, or even upside-down.

From the point of view of traditional literary culture, the Great Stemma would have seemed to break every rule in the canon.

Leaping out of the Page: The Use of Diagram in Greek Mathematics. London: British Academy, 2013. https://www.youtube.com/watch?v=7hzzdLsTb5E&feature=youtube_gdata_player.
Netz, Reviel. Ludic Proof: Greek Mathematics and the Alexandrian Aesthetic. 1st ed. Cambridge; New York: Cambridge University Press, 2009.
The Shaping of Deduction in Greek Mathematics: A Study in Cognitive History. Ideas in Context 51. Cambridge University Press, 1999.
Netz, Reviel, and William Noel. The Archimedes Codex. Revealing the Secrets of the World’s Greatest Palimpsest. London. Orion, 2007.

2014-11-02

Floreffe Bible now Online

The only way that we know that a graphic artist – or studio – in the fifth century CE drew a magnificent chart of biblical history and genealogy is from the later copying of that work onto parchment manuscripts during the medieval period. Fortunately, 24 such copies are still in existence.

Since I began studying the Great Stemma in 2009, there have been constant advances by museums, archives and libraries in bringing digital versions of this chart online. The British Library has now added the Bible of Floreffe Abbey to its collection of over 1,000 online manuscript digitizations. This must have happened rather quietly, as I have not seen it mentioned in the BL blog yet.

For the study of the Great Stemma, this is a rather important development. Until now, just seven of the manuscripts had been issued online, but nowhere on the internet could one study a peculiar evolution of the Great Stemma into what I call the "School Stemma," a revision of the graphic to bring it in line with orthodox, 12th-century doctrine about the ancestry of Christ.

The Floreffe Bible contains a beautifully coloured version of the School Stemma, with exquisite script that faithfully reproduces all the defects and nonsenses of whatever text it was modelled on. Check it out and page through its glories.

We do not yet know where the revision took place. I call it the School Stemma because I guess that it was "cleaned up" so it could be safely taught to impressionable young monks. Charts containing this version also come from Parc (now Belgium), Foigny in France and Burgos in Spain. The chart may also have been copied at Arnstein, Germany as well.

As a result, I have revised my table of Great Stemma manuscripts, doing a good deal of re-arranging to make the layout more user friendly. It will be clear after consulting the table that not only are eight of the 24 manuscripts now available online and readable in their entirety, but that these neatly cover six of the seven recensions (alpha, beta, delta, gamma, epsilon and School). Additionally, the sigma recension is online at the BNF, though the resolution there is too low for a visitor to read the script.

As a result, the Great Stemma can now be seen on the internet in its complete range of forms. By the greatest of good luck, the four manuscripts which offer the best evidence about the fifth-century ur-form (Plutei, Roda, San Millàn and one of the School group) are now all accessible. One could hardly wish for more.

My table tabulates all these online witnesses. The amber-gold squares in the table mark all the high-resolution images where you can click through to each archive's website. The paler yellow squares mark low-resolution images and snippets.

Happy clicking.

2014-10-27

Digital Reviews

Getting manuscript documents (and early printings) online in a form that is electronically readable is only half the battle: keeping it accessible and raising standards to the point that multiple projects could be united into a permanent digital library are the follow-ups we tend to forget.

Germany's Institut für Dokumentologie und Editorik is now publishing RIDE, a review of digital editions (issue one is out and issue two is coming up).

The checklist that reviewers use to assess accessibility and inter-operability is an interesting one. There's a chart of technical issues noted so far and here are some of their criteria:
  • Which licenses are used to determine the copyright of the material published by the projects?
  • Do projects adhere to a standard data model like TEI?
  • Is the raw data accessible, either for the individual parts of the edition or as a whole?
  • Which interfaces do the projects support to allow reuse of their data?
The questionnaire picks out other more general issues like:
  • Is search with wildcards available?
  • Are the methods employed in the project explicitly documented? 
All points that I'll have to figure out. I document my methods, but this documentation is scattered, with part of it appearing here in the blog. It should all be combined in one place.

Like most digital publishers, I am keen to put material in the public domain, but am cautious about the form of declaration. I'm sceptical about TEI. I do want to put the raw data out there, but we are all aware that the rate of uptake is going to be minute: at the most, three or four researchers per century are likely to want to use my MS Excel tables. The interfaces question refers to things like an API or Representational State Transfer principles, which are only vaguely in the mind of independent researchers running single-person projects.

2014-09-16

Line-by-line Poetry

A  nice piece in The New York Times this week points to the increasing traction that electronic formatting for poetry is gaining at small presses.

The report, Line by Line, E-Books Turn Poet-Friendly by Alexandra Alter, explains the macro-typography problem with poetry: lines, not paragraphs, are the fundamental unit. If electronic publishers of poetry attend to the formatting of every line, they can create beautiful presentations.

But in early e-books (and needless to say, in most web presentations of poetry), publishers rushed out poems without appropriate formatting and the results were awful. The Text Encoding Initiative and other projects developed enormously complex standards for historic poetry, but these were not appropriate for contemporary poets, because the rules were so ridiculously complicated to apply.

I worked through these issues 12 years ago in developing the poetry page of my Macro-Typography site. [Note: this shows broken at the moment in Firefox, but loads fine in IE.] Sadly, my efforts to devise a few simply standard units were too far ahead of the curve. At that time, poets did not understand the internet, and web designers did not understand poets.

I argued during the the introduction of HTML 5 that poetry should be taken into account. It wasn't.

Perhaps I was too impatient. It took centuries to establish the conventions used by typesetters in setting up poetry for printed books. It was bound to take a few years for web typographers to be understood.

The solution I propounded in 2002 is now gradually establishing itself. Poetry must be formatted in units of lines so that any spillovers when the screen is too narrow always appear indented, not flush left. The range of formats for lines must include various initial indentations, along with special cases such as the caesura and the flush-right line. Stanzas must be appropriately spaced from one another.

Alexandra Alter describes how eBook Architects, a company in Austin, Texas, is now developing virtuous web-display routines as it formats poetry professionally:
[T]he text was hand-coded and marked up semantically, so that the formal elements were tagged as lines, stanzas or deliberate indentations. When a line runs over because the screen is too small or the font is too big, it is indented on the line below — a convention that’s been observed in print for centuries.
Bravo. It's good that poets are gradually realizing that poetry can look beautiful on an electronic screen, and that Aldine blackletter on creamy white paper is not the sole way to publish. It's a pity it is still so hard to set up electronic publication that specialist companies have to be engaged to provide bespoke solutions. It ought to be part of standard HTML.

2014-05-04

Translating Nonsense

How does one translate nonsense that has arisen when scribes fall into a doze and begin to write drivel? There is a celebrated passage in the Great Stemma which I have been looking at today which marks a low point in scriptorial quality control. It derives from the Septuagint version of the Book of Job. I have tried to collate it in six parallel lines of continuous text below:

LXX  οι δε ελθοντες προς αυτον φιλοι  
VL   Nam qui venerunt ad eum amici  eius hii fuerunt: 
del  Nam qui venerunt ad eum amichi eius hic fuerunt: Sophar Israel
gam  Nam qui venerunt ad eum amici  eius hii fuerunt: Sofar filius
alf  Nam qui venerunt ad eum amici  eius hii fuerunt: Sofar filius
bet  Nam qui venerunt ad eum amihi  eius hii fuerunt: Sofar filius

ελιφας των ησαυ υιων         θαιμανων βασιλευς 
Elifaz, de filiis Esau       Themanorum rex;
Elifaz, de filiis Esau       et Temanorum uxor filiis 
Elifaz, de filiis Esau       et Themanorum uxor et filiis
Elifas, de filiis Esau       et Temanorum uxor filiis 
Elifaz                       et Hemanorum uxor filiis

βαλδαδ ο σαυχαιων τυραννος
Baldad sauceorum tirannus;
Ammon                                 qui fuit Ar-auce tiranni
Ballenon filii Ballac  et filiis Amon qui fuit Cobar-auce tiramni 
Balenon filii Ballac uxor filiis Amon qui fuit Chobar-aucce tiranni 
Balexion             uxor filiis Amon qui fuit Bar-auce tiranni

σωφαρ ο μιναιων βασιλευς
Sophar mineorum rex
et Themas de filiis Elifaz dux Idumee
et Heman     filius Elifaz dux Idumee
et Temans    filius Elifaz dux Idumee
et Themas    filius Elifaz dux Idumee

The six texts are: the Septuagint, the Vetus Latina and the Great Stemma's delta, gamma, alpha and beta recensions. This text does not appear in modern bibles, as it has never been incorporated into the Masoretic or Vulgate versions, and must therefore be referenced as LXX Job 42:17e.

The context is that Job sits on his dung-heap, owning just a broken bit of pottery to wipe the discharge from his ulcers, lamenting and asking God what's up.

A trio of guys -- Eliphaz of Teman (a town in Edom); Bildad of Shuah; and Zophar of Naamath -- show up and make the snarky kind of unhelpful remarks that guys always make, to the effect that it might just be Job's own fault. We men tend to brutality.

Rabbis must have often been asked what rank these guys had. One suspects the three names are simply some kind of impenetrable humour, but a midrash which claims they are monarchs is recorded in the Septuagint.

Here is how the New English Translation of the Septuagint renders the midrash: Now the friends who came to him were: Eliphaz of the sons of Esau, king of the Thaimanites; Baldad, the tyrant of the Sauchites; Sophar, the king of the Minites.

The repetitious, inflated mess this turns into in the Great Stemma was first spotted by the great Yolanta Zaluska. I want to mention the passage in my planned book. It's a dilemma deciding which nonsense version to translate and how to render it. Here's one of the more absurd possibilities:
  • Sofar the son of Elifaz of the tribe of Esau
  • the wife of the Thaimanites
  • the sons of Balenon's son Ballac
  • Amon who was tyrant of Chobar-auce
  • Themas son of Elifaz, duke of Edom
That inflates the list to more than five visitors in the scribal version. Of course, the nonsense could be understood as meaning that a throng of sons came to visit Job too. In which case, the country lane outside Job's country estate in Uz would have had a parking problem with chariots, camels and donkey-carts jamming all available roadside space.

2014-04-30

The Old Latin

Readers of this blog will recall that the names of many biblical characters in the Great Stemma are spelled in the fashion of the earliest Latin bibles, rather than in the forms prescribed by Jerome of Stridon in his Vulgate retranslation of the fifth century. I noted in my Studia Patristica paper that the original names in the diagram would seem to have been immune to Jerome's influence (although one must be cautious in guessing why) and that this was especially emphasized by Yolanta Zaluska:
Among the examples she isolates is Chor (Gen. 36:22), from the personal name Chorri in the Septuagint and Vetus Latina, which Jerome had amended to Horrei. Zaluska also detected the Vetus Latina name of a Horrite chief – Ucan – which Jerome had suppressed, as well as another name, Chat, which had arisen through scribal error in the Great Stemma’s transmission. Both names continued to be reproduced in the Great Stemma into the high medieval period although they were no longer current in the Genesis narrative.
I have undertaken various attempts to quantify just how many of the 540 names are present in the diagram in their Vetus Latina forms. Two years ago I wrote a long blog post on a statistical project to distinguish the variants and seek salient points in the data. The effort was rather inconclusive, partly because my methods were rather basic ones.

A recent article by Philip Burton (all references below) points to a possible way forward. Certain ordinary Greek words can be and were translated in alternative ways in Latin: Burton compares how those words are handled by Vetus Latina translators in seven cases and shows the results in a "distance chart of agreements" (page 187). He mentions that he gave a paper in 2008 offering hundreds of such comparisons. It might be useful for a future scholar to employ Burton's method to compare names in the Great Stemma manuscript.

Many other difficulties remain. Not only is it hard to distinguish Vetus Latina variants from bizarre scribal errors. We also simply don't know in many cases what the Vetus Latina text actually said.

Very little of 1 Kings and 2 Kings, two books that are of central importance to the Great Stemma, even survive. Natalio Fernández Marcos, the great Spanish biblical scholar, estimates in Scribes and Translators (page 44) that 90 per cent of what we know of the Vetus Latina version of 1-2 Kings derives from just three very fragmentary sources - the Opuscula of Lucifer of Cagliari, a Naples palimpsest and marginal glosses to several Spanish bibles.

In addition, the manuscript tradition for more substantial sections of the Vetus Latina is not entirely reliable either. As Burton has pointed out, even the question of whether there was just one translation or many remains a moot point in current scholarship. Even the tradition of the Vulgate, often thought of as fixed, retains uncertainties in its transmission. So it will never be possible to take all 540 of the Great Stemma names and fix what proportion have Vetus Latina origins beyond any doubt.

In the past two weeks I have returned to the issue because I need to summarize this matter concisely in my planned book. Among the results: on my website, there is a new tabulation of Vetus Latina reconstructions of the 14 names of women in 1-2 Kings where each name is traced back to the Septuagint. The extreme corruption in the graphic organization of these names in the Great Stemma makes the passage an especially good pointer to the primitive Vetus Latina forms.

In addition, I have done a little more checking of how five of the most salient changes fared in the Vulgate tradition, checking the names usage in the 1926-1995 Biblia Sacra iuxta Latinam Vulgatam versionem where the manuscripts are compared. I was surprised to find how often the Vetus Latina forms crept back into the text, particularly in Spain.


Septuagint Vetus Latina Vulgate Ref. Changes Notes
μαλελεηλ Maleleel Malalehel Gen 5:12, 5:15 e to a, insert h Printed Vulgates 1530-1592 employed the VL form
φαλεκ Phalech Faleg Gen 10:25 ch (k) to g Faleg in largest number of witnesses, but Falech, Falec, Falleg, Phaleg, Phaleg (first printed Vulgates), Phalech also recorded
χορρι Chorri Horrei Gen 36:22 from ch (χ) to h Horrei in largest number of witnesses, but Horraei, Orrei, Hori (first printed Vulgates), Horri and Orri also recorded
ζουκαμ Zucam Zevan Gen 36:27 middle consonant from c (g) to v (b) Zeuan in largest number of witnesses, but Zeban, Zeuam, Zefan, Zephan, Zenan, Zauan (first printed Vulgates), Euan also recorded
βηρσαβεε Bersabee Bethsabee 2Sam 11:3, 12:24 second consonant from r to th Bethsabee in most witnesses, but Bethsabȩȩ, Bethsabe, Betsabee, Betsabeȩ, Bersabee (6 witnesses), Bersabeȩ, Bersabȩȩ and Bersabe also recorded
The last of these cases is particularly interesting. Jerome preserved Bersabee as the name of the city in the southern desert in Genesis 22 (see a discussion by Rico), but there can be no doubt whatever that he desired to suppress Bersabee as the name of Solomon's mother and amend this to Bethsabee. Yet it re-infected "his" text soon enough. The Bersabee witnesses are: Madrid RAH 2, Burgo 2, Milan Biblioteca Ambrosiana B48, Vat. lat. 10510 (2nd Bovino Bible), Paris 15467, Mazarine 5 (all Bersabee), Madrid Mus. Arch. 485 (Bersabeȩ) and Madrid Bib. Nat. A2 (Bersabȩȩ).

Among the other instances, it is notable that Malaleel even crept back into early printings of the Vulgate before it was expunged.

This exercise has tested and confirmed the hypothesis that most Vetus Latina forms of the biblical names turn up in the Great Stemma, sometimes in quite bizarre transformations. For Zucam, the variety of readings in the diagram include Zugat and Zucat. It remains difficult to imagine any process by which the Great Stemma could have been drafted using a mixed supply of Vetus Latina and Vulgate forms of the names. The hypothesis that it was drawn up with the aid of a Vetus Latina bible and was fitfully altered later to match Jerome's spellings is the only one sustainable in the court of common sense.

Burton, Philip. “The Latin Version of the New Testament.” In The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, edited by Bart D. Ehrman and Michael W. Holmes. 2nd ed. Brill, 2012. Google Books
Fernández Marcos, Natalio. Scribes and Translators: Septuagint and Old Latin in the Books of Kings. Vetus Testamentum / Supplements. Brill, 1994. Google Books
Moreno Hernández, Antonio. Las glosas marginales de Vetus Latina en las Biblias Vulgatas Españolas: 1-2 Reyes. Textos y estudios “Cardenal Cisneros.” Madrid: Editorial CSIC, 1992. Google Books
Rico, Christophe. “La Traduction Du Sens Littéral Chez Saint Jérôme.” In Le Sens Littéral des Ecritures, edited by Olivier-Thomas Venard, 171–218. Collection Lectio Divina, Hors Série. Paris: Editions du Cerf, 2009. Academia.edu.

2014-03-21

Vatican Library to be digitized ... again

These stories about the Vatican Library - the Biblioteca Apostolica Vaticana or BAV - digitizing its manuscript collection are starting to get repetitive.

Repetitive: because we have been here before, and most Rome correspondents never dig into what must be some rather messy background. Readers of this blog will recall a post four years ago about an impending digitization. At that time, a British company named Autonomy Systems was named as contractor. Is this the company which Hewlett Packard took over in 2011 and got into a fight with?

I don't know. I do know I still don't see the promised manuscript library. And I don't see any explanation by the BAV, though Cesare Pasini's rather silly encomium of that project is still online.

Chief prefect Pasini is still at it. He was referring grandiloquently in December to the "vast project" funded by Leonard Polonsky to digitize Greek and Hebrew items, but this too has moved at a snail's pace. The website under the auspices of Oxford University shows only a handful of Greek manuscripts  have actually been digitized so far.

At the Vatican, the digital library as of now is still a very meagre affair: if it were not for the Palatini latini files donated by the German University of Heidelberg, there would only be 24 codices available on the site. A sobering account in the National Catholic Reporter says the BAV is in fact 23 million dollars short of funding, adding that outside groups have helped the BAV digitize 6,800 codices, but it seems fewer than 5 per cent are on its website.

Yesterday, March 20, there seems to have been a news conference at the Vatican to announce a new contractor, NTT Data of Tokyo. The press release is short on specifics. Pasini says the Japanese company will "support the further improvement of the project". Not a word about the four lost years, or why the projects have stumbled.

Most of the news reports I am seeing miss all these subtleties. And trust a Daily Telegraph sub to make the shoddy reporting even worse: the headline "Vatican library plans to digitise 82,000 of its most valuable manuscripts" is plain misleading. The press release clearly says NTT Data will only be working on 3,000 items over the next four years.

Of one thing we can be sure. If the day ever comes when the Vatican has digitized all 82,000 of its codices, the Daily Telegraph will have long ceased to exist.