Reading a single chapter cannot substitute for a course in papyrology, but my hope is that after reading this essay you will be able to view an ancient Greek book with a considerably more informed eye, and will have come to understand many fundamentals about how such books were written and put together in antiquity, and also something about how they were used. In the Graeco-Roman world, the history of books begins with the papyrus roll, and thus there we will start.
1. The Bookroll
We are fortunate that well over 3000 papyrus bookrolls1 survive in fragmentary form from antiquity, almost all from the sands of Egypt. We are even more fortunate that quite a few survive to some significant extent, that is, such that even a non-specialist can get a sense of the original bookroll by looking at what survives. In what follows I will attempt a magisterial tour in some depth for three substantial examples, while looking at a couple of others in more glancing fashion along the way. Our goal, in the first long part of this chapter, is to train ourselves to see more clearly what would have made some of these examples more ‘finely made’ than others and thus presumably more handsome to the ancient eye, and also to understand what it is that makes the third of our three illustrative examples look hardly like a bookroll at all.
The invocation in scare quotes of ‘finely made’ turns our focus at once to a salient feature of ancient bookrolls, that they are almost always made by artisans (whether in-house or trade) who are trained to the task. A finely made bookroll relies not simply on the shaping of the letterforms. Just so, to turn to an analogy, the merit of, say, a handcrafted wooden table relies only superficially on the appearance of wood grain and finish, important as those are. If we take apart a finely made table jointed with mortise and tenon, we expect to find that the angles will measure exactly 90.0 degrees, that the table is therefore exactly level, with legs that have not the slightest rock; also, that the tenons fit squarely and snugly into the respective mortises such that glue seems almost superfluous. Details like this will oblige skilled woodworkers to agree that a table is finely made, regardless of what they think of the design and style. Turning back to the ancient book, our wish then is to learn in some detail how the bookroll was constructed, so as to see it as might appear to an artisan, someone who has studied the craft and knows the particulars; and our goal will be to come to see what features define a bookroll as a craftsman's product that is made just so.
We begin our tour with the Arden Hyperides roll (Fig. 1), which dates to the late first or early second century of the common era, and was preserved largely intact. It contains several speeches by one of the canonical Attic orators, Hyperides, an important author whose works did not survive — that is, until Joseph Arden traveled to Egyptian Thebes and in January 1847 bought this bookroll from locals who had found it in a tomb. The Arden papyrus (as we will call it) is a beautiful bookroll, not calligraphic in its writing, but very well made. When we look at the papyrus, what do we see? First, a few preliminaries to be sure we are properly grounded. We are looking at an image of part of a papyrus roll. Papyrus, from which we get our word ‘paper,’ is a product made from a papyrus plant, which is indeed in many respects much like a sturdy, tough, pliant paper.3 It is manufactured in sheets, but is not sold as such. Instead, at the point of manufacture the sheets are glued together so that it can be sold as pre-made rolls of, typically, 20 sheets, though we also hear of manufactured rolls of 50 and even 70 sheets. The product that a consumer buys, then, is not a ream of paper, but something more like a roll of shelf paper, with the manufactured joins cleverly constructed so as to minimize the bump that the pen has to go over when writing. Creating a bookroll is, then, simple: you write out the text until either you reach the end of the text, at which point you cut off the remainder of the roll and are done; or you write until you reach the end of the pre-manufactured blank roll, in which case you pick up another blank roll, carefully glue it to the end of the one you're writing, and continue writing until you reach the end. Fig. 1 shows but 7 columns of a roll that contained dozens: the extant part of the roll, which is close to complete, has 97 columns and is a roll 7 meters, that is, over 20 feet long. As you see, there is some damage to the roll, particularly along the bottom, the result of rot or ants or worms or the other vicissitudes of a fragile survival that comes from burial in desert sands for 2000 years. This roll in fact is in excellent preservation: think about what it would be like if you dumped a book in the trash and let it fester for hundreds of years; even in the dry sands of Egypt it would not be a pretty sight, and most of the papyrus bookrolls we have come in fact from ancient rubbish heaps, and are much more scrappy and damaged than this example.
In Fig. 2, the Arden papyrus is overlaid onto a model, so that you can better imagine the complete roll, with its parts labeled. Most of the terms I will use are self-evident, but I point out the intercolumn, which is the term for the space between columns; I will reserve ‘margin’ for the upper and lower margins alone. I also point out the sheet, what Classicists call the kollêma, whose joins are only just visible on the roll itself, but which are exaggerated on the model so that you can see them clearly. The sheets in this papyrus are fairly wide (26 cm or so), such that a join happens only every three columns or so; that is a sign of a high-quality papyrus roll, since the wide sheet makes for fewer bumps as the pen moves along; and, since joins become vulnerable with use, the fewer joins also make for less opportunity for damage over time. Note that the columns of writing march along without any consideration of the sheet joins; the scribe is perfectly at ease with a column of writing sitting right on top of a join.
In Fig. 4 I simplify the image by exposing the bare writing and reducing the focus to three columns, with, at left, a modern edition of Hyperides for the leftmost of the three. Even those without Greek should be struck by several things at once. The columns are very narrow, rather like a printed newspaper today. We have already remarked on the regularity of the column along the left and the rough justification along the right. Notice how the scribe effects this: at the bit highlighted in pink, for instance, he makes the ‘Ο’ of ΟΝ (Greek omicron plus nu) tiny to squeeze in the end of the syllable without projecting too far into the intercolumn; or at the point highlighted in green he adds a mark that looks like a right angle bracket (>) to fill out the line of text. We have also mentioned the way that the tops and the bottoms of the columns line up, but again notice how that works: in the leftmost column, you can see that the scribe had to scrunch up the vertical spacing between the last two lines (highlighted in blue), so as to get the bottom to align with the other columns. That's an important detail, since it also tells you that the scribe is doing this freehand; that is, the scribe has not ruled each line in advance in that laborious fashion familiar from medieval parchment codices. You might also notice the regularity of the writing itself, even if you don't know the alphabet; though not calligraphic writing, it is handsome, and you can see the ease and regularity with which this scribe writes; remember that this scribe is writing dozens of such columns (97 are extant), and yet there is no sign as you move along the columns of a hand getting tired or sloppy. That too is an important clue, as it tells you immediately that this bookroll is made by someone trained to the task, and is not a casual, private production.
The other thing that surely you have noticed is the lack of spaces between words. This lack of word division is a famous feature of ancient Greek bookrolls, which goes under the technical term scriptio continua, and it is worth pausing for a moment to look at some details. Our first impression is that the letters march along without any differentiation whatsoever. Certainly, even if you don't read Greek, you can see in the modern text lots of word division, and some other marks too — accents, apostrophe, punctuation. None of that is present in the ancient text. It's as if the column looked like this:
Nowaninterestingexperimentis
toseehowhardoreasyitisforyou
ashighfunctioningliteratesto
readthisundividedEnglishtext
TakeamomentandgiveitatryYou
ArenottrainedtoitbutImguessing
thatyouwillagreethattraining
yourselftoreadthiswouldnotbe
impossiblenorevenallthathard
Itwouldhowevercertainlyslow
youdownandthatsimportanttoo
Now an interesting experiment is to see how hard or easy it is for you, as high-functioning literates, to read this undivided English text. Take a moment and give it a try. You're not trained to it, but I'm guessing that most will agree that training yourself to read this would not be impossible, nor even all that hard. It would, however, certainly slow you down, and that's important too. But, getting back to the Greek, on closer inspection, there does seem to be some articulation of the Greek text. Do you see it? Certain spaces, for example, seem perhaps deliberate (Fig. 5, pink); and undoubtedly deliberate are the small horizontal marks underneath certain lines at the margin (blue highlight). This horizontal mark is called a paragraphus. We get our word ‘paragraph’ from that, but it doesn't mark a paragraph exactly. Rather, in prose texts like this one, it marks a full stop, a point where a sentence is really over and done with, when it's time for a lector (that is, someone reading performatively) to take a full breath pause. That is, the combination of paragraphus and space tells the lector when to pause. In a modern edition, that will usually match up with a period (as it does here – again, blue highlight in the modern edition). The other points of articulation (here spaces without paragraphus, highlighted in pink) mark lesser pauses, a shorter breath for the lector, and these will match up with certain types of comma or colon in modern editions (see pink in modern edition). Note however that not every point of modern punctuation has a presence in the ancient text (green highlight): the scribe seems to want the bare march of the letters to have as little interference as possible.
The column of writing, then, is readable, but reading it might involve some hesitation or backtracking. There is little concern with making the job easy for someone reading the text aloud to a group (as was common in antiquity). The text signaled when to take a short breath pause, when to take a longer one. The paragraphus did make it easy for a lector, when taking that longer pause, to look up from the text at his audience and when looking back to the roll to relocate his place in the text (he just had to remember, third paragraphus down, for example). Beyond that, it was the reader's job to pull the meaning out the text, to decide on how to add intonation and phrasing to what was written, to read fluidly and with understanding.
Now a lot of silly things have been said and written about ancient texts written in scriptio continua. You may have heard, for instance, the old canard about the ancient Greeks being unable to read silently, because scriptio continua supposedly required reading aloud in order to make sense of the text; or the notion that it was only once spaces were added back to Latin texts in the late Middle Ages that the intellectual revolution we associate with the Renaissance became possible. This is, as said, silly: there is unambiguous evidence, for example, that ancient scholars read silently when they wanted to concentrate; and intellectualism does not seem to have been hindered by lack of word spaces in the time of classical Greece or imperial Rome, when in fact great technical and scientific advances went hand in hand with the obvious philosophical and literary flowerings. I encourage you, rather, to look at the situation more from the other way around. The Greeks and Romans knew how to use spaces to help define words or phrases. We see them used when teachers are teaching school children about reading and writing; we see them deployed to mark phrases in documentary papyri. The Greeks and Romans also knew of elaborated punctuation and other marks to assist reading: we have examples of texts just like our bookroll here, where however readers have added many additional marks. But, and this is the crucial point, if a trained scribe picks up a copy with readers’ marks, he will not think it his job to copy these added marks — rather, he will produce a copy that is clean, bare, devoid of any but the most basic marks, like the paragraphus. Moreover —and this is an amazing tale! — the Romans at first had bookrolls with dots between the words or phrases, that is, with rough word division, and it was only in the period when they fell under Greek cultural influence that they made the step of removing all the word divisions from their bookrolls. Think about this for a moment: can you imagine what might prompt us to remove all the spaces between words in our texts? Of course you cannot. That very fact suggests that there is something quite different in the way that the Greeks and Romans used literary texts, something or some things distinct about their system of reading. For a thousand years, the ancient Greeks, and later the Romans, saw no need to alter the basic look and feel of the bookroll as we have now come to understand it. This is not the time or place to go into details of this system of reading— I must there refer you instead to my 2010 book on the subject, Readers and Reading Culture in the High Empire — but as a crude generalization, we can say that the system of reading assumed a reader for literary texts who was trained to the task, and often one who already knew the text; and whose job was to bring out his or her understanding of the meaning of the text, often by reading aloud in groups.
As for the bookroll, it was designed for clarity and beauty, but not for ease of use, much less for mass readership. Strict functionality was not a primary goal, nor was thrift. Upper and lower margins tended to be quite wide, as we have seen, far wider than needed on any practical ground. Typical bookrolls thus used only 40–70% of the available writing surface on the front of the roll, and left the back blank. The cost of copying was substantial, and thus even a less well-crafted bookroll, if not fully elegant, was not cheap: a second-class bookroll cost the buyer twice as much for the copying as did a routine document (which would be copied in a rapid, semi-cursive hand). As I wrote back in 2010, ‘The bookroll seems, then, an egregiously elite product intended in its stark beauty and difficulty of access to instantiate what it is to be educated.’5
Let's pause then to summarize what we have established so far. (1) The scribe was a professional craftsman, working to create a look and feel that was well defined and traditional. (2) The product was traditional and stable, symbiotic with the system of reading. (3) The physical roll had iconographic importance as an expression of literate elitism.
Before leaving the bookroll behind I want, as promised, to look at two further examples that are not such fine bookrolls. Before looking at them, I want to stress that our exemplary bookroll, the Arden papyrus, is not simply illustrative but normative: the huge majority of surviving bookrolls share the features I have pointed to. But there are naturally outliers, both those considerably more calligraphic, and, as with our next two examples, those that show the downside of the range of product we would expect from something so simply hand produced. Such examples can be interesting as a point of contrast, and they also function as a test: these will tell you whether you have learned to look upon these images of the ancient Greek book with different eyes.
The first is a papyrus bookroll from the site of Oxyrhynchus, containing the fifth book of Homer's Iliad and dating to the third century of the Common Era (Fig. 6). The hand to an expert eye appears very much a trained one, writing easily and comfortably in a script known as the Severe Style, but the bookroll itself is a noticeably hasty or casual production. The book is written on the back of a document, and you can perhaps see that the back is coarser in its surface.6 But the most striking feature is that, unlike other Roman-era bookrolls, the layout is not really quite regular. The measurement from the top left of one column to the next is twice the variability that we find elsewhere, a plus or minus of 4 mm, admittedly still not all that much. But very noticeable is the irregular angle of the column tilt, such that the column-to-column measurement varies as much as 16 mm as you move down the column (Fig. 6, arrows). For this reason the layout itself has a far less regular look and feel. You'll also notice in this manuscript that there are a large number of accents and other marks added to the text by a different pen (see Fig. 6, inset). We have quite a few manuscripts, mostly Homeric, with these kinds of markings, and the scholarly consensus is that these markings were made by students practicing accentuation and/or pronunciation. This, in short, is a cheaper bookroll made, apparently, for use in school.
Our final illustrative bookroll (Fig. 7) contains the text of Aristotle's Constitution of Athens, one of our most important sources on the history of the Athenian democracy, and unknown except for small fragments before its appearance on this papyrus, discovered in 1890. This is not a well-made bookroll, and my hope is that now, even after this brief training, you will be able to see (as I said before) ‘what it is that makes this last example look hardly a bookroll at all.’ It too is written on reused papyrus, on the back of a document. But it has very unusual and striking features. Less striking to the non-specialist is that it is written in multiple hands, and that the hands grow evidently weary after only a short amount of copying. But even a Greekless layman may be able to see that the script is not at all the clear capitals we have seen elsewhere, but sloppy and cursive (and with abbreviations), as you see in the inset. In any case, you should by now be able to see what about the layout makes this such an oddball papyrus that in some ways it exemplifies what a bookroll is not. Upper and lower margins are unusually narrow. Some columns are very wide, much wider than any other examples of prose bookrolls. The columns vary wildly in width; the leftmost column pictured at the top in Fig. 7 is more than twice the width of the columns following. The horizontal line of text wavers considerably. The alignment at the left edge of the column is also uneven. The vertical line of the ‘justified’ right edge of the column sometimes does not run in parallel with the vertical line at the left; thus the intercolumn narrows as it proceeds downwards, and both column and intercolumn accordingly look more trapezoidal than rectangular (see highlight for the trapezoidal intercolumns).
With, now, a firm impression of what an ancient bookroll looked like, it is time to turn to the codex.
2. The Codex
I do not plan the same kind of deep dive into the details of codex production as I charted for the bookroll, since other chapters in this volume offer considerable exposure to the codex in its many manifestations. Instead, I will confine my remarks so as to focus on that very interesting period of transition, as ancient literary texts in the Graeco-Roman world move from bookroll to codex. (By ‘literary’ texts I intend a broad definition, that includes poetry of course but also history, philosophy, medical treatises, and, as we move into the Common Era, religious texts as well.)
Let us first get before us a clear image of what a first-rate example of an ancient codex might look like. The book pictured in Fig. 8 is called the Sinaiticus, from the location of the monastery in which it was found. It is an egregiously deluxe parchment codex, a pandect Bible containing the entire Greek Old Testament (the Septuagint) and New Testament, dating to the mid-fourth century CE. The writing material is made from calf and lamb, and it is estimated that 360 animals were required to produce it. Pictures like this of the Sinaiticus (and its sister, the Vaticanus, at lower right) have informed, or rather misinformed, our view of what an ancient codex generally looked like. The huge majority of early codices, as we will see, were much more plain, and early ones were more likely made of papyrus than of calf and lamb. We will return to how best to view the (rare) sort of codex represented by the Sinaiticus at the end.
As a preliminary, I will pause to emphasize that the transition from bookroll to codex was by no means inevitable. Since the codex is our modern book form, we see its advantages too readily and starkly. The bookroll seems to most of us a clumsy device. It holds less than half of what a codex does. It is prone to squashing when stacked or packed, so you have to store it in an upright container when you travel (typically using a device known as the capsa, which looks much like an ice bucket). With a bookroll, you can't turn to a particular page or relocate your place in it very easily. You have to roll the thing back up again when you're finished with the book. Moreover, it is clumsy to use, seemingly requiring two hands to manipulate as you read, and when dropped it has a tendency to unroll and make a considerable mess. A story in the younger Pliny tells us of an elderly friend who dropped a heavy bookroll and bending over to gather up the quickly unraveling roll he lost his footing and broke his hip— and then later died of complications!7 But most of the supposed advantages are in my view exaggerated, and do not take adequate account of the ancient reading system. I've spent a lot of time experimentally reading from bookrolls, and it in fact quite comfortable. You can hold the whole in one hand if it's a small roll. If it is larger, you can sit and let the part you've read spill to the floor as you go along, thus also using but the one hand. Moreover, you can see several columns at once, so you don't need to advance the roll nearly so often as turning the page of a book. True, it is a bother to pack up, or to gather and roll back to the beginning. The difference in capacity is more significant, but the expense of the material was not much of an issue (since bookrolls were largely owned by people of substance), and from the point of view of the individual, one can easily see why one might prefer to read Vergil's Aeneid one book at a time from twelve individual rolls, in the time-honored way, rather than dealing with a heavy codex containing the full 10,000 lines. In just this way, I prefer to read Shakespeare from my set of the Yale Shakespeare with its handy small-format volumes, one per play, rather than from the Riverside Shakespeare, with its 1900 pages and weight of 2.5 kilograms (5.6 pounds). Bookrolls held the throne for a millenium, and it was not by chance.
The typical story of the transition goes like this. The codex first appears in the literary record in the late first century AD. The locus classicus is the Apophoreta of the Roman poet Martial, in which he mentions codices as special Saturnalia presents: a codex with all of Vergil is an example. After Martial is a gap of several generations among our literary sources; meanwhile, the survivals of codices from the late first or early second century are almost nonexistent.8 The best candidate for so early a codex is a Latin parchment that appears to be from the late first or early second century; that such an early codex is one of our few Latin examples may (or may not) be significant. In any case, we don't know that Egyptian practice accorded with what was happening in Rome, but from Egypt we can gather the following approximate data.9
If we focus on the second to the fourth century we see that, once the transition gets underway, it becomes a reality in roughly 200 years. It is very probable, though we cannot prove it, that the scribal shops of the times changed considerably to adapt to the new codex format; and that monasteries and other Church traditions became involved in the latter part of this period is certain.
Since codices, as many will know, are strongly associated with parchment, it was thought for a surprisingly long time that the first codices were parchment. Surprisingly, I say, because the numbers long since have suggested a different conclusion.
As you see, the earliest codices tend towards papyrus. The survivals, once again, are from Egypt, and that may well influence the results, since papyrus is such a common product of Egypt, including for the export trade; on the other hand, there are, as the chart shows, plenty of parchment codices that survive from Egypt in the fourth, fifth, and sixth centuries, when documents continue to be written on papyrus and papyrus manufacture continues to flourish.10 Most scholars these days are non-committal on the question of the priority of papyrus or parchment for early codices. One bit of evidence beyond the chart of surviving codices may help tilt the argument: unnoticed in the scholarly discussion is a telling passage in Jerome.11 Jerome is commenting on the famous Library of Pamphilus in Caesarea, which stemmed from the collecting and scholarly activities of the circles about Origen and, later, Eusebius, in third- and early fourth-century Caesarea. Many of these books, Hebrew as well as Greek, were codices. Jerome interestingly remarks that the famous library had fallen into considerable decay after Eusebius's death, and he mentions that the contents are being systematically copied onto parchment codices. That is, in Jerome's time (the late fourth century), the notion seems to be that one takes papyrus codices, subject to rapid decay, and replaces them with codices of sturdy, durable animal hide, that product we call parchment.12 This, then, is an additional indicator that in fact papyrus codices were normative in the early period — that is, prior to the late fourth century — and that some time in that century the idea of the superior durability of parchment became more prominent, leading to the dominance of parchment that we see in codices from the fifth century and later.
By the end of the fourth century, in any case, most books were codices and many codices were parchment; and by the fifth century parchment codices are the majority. But there is a well-known, indeed famous, wrinkle to all this, which comes through clearly in Chart 3:
The overlap between Christian texts and codex format is manifest from the earliest period, as the chart makes clear; and for scriptural texts it is almost absolute. Of about 150 copies of New Testament and texts analogous to the New Testament (that is, the non-canonical Gospels, such as the Gospel of Thomas) only 5 from the early period are written as bookrolls. (There are a few Old Testament texts on rolls, and these are usually assumed to be the result of Jewish scribes or influence, since the Torah has always written in roll form, to this day; but even in that case there are only a few, and they are almost all very early.)
From here the story gets more complicated, and more contested. The notion of the ‘Christian preference for the codex’ —as demonstrated in the charts above— is often confused or conflated with narratives that are in fact quite different, particularly (1) that early codices were mostly Christian, or (2) that Christians were responsible for the introduction of the codex as a book form. That neither of these statements has good evidence to support it can be seen most succinctly from Chart 4(a) below, which makes clear that even in the fourth century almost half of codices are made up of non-Christian texts.13 In the earliest codices, from the late second and third centuries, non-Christian texts outnumber Christian by a ratio of almost 2 to 1.
The picture is complicated by another factor, that is, the types of text that survive. In the second century, only a tiny percentage of the texts are Christian; by the fourth, that total has skyrocketed to about 40% and by the fifth Christian texts are in the majority.14 (See Chart 4(b) above.) This too can be seen, then, as an important influence on the dramatic increase in Christian codices in the period after Constantine. Another factor is a strong rise in the use of parchment, since though papyrus is widely used for both bookroll and codex, parchment is almost exclusively used for codex manufacture: by the fourth century, fully a quarter of surviving witnesses are written on parchment rather than papyrus, even in Egypt. (See Chart 4(c) above, together with Chart 2.) By the fifth century, we can draw a plausible link, then, between the rise in Christian textual production, the preference for codex, and the consequent heavy use of parchment. Before that time, inferred linkages are far less clear.
We can summarize our conclusions so far as follows: (1) Christians used codices for scripture from the earliest date. (2) Many non-Christian texts were also written in codex form, however, so it is not true that early codices were mostly Christian, nor does it follow that Christians were responsible for the introduction of the codex form. (3) In the early period, many codices are written on papyrus. The preference for using parchment comes late (fifth century), as codices come to predominate alongside the greater percentage of Christian texts, and as the preference for parchment’s durability takes hold. For the early period, then, the relation between Christianity and ‘the rise of the codex’ remains murky. For late antiquity, we can do better. It seems a reasonable inference that the observed trio of changed preferences (Christian texts; codex; parchment) reflects a shift in book production from individual scribes to monastic and other Christian groups, as church institutions consolidate in the late fourth and fifth centuries.
The full story of the ‘Christian preference for the codex’ is, then, a complicated one, and it is harder than one might think to tease out the threads of the narrative — very much, in my view, still an area that needs more and better work. But we can with confidence return to that place where we began (Chart 3), that is, with the observation that Christian ‘scripture’ was pretty much always written in codex form. We will never know why, though speculation abounds. Once it was popular to argue that the codex was used since it allowed one book to hold the four Gospels — but our early codices in fact turn out to be individual Gospels and collections of Gospels come into the story only later. Another popular move has been to imagine an Ur-text written in codex form that then became iconographic for Christian belief: earlier arguments were in favor of this Ur-text being a proto-Gospel, or the first text of Mark. The problem there is that in fact our early survivals very much privilege other Gospels, especially John and Matthew. A popular theory at the moment makes a similar argument, but assumes the Pauline letters as the Ur-text: but this too fails to have evidence from the early codices themselves to support it. Most recently, it has been supposed that the codex was better for travel, which it was, and that early Christian ministries became associated with the miniature codex, which functioned then as a sort of emblem of their missionary travels; a pretty story, but again without firm evidence to support it.15
Normative were early codices that looked like Fig. 12, a reasonably well-preserved full sheet (two folios) from a codex of Luke. You can see that the original maker of this codex cut off a double-wide sheet of papyrus from a roll, wrote out columns that are much wider than those of any prose bookroll (about 16.5 cm as opposed to a normative range of 4.5-7.5 cm for a bookroll), and then folded the sheet over before gathering the other sheets of the quire and sewing them together. These wide columns give a very different look to the page. Even though only some of the margin survives, we can readily see that the column is designed to fit within and mimic the shape of the physical page.
Similarly, in Fig. 13, which shows a full sheet from an early copy of Pauline epistles, the columns are not shaped to anything resembling the idea of a bookroll, but are pragmatically designed to fill the page with enough margin to accommodate the binding. The same is true of early non-Christian texts. One gets the sense of a sort of pragmatism about these early codices, and an analysis of the hands shows an important difference from bookrolls. Codices are typically written in scripts that are informal, ‘workaday’ as Turner described them, ‘reformed documentary’ in the phrase Roberts used, but in any case fairly quickly written, more serviceable than beautiful.18 Only 15% of early codices —20 of the 131 we’ve processed in our early codices database so far— are written in hands that are calligraphic or pretentious; contrast that with the 30% of surviving bookrolls that are written slowly, calligraphically and/or pretentiously, with unmistakable effort at beauty.19 The movement from bookroll to codex may, then, have more to do with sheer pragmatism, a shift in the socio-economic position of the people who were using books, and a shift in their motivations for reading as well. Whether or how that links up to early Christian communities must remain speculative.
That then brings us back to where we began. The two beautiful codices, the Sinaiticus and Vaticanus pandect Bibles pictured at the front of this section (Fig. 8), with their unusual 4- and 3-column format, are no longer thought to demonstrate a development from the bookroll. Rather (and here I follow the lead of Grafton and Williams, who in turn are following a suggestion of Turner's), I think we can now better understand these books as gestures towards the bookroll, and in an important way.20 Both are products of patronage,21 that is, deluxe productions, in that post-Constantine period when Christianity began to hold sway, but when significant numbers of classical, non-Christian books continued to be written in the bookroll format with which we began. These codices mark, then, an iconographic appropriation of the old ways, the seizure of that bookroll beauty associated with Homer and Hyperides and elite culture generally, redeployed upon a codex page of parchment, and containing the stories of the Old and New Testament, the central texts for that new, Christian elite that came to flourish as the latter fourth and then fifth centuries run their course.
Further Reading
The bibliography is, of course, vast, but the following will provide entry points and the more recent works contain copious further bibliography. For more detail on papyrus and its manufacture, see both Chapter 00 in this volume and the classic study by Lewis (Lewis 1974). For close studies of bookrolls, the details of their manufacture, and the conventions of layout, see Turner 1978, Turner/Parsons 1987 (introduction), and especially Johnson 2004. For discussion of ancient cultures of reading in the Roman era, see Johnson 2010 and the growing number of monographs in the Oxford University Press series, Cultures of Reading in the Ancient Mediterranean, edited by Johnson and Chris Keith; on the specific question of scriptio continua and silent reading in antiquity, Johnson 2000. On ancient codices, fundamental is Turner 1977; and there are good discussions of the issues, especially as they revolve around the Christian preference for the codex, in Gamble 1995, Hurtado 2006 (with review by Johnson 2007), Bagnall 2009, and Nongbri 2018; the classic reference on the early Christian codex is Roberts/Skeat 1983. For a more general (and particularly stimulating) discussion of the role of Christianity in changes to the format and function of books, see Grafton and Williams 2006. Trismegistos (www.trismegistos.org) is an open-source site that collects data on, among much else, all the early bookrolls and codices under the rubric of the Leuven Database of Ancient Books, and provides a convenient and reliable entry point for further information on all the manuscripts discussed here, including location and inventory numbers, publications, and links to images.
Bibliography
Bagnall, Roger (2009), Early Christian Books in Egypt, Princeton: Princeton University Press.
Epp, Eldon (1997), ‘The Codex and Literacy in Early Christianity and at Oxyrhynchus,’ in Critical Review of Booiks in Religion 1997, ed. C. Prebisch, 14-37, Atlanta: Scholars Press.
Gamble, Harry Y. (1995), Books and readers in the early church : a history of early Christian texts, New Haven, Ct: Yale University Press.
Johnson, William A. (2000), ‘Toward a Sociology of Reading in Classical Antiquity,’ in American Journal of Philology, 121: 593–627.
Johnson, William A. (2004), Bookrolls and Scribes in Oxyrhynchus, Toronto: University of Toronto Press.
Johnson, William A. (2007), Review of Larry W. Hurtado, The Earliest Christian Artifacts: Manuscripts and Christian Origins (Eerdmans 2006) in The Bulletin of the American Society of Papyrologists, 44: 249-251.
Johnson, William A. (2010), Readers and Reading Culture in the High Empire: A Study of Elite Reading Communities, Oxford and New York: Oxford University Press.
Lewis, Naphtali (1974), Papyrus in Classical Antiquity, Oxford: Oxford University Press.
Nongbri, Brent (2018), God’s Library. The Archaeology of the Earliest Christian Manuscripts, New Haven: Yale University Press.
Orsini, Pasquale and Willy Clarysse, (2012) ‘Early New Testament Manuscripts and Their Dates: A Critique of Theological Palaeography,’ in Ephemerides Theologicae Lovanienses 88/4, 443-474.
Roberts, Colin H., and T. C. Skeat (1983), The Birth of the Codex, London: British Academy.
Romano, Francesco P. (2023), ‘Layout of ancient Greek papyri through lead-drawn ruling lines revealed by Macro X‑Ray Fluorescence Imaging’ in Scientific 13: 6582 | https://doi.org/10.1038/s41598-023-33242-8
Turner, E. G. (1977), The Typology of the Early Codex, Philadelphia: University of Pennsylvania Press.
Turner, E. G. (1978), The Terms Recto and Verso: The Anatomy of the Papyrus Roll, Brussels: Papyrologica Bruxellensia vol. 16.
Turner E. G., rev. P. J. Parsons (1987), Greek Manuscripts of the Ancient World, second ed., London: BICS Supplement 46.
Notes
1 By bookroll I mean a papyrus roll that is written so as to contain a literary text, such as a book of Homer, a play of Sophocles, or a dialogue of Plato. The focus is on Greek because most of what we know in detail of the Graeco-Roman book comes from fragments of ancient books that survive in Egypt, and these are almost all written in Greek (or, in the late period, Coptic).
2 The Trismegistos number is the entry point to the full metadata for each manuscript in the database found at URL www.trismegistos.org.
3 For detail on the production of papyrus, see Chapter 00 in this volume.
4 See Johnson 2004, Chapter 3, §3.1.2, 91-99. Add now Romano et al., 2023 for Macro Florescence confirmation of faint lead-drawn vertical lines setting the columns at a forward slant (Mass’s Law) in certain Herculaneum papyri.
5 Johnson 2010, 21.
6 On the difference between front (‘recto’) and back (‘verso’) of a papyrus roll, see Turner 1978.
7 Pliny, Epistles, 2.1.5.
8 For a wide-ranging discussion on this important detail, see Bagnall 2009, Chapter 1.
9 The following charts are taken from the Leuven Database of Ancient Books, which can be found under the umbrella of www.trismegistos.org. The LDAB is not without inconsistencies — a late amulet is once listed as a NT roll, for example — so, while the overall profile is correct, the numbers are not exact. I limit the results to texts recovered in Egypt for the sake of consistency of comparison: if western manuscripts were included, there would be many late parchment codices, mostly in Latin, which, since papyrus hardly ever survives outside of Egypt, would distort the chart profile. Note that few Latin speakers resided in Egypt even in the Roman era, and thus almost all of the results are books written in Greek and Coptic— but there is no reason to think that Latin books would have profiled differently. Accessed Fall 2016 (the graphing feature is not currently available for the Leuven Database).
10 On parchment and codices, see Bagnall 2009, Chapters 3 and 4, 50-90, esp. p. 79. The archaeological question of what survives and what does not may at first seem an important factor, but the stability in numbers of papyrus codices from the third through sixth centuries suggests that the increase in parchment codices is an actual datum.
11 Jerome, de viris illustribus 113.
12 Papyrus bookrolls were also copied onto parchment for preservation purposes. See Grafton and Williams 2006, 215.
13 Bagnall 2009, 73
14 Dating of non-documentary texts can be perilous and there is a well-exposed tendency to push earlier proposed dates for early Christian texts (second- to fourth-century). For our purposes, however, the distortion introduced by that tendency does not affect the larger argument: correcting the data will simply pull down the small number of Christian texts from the second to fourth centuries, thus reinforcing the conclusion. On the tendentious dating of early Christian codices, see Bagnall 2009, chapter 1, 1-24; Orsini and Clarysse 2012; Nongbri 2018.
15 That the preference for the codex goes back to an Ur-text of Mark: Roberts in the 1954 edition of The Birth of the Codex (withdrawn in the revised 1983 edition, 55-57); that a codex could hold all four Gospels: Roberts and Skeat 1983; that the codex goes back to an Ur-text of the Pauline letters: Gamble 1995; better for travel and linked to missionaries: Epp 1997. For more detail, see Hurtado 2006, 69-83.
16 Kenyon, for example. See Turner 1977, 35-37.
17 See figures in Turner 1977 for published results; my own figures come from the Early Codices Project, a born-digital database not yet public.
18 Turner 1977, 37; Roberts 1983, 46.
19 Again, the figures for codices are based on the Early Codices Project, whose public instantiation is forthcoming; for bookrolls, see Johnson 2004, Chapter 3, §3.9, 155-567.
20 Grafton and Williams 2006, 220.
21 Grafton and Williams 2006 speculate that these are among the 50 codices that Constantine ordered Eusebius to make; if so, these codices are the product of imperial patronage.