*
Q 28.
The one problem with the bracket notation is that it can only be used for add/omit readings. Where two readings are equally good, but one substitutes for the other, there is no way to indicate the degree of uncertainty expressed by the brackets. This has caused some editors (e.g. Bover) to avoid the use of brackets; these editors simply print the text they think best.
The third course, and probably the best, is to do as Westcott and Hort did and have noteworthy marginal readings. But this policy has not been adopted by modern editors.
Today, scholars have excellent tools for dealing with such problems (notably ultraviolet photography, though there are many other techniques in use). That wasn't so in the past, but the desire to read the manuscripts was just as great.
In consequence, scientists developed a number of chemicals for trying to bring out faded or eradicated ink. The first ink restorer seems to have been oakgall (gallic acid or, technically, trihydroxybenzoic acid, C6H2(OH)3COOH), used as early as the early seventeenth century (possibly earlier), but much stronger chemicals were eventually discovered. Some of the reagents used in the nineteenth century include ammonic sulphydrate, potassium nitrate, potassium bisulfate, and Gioberti tincture -- successive coats of hydrochloric acid and potassium cyanide (!).
The problem with these chemicals is that, although they can bring out the writing in the short term, they destroy the manuscript in the slightly longer term. They can cause the ink to blot and the parchment to decay. (As a result, there was a brief period during which scholars applied their glop, photographed the results, and washed the chemicals off. Somehow this doesn't seem much better than leaving it on the manuscripts.) Among New Testament manuscripts, this sort of defacement happened notably to C (though it is not clear whether Tischendorf, who is frequently blamed for it, was guilty; other scholars seem to have been the primary culprits). The problem is especially bad when multiple chemicals are applied (as was done, e.g., to the manuscript of The Poem of the Cid); not only does this damage the parchment, but it also renders ultraviolet photography less effective.
Chemical "enhancement" of manuscripts is now strongly frowned upon, and has effectively stopped. Unfortunately, there are instances of the use of chemicals as late as the 1920s; many manuscripts which survived the Middle Ages have now been permanently damaged by more modern scholars who generally did not learn much as a result of their vandalism.
It's interesting to note that some of these chemical reagents were known long ago. Pliny the Elder was perhaps the first to describe an invisible ink. Of greater significance, perhaps, is a remark by Philo of Byzantium, who refers to an ink of nutgalls which could be developed with what we would now call copper sulfate. Since many ancient inks contained nutgall, Philo deserves credit, in a sense, for the first method of "developing" palimpsests.
Another application of chemistry to textual criticism is in the dating and verification of manuscripts. Spectroscopy and other tests can reveal chemicals contained in inks or paintings without damaging the manuscript. And if a manuscript contains a chemical not in use at the time it was thought to have been written, well, that implies a problem. This line of argument has been used, e.g. to implicate 2427 as a forgery, since it contains Prussian Blue, a dye not invented until the eighteenth century, well after 2427's alleged date. The problem with such arguments is that they depend to a strong extent on our knowledge of history of chemical use; there is currently a major argument about another chemical, titanium dioxide, thought to be modern but now found in small amounts in ancient inks.
Another recent surprise came when a technique called Raman spectroscopy was used on the British Library's King George III copy of the Gutenberg Bible. According to a (non-technical) article in Renaissance magazine (issue #45, p. 18), the inks used to illuminate that printed book (which of course is contemporary with some late manuscripts) included cinnabar for bright red (as expected), carbon for black, azurite for blue (not a surprise, though some blues use lapis lazuli), calcium carbonate (chalk) for white, malachite for olive green, and verdigris (copper ethanoate) for dark green. More notably, the Göttingen copy was found to contain anatase and rutile, which had been regarded as modern compounds. This may be the result of contamination, but it may be a hint that we may still have more to learn about ancient inks.
The codex was in fact what moderns think of as a book -- a series of leaves folded and bound together, usually within covers. Codices could be made of parchment or papyrus (or, of course, paper, once it became available). Whichever writing material was used, a series of sheets would be gathered and folded over, meaning that each sheet yielded four pages. These gatherings of leaves are normally referred to as quires.
Many of the earliest codices consisted of only a single quire of many pages. Examples of single-quire codices include P5, P46, and P75. Single-quire codices, however, are inconvenient in many ways: They do not fold flat, they often break at the spine, and the outside edges of the page are not even. Still more troublesome is the fact that the scribe had to estimate, before the copying process began, how many leaves would be needed. If the estimate were inaccurate, the codex would be left with blank pages at the end, or -- even worse -- a few extra pages which would have to be somehow attached to the back of the document. As a result, it became normal to assemble books by placing smaller quires back to back. This can be seen as early as P66, which uses quires of from four to eight sheets (16 to 32 pages). Quires of four sheets (16 pages) eventually became relatively standard, although there are many exceptions (B, for example, uses five-sheet quires).
It is sometimes stated that the Christians invented the codex. This is of course not true; the word itself is old (Latin caudex properly refers to a tree trunk, hence to anything made of wood, and hence came specifically to mean a set of waxed tablets hinged together. E. Maunde Thompson, An Introduction to Greek and Latin Palaeography, p. 51, notes that Ulpian in the third century makes reference to literary codices). Indeed, we have quite a few examples of pagan literature on codices in the early centuries of the Christian Era; David Diringer (The Book Before Printing, p. 162) surveyed known manuscripts from Oxyrhynchus (as of half a century ago), noting that of 151 pagan documents known to him from the third to sixth centuries, fully 39 were codices. But the church does seem to have been responsible for the popularity of the codex format (e.g. in Diringer's example, of 82 Christian documents, 67 were codices).
We should also note that a sort of proto-codex existed in the form of the orihon.
We observe that the codex has both advantages and disadvantages for literature, especially when dealing with papyrus codices. It requires less material (which may be why the Christians adopted it), and it's easier to find things in a codex. But it's rather harder to write (since one must write against the grain on a papyrus, or on the rough side of a piece of vellum), and one also has to estimate the length of the finished work more precisely. The latter disadvantages probably explain why the Christians were the first to use the codex extensively: They needed a lot of books, and didn't have much money; pagans didn't need so many books, so they felt the disadvantages of the codex more, and the advantages less.
Codices have another advantage, though it wasn't realized at the time: They survive abuse better. Being flat, there are no air pockets to collapse, and they protect their contents better. At Herculaneum, thousands of scrolls were discovered, rolled up and damaged by the conditions that buried them. Centuries of efforts to open and read them accomplished little except to ruin the documents involved. Had the documents been stored in codex form, they would almost certainly have survived better.
In fullest form, the Alexandrians used six symbols:
| Symbol | Name | Purpose |
| - | Obelus | Oldest and most basic (and occasionally shown in other forms); indicates a spurious line. (Used by Origen in the Hexapla to indicate a section found in the Hebrew but not the Greek. For this purpose, of course, it had sometimes to be inserted into the text, rather than the margin, since the LXX, unlike Homer, was prose rather than poetry.) |
| Diple | Indicates a noteworthy point (whether an unusual word or an important point of content). Often used in conjunction with scholia. |
| periestigmene (dotted diple) | Largely specific to Homer; indicates a difference between editions |
| Asteriskos | A line repeated (incorrectly) in another context (the location of the repetition was marked with the asterisk plus obelus). (Used by Origen to note a place where the Greek and Hebrew were not properly parallel.) |
| -
| Asterisk plus obelus | Indicates the repetition of a passage which correctly belongs elsewhere (the other use, where the passage is "correct," is also marked, but only with the asterisk) |
| Antistigma | Indicates lines which have been disordered |
Many mistakes in copying arise when a scribe misreads the exemplar. Handwriting
being what it is, chances are that, on occasion, almost everything has been read as
something else. But some errors are much more likely than others. In Greek uncials,
for example, the shown at right were frequently and easily confused:
In Greek minuscule hands, with many different styles and vast numbers of ligatures, there were many more combinations which might be confused occasionally. Some of the most common confusions, however, include
b k m
m n
eu
It will be noted that errors which could occur in uncials are more important for the history of the text, as these errors could have arisen early in the history of copying.
Similar confusions could, of course, occur in other languages. The list for Coptic, for instance, closely resembled the Greek list, as Coptic letters were based on the Greek. Latin had its own list. In uncials, the primary problems were:
I L T
F P R
C E O G U
EU COG
(the list for inscriptional capitals is somewhat different, as E, for example, was straight in capitals but curved in uncials. Since, however, there are no known copies of the New Testament inscribed on stone tablets, this is of little concern.)
Easily confused letters in Latin minuscule script include
a u
o e
cl d
n u
s f
c t
In addition, almost any combination of letters with many vertical strokes (such as i l m n t) could cause confusion. Particular scripts might add additional confusions; Beneventan script, for instance, used an odd form of the letter t which closely resembles the letter a!
Also, it's worth remembering that the above lists are based on book hands. In the days when almost all copying was done by trained copyists, one could expect nearly everything to be written in such hands. But as literacy became widespread, this tended to break down. Casual writers could produce almost anything. A book on English letterforms, for instance, gives samples of sixteenth century writing which show forms of the letter a which look like b, n, u, and w; many writers made c resemble t; d and e could both look like a theta (!), and so forth.
A final reminder concerns numbers. In Greek as in most modern languages, a number could be written as a numeral or spelled out (e.g. in Rev. 13:18, the "number of the beast" could be exakosioi exhkonta ex or CXC'. It will be evident that this can produce different confusions. (Though this error is perhaps more likely in Latin, with its repeated I and X symbols, than Greek.)
This is one of the arguments often cited by those who favor the Byzantine or the Majority Text. It is, however, simply invalid. Hence the term Fallacy of Number.
To be clear, a fallacy is something which does not follow logically. For example, it's easy to prove that 1=2 if you allow division by zero. But you can't divide by zero. The proof is false because it relies on impermissible methods.
Why is counting numbers of manuscripts invalid? Because it only works if all manuscripts are copied and destroyed at the same rate. If simply counting numbers were sufficient, then the Latin Vulgate would be the original New Testament -- there are more Latin than Greek New Testaments in existence.
We can in fact demonstrate that there are cases where the majority is not the original. Almost all our manuscripts of Euclid are of Theon's recension. It says in the manusripts that they're rewritten! But they are still the majority.
There are lots of ways in which an un-original text can become common. It might look more authoritative for some reason. A particularly strong church figure might promulgate it. It might come from a region where persecutions against Christians were few, so manuscripts weren't destroyed. It might be the local text of a region where the Christian population is particularly large. Most of these have been urged as arguments for and against the Byzantine text. We do not, at present, know which of them are true -- if any. We do know that they are sufficient to disallow us from countine manuscripts to determine which text is original.
The fallacy is sometimes called the "Democratic Fallacy." The Democratic Fallacy is that, just because people believe something, it's true. For most of history, the majority of people believed that the sun moved around the earth -- which is, simply, false. The fact that lots of people believed it doesn't make it true. A more recent example, which shows the fallacy even more clearly, is the American war in Iraq. In 2003, most Americans believed it was right. In 2007, most believed it wrong. Was the war right? Wrong? People will probably disagree for as long as it is remembered. What is certain is, if it was right in 2003, it was right in 2007; if it was wrong in 2007, it was wrong in 2003. In one year or the other, the majority was wrong.
Note that the fallacy of number is merely a fallacy. That is, number has absolutely no bearing on what was the original text. The Byzantine text may be original. Most think not, but the fact that the advocates of the Byzantine text cite numbers should not be held against it (except in the indirect sense that the Byzantine text advocates cite numbers, implying that they are sorely lacking in valid arguments. To me, the fact that they even cite numerical preponderance is proof of desperation -- they want the Byzantine text, for whatever reason, and so grasp at straws. But this is no more evidence of the falseness of the Byzantine text than is numerical preponderance evidence for it).
But I must emphasize: The Fallacy of Number is a fallacy. It is an argument that should be retired, forever. Most arguments in textual criticism are about data or interpretations. This one is not. It is purely about mathematical logic. There is a right answer -- and the right answer is that counting noses doesn't work.
Hort's use of the genealogical method led him to the theory of "Neutral," Alexandrian, "Syrian" (Byzantine), and "Western" texts which formed the basis of the Westcott-Hort edition. This textual theory has been modified in some instances, with the result that the "genealogical method" is now rather in dispute. This is rather unfair; although Hort's results cannot stand, and his description of his method is too theoretical (and was not, in fact, the entire basis of his text), the principle of grouping and editing by text-types has by no means been disproved. See, e.g., the section on The Use of Text-Types in the article on Text-Types.
Like homoioteleuton errors, homoioarcton errors can produce nonsense, but can also be sensible (and therefore perhaps difficult to tell from other sorts of errors).
Homoioarcton is noted in the Nestle-Aland apparatus with the notation h.a., but observation shows that this notation is not used nearly as often as it might be (e.g. none of the omissions in Luke 3 are noted as possible homoioarcton errors). Students are therefore advised to note this possibility in examining variants.
Original reads "Pete went to the store. When he reached the store he bought bread and milk." The scribe, skipping from the first instance of "store" to the second, would write "Pete went to the store he bought bread and milk."
Homoioteleuton errors can occur almost anywhere, and are often easily detected as they produce nonsense. There are, however, exceptions, as e.g. in 1 John 2:23, where the Majority text has skipped ton patera ecei...ton patera ecei, leaving a text which is incomplete but perfectly sensible.
Homoioteleuton is symbolized in the Nestle apparatus by the symbol h.t. (which indicates either that a manuscript has a homoioteleuton error or that a variant is or might be caused by homoioteleuton). Others such as Merk use a "leap" symbol,
,
similar to a sideways parenthesis or a musical slur.
Exactly how common are h.t. errors? This is complicated. Examining the NT auf Papyrus apparatus of Philippians shows that the 17 papyri and uncials cited there display a total of 12 clear h.t. and h.a. errors. This is if anything a low rate of such errors -- and there are at least four other errors not directly attributable to h.t. which may result from skipping lines. And skipping lines can be far more common than the above statistics would indicate. Thomas C. Knott and David C. Fowler's edition of the A text of Piers Plowman includes a table of omitted lines. Their text is 2418 lines long. The manuscripts they cite have (apart from defects and long stretches omitted presumably for other reasons) a total of 606 lines omitted. That's out of an average of about fifteen manuscripts for each portion of the text. Thus, the manuscripts average out to omitting about one line in sixty. This rate is naturally higher than in the NT tradition, because these manuscripts aren't as familiar to scribes and aren't as heavily corrected and used. But it's an indication of the potential of haplographic errors.
An imitation is a written work deliberately done in the style of an earlier work. A typical example in English is the "Thou shalt not" stricture: The King James Bible uses this formulation for the Ten Commandments, so moderns may say anything from "Thou shalt not speak ill of another Republican" to "Thou shalt not be the first to start a war." These are, of course, trivial examples, but the King James Bible has inspired many non-trivial examples, e.g. the Book of Mormon, which is in a pseudo-Biblical English which is in fact neither Jacobean nor English; similarly, Spencer's Fairie Queen is intended to imitate Chaucer but -- because Chaucerian English was long dead -- instead imitates gobbledigook.
Another example may be familiar to some English readers: Lancelot C. L. Brenton's translation of the Septuagint. This is a more modern equivalent of the Spenser/Chaucer situation: Brenton's translation is in pseudo-King James English, much influenced by the KJV. If one has the Hebrew, the Greek, and Brenton, one can at times retrovert to the KJV text. Brenton's translation is in fact less competent than it could be (as well as irritating to read) because it's so much a KJV imitation.
An imitation is not quite the same as an allusion, though the resemblance is obvious; Spencer, e.g., had spent so much effort reading Chaucer that he took on some of his speech patterns without actually understanding Chaucerian grammar. Extremely careful and cautious use of such references might enable us to occasionally see a hint as to how a damaged passage evolved.
Such a method is probably not needed for the Greek New Testament; the materials available to us are too slight. But I can imagine it coming up with regard to one of the more obscure versions, such as the Gothic or the Palestinian Syriac or perhaps even the Sahidic Coptic.
The adjective lacunose may refer to a manuscript with many lacunae.
In textual criticism, "lemma" usually is used to describe the text of a running commentary or commentary manuscript. So, for example, we might cite Origenlem and Origencomm, with the lemma being the reading found in the biblical text of the manuscript and the commentary being found in the margin.
Since the biblical text seems more liable to correction than the commentary, the value of a lemma is usually less than the reading(s) in the margin. Thus certain editions will only cite a lemma where the commentary is missing or unclear.
It should perhaps be added that the Alands, in their work on the United Bible Societies Edition, do not appear to have followed this method, as the UBS text is overwhelmingly Alexandrian. A text proceeding purely from local-genealogical work (i.e. from internal criteria only) would without doubt be more eclectic. This leads to the suspicion that the Alands have not correctly described their method, which instead consists of using "local genealogy" as assisted by the history of the text (so, e.g., a reading found only in a late text-type cannot be earlier than one found in an early text-type, no matter how original it may appear on internal grounds). This is, in the author's opinion, the best and most proper form of criticism -- but it requires a truly accurate history of the text, something which the Alands (on the evidence) had not achieved -- or at least had not enunciated in a way usable by other scholars. Which, if one wishes to follow the rules of scientific work, is hte same thing.
C L
33 Sahidic Bohairic etc.), the Cæsarean
(Q family 1 family 13
28 565 700 Armenian Georgian), the Antiochian (Old Syriac), the
Italian or Gaulish (D a b), and the African (WMark k e) (see
The Four Gospels, p. 26, etc.).
Direct evidence for the theory of local texts is largely lacking; except for the Egyptian papyri, we generally cannot correlate texts with the place of origin of manuscripts. There is some evidence of local texts on a lower level; we tend to find, e.g., that if a particular scribe copies several manuscripts, they tend to be of a single type. (Consider the work of Theodore of Hagiopetros, who is almost single-handedly responsible for Wisse's Kx Cluster 74, or George Hermonymos, who game us manuscripts of Kx Cluster 17). There is also evidence from non-Biblical manuscripts; in works such as Piers Plowman, we find significant correlation between the place a manuscript was copied and the text it contains. (The vast majority of manuscripts of the "C" recension are found in the general area of Gloucester and the southwest; the "B" recension is common around London; the "A" recension is scattered but has several representatives near Cambridge.)
With the discovery of the papyri and the realization that not all manuscripts from Egypt have Alexandrian texts, the theory of local texts has lost some of its favour. We also find that not all the texts at large ancient repositories (Athos, Sinai) are of the same type. The truth is, however, that even in Egypt a single text (the Alexandrian) is dominant. At the very least, we could expect local texts to flourish in isolated areas, and also to find particular sorts of texts associated with particular localities. There was much commerce in the ancient world, and so not all manuscripts in an area will automatically have the local text -- but this does not invalidate the theory; it merely means that we must investigate manuscripts to see if they belong with their local type.
Still, caution must be used in assessing the value of local texts. If two local texts are indeed independent, then their common readings do have extra value. But the texts must indeed be independent! If, as some have charged, the "Cæsarean" and/or Byzantine texts are the result of editorial conflation of the Alexandrian and "Western" texts, they have no value as diverse witnesses. In addition, we must be alert to the possibility that one local text is derived from another. If, e.g., the texts at Athos are ultimately derived from Constantinople (a real possibility), then the local text of Athos has no independent significance.
As far at the quotations itself are concerned, it should be kept in mind that most scribes knew them in their own language. Thus copies of the Greek Bible tended to use the Septuagint text, and scribes would tend to conform passages to the Septuagint if by some chance they differed. This phenomenon doubtless occured also in the other versions (e.g. a Vulgate quotation might be assimilated to the Vulgate Old Testament), though this is not normally a matter of great concern for textual critics.
An orihon was essentially an ordinary scroll folded as a codex. That is, a papyrus scroll was prepared, written in columns in the ordinary manner. Once finished, however, it was not rolled but folded in a concertina fold, with one or two columns per fold. That is, if we started with a scroll looking like this (where each symbol _ represents a column of print), the process would look like this:
_ _ _ _ _ _ _ _ <-- initial flat layout
/\/\/\/\ <-- partially folded
|| || || || <-- fully folded
_\/_ <-- after end is bound
Thus you ended up with something which, to all intents and purposes, was an ordinary book, openable to any page. In some cases, the backs of the inner folds were pasted together and the whole inner margin bound with thongs to make it even more book-line. In terms of reader convenience, the orihon came close to being the equal of the codex (its only drawback was that it was rather bulkier). But, of course, it required more material than a codex. And it wasn't a particularly suitable form for vellum books.
There were other advantages to the form. The backs of the pages did not have to be glued together, meaning that the orihon could be flattened. This produced, in effect, a scroll -- but one which did not have to be unrolled and rerolled to find a particular passage. It was much more convenient for finding pasages. And, folded, it was typically smaller than a scroll and less subject to damage. If one were forced to use a cheap paper which could be written on only one side, an orihon was a very useful form.
I don't know of any surviving Biblical orihon, but it is possible that one of the handful of surviving papyrus fragments we regard as being from scrolls might in fact be from an orihon. (I have this strange, completely unscientific feeling that P12 is such. But I have absolutely no evidence for that proposition.)
The term "paleography" was coined by Dom Bernard de Montfaucon, who in 1708 published the Paleographia graeca -- not actually the first book on dating manuscripts, but the first one to develop the tools of the discipline; soon after, Scipione Maffei discovered many old documents in Verona, and on this basis developed Latin paleography and added greatly to the knowledge of the field.
Palaeography uses many tools to make its judgements (far too many to be covered here!); of these, shapes of the letters is perhaps the most important (for examples of the evolution of uncial letterforms, see the article on and examples of Uncial Script). However, a paleographer will also examine the way the manuscript is prepared -- both the material (papyrus, parchment, paper; scroll or codex) and the method of writing (reed, quill, metal pen; ink type), plus the way the lines are ruled (sharp or blunt point, etc.) Word forms as well as letter forms must be examined, as well as the shape of the page and the arrangement of the columns, plus any marginalia or artwork or even unrelated scribbles.
Care must be taken with the results of paleography, however. It is not an exact science, and all its judgments are approximate (so, e.g., the enthusiasm about the early date of P52 should be treated with a certain amount of caution; it is simply not possible to date a manuscript to the fifteen or so year span some have proposed for P52). Book hands are more datable than casual hands (an advantage, obviously, to Biblical scholars) -- but most scribes will stick with a particular style as long as they live, meaning that even if we can accurately date a style to 125 C.E. (the typical date for P52), that scribe could still have been working forty or fifty years later. Housman writes, wisely, that "...even when palaeography is kept in her proper place, as handmaid, and not allowed to give herself the airs of mistress, she is apt to be overworked." It is perfectly possible for old handwriting styles to be preserved long after new ones have evolved. Sometimes this is the result of isolation -- but sometimes it is the result of peculiar needs. (An example of this is Old English hands. Old English used three letters not in the Roman alphabet -- eth (
),
thorn (
),
and yogh (
).
This led to preservation of an older script for
Old English documents even as new ones evolved for Latin (we see instances,
even from the same scribe, of Old English documents written in an insular
hand even as Latin works are copied in a Caroline minuscule). We see something
rather analogous in the case of Codex Bezae, where the Greek and Latin hands
have been conformed to each other (this is the chief reason why Bezae is so
difficult to date).
It should also be noted that paleography does not concern itself solely with
manuscript dating, although this seems to get all the "press" in
most English-language volumes on TC. Paleographers concern themselves also
with the place of the writing, the scribe, etc. (E. Maunde Thompson, for
instance, was perhaps the most famous of all students of classical
paleography -- and he was called upon to examine the manuscript of the play
"Sir Thomas More" to see if a particular scene was indeed in
Shakespeare's own hand.) These other considerations can be very important:
Consider the implications, e.g., if Tischendorf had been right and the same
scribe had worked on B and
,
or if it could be proved that one of those manuscripts had been written
in an unexpected place (e.g. Rome).
From Greek roots meaning "again-scraped." A palimpsest
was a manuscript which was re-used. Presumably the original writing
was no longer valued and/or easily read, and a scribe decided that the
expensive parchment could be better used for something else (almost all
palimpsest are parchment; papyrus and paper are not suitable for re-use).
In most instances the parchment would be washed and/or scraped and
resurfaced, then overwritten, although there are instances of manuscripts
which were overwritten without being cleaned. (As inks evolved, they became
harder to erase, so some documents reportedly were actually written between
the lines of the old manuscripts -- quite possible, given the size difference
between literary uncials and late minuscules.) The most thorough method
of cleansing is said to be scraping (with a knife or pumice), followed by
soaking with cheese, milk, and lime.
The under-writing of palimpsests is, of course, often difficult to read, although modern tools such as ultraviolet photography help somewhat, particularly when two different formulations of ink have been used which produce different degrees of flourescence. (Earlier chemical reagents often damaged manuscripts without doing much to improve their legibility.) But almost all palimpsests are illegible at certain points (and, ironically, the remnants of the under-writing can sometimes make the upper writing equally unreadable).
Among the more important New Testament palimpsests are C (sometimes listed as the first palimpsest "discovered"), Pe, Papr, Q, and 048 (the latter a double palimpsest -- it was overwritten twice). It should be noted that none of these documents is intact (Papr is about 90% complete, which is about as high a fraction as one can reasonably expect); since the erased leaves were simply raw material, they would end up being used out of order, and some leaves would generally be used for other purposes).
Classical palimpsests are, if anything, even more common, since Christians had a tendency to erase these works to use for religious works. I know of no comprehensive catalog, but just for comparison, Harold W. Johnston at the end of the nineteenth century listed all classical Latin documents of the sixth century and earlier. He counted 24 such documents -- and 14 of them palimpsests. The illustration above right is typical of the form: The image is of (Johnston's facsimile of) Codex Palimpsestus Vaticanus 5757, the Schedae Vaticanae,. The under-writing is a fourth or fifth century Latin uncial copy of Cicero's De Re Publica (I.xvii.26), once at Bobbio, now in the Vatican; the upper writing is of the eighth century. Note how clearly visible both still are.
This sort of vandalism is particularly regrettable when we have no intact copies of a particular work (though it might be argued that the documents might not have survived at all were it not for the palimpsests); much of the work of Archimedes, e.g., survives only in a single tenth century manuscript that was over-written for church use in the thirteenth century.
We might also note, in passing, that palimpsests often we not neatly rewritten one atop the other. It was quite typical to see the parchment -- which would be much worn at the seam where the quire was folded -- cut in half, and then re-folded. This would usually cause the upper writing to be at right angles to the lower. This is the case, e.g. with Codex Z, shown at right in exaggerated colour.
Note: One will occasionally see the usage "primary version" applied to the versions of greatest significance for TC. (Under this definition, the Latin is still a primary version, but the Gothic becomes secondary.) Such usage is to be discouraged as it can cause confusion.
It should be noted that these four are not the only purple manuscripts in existence. 565 is not an uncial, but it is probably the most famous purple manuscript. There are a handful of manuscripts with some purple pages, plus there are purple manuscripts of the versions -- the Old Latins a, b, e, f and j of the gospels, among others, plus the Vulgate codices em, per, reg, theod; also the Gothic Codex Argenteus.
Many of these manuscripts are in a rather poor state of repair; gold ink was often hard on parchment -- observe the state of the manuscript at right (Codex j of the Old Latin; colours exaggerated).
Volumes fall into two basic types: Single-quire codices and multi-quire codices. Multi-quire codices have the quires set back to back, with relatively small numbers of sheets per quire (usually four sheets, or sixteen pages, though other numbers are known), arranged so that sheets of similar type face each other (for papyri, e.g., vertical strips facing vertical strips and horizontals facing horizontals; for uncials, flesh side facing flesh and hair facing hair). Multiple-quire codices were easier to assemble (since one didn't need to guess how many leaves one would need), and generally more attractive, but required binding, meaning that at least some codices (such as P46 and P75) were single-quire codices: One huge gathering of dozens of sheets folded over. This has its conveniences for critics: We don't have the outermost leaves of either P46 or P75, but we know the overall length of both manuscripts, because we can locate the center leaf and calculate from there. (This is possible even if we have only a single leaf of a single-quire codex, as long as page numbers can be found on both sides.) And knowing the overall length, we can at least estimate the extent of the contents (it is by this means, e.g., that we calculate that P46 can never have contained the Pastoral Epistles). Of course this is also possible with multi-quire codices, but only in the special case where we have quires before and after the lacuna. If a multi-quire codex simply ends (as is the case, e.g., with B), there is no way to estimate how many leaves are missing.
Another problem with single-quire codices is length. A single quire can only contain so many leaves -- a few dozen at most. So to assemble a full Bible, or even a complete set of the Four Gospels or the Acts and Epistles, requires a multiple-quire codex.
Most fragments, of course, consist of only a single sheet (not even a complete leaf; it's quite common for the page to break at the fold, and only one half of the broken leaf to survive), making it impossible to tell whether they come from single-quire or multiple-quire codices.
For more on the significance of quires, see the entry on codices.
In classical criticism, the term is much more precise: it refers to two texts of the same work which are distinct as a result of editorial work. For instance, we have two different texts of Euclid's Elements, one edited by Theon ("Theon's Recension," found in the large majority of surviving manuscripts) and one believed to be closer to the original. Similarly, it is now believed that the two texts of Shakespeare's King Lear (Quarto and Folio) represent two distinct stage settings of the same play, and as such we find editions such as the Pelican Shakespeare actually printing both texts in parallel columns (and then a conflated version).
The differences between these two usages is somewhat unfortunate, since it is now generally agreed that the Byzantine text of the New Testament is not recensional in the proper sense (i.e. it was not edited, by Lucien or anyone else). But there are edited texts of the New Testament -- Marcion's, obviously, and also, in a way, the Diatessaron; in addition, the D text of Luke-Acts has unquestionably been edited at some points (e.g. the use of Matthew's genealogy of Jesus in Luke 3). Thus when one encounters the word "recension" in New Testament situations, one must always be careful to learn the precise sense in which the word is being used by a particular author.
Our descriptions of scriptorium differ, and it is likely that their nature changed over time -- depending, among other things, on whether manuscripts were copied individually or in bulk.
If the former, then the scriptorium was kept entirely silent; David Diringer (The Book Before Writing, p. 207ff.) reports that the scribes even evolved a series of hand gestures by which to communicate the volumes or materials they needed so that they could do their work without speaking.
The policy was to make the scriptorium a large, open room with good natural light; artificial light was banned to prevent fires. The scriptorium was managed by the armarius, who supplied the tools required by the scribes: desks, pens, ink, parchment or papyrus, and the other tools used by the scribe: knife, awl (to prick the parchment for lining), stylus (to score the lines), ruler, pumice (to smooth the parchment), perhaps weights (to flatten pages) and sponges (to erase, though this was sometimes done with the knife).
None of this, of course, was possible when a manuscript was copied by dictation. There is little evidence of the procedure for this, but that manuscripts were copied by dication can hardly be denied; there are simply too many errors of hearing in certain manuscripts. (We should add, however, that some errors which appear to be errors of hearing may not be: Recall that most ancients read by sounding out the words before them. Thus they could sound out a word, turn to copy it, and mis-copy it because they mis-heard what they themselves had read!) But it is not at all unlikely that copying by dictation died with the Roman Empire in the west, and perhaps fell into decay in the east also; demand for manuscripts would have fallen both with the adoption of parchment (which lasted longer and reduced the need for replacement manuscripts) and with the decline of literacy. In these circumstances, individual copying of manuscripts would suffice, and copying by dictation might well have died out.
Many of these defects are modern, but many are old, as well. Today, if a book is damaged, we will likely just go out and buy another copy. When manuscripts were copied by hand and expensive, this was not a reasonable option. Far easier to copy off enough pages to fill the gap, and re-insert that into the binding. This is very common among the early uncials. B was supplemented by the minuscule 1957. But this is an unusual supplement, coming much later and in another style of writing. Usually we see supplements in the same sort of script. So Dea, for instance, has supplements in Matt. 3:7-16, Mark 16:15-20, John 18:14-20:13. If a critical apparatus notices this (not all do), the supplement will be marked with the superscript s or supp. So in John 19, for instance, the Nestle-Aland apparatus does not cite D but Ds. Other important manuscripts with supplements include Dp (in 1 Corinthians), W (in John), 565 (various places), 892 (in John), and 1241 (portions of Paul and the Catholics).
There are instances where it appears the supplement may have been copied from the original manuscript, in whole or in part (this could happen, e.g., if a portion of a page had been damaged by damp or torn). Usually, however, another exemplar had to be consulted. This can result in a change in text-type. Usually this will mean a shift toward the Byzantine text (892supp, for instance, is noticeably more Byzantine than 892 proper). But not always! In Paul, 1241's basic run of text is purely Byzantine, while the supplements are an Alexandrian/Byzantine mix.
Most supplements appear to be a response to accidental damage. But this is not always the case. Codex Vercellensis (a) of the Old Latin appears to have been deliberately supplemented: The ending of Mark is missing, cut away, and a portion restored. C. H. Turner calculated that the missing leaves could not have contained the "longer ending" 16:9-20. Thus the logical conclusion is that a was deliberately mutilated and a supplement added to supply this ending.