A few words of introduction are necessary to the investigations contained in the following pages, in order to remove some of the perplexity which may hang around the enunciation of the theory which they contain.

In the course of an examination of the columnar arrangement of the text of the oldest MS of the New Testament, my attention was drawn to a remarkable numerical peculiarity in the arrangement of the lines and columns of the several books, and from this my mind was forced to the conclusion that the scribes of the New Testament produced epistles more uniformly written and at the closing page more frequently filled than is the custom at the present day ; and that it was, in fact, possible to reproduce the original pages by a simple process of numerical subdivision, if only the MS had preserved the lines of the original writing. Further study of the Vatican Codex showed that a large number of the books of the New Testament were capable of this subdivision (by the very simple process of dividing the column of the MS into three equal parts), and that the pages resulting from the subdivision were very closely related to the original pages.

Perhaps this will become easier to apprehend by a simple variation of the statement. Imagine a printed book, in which there are, let us say, ten equal pages, of thirty lines to each page, printed uniformly. If a reprint be made of this book in any other form, i. e. on pages and with lines of a different size to the copy, it is evident that the original arrangement of the book will be lost, and it is very unlikely that the last page of the new book will be a complete one. If, however, the printer adheres to the original lines, no matter how he may change his pages or his type, we shall always be able to restore the book to its original shape by simple subdivision of its 300 lines into ten pages, although, of course, the subdivision may not be easy to detect, nor to demonstrate.

This is what has happened in the Vatican MS; the scribe has retained the original line, and in a certain sense has preserved the original page also, since he made his column (as the investigation will show) by placing three of the original pages in a vertical line. This fundamental fact is the key to the method of textual criticism to which these pages form an introduction.

New Testament Autographs

Section A.

1. In the course of the first lecture, which I had the honor of delivering in this University, on the Textual Criticism of the New Testament, I pointed out that the material of the second and third Epistles of St. John was probably a sheet or series of sheets of papyrus ; and not only so, but that in the two documents mentioned, the sheet of paper was of a given size, capable of holding a given quantity of uncials. The first of these statements was based upon the allusion which the writer makes to paper, pen, and ink

δια χαρτου και μελανος,  2nd John 12;
 δια μελανος και καλαμου,  3rd John 13;

- while the second statement was an inference from the equality in the contents of the two Epistles, which in Westcott and Hort's edition of the New Testament occupy twenty-nine lines of type apiece, and from the evidence that in each case the writer had completely filled the sheet on which he was writing, since he complains of the insufficiency of his writing materials:

πολλα εχων υμιν γραφειν,
  πολλα ειχον γραψαι σοι.

From this point we are led to the enquiry as to the usual size of the sheets of paper employed in the New Testament documents, and the number employed in the autographs of the several books.

2. In order to make the enquiry carefully, we will first tabulate the number of columns and lines occupied by the uncial letters of the separate texts, as they are presented in the oldest known manuscripts. We begin, then, with the Vatican Codex B. This manuscript is written three columns to the page, and each column contains 42 lines of uncial writing. Omitting the Epistle to the Hebrews, the latter part of which is in a later cursive hand, and the Apocalypse which is also supplied in cursive character,¹ we construct the following table :

Book	Columns	Lines	Line Total
Matt	127	9	5343
Mark	77	31	3265
Luke	136	41	5753
John	97	6	4080
Acts	130	3	5463
Rom	49	16	2074
1 Cor	46	6	1938
2 Cor	31	28	1330
Gal	15	27	657
Eph	16	22	694
Phil	11	0	462
Col	11	15	477
1 Thess	10	28	448
2 Thess	5	34	244
James	12	26	530
1 Pet	12	30	534
2 Pet	8	32	368
1 John	13	27	573
2 John	1	27	69
3 John	1	27	69
Jude	3	27	153

The first thing that strikes us on examining this table is that the compositions do not end, as one might suppose, at different points of the page according to random distribution, but they show a preference for ending at particular points, and especially at the 27th line. Out of the 21 documents cited, five end on the 27th line of the page, two on the 28th and one on the 26th. This is very remarkable.

3. If the compositions were of arbitrary length, the probability that five out of the twenty-one should end on the same particular line is small indeed. Unless I am mistaken, it would be repre- sented by the fraction

which is evidently much less than 1/1x2x3x4x5 x 1/32 or 1/3840. We may be sure then that the odds are at least four thousand to one against such a conjunction of endings being the work of chance.

It is evident that the eight compositions alluded to, viz. II Corinthians, Galatians, I Thessalonians, James, the three Epistles of John, and Jude, are each written on an integral number of sheets of a given size ; and further, this sheet of given size must bear a peculiar relation both to the whole column of the Vatican Codex consisting of 42 lines, and to the fractional column of 27 lines ; for, otherwise, it would not be possible for documents of different length, even though written on sheets of given size, to end at the same place on the Vatican page.

The "V-Page"

If we allow a line for the subscription of those Epistles which end at the 27th line, we have to seek a submultiple of 28 and 42 ; and we at once see that 14 lines of the Vatican Codex bears some multiple proportion to the size of a page of the original writing, and in all probability, in the cases referred to, we may say that 14 lines of the Vatican Codex represents exactly one page of the autograph, ², the only submultiples of 14 being 7 and 2. This provides us with a unit upon which to base our calculations, which for convenience we will denominate a V-page [i.e., "Vatican-page"].

---

1. Scrivener adds the Pastoral Epistles (Introduction, p. 96), apparently following Cardinal Mai, but I can find no trace of them in the Roman edition. The Palaeographical Society, in the description accompanying their facsimile, follow Scrivener. [at this point in time, Harris is working with the Roman facsimile.]

2. [modern note: a better choice of word here might be 'exemplar' rather than 'autograph']

Codex Vaticanus (cont.)

4. We see, then, that of the Epistles especially referred to,

• II Corinthians = 95 V-pages exactly.
• Galatians = 47 V-pages, wanting one line.
• 1 Thessalonians = 32 V-pages exactly.
• James = 38 V-pages, wanting two lines.
• 1 John = 41 V-pages, wanting one line.
• 2 John = 5 V-pages, wanting one line.
• 3 John = 5 V-pages, wanting one line.
• Jude = 11 V-pages, wanting one line.

With regard to these conclusions, the single line left blank in the letter is probably left for subscription ; in the case of the Epistle to the Galatians we have the additional explanation that there was a sentence in it written in large letters by the Apostle Paul's own hand, and when this sentence is copied there is a slight contraction in the copy as compared, with the original.

With regard to St. James, we find two lines wanting; either, , therefore, his handwriting is larger than ordinary, or we may assume that he actually left a somewhat larger blank space than was usual with the other writers, who evidently economized every inch of paper. The sheet of paper, too, is noticeably a small one ; it is only capable of containing 14 lines of average length, about 17 letters each : this also is explicable by the supposition of economy, for the cost of a sheet of papyrus increases with the size of a sheet, but in a much greater ratio than the sheet, on account of the difficulty of finding plants or reeds of a very great length and section. We can see, then, that the cheapest paper is used, and no space spared. Now turn to the table again, and observing that our manuscript- unit is fourteen lines of the Vatican Codex, we see that in the autograph

Philippians = 33 V-pages exactly.

We come, then, to a group of three Epistles which run slightly over an exact number of pages ; thus :

• Romans = 148 V-pages and two lines.
• Colossians = 33 V-pages and one line.
• 1 Peter = 38 V-pages and two lines.

With regard to the Epistle to the Romans, it is not inconceivable that in 148 pages the copy should have gained two lines on the autograph ; the study of the Epistle is, however, complicated by the existence of important various readings, and by the doubtful character of its concluding portion, which seems rather to be addressed to an Ephesian than a Roman community, and by the questionable authenticity of its doxologies. We content ourselves, for the present, by saying that the Epistle, as it stands in Codex B, probably represents 148 pages of the autograph.

With regard to the Epistle to the Colossians the question is more simple, as the document is shorter. Four lines of this Epistle, at least, are from the hand of Paul himself, and would therefore be in larger characters than usual ; this would make the original document longer than 33 V-pages and one line. Either, therefore, the greater part of a page was left blank, which is unlikely ; or Codex B has inserted words in the text, or the amanuensis of Paul (Tychicus, Onesimus ?) must have written a smaller hand than was normal.

We leave the matter for the present undecided.

Similar remarks will apply to the ist Epistle of Peter.

We annex the 2d Epistle of John, as we imagine it to have stood on the original sheets.

When we turn to the Gospels we have a much more difficult question to examine, on account of the multitude of various readings. We shall simply remark that the Gospel of Luke, in Codex B, is within a line of the end of a column, so that

Luke = 411 V-pages, wanting a single line.

In the Gospel of St. John, if we omit the last verse, we find ourselves at the end of a page, and

John = 291 V-pages exactly.

It will have been noticed that the number of V-pages occupied by the documents discussed is more often odd than even, which is more consistent with the hypothesis of papyrus sheets written on one side only, than with the supposition of a material capable of being written on both sides.¹

---

1. The more delicate papyri are quite unsuited to the reception of writing on both sides : that species, in particular, which was held in the highest Roman estimation, and honored with the name of Augustus, was so fine as to be almost transparent, so that its extreme tenuity came to be regarded as a defect.

For a document to be written on both sides seems to be a mark of the poverty of the writer or the over-productiveness of his brain : thus we find in Juvenal I 5 :

" Summa pleni jam margine libri Scriptus et in tergo, necdum finitus Orestes."

Lucian, Vit. Auct. 9, represents Diogenes as saying ?/ iriipa ^t cot depftuv eoTCi) ueary Kal oKLadoyfidcbuv [iip?Juv.

Scripture students will call to mind an illustration of a similar kind in the Apocalypse, where the plenitude of coming judgments and tribulations is represented by a book or paper-roll written both outside and inside (Rev. Vi).

---

Codex Sinaiticus

5. We shall now turn our attention to the Sinaitic Codex, which is written in columns, four to each page, and in lines, 48 to each column. ² The difficulty in this case will arise from the fact that the lines of the text are not nearly so uniform as in the Codex Vaticanus, and in the first two Gospels in particular the text is broken up into paragraphs, and the recurrence of short lines, unless it be a genealogical feature of the successive MS, will prevent us from tracing the structure of the original documents. We proceed, however, to form our second table, constructed in the same way as the previous one, and containing a larger collection of books. The lines in this manuscript are shorter than in B, by several letters.

---

2. This is not always true ; in the Catholic epistles the scribe has frequently contented himself with a column of 47 lines. I do not know whether this peculiarity has ever been noted. Scrivener, in his collation of the Sinaitic MS, does not seem to allude to it. Our results, as given in the table, must be corrected for the aberration of the scribe, when we come to analyse the documents more closely.

---

The first thing we notice is that the distribution of the concluding lines of the books is much more varied and irregular. The only thing that is remarkable is the recurrence of the multiples of twelve ; three books end at the twelfth line, viz. I Corinthians, I John, Revelation ; four end on the 24th line : Luke, Hebrews, Philemon, and II Peter ; the Gospel of John ends on the 35th line, which may practically be counted as the 36th.¹

This, again, can hardly be accidental ; we may assume that in the cases alhided to, with the exception of the ist Epistle of John, which, on account of the irregular length of the columns, furnishes an accidental coincidence, there is a unit sheet of paper employed, capable of containing 1 2 lines of the Sinaitic Codex ; we shall therefore have a new leaf of paper, (for reference to which we adopt the expression S-page, in order to distinguish it from the previous V-page), by means of which to measure our documents.

With regard to the comparative sizes of the two pages, it is evident at a glance that the S-page is smaller than the V-page, for it contains twelve lines where the other has fourteen, and has a smaller number of letters to the line.

---

1. It may be asked why, in discussing this table, we pay no attention to the repetition of the sixth line as an ending of three books, nor to the double recurrence of the number three. I have no theoretical objection to urge against either of these numbers, seeing that they are both submultiples of the whole column of 48 lines ; but practically they are too small subdivisions, and their recurrence is accidental. The probability that out of 28 books, one number should recur in the line-endings three times (I do not say this time a particular number) is represented by

[insert equation]

- whose value is nearly 21/25.

It is almost certain, then, that such an event as the recurrence alluded to will be found in our table. Those who are interested in observing these recurrences may study the following table from the Codex Sinaiticus :

[insert table]

Here every ending is formed by random distribution (unless we except the book of Judith and the Maccabees), for the works referred to are translations, and have therefore no pattern ; yet there is a double recurrence of the 3, and of the 38 with its submultiple 19. These are, of course, purely accidental. The recurrence would have to be more frequent before we should notice it, or look for any concealed cause at work to produce such a result.

---

Codex Sinaiticus (cont.)

6. We thus get the key to the method by which the text of the papyrus leaves was reduced into the shape in which we find it in the oldest manuscripts. Codex B selects the larger type of page, and arranges them nine on a page, or three in a side ; while the Sinaitic Codex selects the smaller leaf, and arranges them sixteen on a page, four in a side. And it is this arrangement which Eusebius¹ describes when he says that the accurate MSS, prepared by order of Constantine, were written τρισσα και τετρασσα as we should say, in a square whose side is three, or in a square whose side is four. The V-pages, then, are arranged τρισσα, and the S-pages τετρασσα.

7. Now, examining our second table, we see at once that the Sinaitic Codex gives

• Gospel of Matthew = 556 S-pages, and three letters.
• Gospel of Luke = 598 S-pages.
• I Corinthians = 205 S-pages exactly.
• Hebrews = 162 " "
• Philemon = 10 " "
• II Peter = 38 " "
• Revelation = 273 " "

We may perhaps conjecture that Titus should be added to the list, as containing 23 S-pages and one line ; while the Epistle to the Colossians is again doubtful, comprising 49 leaves and one line. We have thus deduced the type of almost all the Epistles, some of them with great exactness ; and we observe that they fall into two groups, with the exception of some four or five Epistles, which either are not written so as to fill the paper, or are written on paper of a different size to the two sorts we have been considering, or on a different pattern.

---

1. Eusebius, Vit. Const. IV 37. (Life of Constantine)

---

The Gospels, Omissions

8. When we turn to the Gospels we have a harder problem to solve, but I think we may say that if the two principal types of the early MSS are those indicated as τρισσα and τετρασσα, then it is far more likely that those types were found in the Gospels than that they were merely adopted from the Epistles. We may therefore expect to find some of the Gospels written τρισσα and some τετρασσα, or rather some on the V-page and some on the S-page.

The question is, how shall we determine the type of the autograph for any particular Gospel? And here an important remark must be made. I am aware that every one of these results and suggestions is subject to a disturbing factor of the greatest moment, viz. the question of various readings in the text, and of accidental omissions or insertions of passages or lines in the great Codices. The disturbance will be most to be apprehended in the case of the longer compositions, and with regard to these all our results must be looked upon at first as tentative. But in the smaller writings the various readings are generally so few and unimportant that the majority of our results may be regarded as unaffected by them. We will, however, examine the effect of these various readings in each of the separate books. It is the more important to do this carefully, because the Sinaitic and Vatican Codices are known to contain a number of apparent insertions and omissions and repetitions, which have been held up by a certain school as convincing proof of their unreliable character as witnesses to the text of the New Testament. Dr. Dobbin gave in the Dublin University Magazine for November, 1859, a calculation of the omissions of Codex B in the different books of the New Testament, in which we find:

[insert table]

An appalling table, certainly, and one which, if we did not remember that the figures are the result of a collation with the Textus Receptus [the Traditional Text = Byz], and that the majority of them refer to wholly insignificant readings, would almost make us despair of finding in the Vatican or Sinaitic MSS any traces of the original style and size of the books of the New Testament. We will, however, discuss any important readings that may occur, and after having first carefully dissected the text of St. John, and examined the bearing, of our investigation upon the stichometry of the New Testament, we will proceed to the Epistles, beginning with the smaller ones, and so working up to the longer Epistles, the Acts and the Gospels. And no result of the previous tentative examination is to be allowed to pass unchallenged or unverified.

---

John's Gospel and the PA

9. We begin with the Gospel according to John. In the Vatican Codex this occupies 97 columns and six lines. In the Sinaitic Codex it occupies 107 columns and 35 lines. At first sight, therefore, it seems that the Gospel is written on the S-page, with only a deficiency of one line from a total of 431 S-pages. But here comes in the question of the last verse of the Gospel, which Tischendorf observed to be written in the Sinaitic MSS by a different hand, and many scholia to different MSS affirm to be an addition. Removing this verse, eight lines of the Codex, the S-page is of course no longer apparent. But strange to say, when the verse is also removed from Codex B, in which it occupies six lines at the top of a page, we are left with a Gospel terminating at the end of a page, and in our notation occupying exactly 291 V-pages. The Gospel of John is, therefore, probably written on the V-page, and the apparent contradiction of this statement by the Sinaitic Codex may be due to the fact that in the type of MSS which that Codex has been following some one has utilized part of the blank space at the latter half of a column for the insertion of a sentence as to the number of books that might have been written. The addition must have been earlier than the age of vellum MSS, and may have arisen in the transcription of the Gospel of John from the larger-sized paper to the smaller, since it nearly fills the blank in a smaller sheet, and that sheet not the lowest in a Sinaitic column.

The Pericope Adultera (Jn 7:53-8:11)

10. This conclusion with regard to the autograph of St. John leads to very important consequences with regard to the celebrated pericope of the woman taken in adultery. An examination of this passage shows that there are 908 letters either inserted in the text or dropped from it. Now the average number of letters to the line in St. John's Gospel in the Codex Vaticanus is 16.4, from whence we conclude that the passage in question is equivalent to about 56 lines of Codex B, i. e. to four V-pages exactly.

Now it is obvious that four such pages could not by any possibility have been excised from a document in which the V-pages are arranged nine in a square. They must, therefore, have been lost from the original document before it came into the shape represented by Codex B. Their reinsertion has been characterized by great awkwardness in later manuscripts, and breaks the continuity of the narrative. They have been, in fact, restored to a place which they did not previously occupy.

The PA (reconstructed exemplar)

Before going further we insert a reproduction of the four pages which we have reason to believe the lost passage to have occupied.

As a restoration of the text of B, it is not quite a successful effort. I have not, I find, done justice to the syllabic division followed by the scribe, who has a distinct custom in ending his lines and dividing his words, and prefers, if possible, to write a seven-syllabled line. Moreover, some of the most capriciously concluded lines are meant to be syllabically. divided, such as those which end with ου and leave the κ of the ουκ to be carried to the next line. This division occurs so frequently that it is evident that the scribe, in writing such words as ουκ εστιν, really regards the κ as a sort of prefix to the verb.

We may now proceed to determine the place where the celebrated pericope should be reinserted. Turning to the end of the fifth chapter, we find that it closes with the words :

The scene then changes abruptly to Galilee:

It is between these chapters that I would locate the pericope. The 5th chapter narrates how Jesus found in the temple the man whom he had healed at the pool of Bethesda ; it describes the long subsequent discussion with the Pharisees, which must have taken nearly all day, after which they depart, each man to his own house, but Jesus to the Mount of Olives. Appropriately the Pharisees bring him next morning the woman for judgment, with the remark that " Moses in the law said . . . but what say est thou?".

Codex D, which gives the pericope in somewhat shorter form, is even more forcible, τι δε νυν λεγεις; we conclude, then, that this is a far more likely place to locate the pericope than at the end of the seventh chapter.

This readjustment of the text at once removes many of the objections urged against its authenticity, and it also helps to fill up that unsightly chasm at the close of the fifth chapter. It is unnecessary to discuss in detail the objections which had been raised by critics to the passage as it originally stood, but we will quote a single one out of many difficulties urged, as given by S. Davidson. He says :

I hope the obscurity disappears in the new arrangement of the text, and that the passage is more harmoniously placed with regard to the context than previously.

Moreover there is this difficulty, that in the ordinary supposition these lost V-pages would begin four lines from the top of the page, and we should have to assume that Codex B had either added four lines to the autograph, or lost ten lines in the first seven chapters, before we could rectify the pages so as to reintroduce the lost columns of the papyrus. Neither of these suppositions seems likely, as the text of John in these chapters is remarkably good, and the text of B is more likely to be marked by omissions than insertions.

On our hypothesis they begin on the last line of the left-hand column of the page, and we have only to assume that a single line has been lost from Codex B in the first five chapters. We proceed to go in search of this lost line. The Gospel of John in B has comparatively few various readings in the shape of insertions or omissions. The majority of them consist of transpositions and changes of rarely verbal importance. We proceed to tabulate those of them which affect our enquiry, from the principal editors and MSS.

Reviewing the variants of the text of B thus far, we find four cases of probable omission, and two of insertion. If we allow that B is right in omitting τον κραβαττον σου (Jn. 5:12), the result is a balance of a line to be added, which suits our case exactly.

---

John's Text after the PA (8:12-end)

11. We must now examine the remainder of the Gospel in the same manner.

The total result of our examination of this passage is that perhaps one or two lines might be added to the text of B, but the text has repeated more than five lines and dropped only three, so the total result is hardly affected.

It will be seen that we have made no allusion to the account of the troubling of the waters at Bethesda, which does not occupy a distinct number of V-pages.

But we must not altogether pass the passage by, for it enables us to see why the PA came to be inserted in the wrong place. There is no doubt whatever that the gloss in question is very early, seeing that we find a striking reference to it in Tertullian, De Baptism 9. Written on the V-pattern, the passage John V 3, 4 would occupy about 10 lines of manuscript. Bearing in mind that the passage to which the pericope de adidicra has been wrongly restored is four lines from the beginning of a column, and adding the gloss on the Troubling of the Water to the fifth chapter, we have now moved the inserted pericope to the beginning of a V-page. Each of the three errors, viz. the omission of the pericope, its reinsertion, and the insertion of the gloss in chapter V, is therefore anterior to the age of vellum manuscripts, and we can even arrange the errors in their proper chronological order. Perhaps we ought to have added that in the same interval of time a balance of a single line was lost from the first five chapters of B.

The majority of the errors are of the V-type, that is, there are more V-lines than S-lines inserted or omitted. And this is just what we should expect, if the MSS were originally of the V-pattern ; and we may lay down the following general principle : A manuscript originally written on a certain pattern will generally show a majority of errors of the pattern on which it is written. The advantage of this proposition is that it will help us to determine the original character of a MS, whether the MS occupy an exact number of pages of its pattern or not. We are now in the position to print the Gospel of John, approximately, from the original sheets.

No one can study the Gospel carefully without noticing the discontinuity of many of its sequences. The probability is that some passages are still lost from the 500 original sheets of the Gospel.

12. Now let us turn to the close of the Gospel and examine the endings of the 20th and 2 ist chapters : the similarity of the 30th verse of the 20th chapter to the last verse of the 21st chapter is unmistakable. The Gospel has apparently two endings. And here comes in the remarkable fact that Tertullian calls the 30th verse of the 20th chapter the close of the Gospel, although he quotes from the 21st chapter in at least two places :

The proper place for the two closing verses of the 20th chapter is most likely at the end of the 21st chapter.

For the expression that there were "many other signs not recorded which Jesus wrought" implies (just as the expression "I had many things to write to you" in the II and III of John) an insufficiency of writing material; we are close to the end of the roll of paper.

In the next place, the restoration of the closing verses of the 20th chapter to the end of the 21st is strikingly harmonious with the introduction of the Gospel, to which it returns as a keynote, and with the 24th verse of the 21st chapter which precedes it.

And thirdly there is room for a single conjectural emendation which adds vividness to the narrative. In Jn 21:30, after ενωποιν των μαθητων, many important MSS, especially those which exhibit a Western text, insert αυτου. It is a lawful suggestion that the original reading was simply ενωπιον αυτου, which was altered as soon as the verse had become severed from its proper connection.

The gospel now closes as follows:

---

1. Tertullian, Adv. Praxeam, 25.

---