One of the vital and until recently, more tedious, tasks in the work of textual criticism was that of collating every extant Greek manuscript or fragment of the New Testament. We may be overjoyed at the abundance of sources available to us, which include the papyri, the codices, and even citations in the fathers; without collation, however, we would have no practical way to access and use them.
Comparing One Text with Another
If you have a copy of the UBS Greek New Testament, or better, the Nestle-Aland (NA) text, you have the means to compare differing readings found in various sources with the reading in the text that has been chosen by the editors of that text. This is the end result of collation, and if you ever meet a collator on the street or in a coffee shop, be sure to thank that person for a hard job well done. Were it not for collation, most of us would not be aware, for example, that some important manuscripts have “nothing came to be” in John 1:3 while others have “not one thing came to be.” There are thousands of other variations across the many ancient manuscripts and fragments of the New Testament in existence today. It would be an impossible task for any one of us to somehow get physical access to all these manuscripts and compare their readings. Collators have collected all the data for us and organized the results, reporting in a concise manner which manuscripts have which readings.
Why Collation Is Important
I have little doubt that for many readers of the UBS or NA Greek texts, not to mention MT or TR proponents, only the text chosen by the editors of the Greek New Testament that they use matters. This could be the result of a bad habit: that of ignoring footnotes. I refrain from being judgmental because for many years, I was a lazy student and I tended to skip footnotes, let alone endnotes. Others may have a seemingly defensible argument, one that strikes me as analogous to the old saying, “A man with one clock knows what time it is; a man with two isn’t sure.” That is, these readers are content simply to let the text preferred by the editors be their own as if other editions of the Greek NT did not exist. Perhaps there is nothing wrong with that, but even in this case collation has already played a very large role by supplying the editors of the chosen text with the sources from which they edited their text.
Moreover, if I may use the old saying once again, there is something that the man with two clocks knows, of which the man with only one is blissfully ignorant: in reality he cannot be sure of what time it is (unless of course, he owns a modern atomic clock), and the man with just one (conventional) clock is only fooling himself. Similarly, the readers aware and in possession of only one Greek NT text might imagine that they are in possession of the complete, unadulterated word of God. However, the rest of us know better: God inspired the autographs, not the texts ultimately edited from them. The fact that the extant sources do not entirely agree denies us the right to claim that we have even one manuscript with a perfect copy of the autograph, and the most prudent conclusion seems to be that all of the autographs have perished.
Still, we want perfect copies of all the autographs of the Bible books, both New Testament and Old, and unless one (and only one) edition of the text is accepted by faith as perfect, the task of reconstructing the autograph falls to textual critics and to translators. For anyone unfamiliar with the specialized use of “critic” employed here, it is not a person hostile to the Bible, at least not by definition. TC’s, to abbreviate the term, compare all available readings and attempt to determine the reading most likely to be the “original” according to a set of criteria that are generally accepted by textual scholars. The text at which they arrive usually becomes the “base” text in the collation.
Translators are not textual critics primarily, and I suspect that a good number of them work only from the base text of a published Greek NT, because they either choose to work this way or are assigned this method by superiors. However, it would surprise me if the majority of translators did not do some textual criticism in the process of translation. This would be utterly impossible without collations. A good collation notes all variations of any significance and provides an acceptable list of sources supporting the variations. If the translators are permitted to choose a reading differing from that in the base text, they must have a thorough familiarity with the criteria of textual criticism and a particular translation philosophy that will probably emphasize certain criteria. Textual criticism is very demanding at the expert level, and most translators are not expert as both translators and TC’s. Increasingly, however, textual criticism is claiming a place in the translation process that demands competence in TC on the part of translators.
The Process and History of Collation Summarized
In the bad old days before computerized word processing, until fairly recently, everything in collation had to be done by hand. Collators often had to travel to obtain access to NT sources and had to pore over their sources and painstakingly write down every variation from a chosen base text, letter by letter. Choosing a base text was practical and the logical first step; instead of writing out in longhand how each manuscript reads, the collator could copy the base text, and then record only the variations from it found in other sources, listing those variations as notes on the text. In some cases, of course, it would be necessary to copy a source by hand.
A bare minimum of historical background will help to bring us where we find ourselves today. I believe everyone would agree that the first substantial collation of the New Testament was produced by John Mill and published by Oxford University in 1707. Mill died just two weeks after the publication, and scholars who thought that it put the veracity of Scripture in doubt, harshly criticized his collation. Others had to rise to his defense.
All in all, Mill spent about 30 years on this work. He had acquired 100 Greek manuscripts, various versions, and some writings of church fathers, but he apparently did not intend to produce a collation until the Bishop of Oxford saw his collection and urged him to do so, covering his expenses. Completion of the collation, which was very slow work, was further postponed when the bishop died, and Mill’s costs were no longer met. When the work was finally finished, the collation cited 30,000 variant readings found among the manuscripts, taking Stephanus’ Greek New Testament (eventually to become the Textus Receptus) as the base text. Excellent images of the Mill’s collation are available online thanks to the Center for the Study of New Testament Manuscripts.
Other editors and editions deserve to be mentioned, among them Tischendorf, Tregelles, and of course Wescott and Hort. Since our purpose here is only to provide a brief introduction to collations, however, I am venturing to commit a great sin of omission and jump well ahead to Eberhard Nestle. In 1898 Nestle published the first edition of Novum Testamentum Graece, which has become the standard text for NT scholars and (most) translators, at least those who produce English Bibles. This first edition was quite crude by current standards. The base text was the product of three other existing texts: Tischendorf, Westcott and Hort, and Weymouth (later that of Weiss). Wherever the texts of Tischendorf and Westcott and Hort agreed, Nestle put these readings into the text. When they disagreed, he put the reading that was supported by Weymouth or Weiss into the text and cited the other in the apparatus. His son Erwin later took over the work and supplied more information about the original sources along with needed corrections in the thirteenth edition (1927). In the editions that followed, he continued to expand the apparatus, but the information was second-hand, taken from other apparatuses.
We shall fast-forward again to the arrival of Kurt Aland on the scene. He brought a more acceptable methodology to the work as he checked and ensured that the information in the apparatus agreed with the original sources, and he introduced the readings of papyri that were discovered in the twentieth century. This culminated in the publication of the twenty-fifth edition (1963) of what was then to be known as the Nestle-Aland text. The text remained in principle the product of the majority of the three upon which it was based. The most significant changes were the result of replacing the text of Weymouth with that of Weiss, which Eberhard Nestle had intended but neither he nor his son had yet carried out completely.
More than a decade later, the NA text underwent major changes, with the twenty-sixth edition of 1979. Guided by an eclectic approach which they called the “local-genealogical method,” Aland and his colleagues completely overhauled the text, jettisoning the system that the Nestle’s had used and choosing readings based on a set of TC criteria, which we will discuss in the next chapter. In the process, they also jettisoned all positions both pro and con took by Westcott and Hort on the value of various key manuscripts. To aid readers, and perhaps to emphasize the fact that their new base text was uniquely theirs, they provided an appendix listing the differences between this text and earlier texts, including NA25. I myself have consulted it many times, whenever I had the impression that something in the text was different.
Before concluding our excursion into the past, let me emphasize two trends and what we see happening in each. In one camp, we have those who abide by the text behind the KJV, a text they consider unchanged throughout the centuries, i.e. the Textus Receptus (TR). This, again, is the one-clock view of the Scriptures, and it is a matter of faith. I pass over its sister text the Majority Text, for which a defense based largely on non-standard criteria is offered. In the other camp, we see a search for the original text, a text that changes over time as discoveries of manuscripts are made. Theoretically, the text of every manuscript ever discovered could have been published line by line in interlinear fashion, with the understanding that it would be up to the reader to determine the value of each. In practice, though, this is unrealistic (or at least has been considered impractical), so a respected text traditionally has been chosen as a base text, with variations from it in other manuscripts noted.
We have already seen that the original base text essentially was an early version of the Textus Receptus, which may seem ironic, but actually makes sense because there was no other established text, and the TR had not yet come under criticism. What is important for us in this discussion is that with the Nestle-Aland (NA) text of the twenty-sixth edition, we were given a new base text which, like those before it, was the product of the editors. However, it was not a previously-established text. Its value depended entirely on the judgment of its editors in making textual decisions that resulted in changes. This has become a trend in the present century.
Moreover, if the full implications of the NA collation with a base text constructed by and credited to the editors were not clear at once, they were spelled out in the introduction of the twenty-seventh edition, whose text was identical to that of the twenty-sixth. This is well worth quoting:
The purpose of the 27th edition remains the same as that of the 26th edition. It intends to provide the user with a well-founded working text together with the means of verifying it or alternatively of correcting it. Correspondingly the edition contains all the variants necessary for this purpose in as complete a form as possible within its limitations.
One could hardly compose a better definition of a collation, not that this was intended as such. As confident as the editors were in the great majority of chosen variants in their text, they harbored no illusions that it was a perfect reconstruction of the original. For better or worse, it was simply the NA text. They also conceded that they could not provide the reader with all the variant readings in all the extant manuscripts. Today, as the result of work using the Coherence-Based Genealogical Method (CBGM, to be discussed later), we have a new, NA28 text which remains substantially the same as NA26/27. That is due, however, to the fact that only the General Epistles have thus far been reedited.
As the NA text will continue to change, one of the features about it that will remain the same is the fact that, from the beginning, it has never been identical to any existing Greek manuscript. This fact is seen quite explicitly in the quotation above where it is called a “working text.” Yet it has been and probably will continue to be the prevailing Greek text among scholars and translators alike (though see below). At the same time, however, the NA editors gave all users what amounts to an open invitation to construct their own versions of the Greek New Testament. Two publishers undertook this opportunity on a major scale.
Other Base Texts
Zondervan did this in an indirect way by having two scholars construct the Greek text underlying Today’s New International Version. Evidently, neither man worked as a translator on the TNIV, nor we do not know the exact process by which they constructed the Greek, which was titled A Reader’s Greek New Testament. I would assume that whenever there was a question about the variant reading chosen–and there must have been many given the thought-for-thought style of translation–the two scholars would simply consult a TNIV secretary or translator who would know. For our purposes it, fortunately, does not matter; we are only interested in the fact that the TNIV (and I would assume NIV) translators did in effect choose to construct their own version of the Greek text from the NA collation and other sources. I need to add that no significant attempt was made by Zondervan to produce their own collation. The only apparatus similar to that is one that points out disagreements with the UBS text (=NA, see below) and makes references to various unnamed manuscripts.
A formal construction of a different text was performed Michael W. Holmes for the Society of Biblical Literature. While it is probably fair to say that he accepted the invitation of the NA editors, at least in spirit, he officially used other edited NT Greek texts as his primary sources, much as the Nestles originally did. Holmes did, however, compare his work with the NA text, which is cited in the apparatus when he disagrees with it. His text was published as The Greek New Testament SBL Edition or SBLGNT. Holmes is an admirer of Günther Zuntz and his view of NT textual history, which gives full credibility to the Western textual tradition and accords value to various Byzantine readings as well. While Holmes includes an apparatus, he provides only a primitive collation of his sources by modern standards and does not mention actual manuscripts. Particularly odd is his apparatus notation “Holmes,” referring to any source that is not to be found in one of his listed sources and is otherwise unnamed. The curious reader is left to find the preferred reading in a collation elsewhere.
To some readers, the choice or construction of an edited text as the base text in a collation–not to mention such a text as the source of Bible translation–is probably unsettling. The late Yale scholar Reuben Swanson not only expressed his own misgivings about the existing edited texts but also went to great efforts to do something about it. He maintained that all the standard Greek texts ultimately rested on the Textus Receptus of 1633 which was a very mixed text, and he deplored the eclectic method of editing the text used ever since then, suggesting that the reader does not always know what has been edited.
As a solution, Swanson published fascicles of individual NT books in which he featured codex Vaticanus (B) as the base text, and he provided interlinear parallel readings of the sources provided in the UBS text (more about this below). He was well aware that of the early manuscripts with solid credentials only codex Sinaiticus is complete, but he emphasized that Vaticanus is recognized among scholars as the best, and he planned to decide on another source to cover the missing sections in Vaticanus. Before his death, he published the Gospels through Galatians.
I think there is value in Swanson’s decision to use Vaticanus as the base text. Among the more voluminous manuscripts, it has an early date, and it consistently ranks highest among extant manuscripts in exhibiting variant readings that are most likely to be original, at least by the generally accepted criteria. It, therefore, commends itself as a good candidate for the honor of extant manuscript that contains the autograph in large part. It is not perfect, however, and we would be naïve to assume that it escaped all the same effects of eclectic editing that Swanson condemned when he chose it.
I think we have surveyed enough for a general understanding of the concept of a base text in a collation. Now I propose that we look at the apparatus, which is really the substance of a collation. You will recall that when we discussed the Reader’s GNT and the SBLGNT, I pointed out that neither comes with a collation of manuscripts, even though each has an apparatus. Of course, I cannot say for certain, but I suspect that this is due in large part to the simple fact that constructing and presenting an acceptable collation is very difficult. It has always been assumed that to be useful; a collation needs to be available in a manageable form for the user. Typically, in a modern collation, this calls for an apparatus of footnotes at the bottom of each page that provides two vital pieces of information: a listing of differing readings whenever one or more manuscripts has a different reading from that of the base text; and a list of the manuscripts and other sources that have each of the readings cited. Usually, some system is provided to inform the user of which sources have the reading chosen for the text.
Now, given the large number of Greek NT manuscripts extant, one usually does not read more than two or three lines of text without encountering a place where there are variant (i.e. differing) readings, and it is not unusual to find three or more variants in a single line. Consequently, editors and publishers of a Greek text who are providing genuine collations have had to become creative in providing sufficient information in so cramped a space as is available in a hand-carried New Testament. So once the base text has been established, the editors must compile all the differences that exist among other manuscripts that are chosen for inclusion, and a system of presentation must be developed that is suitable for the user’s purposes.
In the bad old days, which I again use as our backdrop, compiling the textual information was incredibly painstaking and eye straining. I am sure that different people and organizations used different methods, but essentially every letter of every word had to be checked in corresponding manuscripts, and lists of variants relative to the base text, along with which manuscripts displayed them, had to be compiled. Once individual manuscripts took the place of edited texts in the apparatus of the NA text, as much information as possible had to be included because textual scholars needed to know every place where a variant existed, what the variants were, and what manuscripts supported each variant, if they were to make intelligent choices of readings. The only way to do this was to use single-character symbols etc. in the text and apparatus to save precious space, and the result seemed user-hostile to many readers. Even then, not every possible variant and source could be listed by any means. However, there was enough manuscript evidence presented so that users who were familiar with traditional TC could make satisfactory choices. The traditional lines of transmission of the NT text were reasonably clear.
In the mid-1960s Eugene Nida assumed the role of spokesman for translators worldwide who needed a Greek text with an apparatus that was not cryptic, as the NA apparatus could be described. There was no way to do this without adversely affecting the portability of the text, or reducing the amount of information conveyed. The latter approach was taken, with the result that many variants that were not considered essential to the translation process were eliminated. The rating system which I mentioned above (p. 258), representing the level of confidence on the part of the editors in their choices of some readings (from A for certainty to D for great difficult in deciding), was also introduced. This eventually became The Greek New Testament of the United Bible Societies and is now prepared by the Institute for New Testament Textual Research.
You might wonder whether the UBS GNT was really necessary, given that its text is the same as that of NA and it has only a subset of the NA’s variant readings. I can attest that it is entirely inadequate for me as a translator of English Bibles and that I have managed to become fairly comfortable with the symbols and abbreviations in the NA. However, I cannot speak for translators worldwide, and I certainly can agree with them that the NA text and apparatus is challenging at times, and perhaps even overwhelming for someone unfamiliar with it and lacking the time to master it. The rating system is only included in the GNT, but then that pales in value in comparison to Metzger’s Commentary, since the GNT has only the letter grades. The additional commentary provided by Metzger in the companion book makes it indispensable.
The main reason I mention the UBS GNT is to illustrate that not all collations are equal, and perhaps they do not need to be. In this particular case, an existing collation has been highly abbreviated to serve a specific purpose, with the pro’s and con’s all acknowledged. I will now close this chapter with a brief look at the future of NT collation.
The Future of Collation
One thing that is immediately apparent for collation from the present on is that we can say goodbye to much of the tedious work because computers are very good at doing these jobs. We, humans, were created for better things–not that I do not honor and admire the dedication of those who did such work in the past! We cannot eliminate all of the tedium, unfortunately. For example, humans still have to input or at a minimum verify the accuracy of inputting ancient handwritten text into machine-readable form. But once the text is accurately recorded for all time, we can have computers check the texts of different manuscripts against each other for agreements and disagreements and perform most other tasks of collation that previously had to be done by humans.
Now then, if you have purchased NA28 or GNT5 recently and have discovered that only James through Jude has been reedited, you may feel a little short-changed. You probably should not. If nothing else, you at least have a window into the future of collation. What you find in these short books is the result of the ECM2 committee’s work applying the massive collation of the Institute for New Testament Textual Research and using the tools of the Coherence-Based Genealogical Method. In the NA28 and GNT5 apparatuses, you have additional citations of papyri, but also the loss of citations of various variant readings, so do not throw away your previous editions of these texts unless you also plan to purchase the ECM2 itself.
Permit me to do some forecasting and advisement beyond this. If you want to know as much as possible about the Greek text, and if NA28 and GNT5 are representative of what the final NA and GNT texts and apparatuses will be (as I suspect they are), then NA28 and GNT5 will be inadequate for you. For a full reporting of the text and accompanying collation in printed form, you will need the ECM2. On the other hand, if you are content with the methods of traditional TC, then NA26/27 and GNT4 will continue to serve you well, and if you wish, you can consult the online CBGM tools or purchase NA28 or GNT5 to see where the editors have changed their previous decisions in the General Epistles. GNT5, of course, will supply the additional information of ratings and have a clearer apparatus, but at the expense of omitting numerous variant readings.
Whether you purchase the ECM2 or not, I strongly encourage you to consult the online CBGM tools, which I will discuss in chapter xii. They provide invaluable information and are free to all users, at least for now. These tools are actually much better as a source of information than the NA28 and GNT5 apparatuses, and even extend the usefulness of the ECM2 considerably.
As to other collations, I can hardly think of any organizations more resourceful than Zondervan and the SBL, and so far neither has made any discernable attempt to add collations to their published Greek texts (A Reader’s Greek New Testament and the SBLGNT respectively). Indeed, as we have seen, they have not even clarified for their readers what individual manuscripts their texts are based upon when they differ from the NA.
The closest thing we have seen to a rival collation, at least in my experience, is Swanson’s work. No one has stepped in to complete that project after his death, and for good reason, I think. Swanson had complained about unidentified mixture in base texts, but now what we have with all of our resources and computing power is common material among all the Greek manuscripts in existence. When we had only print resources, using Swanson’s interlinear layout was a pleasure compared to the relative eyesore encountered in the NA apparatus, and Swanson had the extant sources accounted for just as well. Now, however, the Institute for New Testament Textual Research (INTTR) has virtually all the Greek NT manuscripts in existence online, and one can easily see clean representations of all the variants and can even choose any variant temporarily as the base text for comparison.
I think it would be impossible to compete with such a resource and organization, at least so long as the organization provides convenient, free access to its sources. It has flaws, certainly, some of which I will point out in the chapter on the CBGM, and we can only hope that these will be addressed. The intention has also been expressed to make the ECM available online, and that has yet to be implemented. So long as the database is properly managed and good access provided, however, I expect that it will be the one everyone uses in place of everything else that may still be available.
Ultimately, then, I foresee everyone who edits a Greek text of the New Testament linking it to the ECM in some way. Today, probably the only way to do this, at least off-site, would be to set up one’s preferred readings as the chosen text and copy and record the results from the CBGM tools, variant by variant. It would undoubtedly be better for someone planning a new edition of the Greek NT to do it by consulting the CBGM, taking full advantage of the information available nowhere else.
In the publication of a new Greek text, we should then expect to see some accounting of the sources behind the editor’s (or editors’) choices relative to the ECM. I do not know what, if any, copyright issues might be involved at present in listing the actual manuscripts, but I would think that at least it would be possible for editors to note that a particular choice was the ‘b’ or ‘c’ etc. reading in the ECM.
Please Support the Textual and Bible Translation Work
SCROLL THROUGH DIFFERENT CATEGORIES BELOW
BIBLE TRANSLATION AND TEXTUAL CRITICISM
BIBLICAL STUDIES / INTERPRETATION
CHRISTIAN APOLOGETIC EVANGELISM
CHURCH HEALTH, GROWTH, AND HISTORY
 It may be necessary at times to distinguish the term “original” from “autograph,” terms also discussed in the Glossary. The autograph clearly is the text penned by the NT author himself, though some TC’s maintain that there could be two or more versions of an autograph if the author later rewrites portions of the text. “Original” most often will be synonymous with “autograph,” but could refer loosely to what is now called the “initial text,” i.e. the text to which all extant manuscripts can ultimately be traced. This “initial text” can also be understood as the autograph, but many TC’s prefer “initial” to provide a kind of buffer to avoid dispute about the autograph. That is, the “initial text” by definition may or may not be the autograph; it is, however, by definition the text from which the very first copy was made. Everyone can agree about that, whether or not they can agree on how the autograph was composed, whether it is really possible to reconstruct the autograph, etc.
John Mill, Novum Testamentum. http://www.csntm.org/printedbook/viewbook/JohnMillNovumTestamentum1707.
 If all three disagreed, Nestle selected a reading for the text that represented a compromise between them.
 See the Introduction to the twenty-sixth edition of the NA text, p. 43*.
Barbara and Kurt Aland et al., Novum Testamentum Graece et Latine, 27th Ed. (Stuttgart: Deutsche Bibelgesellschaft, 1994), 2*-3*. My apologies for citing only the Greek-Latin edition here (I recommend it for anyone interested in learning Latin).
 A Reader’s Greek New Testament: Third Edition, 3rd Ed. (Grand Rapids, Mich: Zondervan, 2015).
 Actually, differences with the NIV are also cited, and disagreements with the NA are specifically cited only when the NIV also differs with the NA, which I find odd. It would seem that pride of place should go to the NA text, given that the NIV is based on it, and one would expect the NIV to be cited only secondarily (if at all).
 This includes only a selected group of representative Byzantine manuscripts.