Rules of Textual Criticism

When the manuscripts differ, how do scholars decide which words are the original ones? There is more to it than simply choosing the readings of the oldest available manuscripts. Here are three historically important sets of rules published by some influential scholars of textual criticism: Bengel, Griesbach, and Hort.

Critical Rules of Johann Albrecht Bengel

In his essay Prodromus Novi Testamenti recte cauteque ordinandi [Forerunner of a New Testament to be settled rightly and carefully], (Denkendorf, 1725), Johann Albrecht Bengel, a Lutheran schoolmaster, published a prospectus for an edition of the Greek Testament which he had already begun to prepare (published in 1734). In it he outlines his text-critical principles, which included a novel classification of manuscripts into two primitive groups: the Asiatic and the African. The first group he supposed to be of Byzantine origin, and to it belonged the majority of modern manuscripts and the Syriac version; the second, of Egyptian provenance, was represented by Codex Alexandrinus and the manuscripts of the early Latin and Coptic versions. In this work Bengel also set forth a very influential rule of criticism: a preference for harder readings. This rule he expressed in four pregnant words:

proclivi scriptioni praestat ardua. "before the easy reading, stands the difficult."

The "Monita" of Bengel

In Bengel's Preface to his Gnomon Novi Testamenti (Tubingen, 1742) he includes an enumerated list of 27 "suggestions" (Monita) which may be taken as a summary of his critical principles. The following extract of these is taken from pages 13 through 17 of Fausset's translation:

"1. By far the more numerous portions of the Sacred Text (thanks be to God) labour under no variety of reading deserving notice.

2. These portions contain the whole scheme of salvation, and establish every particular of it by every test of truth.

3. Every various reading ought and may be referred to these portions, and decided by them as by a normal standard.

4. The text and various readings of the New Testament are found in manuscripts and in books printed from manuscripts, whether Greek, Latin, Graeco-Latin, Syriac, etc., Latinizing Greek, or other languages, the clear quotations of Irenaeus, etc., according as Divine Providence dispenses its bounty to each generation. We include all these under the title of Codices, which has sometimes as comprehensive a signification.

5. These codices, however, have been diffused through churches of all ages and countries, and approach so near to the original autographs, that, when taken together, in all the multitude of their varieties, they exhibit the genuine text.

6. No conjecture is ever on any consideration to be listened to. It is safer to bracket any portion of the text, which may haply to appear to labour under inextricable difficulties.

7. All the codices taken together, should form the normal standard, by which to decide in the case of each taken seperately.

8. The Greek codices, which posses an antiquity so high, that it surpasses even the very variety of reading, are very few in number: the rest are very numerous.

9. Although versions and fathers are of little authority where they differ from the Greek manuscripts of the New Testament, yet, where the Greek mauscripts of the New Testament differ from each other, those have the greatest authority, with which versions and fathers agree.

10. The text of the Latin Vulgate, where it is supported by the consent of the Latin fathers, or even of other competent witnesses, deserves the utmost consideration, on account of its singular antiquity.

11. The number of witnesses who support each reading of every passage ought to be carefully examined: and to that end, in so doing, we should separate those codices which contain only the Gospels, from those which contain the Acts and the Epistles, with or without the Apocalypse, or those which contain that book alone; those which are entire, from those which have been mutilated; those which have been collated for the Stephanic edition, from those which have been collated for the Complutensian, or the Elzevirian, or any obscure edition; those which are known to have been carefully collated, as, for intance, the Alexandrine, from those which are not known to have been carefully collated, or which are known to have been carelessly collated, as for instance the Vatican manuscript, which otherwise would be almost without an equal.

12. And so, in fine, more witnesses are to be preferred to fewer; and, which is more important, witnesses who differ in country, age, and language, are to be preferred to those who are closely connected with each other; and, which is most important of all, ancient witnesses are to be preferred to modern ones. For, since the original autographs (and they were written in Greek) can alone claim to be the well-spring, the amount of authority due to codices drawn from primitive sources, Latin, Greek, etc., depends upon their nearness to that fountain-head.

13. A Reading, which does not allure by too great facility, but shines with its own native dignity of truth, is always to be preferred to those which may fairly be supposed to owe their origin to either the carelessness or the injudicious care of copyists.

14. Thus, a corrupted text is often betrayed by alliteration, parallelism, or the convenience of an Ecclesiastical Lection, especially at the begining or conclusion of it; from the occurence of the same words, we are led to suspect an omission; from too great facility, a gloss. Where the passage labours under a manifold variety of readings, the middle reading is the best.

15. There are, therefore, five principal criteria, by which to determine a disputed text. The antiquity of the witnesses, the diversity of their extraction, and their multitude; the apparent origin of the corrupt reading, and the native colour of the genuine one.

16. When these criteria all concur, no doubt can exist, except in the mind of a sceptic.

17. When, however, it happens that some of these criteria may be adduced in favour of one reading, and some in favour of another, the critic may be drawn sometimes in this, sometimes in that direction; or, even should he decide, others may be less ready to submit to his decision. When one man excels another in powers of vision, whether bodily or mental, discussion is vain. In such a case, one man can neither obtrude on another his own conviction, nor destroy the conviction of another; unless, indeed, the original autograph Scriptures should ever come to light."

Following this are ten more paragraphs, numbered 18 through 27, which do not pertain to the evaluation of various readings, but instead contain sundry remarks relative to the design and use of his critical edition. The seventeen given above may therefore be taken as Bengel's formally stated canons of criticism.

Griesbach's Fifteen Rules

In the Introduction to his second edition of the Greek New Testament (Halle, 1796) Griesbach set forth the following list of critical rules, by which the intrinsic probabilities may be weighed for various readings of the manuscripts. Rules for the prior evaluation of documentary evidence, such as the ones formulated by Bengel, are implicit in Griesbach's theory of the manuscript tradition, and so they are not taken up here. What follows is a translation of Griesbach's Latin as it was reprinted by Alford in the Introduction of his Greek Testament (London, 1849. Moody reprint, page 81).

1. The shorter reading, if not wholly lacking the support of old and weighty witnesses, is to be preferred over the more verbose. For scribes were much more prone to add than to omit. They hardly ever leave out anything on purpose, but they added much. It is true indeed that some things fell out by accident; but likewise not a few things, allowed in by the scribes through errors of the eye, ear, memory, imagination, and judgment, have been added to the text. The shorter reading, even if by the support of the witnesses it may be second best, is especially preferable-- (a) if at the same time it is harder, more obscure, ambiguous, involves an ellipsis, reflects Hebrew idiom, or is ungrammatical; (b) if the same thing is read expressed with different phrases in different manuscripts; (c) if the order of words is inconsistent and unstable; (d) at the beginning of a section; (e) if the fuller reading gives the impression of incorporating a definition or interpretation, or verbally conforms to parallel passages, or seems to have come in from lectionaries.

But on the contrary we should set the fuller reading before the shorter (unless the latter is seen in many notable witnesses) -- (a) if a "similarity of ending" might have provided an opportunity for an omission; (b) if that which was omitted could to the scribe have seemed obscure, harsh, superfluous, unusual, paradoxical, offensive to pious ears, erroneous, or opposed to parallel passages; (c) if that which is absent could be absent without harm to the sense or structure of the words, as for example prepositions which may be called incidental, especially brief ones, and so forth, the lack of which would not easily be noticed by a scribe in reading again what he had written; (d) if the shorter reading is by nature less characteristic of the style or outlook of the author; (e) if it wholly lacks sense; (f) if it is probable that it has crept in from parallel passages or from the lectionaries.

2. The more difficult and more obscure reading is preferable to that in which everything is so plain and free of problems that every scribe is easily able to understand it. Because of their obscurity and difficulty chiefly unlearned scribes were vexed by those readings-- (a) the sense of which cannot be easily perceived without a thorough acquaintance with Greek idiom, Hebraisms, history, archeology, and so forth; (b) in which the thought is obstructed by various kinds of difficulties entering in, e.g., by reason of the diction, or the connection of the dependent members of a discourse being loose, or the sinews of an argument, being far extended from the beginning to the conclusion of its thesis, seeming to be cut.

3. The harsher reading is preferable to that which instead flows pleasantly and smoothly in style. A harsher reading is one that involves an ellipsis, reflects Hebrew idiom, is ungrammatical, repugnant to customary Greek usage, or offensive to the ears.

4. The more unusual reading is preferable to that which constitutes nothing unusual. Therefore rare words, or those at least in meaning, rare usages, phrases and verbal constuctions less in use than the trite ones, should be preferred over the more common. Surely the scribes seized eagerly on the more customary instead of the more exquisite, and for the latter they were accustomed to substitute definitions and explanations (especially if such were already provided in the margin or in parallel passages).

5. Expressions less emphatic, unless the context and goal of the author demand emphasis, approach closer to the genuine text than discrepant readings in which there is, or appears to be, a greater vigor. For polished scribes, like commentators, love and seek out emphases.

6. The reading that, in comparison with others, produces a sense fitted to the support of piety (especially monastic) is suspect.

7. Preferable to others is the reading for which the meaning is apparently quite false, but which in fact, after thorough examination, is discovered to be true.

8. Among many readings in one place, that reading is rightly considered suspect that manifestly gives the dogmas of the orthodox better than the others. When even today many unreasonable books, I would not say all, are scratched out by monks and other men devoted to the Catholic party, it is not credible that any convenient readings of the manuscripts from which everyone copied would be neglected which seemed either to confirm splendidly some Catholic dogma or forcefully to destroy a heresy. For we know that nearly all readings, even those manifestly false, were defended on the condition that they were agreeable to the orthodox, and then from the beginning of the third century these were tenaciously protected and diligently propagated, while other readings in the same place, which gave no protection to ecclesiastical dogmas, were rashly attributed to treacherous heretics.

9. With scribes there may be a tendency to repeat words and sentences in different places having identical terminations, either repeating what they had lately written or anticipating what was soon to be written, the eyes running ahead of the pen. Readings arising from such easily explained tricks of symmetry are of no value.

10. Others to be led into error by similar enticements are those scribes who, before they begin to write a sentence had already read the whole, or who while writing look with a flitting eye into the original set before them, and often wrongly take a syllable or word from the preceding or following writing, thus producing new readings. If it happens that two neighbouring words begin with the same syllable or letter, an occurance by no means rare, then it may be that the first is simply ommitted or the second is accidentally passed over, of which the former is especially likely. One can scarcely avoid mental errors such as these, any little book of few words to be copied giving trouble, unless one applies the whole mind to the business; but few scribes seem to have done it. Readings therefore which have flowed from this source of errors, even though ancient and so afterwards spread among very many manuscripts, are rightly rejected, especially if manuscripts otherwise related are found to be pure of these contagious blemishes.

11. Among many in the same place, that reading is preferable which falls midway between the others, that is, the one which in a manner of speaking holds together the threads so that, if this one is admitted as the primitive one, it easily appears on what account, or rather, by what descent of errors, all the other readings have sprung forth from it.

12. Readings may be rejected which appear to incorporate a definition or an interpretation, alterations of which kind the discriminating critical sense will detect with no trouble

13. Readings brought into the text from commentaries of the Fathers or ancient marginal annotations are to be rejected, when the great majority of critics explain them thus. ("He proceeds at some length to caution against the promiscuous assumption of such corruptions in the earlier codices and versions from such sources." - Alford)

14. We reject readings appearing first in lectionaries, which were added most often to the beginning of the portions to be read in the church service, or sometimes at the end or even in the middle for the sake of contextual clarity, and which were to be added in a public reading of the series, [the portions of which were] so divided or transposed that, separated from that which preceeds or follows, there seemed hardly enough for them to be rightly understood. ("Similar cautions are here added against assuming this too promiscuously." - Alford)

15. Readings brought into the Greek manuscripts from the Latin versions are condemned. ("Cautions are here also inserted against the practice of the earlier critics, who if they found in the graeco-latin MSS. or even in those of high antiquity and value, a solitary reading agreeing with the Latin, hastily condemned that codex as latinizing." - Alford)

Latin text of the above

1. Brevior lectio, nisi testium vetustorum et gravium auctoritate penitus destituatur, praeferenda est verbosiori. Librarii enim multo proniores ad addendum fuerunt, quam ad omittendum. Consulto vix unquam praetermiserunt quicquam, addiderunt quam plurima: casu vero nonnulla quidem exciderunt, sed haud pauca etiam oculorum, aurium, memoriae, phantasiae ac judicii errore a scribis admisso, adjecta sunt textui. In primis vero brevior lectio, etiamsi testium auctoritate inferior sit altera, praeferenda est-- (a) si simul durior, obscurior, ambigua, elliptica, hebraizans aut soloeca est, (b) si eadem res variis phrasibus in diversis codicibus expressa legitur; (c) si vocabulorum ordo inconstans est et instabilis; (d) in pericoparum initiis; (e) si plenior lectio glossam seu interpretamentum sapit, vel parallelis locis ad verbum consonat, vel e lectionariis immigrasse videtur.

Contra vero pleniorem lectionem breviori (nisi hanc multi et insignes tueantur testes) anteponimus-- (a) si omissioni occasionem praebere potuerit homoeoteleuton; (b) si id quod omissum est, librariis videri potuit obscurum, durum, superfluum, insolens, paradoxum, pias aures offendens, erroneum, aut locis parallelis repugnans; (c) si ea quae absunt, salvo sensu salvaque verborum structura abesse poterant, e quo genere sunt propositiones, quod vocant, incidentes, praesertim breviores, et alia, quorum defectum librarius relegens quae scripserat haud facile animadvertebat; (d) si brevior lectio ingenio, stylo aut scopo auctoris minus conveniens est. (e) si sensu prorsus caret; (f) si e locis parallelis aut e lectionariis eam irrepsisse probabile est.

2. Difficilior et obscurior lectio anteponenda est ei, in qua omnia tam plana sunt et extricata, ut librarius quisque facile intelligere ea potuerit. Obscuritate vero et difficultate sua eae potissimum indoctos librarios vexarunt lectiones-- (a) quarum sensus absque penitiore graecismi, hebraismi, historiae, archaeologiae, &c. cognitione perspici non facile poterant, (b) quibus admissis vel sententia, varii generis difficultatibus obstructa, verbis inesse, vel aptus membrorum orationis nexus dissolvi, vel argumentorum ab auctore ad confirmandam suam thesin prolatorum nervus incidi videbatur.

3. Durior lectio praeferatur ei, qua posita, oratio suaviter leniterque fluit. Durior autem est lectio elliptica, hebraizans, soloeca, a loquendi usu graecis consueto adhorrens aut verborum sono aures offendens.

4. Insolentior lectio potior est ea, qua nil insoliti continetur. Vocabula ergo rariora, aut hac saltem significatione, quae eo de quo quaeritur loco admittenda esset, rarius usurpata, phrasesque ac verborum constructiones usu minus tritae, praeferantur vulgatioribus. Pro exquisitioribus enim librarii usitatiora cupide arripere, et in illorum locum glossemata et interpretamenta (praesertim si margo aut loca parallela talia suppeditarent) substituere soliti sunt.

5. Locutiones minus emphaticae, nisi contextus et auctoris scopus emphasin postulent, propius ad genuinam scripturam accedunt, quam discrepantes ab ipsis lectiones quibus major vis inest aut inesse videtur. Erudituli enim librarii, ut commentatores, emphases amabant ac captabant.

6. Lectio, prae aliis sensum pietati (praesertim monasticae) alendae aptum fundens, suspecta est.

7. Praeferatur aliis lectio cui sensus subest apparenter quidem falsus, qui vero re penitus examinata verus esse deprehenditur.

8. Inter plures unius loci lectiones ea pro suspecta merito habetur, quae orthodoxorum dogmatibus manifeste prae caeteris faciet. Cum enim codices hodie superstites plerique, ne dicam omnes, exarati sint a monachis aliisque hominibus catholicorum partibus addictis, credibile non est, hos lectionem in codice, quem quisque exscriberet, obviam neglexisse ullam, qua catholicorum dogma aliquod luculenter confirmari aut haeresis fortiter jugulari posse videretur. Scimus enim, lectiones quascunque, etiam manifesto falsas, dummodo orthodoxorum placitis patrocinarentur, inde a tertii saeculi initiis mordicus defensas seduloque propagatas, caeteras autem ejusdem loci lectiones, quae dogmati ecclesiastico nil praesidii afferrent haereticorum perfidae attributas temere fuisse.

9. Cum scribae proclives sint ad iterandas alieno loco vocabulorum et sententiarum terminationes easdem, quas modo scripsissent aut mox scribendas esse, praecurrentibus calamum oculis, praeviderent, lectiones ex ejusmodi rhythmi fallacia facillime explicandae, nullius sunt pretti.

10. Hisce ad peccandum illecebris similes sunt aliae. Librarii, qui sententiam, antequam scribere eam inciperent, totam jam perlegissent, vel dum scriberent fugitivo oculo exemplum sibi propositum inspicerent, saepe ex antecedentibus vel consequentibus literam, syllabam aut vocabulum perperam arripuerunt, novasque sic lectiones procuderunt. Si v.c. duo vocabula vicina ab eadem syllaba vel litera inciperent, accidit haud raro, ut vel prius plane omitteretur, vel posteriori temere tribueretur, quod priori esset peculiare. Ejusmodi hallucinationes vix vitabit, qui libello paullo verbosiori exscribendo operam dat, nisi toto animo in hoc negotium incumbat: id quod pauci librarii fecisse videntur. Lectiones ergo, quae ex hoc errorum fonte promanarunt, quantumvis vetustae ac consequenter in complures libros transfusae sint, recte rejiciuntur, praesertim si codices caeteroqui cognati ab hujus labis contagio puri deprehendantur.

11. E pluribus ejusdem loci lectionibus ea praestat, quae velut media inter caeteras interjacet; hoc est ea, quae reliquarum omnium quasi stamina ita continet, ut, hac tanquam primitiva admissa, facile appareat, quanam ratione, seu potius quonam erroris genere, ex ipsa caeterae omnes propullularint.

12. Repudiantur lectiones glossam seu interpretamentum redolentes, cujus generis interpolationes nullo negotio emunctioris naris criticus subolfaciet.

13. Rejiciendas esse lectiones, e Patrum commentariis aut scholiis vetustis in textum invectas, magno consensu critici docent....

14. Respuimus lectiones ortas primum in lectionariis, quae saepissime in anagnosmatum initiis ac interdum in clausulis etiam atque in medio contextu claritatis causa addunt, quod ex orationis serie supplendum esset, resecantque vel immutant, quod, sejunctum ab antecedentibus aut consequentibus, vix satis recte intelligi posse videretur....

15. Damnandae sunt lectiones e latina versione in graecos libros invectae....

Theories of Westcott and Hort

In 1881 two English scholars, B.F. Westcott and F.J.A. Hort, published a very influential edition of the Greek Testament: The New Testament in the Original Greek (New York: Harper & Brothers, 1881). The Introduction and Appendix of Notes on Select Readings volume of the original edition was written by Dr. Hort, and in it he set forth the arguments and general theories upon which the text was reconstructed, and provided explanations for many specific textual decisions.

Westcott and Hort brought the main tendency of nineteenth century textual criticism—the exaltation of the oldest Greek copies—to its culmination. They firmly set aside the Latin witnesses along with the later Greek manuscripts; but the oldest known Greek copies, Codex Vaticanus and Codex Sinaiticus, they elevated to a pristine class called "Neutral," and very nearly identified them with the original manuscripts. It cannot be said, however, that Westcott and Hort were simply following a tendency here, for they realized that if such weight were to be given to only two manuscripts, a theory must be offered to explain how the text given in them had so early disappeared from the manuscript tradition. And so Hort offered in the Introduction of their text a theoretical history of the manuscript tradition that met the needs of the case, or at least so it seemed to many scholars.

They theorized that the "Neutral" text was the most primitive type, carefully copied for use in the worship services of the churches. The "Western" text-type arose early on as an uncontrolled popular edition, and persisted mainly in the Latin witnesses after Greek copies were no longer being produced in Italy. The "Byzantine" group, which includes the mass of later copies, began in the fourth century as an official church-sponsored edition of the New Testament, written probably in Antioch, which combined the various readings of the Western and Neutral groups. This edition was so effectively propagated throughout Europe that both the older "Neutral" and "Western" text-types ceased to be copied in the European scriptoriums, and eventually decayed. The Neutral text survived for a while in Egypt, but then suffered corruption and became the "Alexandrian" type. Codex Vaticanus and Codex Sinaiticus are relics of the Neutral type. A considerable amount of speculation is involved in this argument, but Westcott & Hort further bolstered their text with detailed arguments from two other directions, presenting "external" arguments (from the oldest manuscripts, as in Lachmann) and "internal" arguments (from the tendencies of scribes, as in the rules of Griesbach). External and internal arguments were also made to support one another by the principle, "Readings are to be preferred that are found in a manuscript that habitually contains superior readings:" superior, that is, as determined by the rules of internal criticism. The text of Westcott & Hort therefore had the appearance of resting firmly upon three-legged arguments, and it was considered by many scholars to be the best possible text.

Whatever may be the merits of Westcott and Hort's theory, the success of their text was largely due to personal influence and advantageous timing. In the 1860's the two most ancient copies, Sinaiticus and Vaticanus, were both published for the first time, creating a public sensation. At about that time, Westcott and Hort began work on their text, and so in 1870, the year that a critical revision of the King James version was commissioned by the church authorities in England, they were able to distribute to the members of the revision committee a draft copy of their text. They both served on the revision committee, and they published their text in 1881, the same year that the revision was published. For ten years, then, Westcott and Hort continually advocated their views in favour of the texts of Sinaiticus and Vaticanus in regular meetings of the most influential scholars of Great Britain and America, and it is hardly surprising that their text should be so well regarded when it appeared. In fact two generations passed before most scholars would recognize that the genealogical theories of Westcott and Hort were without adequate empirical foundation.

The text of Westcott & Hort was most vigorously assailed by John William Burgon, Dean of Chichester, and more temperately criticized by many others. The common theme of criticism was the lack of historical basis for their hypothesis of an early "Byzantine" recension in Antioch.

Critical Rules of Westcott & Hort

The following summary of principles is taken from the compilation in Epp and Fee, Studies in the Theory and Method of New Testament Textual Criticism (1993, pages 157-8). References in parentheses are to sections of Hort's Introduction, from which the principles have been extracted.

1. Older readings, manuscripts, or groups are to be preferred. ("The shorter the interval between the time of the autograph and the end of the period of transmission in question, the stronger the presumption that earlier date implies greater purity of text.") (2.59; cf. 2.5-6, 31)

2. Readings are approved or rejected by reason of the quality, and not the number, of their supporting witnesses. ("No available presumptions whatever as to text can be obtained from number alone, that is, from number not as yet interpreted by descent.") (2.44)

3. A reading combining two simple, alternative readings is later than the two readings comprising the conflation, and manuscripts rarely or never supporting conflate reading are text antecedent to mixture and are of special value. (2.49-50).

4. The reading is to be preferred that makes the best sense, that is, that best conforms to the grammar and is most congruous with the purport of the rest of the sentence and of the larger context. (2.20)

5. The reading is to be preferred that best conforms to the usual style of the author and to that author's material in other passages. (2.20)

6. The reading is to be preferred that most fitly explains the existence of the others. (2.22-23)

7. The reading is less likely to be original that combines the appearance of an improvement in the sense with the absence of its reality; the scribal alteration will have an apparent excellence, while the original will have the highest real excellence. (2.27, 29)

8. The reading is less likely to be original that shows a disposition to smooth away difficulties (another way of stating that the harder reading is preferable). (2.28)

9. Readings are to be preferred that are found in a manuscript that habitually contains superior readings as determined by intrinsic and transcriptional probability. Certainty is increased if such a better manuscript is found also to be an older manuscript (2.32-33) and if such a manuscript habitually contains reading that prove themselves antecedent to mixture and independent of external contamination by other, inferior texts (2.150-51). The same principles apply to groups of manuscripts (2.260-61).