(For an in-depth examination and refutation of "Colwell's Rule" see the 'Definite John 1:1c' study on this blog.)
Journal of Biblical Literature [Vol. 52, 1933.]:
A DEFINITE RULE FOR THE USE OF THE ARTICLE IN THE GREEK NEW TESTAMENT
E. C. Colwell
University of Chicago
Although Walter Bauer cautiously asserts that hard and fast rules for the use of the article in Hellenistic Greek are an impossibility, the grammarians have formulated many rules for which they claim various degrees of reliability. This is true of the use of the article with predicate nominatives, the use with which this study is concerned.
The discussion in A. T. Robertson’s mammoth grammar may be regarded as representative since it is built upon the opinion of other grammarians and quotes largely from them. He emphasizes most strongly the two following rules: (1) predicate nouns tend to omit the article; (2) predicate nouns occur with the article in convertible propositions.
Professor Torrey goes further and claims that in some New Testament passages the article is omitted because of the anarthrous construct state in the Semitic original. He has claimed that six nouns in John are without the article for this reason. Three of his anarthrous nouns (in John 1 49; 5 27; 9 5) [Jn 1:49; 5:27; 9:5] are predicate nominatives, and in each of these cases the predicate noun precedes the verb.
It was a study of these passages, especially John 1 49, that suggested the rule which is advocated in this study. In this verse Nathanael ascribes to Jesus two titles; in one of them he uses the article, in the other he does not: σὺ εἶ ὁ υἱὸς τοῦ θεοῦ, σὺ βασιλεὺς εἶ τοῦ Ἰσραήλ [su ei ho huios tou theou. su basileus ei tou Israel: ‘you are the son of the god. you king are of the Israel.’]. What reason is there for this difference? When the passage is scrutinized, it appears at once that the variable quantum is not definiteness but word-order. “King of Israel” in this context is as definite as “Son of God.” It seems probable that the article [ho] is used with “Son of God” because it follows the verb, and is not used with “King of Israel” because it precedes the verb. If this can be established generally in the New Testament, it will of course involve only those sentences in which the copula is expressed. And for such sentences the rule may be stated briefly as follows: A definite predicate nominative has the article when it follows the verb; it does not have the article when it precedes the verb. Of course, this can be claimed as a rule only after it has been shown to describe the usage of the Greek New Testament as a whole or in large part.
An important part of this demonstration is found in those passages in which a phrase is used now with the article and now without it. In John 19 21 the title “King of the Jews” is used of Jesus both with and without the article: μὴ γράφε· ὁ βασιλεὺς τῶν Ἰουδαίων, ἀλλ’ ὅτι ἐκεῖνος εἶπεν· βασιλεὺς τῶν Ἰουδαίων εἰμί [Me graphe ho basileus ton Ioudaion, all hoti ekeinos eipen Basileus ton Ioudaion eimi: ‘not be writing the king of the jews, but that that (one) said king of the jews I am.’]. It is plain that the Jews are objecting to the superscription because it may be read as a statement of fact; they want it changed so that the readers will know that this title is only a claim made by Jesus. But in each case the title itself remains the same; the article does not occur in the second instance because the predicate precedes the verb. In half a dozen New Testament passages, the same phrase appears with the article after the verb. And in Matt. 27 42, where the title “King of Israel” is applied to Jesus, this appears without the article and before the verb.
The words “Son of God” appear approximately thirteen times as a predicate with the article, in each of the thirteen passages they follow the verb. These words also appear ten times as predicate nominatives without the article; in nine of these passages they precede the verb, and in the tenth (Matt. 27 43) it may be significant that θεοῦ precedes the verb.
The title “Son of Man” appears twice in the New Testament as a predicate nominative: once with the article (Matt. 13 37) and once without the article (John 5 27). In the Matthean passage, where it has the article, it follows the verb. In the Johannine passage, where it lacks the article, it precedes the verb.
This variation in the use of the article frequently occurs with the same phrase in the same gospel. In John 8 12 Jesus says, ἐγὼ εἰμι τὸ φῶς τοῦ κόσμου; [Ego eimi to phos tou kosmou: ‘I am the light of the world’]; in John 9 5 he says, φῶς εἰμι τοῦ κόσμοῦ [phos eimi tou kosmou: ‘light I am of the world]. A similar variation occurs in Matt. 12 48 and 50, where Jesus uses the words “my mother” with the article after the verb and without the article before the verb.
One of the most impressive examples of the correlation between word-order and use of the article occurs in Matt. 13 37-39, the explanation of the parable of the tares: ὁ σπείρων … ἐστὶν ὁ υἱὸς τοῦ ἀνθρώπου, 38 ὀ δὲ ἀγρός ἐστιν ὁ κόσμος. … οὗτοι εἰσιν οἱ υἱοὶ τῆς βασιλείας· τὰ δὲ ζιζάνια εἰσιν οἱ υἱοὶ τοῦ πονηροῦ, 39 ὁ δὲ ἐχθρὸς … ἐστιν ὁ διάβολος, ὁ δὲ θερισμὸς συντέλεια αἰῶνος ἐστιν, οἱ δὲ θερισταὶ ἄγγελοι εἰσιν. [ho speiron...estin ho huios tou anthropou. ho de agros estin ho kosmos. ... houtoi eisin hoi huioi tes basileias. ta de zizania eisin hoi huioi tou ponerou, ho de echthros... estin ho diabolos. ho de therismos sunteleia aionos estin, hoi de theristai aggeloi eisin: ‘the (man) sowing the fine seed is the son of the man; the but field is the world. ... these are the sons of the kingdom the but weeds are the sons of the wicked (one), the but enemy ... is the devil. the but harvest conclusion of age is, the but harvesters angels are’]. Here in a series of seven clauses the predicate nouns follow the verb and take the article five times; while in the last two clauses equally definite predicate nouns precede the verb and do not have the article. [But see RSV, NIV, ASV, NASB, NEB, JB, and AT, esp. for translation of “non-prepositional” aggeloi “angels.” - RDB]
That Matthew changed from one word-order to the other in a definite attempt to secure variety of style is suggested by another series of clauses similar to the one discussed above: Matt 23 8-10, εἷς γὰρ ἐστιν ὑμῶν ὁ διδάσκαλος … εἷς γάρ ἐστιν ὑμῶν ὁ πατὴρ ὁ οὐράνιος … ὅτι καθηγητὴς ὑμῶν ἐστιν εἷς ὁ Χριστός [heis gar estin humon ho didaskalos... heis gar estin humon ho pater ho ouranios ... hoti kathegetes humon estin heis ho Christos: ‘one for is of you the teacher ... one for is of you the father the heavenly... because leader of you is one the christ.’]. Here he twice puts the predicate after the verb with the article, but the third time places it before the verb without the article. In neither of these Matthean passages can it be claimed that the predicates which close the series are less definite or concrete than those which precede; nor are the final clauses of the series less convertible than the others.
Further proof of the significant relation between word-order and the use of the article with predicate nouns is easily obtained from the very grammarians who are unaware of its existence. For the examples which they offer of predicate nouns with the article or of predicate nouns without the article fall into the categories established by this rule almost without exception. Thus in Robertson’s list (pp. 768-769) of forty-one predicates in the New Testament which have the article, there are thirty-eight which follow the verb. Again Robertson lists (p.794) the constructions in which the article is omitted. One of these constructions is the predicate nominative, and in both of his examples of anarthrous predicate nominatives the predicate precedes the verb. Blass-Debrunner lists eighteen predicate nouns that take the article, and every one of them follows the verb. No one will be so unkind as to insinuate that these lists were compiled to support a theory of which the compilers had never heard. Robertson and Debrunner were merely looking for examples of predicate nouns with the article; it is significant that they found them after the verb.
Further support for the claim that there is a connection between word-order and use of the article can be found in the nature of the variants which occur in the MSS of the New Testament. That the MSS vary greatly in adding and omitting the article, every scholar who has done any work in textual criticism is aware. Such a simple omission or addition would of itself prove little for the theory advocated here, but when the omission or addition of the article is accompanied by a change of word-order, we have evidence that the relation between word-order and the use of the article was as real to the scribes who copied the MSS as it was to the original authors.
In the course of this study I noted three passages in which the article is used by one group of MSS and omitted by another group with a change in word-order. In each of these passages Westcott and Hort’s Heavenly Twins (Codex Vaticanus and Codex Sinaiticus) disagree; yet both of them support the rule stated in this paper. That is to say, their variation is from one to the other of the alternatives described in this rule. The MSS differ as follows:
B [4th century] א [4th century]
σὺ βασιλεὺς εἶ τοῦ Ἰσραήλ (1) John 1 49 σὺ εἶ βασιλεὺς τοῦ Ἰσραήλ
[of Israel] [of Israel]
καθηγητὴς ὑμῶν ἐστιν εἷς (2) Matt. 23 10 εἷς γάρ ὑμῶν ὁ καθηγητὴς
[of you] [of you]
εἷς θεὸς ἐστιν (3) Jas. 2 19 εἷς ἐστιν ὁ θεὸς
It is interesting that B each time has the predicate before the verb without the article, while א [aleph] each time has the predicate after the verb with the article. Further study will doubtless multiply examples of this sort, but these are enough to indicate that the scribes felt that a definite predicate noun did not need the article before the verb and did need it after the verb.
The predicate noun is used in the New Testament with the article 254 times in sentences in which the verb is expressed. It follows the verb 239 times, and precedes the verb 15 times. Predicate nouns which are indubitably definite appear without the article in sentences in which the verb occurs 139 times. 99 times the predicate precedes the verb, and it follows the verb 40 times.
These totals include predicates in relative clauses. But the word-order in relative clauses seems to be so definitely fixed as to justify the exclusion of such clauses from this study. Ten definite predicates appeared with the article  in relative clauses, all after the verb. Sixteen definite predicates are used without the article in relative clauses, two before the verb (with the relative in the genitive) and fourteen after the verb (with the relative in the nominative). Thus only two out of twenty-six predicates precede in relative clauses.
If the predicates which occur in relative clauses are subtracted, the totals are as follows:
I. Definite Predicate Nouns with Article.............. 244
A. After Verb ................................................... 229 (94%)
B. Before Verb ................................................ 15 (6%)
II. Definite Predicate Nouns without the Article ...123
A. After Verb .................................................... 26 (21%)
B. Before Verb ................................................. 97 (79%)
The close relation between word-order and the use of the article in these cases can be shown by a different grouping of the same figures:
I. Definite Predicates after the Verb ................ 255
A. With the Article .................................... 229 (90%)
B. Without the Article ................................ 26 (10%)
II. Definite Predicates before the Verb ............ 112
A. With the Article ..................................... 15 (13%)
B. Without the Article ................................ 97 (87%)
It is obvious that the significance of these figures rests upon the accuracy with which definite predicate nouns without the article have been identified. There are bound to be mistakes in the list of definite predicate nouns without the article; but an attempt has been made to exclude all nouns as to whose definiteness there could be any doubt. This means, of course, that “qualitative” [abstract?] nouns have been omitted, since all such nouns (and their total in the New Testament is small) are not definite. An inspection of some of the definite predicate nouns without the article will demonstrate that they are definite even though they lack the article.
None of the predicates in the following passages has the article in the original, and each of them precedes the verb. In Hebrews 9 15 it is claimed that Jesus is (the) mediator of a new covenant; in I Timothy 6 10 the love of money is identified as (the) root of all evil; in John 10 2 the one who enters by the door is (the) shepherd of the sheep; in Matt. 5 35 Jerusalem is (the) city of the Great King; in Mark 2 28 the Son of Man is also (the) lord of the Sabbath; in I Cor. 4 4 Paul says, “It is (the) Lord who must examine me;” in Rev. 21 22 it is said of the New Jerusalem that God is (the) temple in it, etc. etc. This is a fair sample of what lies behind the statistics given above.
The rule that it is the exception that proves the rule finds no exception here. The exceptions to the rule that definite predicate nouns before the verb omit the article are about 15 in number. Half of them are scattered around in Luke, John, II Peter, and Revelation; and in five of these passages there is serious manuscript evidence for the omission of the article according to the rule. The other half of these exceptions (7) are grouped in I and II Corinthians; and there is no significant manuscript evidence for variation here. Five of the seven put the predicate not only before the verb, but also before the subject; e.g., I Cor. 9:1 οὐ τὸ ἔργον μου ὑμεῖς ἐστε ἐν κυρίῳ [ou to ergon mou humeis este en kurio: ‘not the work of me you are in lord?’]. Thus the order in these five Corinthian passages is predicate-subject-verb. This is obviously a stylism temporarily affected by the Apostle to the Gentiles, possibly for the sake of greater emphasis. Aside from these five passages, Paul’s exceptions do not loom up as especially significant.
The other class of exceptions - the omission of the article after the verb - contains more examples. Approximately twenty-six definite predicate nouns omit the article after the verb. Two of these are quotations from the Septuagint; five are adjectives used as substantives; but the remainder have no common characteristics. In I Cor. 12 27 and I Thess. 4 3 the textual evidence gives some basis for reading these passages according to the rule, but elsewhere the MSS are practically unanimous for the exception. The large number of exceptions in Romans is the most notable feature in the list.
I have made a hurried sampling of Greek usage outside the New Testament, with results that vary somewhat and yet on the whole support the figures quoted for the New Testament. In Genesis, chapters 1-41, about fifty-eight definite predicate nouns were counted: forty-five support the rule (4 20; 12 12; etc.); thirteen are exceptions. Eleven of the exceptions are after the verb without the article (e.g., 9 18). In Genesis 9 19 the same sort of manuscript variation as was noted for the New Testament occurs: Codex Alexandrinus reads the predicate after the verb with the article and the Cotton Genesis MS reads it before the verb without the article.
In the Didache there are six definite predicate nouns used with a verb, and all of them support the rule. One comes before and five after the verb. In P. Oxy III four support the rule, and there is one exception. In the Discourses of Epictetus IV. i-iv (about 50 pp. of Schenkl’s text) out of seventeen examples fourteen support the rule and three do not. In all of these sources the percentage of support for the rule is about the same: of the predicate nouns with the article about 90% follow the verb; of the definite predicate nouns without the article about 80% precede the verb.
These dead figures have certain vital implications in at least three fields of New Testament study: grammar, text, and translation or interpretation. The New Testament grammars of the future, when they say that predicate nouns regularly omit the article, will point out that this is not the case in sentences in which the verb occurs, for in such sentences two-thirds of the definite predicate nouns have the article. Nor can this use of the article with predicate nouns be attributed to the presence of a large number of participles as predicate nominatives; for of the 244 predicates with the article only 61 are participles.
The following rules may be tentatively formulated to describe the use of the article with definite predicate nouns in sentences in which the verb occurs. (1) Definite predicate nouns here regularly take the article. (2) The exceptions are for the most part due to a change of word-order: (a) Definite predicate nouns which follow the verb (this is the usual order) usually take the article; (b) Definite predicate nouns which precede the verb usually lack the article; (c) Proper names regularly lack the article in the predicate; (d) Predicate nominatives in relative clauses regularly follow the verb whether or not they have the article.
In the field of textual criticism the rule here advocated has an equally definite contribution to make. It shows in certain specific cases what the probabilities are as to the author’s use or non-use of the article. A fine example of this is II Peter 1 17, cited as an exception to the rule since Westcott and Hort follow Codex Vaticanus [B] in reading the predicate with the article before the verb: Ὁ υἱός μου ὁ ἀγαπητός μου οὗτος ἐστιν [Ho huios mou ho agapetos mou houtos estin: ‘the son of me the beloved of me this is.’] The evidence given in this study as to the extreme rarity of this construction in the New Testament reinforces Tischendorf’s judgement that the variant read by practically all the rest of the MSS is to be preferred. They read the predicate after the verb with the article, οὗτος ἐστιν ὁ υἱός μου ὁ ἀγαπητός μου, [houtos estin ho huios mou ho agapetos mou: ‘this is the son of me the beloved of me.’], the usual Greek construction.
But it is in the realm of translation and interpretation that the data presented here have their most valuable application. They show that a predicate nominative which precedes the verb cannot be translated as an indefinite or a “qualitative” noun solely because of the absence of the article; if the context suggests that the predicate is definite, it should be translated as a definite noun in spite of the absence of the article. In the case of a predicate noun which follows the verb the reverse is true; the absence of the article in this position is a much more reliable indication that the noun is indefinite. Loosely speaking, this study may be said to have increased the definiteness of a predicate noun before the verb without the article, and to have decreased the definiteness of a predicate noun after the verb without the article.
The opening verse of John’s Gospel contains one of the many passages where this rule suggests the translation of a predicate as a definite noun. καὶ θεὸς ἦν ὁ λόγος [kai theos en ho logos] looks much more like “And the Word was God” than “And the Word was divine” when viewed with reference to this rule. The absence of the article does not make the predicate indefinite or qualitative when it precedes the verb; it is indefinite in this position only when the context demands it. The context makes no such demand in the Gospel of John, for this statement cannot be regarded as strange in the prologue of the gospel which reaches its climax in the confession of Thomas.
The passages in which υἱὸς τοῦ θεοῦ [huios tou theou: ‘son of the god’] appear have often occasioned debate: Is the word “son” definite or indefinite? It is certainly significant that when used without the article these words regularly precede the verb. Nor can it be claimed that the phrase is “qualitative” or indefinite. In John 10 36 Jesus says of himself, υἱὸς τοῦ θεοῦ εἰμι; [huios tou theou eimi: ‘son of the god I am.’]. This is translated as “the son of God” by Goodspeed, Moffatt, the Authorized Version, the American Revised Version, Weymouth, etc. In Matt. 14 33 the disciples exclaim ἀληθῶς θεοῦ υἱὸς εἶ [Alethos theou huios ei: ‘truly of god son you are.’]. This likewise is definite, and is so rendered in the translations listed above. But in Matt. 27 54 (equals Mark 15 39), Goodspeed and Moffatt translate as indefinite an almost exactly parallel phrase: ἀληθῶς θεοῦ υἱὸς ἦν οὗτος. [Alethos theou huios en houtos: ‘truly of god son was this.’]. The evidence given in this paper as to the use of the article with the predicate nouns strengthens the probability that the centurion recognized Jesus as the Son of God (so Weymouth and the older English translations), rather than as a son of God.
[[ There are a number of problems with "Colwell's Rule." Probably the greatest of these problems is the use of "prepositional" constructions. - see HARNER: JBL 'Qualitative' Article Refuted from beginning page down to  (indicator for end note #3).]]
1. Griechisch-deutsches Worterbuch zum Neuen Testament, GieBen, 1928, s.v.o.
2. Grammar of the Greek New Testament in the Light of Historical Research,3 New York,
1919, p. 767f. I regret that I did not have access to the 4th edition, 1923, but the material
used is unchanged in the first three editions.
3. C.C. Torrey, “The Aramaic Origin of the Fourth Gospel,” Harvard Theological Review
XVI, 1923, p. 323f. I have attempted to make a general answer in The Greek of the
Fourth Gospel, Chicago, 1931, p. 78f.
4. Matt. 27 11,37; Mark 15 2; Luke 23 3,37; John 18 33.
5. Matt. 16 16; 26 63; Mark 3 11; 15 39; Luke 4 41; 22 70; John 1 39,49; 11 27; 20 31; Acts 9 20;
I John 4 15; 5 5.
6. Matt. 4 3,6; 14 33; 27 40,54; Luke 4 3,9; Mark 15 39; John 10 36.
7. A similar variation occurs in Matt. 18 1 and 4.
8. A. Debrunner, Friedrich Blass’ Grammatik des neutestamentlichen Griechisch,6
Gottingen, 1931, p. 156.
9. Tischendorf’s attestation: with B (1) AL 1.33 (2) DGL 1.33. 124. 26ev 48ev xscr (3) 69.ascr
cscr; with ! [aleph] (1) the great majority of MSS (2) GDQP unc8 al pler (3) A 68.
10. The Greek text used was that of Westcott and Hort. No claim of absolute accuracy is
advanced for these totals, nor is there any need for absolute accuracy, as no
significance is, or can be attached to a definite mathematical proportion.
11. The distribution is interesting: Col. 3, Eph. 2, Rev. 5.
12. The least definite nouns included are found in a group of about ten phrases such as
“(the) seed of Abraham,” e.g., John 8 33. Practically all such expressions as ὁ θεὸς ἀγάπη ἐστίν
[ho theos agape estin: ‘the god love is’] have been excluded from this study [abstract predicate
nouns - RDB]; their inclusion would greatly increase the total for predicate nouns without the
article before the verb. Proper names, also, have been excluded because they regularly lack the
article whether they appear before or after the verb.
13. The Septuagint of Psalm 47 3 reads ἡ πόλις τοῦ βασιλέως τοῦ μεγάλου [he polis tou
basileos tou megalou: ‘the city of the king of the great’] without a verb.
14. Luke 4 41; John 1 21; 6 51; 15 1; II Peter 1 17; Rom. 4 13; I Cor. 9 1,2; 11 3,25; II Cor. 1 12; 3
2,17; Rev. 19 8; 20 14.
15. Proper Names regularly omit the article in the predicate when after the verb; they are
not included here. For relative clauses, cf. p.7.
16. Matt. 20 16; Mark 4 32; 9 35; 12 28; Luke 20 33; 22 24; John 4 18; 18 13,37; Acts 10 36; Rom. 4 11,18; 7 13; 8 16,29; 11 6; I Cor. 12 27; 16 15; II Cor. 5 21; 6 16; Gal. 4 31; I Thess. 4 3; I Peter 5 12; Heb. 11 1
17. Swete’s text was used, and the evidence of the variants given in his brief apparatus is
18. The text used was Kirsopp Lake’s edition of the Apostolic Fathers in the Loeb Classical
Library, iv. 7, 14; vi. 3; xiii. 3; xiv. 3; xv. 2.
19. After verb with article: IV i. 42, 62, 63, 114, 132, 158, 166; iii. 12; iv. 44; after verb
without article: IV i. 118; before verb without article: IV i. 60, 71, 73, 94, 158; before
verb with article; IV iv. 22, 48.
20. John 20 28.
For an in-depth examination and refutation of "Colwell's Rule" see the 'Definite John 1:1c' study on this blog.