The Expositor 6th Series vol. IX  (1904):  359-68.

The digital form was graciously edited by Christopher Pfohl at Gordon College, 2006.








                                J. H. Moulton




BEFORE we begin to examine the conditions of Hellenistic

syntax, which must obviously hold the first place for the

student of New Testament exegesis, it will be well to spend

some time upon the forms, which give us the surest evi-

dence as to the position occupied by the sacred writers

between the literary and the illiterate Greek of their time.

The question naturally arises, how far we can be sure that

we possess the exact forms that were used by the writers

themselves. May not our best MSS. have conformed the

orthography to the popular style, just as those of the

“Syrian” text conformed it in some respects to the literary

standards? We cannot give a universal answer to the

question, for, as we have seen already, the rise of an

artificial orthography undoubtedly left the door open for

not a few uncertainties. But there are some suggestive

signs that the great uncials, in this respect as in others,

are not far away from the autographs. A very instructive

phenomenon is the curious substitution of e]a<n for a@n after

o!j, o!pou, etc., which W.H. have faithfully reproduced in

numberless places from the MSS. This was so little recog-

nized as a genuine feature of vernacular Greek that the

editors of the volumes of papyri began by gravely subscrib-

ing “1. a@n” wherever this abnormal form showed itself.

They were soon compelled to save themselves the trouble.

Deissmann (p. 204) gave a considerable list from the papyri,

which abundantly proved the genuineness of this e]a<n;

and four years later (1901) the material had grown so much

that it was possible to determine the time-limits of the

peculiarity with fair certainty. If my count is right,1 the


                1 Class. Rev. xv. 32. I have not brought the count up to date in the two

subsequent articles (xv. 434, xviii. 106), but the results would not be

weakened if this were done.




proportion of e]a<n to a@n is 1:2 in papyri dated B.C. But

the estimate was based on only 12 occurrences. The pro-

portion was soon reversed, being 25:7 in the first century

A.D., 76:9 in the second, 9:3 in the third, 4:8 in the fourth.

e]a<n occurs last in a sixth century papyrus. It will be seen

that the construction itself was specially common in the first

two centuries A.D., when e]a<n greatly predominated, and that

the fashion had almost died away before the great uncials

were written. It seems to follow that in this small point

the uncials faithfully reproduce originals written under

conditions which had passed away in their time.1  This

particular example affords us a very good test, but we may

reinforce it with a variety of cases where the MSS. accu-

rately reproduce the spelling of the first century. I will

follow the order of the material in W.H. App. 141 ff.

(“Notes on Orthography”): it will not be necessary to

give detailed references for the papyrus evidence, which

will be found fully stated in the three Classical Review

papers already cited. We must bear in mind from the first

Hort's caution (p. 141) that “all our MSS. have to a

greater or less extent suffered from the effacement of un-

classical forms of words,” and his statement that the

Western MSS. show the reverse tendency. “The ortho-

graphy of common life, which to a certain extent was used


                1 The case of a@n, if, is separate. In the New Testament it is confined

apparently to the Fourth Gospel, where it occurs six times. In the

papyri it is decidedly a symptom of illiteracy. With this agrees what

Meisterhans3 255 f. says: “Only six times is a@n found from the 5th to the

3rd cent. B.C. The form a@n, is entirely foreign to the Attic inscriptions,

though it is often found in the Ionicising literary prose of the 5th cent.

(Thucydides, cf. the tragedians).” Since a@n is the modern form, we may

perhaps regard it as a dialect variant which ultimately ousted the Attic

e]a<n, but it is hard to say why the Gospel has it and why the Apocalypse

has not. There is some difficulty in determining the dialect to which it

is to be assigned. Against Meisterhans’ suggestion of Ionic stands the

opinion of H. W. Smyth (Ionic Dialect, p. 609) that its occasional appear-

ances in Ionic are due to Atticising! Certainly h@n is the ordinary Ionic

form, but a@n may have been Ionic as well, though rarer. (So Mr. P. Giles.)





by all the writers of the New Testament, though in unequal

degrees, would naturally be introduced more freely in texts

affected by an instinct of popular adaptation.” He would

be a bold man who would claim that even Hort had said

the last word on the problem of the Western Text; but

with our new knowledge of the essentially popular character

of New Testament Greek as a whole, we shall naturally

pay special attention to documents which desert the

classical spelling for that which we find prevailing in

papyri written by men of education approximately parallel

with that of the apostolic writers.

The case of lh<myomai, comes first (p. 142). The intrusion

of the m from the present stem of lamba<nw into various parts

of the verb, and into derivative nouns, is well set after the

Ptolemaic period, in which there is still some lingering of

the older forms. It is therefore unnecessary to show that

the late uncials, in restoring the classical forms, are desert-

ing the unquestioned pronunciation of the first century.

The “unusual aspirated forms” (p. 143) e]f’ e]lpi<di, kaqi[di<an,

a@fide, etc., and ou]x o]li<goj are supported by a large body

of evidence from papyri. It is rather strange that kaqe!toj

does not appear in the MSS.; as in the other cases, there

is a struggle between the two types, but the modern e]fe<to

shows that the aspirate here triumphed. It is of course

impossible to set this phenomenon down to the defunct

digamma: it doubtless originates from analogy processes

within the Koinh< itself (so Thumb), which accounts for the

uncertain tradition. We cannot prove either one or the

other for the New Testament autographs, but we have

already seen good reason for trusting the uncial tradition

in places where we have the means of checking it.

Occasional deaspiration (p. 144) is part of the general

tendency towards psilosis which started from Ionic influ-

ences and became universal, as Modern Greek shows.

The mention of tamei?on (p. 146—add pei?n from p. 170)



brings up a universal sound-change of Hellenistic, the

coalescence of two following i sounds.  Tamei?on, pei?n and

u[gei<a are overwhelmingly attested by the papyri, where

there are only rare examples of a curious reversion like that

in Matthew xx. 22. In the form a]leei?j (Mark i. 17 al.) we

have dissimilation instead of contraction. Three isolated

spellings on p. 148 are instructive.   ]Arabw<n “seems to be

only Western.” In the papyri I counted 11 exx. of this

against 12 of rr, a curious modification of the results of

Deissmann (p. 183), which were obtained from the Berlin

and Rainer papyri only. The word will serve as evidence

of the inaccessibility of the autographs’ spelling except

where the papyri are unanimous: cf. Deissmann’s observa-

tions, p. 181. Next comes sfuri<j, which is invariable in

the papyri after the Ptolemaic period. Zmu<rna is regarded

by W.H. as Western; but though the papyri and inscrip-

tions waver (Deissmann, 185), it surely ought to be trans-

ferred from margin to text on the evidence of the first

century Smyrnaean coins. The next cases of importance

appear on p. 150.   ]Erauna<w is certain for the first century

and after. Hort's account of te<sserej and tessara<konta

gives us our first example of dissonance between the papyri

and the uncials. The forms with e are in the papyri

relatively few, and distinctly illiterate, in the first centuries

A.D. Indeed the evidence for forms of te<sserej is virtually

nil before the Byzantine age, and there is not the smallest

probability that the Apostles wrote anything but the Attic

form. For tessara<konta the case is a little better, but it is

hopelessly outnumbered by the -ar- form in documents which

antedate the uncials; the modern sera<nta, side by side

with sara<nta, shows that the strife continued. No doubt

before the fourth century te<sserej -a (not tesse<rwn) had

begun to establish themselves in the place they hold to-day.

Finally might be mentioned one or two notable matters of

pronunciation to which Hort does not refer. The less



educated papyrus writers very frequently use ā for au, from

the first century B.C. onwards. Its frequent appearance in

Attic inscriptions after 74 B.C. is noted by Meisterhans

(Gramm. d. Att. Inschr.3 154). In Luke ii. 1 ( ]Agou<stou)

this pronunciation shows itself, according to x C* D; but

we do not seem to find a]to<j, e[ato<n, etc., in the MSS., as we

should have expected.1

We pass on to the noun flexion (p. 156). Nouns in – ră

and participles in – ui?a in the papyri regularly form genitive

and dative in –hj -^, except that –ui<ajui<% are still found in

the Ptolemaic period. Here again the oldest uncials alone—

and even they are not without lapses—support the unmis-

takable verdict of the contemporary documents of the Koinh<.

It seems best on the whole to regard this as the analogical

assimilation of -ră nouns (and—somewhat later and less

markedly—ui?a participles) to the other -ă flexions of

the 1st declension, rather than as Ionic survivals.2 It may

be added that as ma<xaira produced maxai<rhj on the model

of do<ca and do<chj, so Nu<mfhj as a proper name produced

what is best read as Nu<mfă Nu<mfan in nom. and acc. (Col.

iv. 15): it is quite feasible to keep the best reading here with-

out postulating a Doric Nu<mfān, the improbability of which

decides Lightfoot for the alternative. The heteroclite proper

names, which fluctuate between 1st and 3rd decl., are

paralleled by Egyptian place-names in papyri. In contracted

nouns and adjectives we have abundant parallels for forms

like o]ste<wn, xruse<wn, and for xrusa?n (formed by analogy of


                1 In Modern Greek (see Thumb, Grammatik, p. 59) we find au]to<j (pro-

nounced aftós) side by side with a]to<j (obsolete except in Pontos), whence

the short form to<, etc. There was therefore a dialectic difference in the

Koinh< itself.

                2 In connexion with this I might mention an Ionic Koinh< feature which

I expected to find more often in New Testament MSS., the spelling

kiqw<n, which (like ku<qra and e]nqau?ta) occurs not infrequently in papyri.

I can only find in Tischendorf's apparatus xeiqw?naj D* (Matt. x. 10) and

kitw?naj B* (Mark xiv. 63— “ut alibi X,” says the editor, but not stating




a]rgura?n). The fact that we do not find short forms of

nouns in –ioj -ion (e.g. ku<rij, paidi<n) is a noteworthy test of

the educational standard of the writers, for the papyri show

them even as early as the third century B.C., and always in

company with other indications of comparative illiteracy.

These forms, the origin of which is as dark as ever, despite

the various efforts of Hatzidakis, Brugmann and others to

unravel it, ultimately won a monopoly, as modern Greek

shows everywhere. Passing lightly over the exact corre-

spondence between uncials and papyri in the accusatives of

klei<j and xa<rij (p. 157), we may note the case of xei?ran in

John xx. 25 x*AB. The great frequency of this formation

in uneducated papyri, which adequately foreshadows its

victory in modern Greek,1 naturally produced sporadic

examples in the MSS., but it is not at all likely that the

autographs showed it, unless possibly in the Apocalypse.

Gregory (Tisch.-Gregory, 118 f.) adds notes of forms

like a]sfalh?n and podh<rhn, which have also papyrus parallels,

but could be explained more easily from the analogy of 1st

decl. nouns. Mei<zwn acc. (John v. 36 ABEGMD) is a good

example of the irrational addition of n, which seems to

have been added after long vowels almost as freely as the

equally unpronounced i.2 Before leaving the nouns and

adjectives we must mention the indeclinable plh<rhj, which

should be read in Mark iv. 28 (C*, Hort) and Acts vi. 5

(xAC*DEHP al.), and is probably to be recognized in John

i. 14. Cf. 2 John 8 (L), Mark viii. 19 (AFGM al.), Acts vi. 3

(AEHP al.), xix. 28 (AEL 13), which show that in every

New Testament occurrence of an oblique case of this word

we find the indeclinable form recognized in good uncials.


                1 It seems most probable that the modern levelling of 1st and 3rd decl.

started with this accusative : the n has vanished again now. See Thumb,

Grammatik, pp. 28, 35.

                2 Thus a!lwi is acc. sing., while h#n (=^#) may be subjunctive. For exx.

see Class. Rev. xviii. 108.



My papyrus citations for this1 virtually begin, however, with

the second century, and I should hardly credit the New

Testament autographs with the form. This probably means

that in John i. 14 an original plh<rh was corrupted to the

vulgar plh<rhj in an early copy. Weiss and others would

make it depend in sense upon au]tou?, but do<can seems more

appropriate, from the whole trend of the sentence: the

glory” or “self-revelation” of the Saviour is “full of grace

and truth.” One may doubt whether it would have occurred

to any one to make a parenthesis of -kai e]qeasa<meqa. . .

patro<j, had it not been for the supposed necessity of

construing plh<rhj with a nominative. In fine, we regard

the Codex Bezae as having either preserved or successfully

restored the true reading.2

I might cite very many more noun forms in which the

MSS. prove to have retained the genuine Hellenistic, as

evidenced by the papyri; but these typical examples will

serve. Verbs naturally produce yet more abundant material,

but we need not cite it here, as our present purpose is only

to show how such a text as Westcott and Hort's, scrupulously

reflecting the best uncials, is in all important features, and

in most of the minutiae, supported as genuinely Hellenistic

by papyrus evidence published long after their text was

made—a conclusion valuable because of the criteria it gives

us for estimating the general grammatical condition of our

texts. Pursuing the order of W.H. app., we pause a

moment on the dropped augments, etc., in pp. 161 f., which

are well illustrated in papyri.  The attachment of 1st


                1 See also C. Turner in Journ. Theol. Stud., i. 120 fr. and 561 f.;  

Rademacher in Rhein. Mus., lvii. 151 ; Reinhold De Graecitate Patrum,


                2 Winer, p. 705, compares the “grammatically independent” plh<rhj

clause with the nom. in Phil. iii. 19, and Mark xii. 40. Dr. Moulton

makes no remark there, but in his joint commentary with Dr. Milligan

he accepts the construction of John i. 14 found in the R.V., or permits his

colleague to do so. Of course the case for the indeclinable plh<rhj was.

before him only in the LXX. (as Job xxi. 24 BxAC).



aorist endings to 2nd aorists is universal in our Koinh<;

documents, and the MSS. here undeniably reproduce in

general the forms of the autographs. Whether the intrusion

should be allowed in the imperfect (as ei#xan Mark viii. 7)

is more than doubtful, as the papyri give hardly any war-

rant. The imperfect and aorist 3rd pl. -osan receives little

encouragement, and the 2nd sing. perf. -ej still less: they are

both marks of illiteracy. The 3rd pl. perf. -an makes a much

better show in the papyri, but though already common in

Ptolemaic documents can hardly be regarded as established

for the New Testament autographs: like the perf. -ej, it

might be allowed in the Apocalypse. Passing on to con-

tract verbs, we note how the confusion between -aw) and –e<w

forms (p. 166) are supported by our external evidence, and

by Modern Greek. Our first serious revolt from Westcott

and Hort will be in the infinitive in –oi?n (and by analogy

-%?n). The evidence for it is “small, but of good quality”

(p. 166—cf. Introd. § 410): it is in fact confined to B*D in

Matthew xiii. 32, B* in Mark iv. 32, x* in 1 Peter ii. 15, BD*

in Hebrews vii. 5 (where see Tischendorf's note), and a lection-

ary in Luke ix. 31. This evidence might pass if the object is

merely to reproduce the spelling of the scribe of B, but there

is absolutely no corroboration that I know of earlier than

the date of B itself, except a second century inscription cited

in HatzidakisEinleitung, p.193.1 Blass, Gram. 48, does not

regard the form as established for the New Testament. I

can quote against it from centuries 1-4 eleven examples

of -ou?n in papyri. That -ou?n and -a?n (not -%?n) are the correct

Attic forms may be seen from Meisterhans3 175 f., which

Hort's hesitation as to – a?n prompts me to quote: for the

reason of the apparent irregularity see Brugmann, Griech.

Gramm.3 61, or Winer-Schmiedel 42. Next may be named

for -aw verbs the 2nd sing. pres. mid. in -asai (kauxa?sai, o]dun-


                1 So Winer-Schmiedel, p. 116 (note). There are two other inscriptions

cited by Hatzidakis, but without dates.



a?sai), which has been formed afresh in the Koinh< with the

help of the -sai that answers to 3rd sing. -tai in the perfect.1

It is well paralleled by the early Ptolemaic future xariei?sai.

I have, unfortunately, no examples of the subjunctive of –o<w

verbs, with which to attack the parsing of  i!na zhlou?te and

the like (p. 167). Blass (Kühner3 i. 2. 587, and New Testa-

ment Gram. 48) accepts Hort's view that the subjunctive of

these verbs became identical with the indicative, just as it

always was in the –a<w verbs. But he, rightly I think, rejects

the supposition that eu]odw?tai, (1 Cor. xvi. 2) is anything but

a pres. subj. To read eu]o<dwtai, as perf. indic., is possible,

though the editors do not seem by their printing to have

favoured that alternative. That it is a perfect subjunc-

tive is extremely unlikely. The parallels on which Hort

(p. 172) relies—set forth with important additions in Blass's

Kühner, i. 2. 100 f.—do nothing to make it likely that the

Koinh< had any perf. subj. apart from the ordinary peri-

phrastic form.2 It is hard, moreover, to see why the present

subjunctive is not satisfactory here: see Dr. Findlay's note

in loc.

The verbs in -mi, were naturally in Hellenistic pursuing

the process of painless extinction which began even in

Homeric Greek, and in modern Greek has eliminated every-

thing outside the verb “be.” The papyri agree with the

New Testament uncials in showing forms like du<nomai, and

-e<deto (as well as –e<doto), and various derivatives from con-

tract verb types. New verbs like i[sta<nw are formed, and

new tenses like e!staka, and the doubly augmented form


                1 To suppose this (or fa<gesai, similarly formed from fa<getai) genuine

survivals of the pre-Greek -esai, is a characteristic feat of the antediluvian

philology which still frequently does duty in this country.

                2 To argue this would demand a very technical discussion. It is

enough to say that the Attic kektw?mai and memnw?mai are not derivative

verbs, and that the three derivative verbs which can be quoted, from

Doric, Cretan, and Ionic respectively, are very small encouragement for a

supposed Koinh< parallel.



a]pekatesta<qhn is well attested. What is more important

the subjunctives didoi?, and doi?, are set on a completely satis-

factory basis, so that the idea that they are irregular

optatives (as they may possibly be in late documents) need

trouble us no more. From oi#da we have as in New Testa-

ment the flexion as an ordinary perfect, but there are rarely

found survivals of the old forms. Finally there is ei]mi< which

shows middle forms h@mhn, etc., and h@tw parallel with e@stw,

just as in the New Testament.

            With this we may leave spelling and inflexions and push

on to the syntax, which will compensate the New Testa-

ment student, I hope, for the dry bones he has had to be

satisfied with in this chapter of our subject. But though

the minutiae of accidence may be dull to those who are not

professed philologists, it will be allowed that forms must be

settled before we can start discussing their uses; and it is

also very clear that they give us our surest criteria for local-

izing texts and for testing the detailed accuracy of our

documents. With this plea I hope to be forgiven on promise

of an effort to be more interesting next time.


James Hope Moulton.