CHAPTER III. ST. LUKE AND THE VIRGIN BIRTH

While, in the preceding chapter, we concluded that the Virgin Birth is a later stratum in the Third Gospel, we were unable to say to whose hand its presence is due. There was nothing to show that St. Luke could never at any time have known of the doctrine, but only that he could not have known of it at the time when he first drafted and wrote his Gospel. We are free, then, to make a new beginning, and to ask: _Did St. Luke teach the Virgin Birth?_

The question is most conveniently treated by discussing the authorship of Lk. i. 34 f. As we have seen, this is the crucial passage. If we can believe St. Luke himself to have written these verses, we must also attribute to his pen the words, “as was supposed” in iii. 23; in a word, we must conclude that he taught the Virgin Birth of Jesus, and we must leave the question, how this result is to be co‐ordinated with those reached in the previous chapter, to be considered later.

That St. Luke and no other did write these verses, is the considered view of the present writer. There are two lines of argument which converge in this direction. The first argument is _textual_, but it is more than a matter of weighing documents; the second is _linguistic_ and stylistic. Neither is completely conclusive in itself, and, when taken together, they do not admit of a result so stringent as rigid demonstration. They are complementary each to the other. Either would be weakened in force in the absence of the other, but their agreement is sufficient to establish a result for which a very high degree of probability can justly be claimed.

I. Lk. i. 34 f. and the Textual Question

It is well known that no exception to Lk. i. 34 f. can be taken on strictly _textual_ grounds. The external evidence for the passage is practically complete. The sole exception, which only serves to throw into relief the overwhelming mass of positive evidence, is found in the Old Latin MS. known as b, which substitutes i. 38 for i. 34 and omits verse 38 after verse 37.(42)

In Great Britain, a generation ago and less, this weight of external evidence would have been thought sufficient to settle the question, and there are probably very many scholars who would still take this view. But within recent years a change has come to be discernible among leading theological writers on the general question of attestation. Much more than in former times it is now recognized that during the first half of the second century the text of the New Testament, and especially that of the Gospels, was subject to rather free handling, and the possibility has to be faced that interpolations may have crept into the text in places where formerly the external attestation would have been thought sufficiently strong.

Dr. George Milligan(43) traces the danger of textual corruption to which the New Testament writings were exposed to a threefold cause, (i) the material upon which the autographs were written, (ii) the employment of non‐professional scribes, (iii) the fact that the thought of the need of absolute verbal reproduction was strange to early scribes. The last named fact led, not only to attempts to improve the grammar and to add “explanatory words”, but also to the insertion “even of deliberate changes in the supposed interests of historic or dogmatic truth”. Milligan instances the case of Dionysius of Corinth who, “in view of the circulation of his epistles in a falsified form”, is found “naïvely comforting himself with the thought that the same fate had befallen the Scriptures” (p. 179 n.). “The general result”, Dr. Milligan concludes, “is, that instead of assigning textual corruption to a comparatively late date ... everything rather points to the conclusion that, the nearer we get to the original manuscripts, the greater were the dangers to which their text was exposed” (p. 180).

In view of this position, it is important to ask whether interpolations may not exist which have left no trace whatever of their origin in the abundant documentary evidence we possess. A representative statement of this view may be found in the words of Dr. James Moffatt (INT., p. 36 f.): “Even where the extant text does not suggest any break, the possibility of interpolations cannot be denied outright; the distance between the oldest MSS., or even the oldest versions, and the date of composition leaves ample room for changes to have taken place in the interval between the autograph and the earliest known text” (p. 38). “The extent of interpolations varied from a word or two to a paragraph, and the motives for it varied equally from sinister to naïve” (p. 38).(44)

One argument in favour of this view may be drawn from the state of the existing MSS. and versions. The multitudinous variations which occur in these documents cannot be explained without admitting the free treatment which has been mentioned, and which was natural at a time when the Gospels were not yet looked upon as “sacred books”. In large measure such additions as we find were drawn from floating Christian tradition, and in many cases, e.g. the _pericope adulteriae_, they probably reflect historic fact.(45) Nevertheless, they are not genuine parts of the New Testament. The further argument is an inference: if such variations from such causes occur in the MSS. and versions we possess, may there not be interpolations of which we have no external indication in the existing texts?

Stated in this way the question invites an affirmative answer, but there are other factors which have yet to be considered. As a matter of fact, there is little profit in a broad and general discussion. We touch the heart of the problem only when we consider the _types_ or classes into which such insertions might conceivably fall. On the whole it is best, even if only for purposes of argument, to admit the possibility that insertions unmarked by signs of textual variation exist, and to ask: Of what character may we suppose these insertions to be, and can we define any limits within which they are more probable than others? In particular, is Lk. i. 34 f. a likely or probable instance? It is obvious that hard and fast lines cannot be drawn in individual cases. Nevertheless, it ought to be possible to say whether or not a passage like the one we are considering is, or is not, the work of a redactor.

Those instances of insertions, _where textual variations can be cited_, supply us with the safest criterion for other suspected cases. Of these instances many, as we have seen, were drawn from the floating tradition of the Christian communities. An interesting case is suggested by Dr. J. H. Moulton (_From Egyptian Rubbish Heaps_, p. 101 f.). He traces the saying, “Father, forgive them; for they know not what they do”, to the reminiscences of the centurion who was present at the death of Jesus. “The words are not in Luke’s original Gospel, but as the great Professor Hort said in regard to the fact that these words cannot be textually defended, ‘Few if any words in all the Gospels bear more intrinsic witness to the truth of what they relate than these’ ” (p. 103). On general grounds, it may very well be, that similar items of tradition have found their way into the existing texts, leaving the surface of the textual stream unruffled. But it is clear that, in any suspected case, the insertion could be the act of the author himself and not the reader. If the latter really is the case, the insertion must have been made very early, and must have been of such a kind as not to awaken comment or dissent.

A second kind of insertions may possibly be found in _explanatory words or phrases_, introduced with the intention of bringing out the original writer’s meaning. We may take as an instance Rom. iv. 1 (“What then shall we say that Abraham, our forefather according to the flesh, hath found?” εὐρηκέναι), where Sanday and Headlam say that they “regard the omission of εὐρηκέναι as probable with WH. text Tr. RV. marg.” (ICC., Rom., p. 99).(46) In this case, however, as in so many others, the gloss, if gloss it is, is reflected in the textual evidence. Nevertheless, the possibility may be allowed, that such glosses exist even where variants cannot be cited. In these cases, however, it is clear that the insertions must have been very early and very happy, and that in specific cases their presence can rarely be conceded with complete confidence.

Yet another class of interpolations may possibly be found in certain passages in the Gospels which later conditions obtaining within the Christian Church have shaped. That later experience did interpret the words of Jesus and give the sense of them in its own terms, need not be questioned. But it should always be remembered that in any suspected case, the process may well have been complete by the time that the Evangelists wrote, and that the passage is not an interpolation at all. There are very good grounds for this opinion even in cases in which variations in rendering can be cited from patristic and other sources, as, for example, in the case of the Great Commission in Mt. xxviii. 19. This fact makes it all the more difficult to concede an interpolation where the textual record is unbroken, though again the possibility that such cases do exist may well be left open.

The cases just considered help us when we come to think of _doctrinal modifications_. As regards these, it is important to draw again a distinction which has been already made. We must distinguish, on the one hand, between those instances of doctrinal modification that are due to the Evangelists themselves, and which are in no sense interpolations, and, on the other hand, those which may subsequently have been made by later scribes or readers. Cases of the former kind unquestionably occur in the Gospels. We have only to examine the way in which the First and Third Evangelists have treated the Second Gospel, which lay before them, to be assured of this. Alterations, e.g., are made out of a sense of reverence for the person of Jesus (cf. Allen, ICC., St. Mt., p. xxxi f.). Mt. xix. 17 (“Why askest thou me concerning that which is good?”), and the changes which Mk. vi. 5 f. has been subjected to, both in Mt. and Lk., will serve as illustrations.

Modifications of this kind are not, however, the sort we have specially in mind. It is the second type, those which are interpolations proper, that we have particularly to consider. The existence of these has frankly to be admitted. It is beyond question that doctrinal insertions were introduced into the text of the Gospels by later scribes and readers. The one case of Mt. i. 16 is proof positive of this (see pp. 105 ff.). If the opinion, that the original ending of our Second Gospel was deliberately suppressed, is correct, Mk. xvi. 9‐20 may be cited as another instance.(47) An important qualification, however, requires to be made. In the two cases mentioned there is a conflict of textual evidence, and, as regards the latter, the objections are reinforced by the internal evidence, arising from the vocabulary, the style, and the subject‐matter. The present writer must needs conclude that _the presence of textual variation is an almost necessary condition in the case of a doctrinal insertion_. It is more difficult to say how far this requirement should be pressed in the other types of interpolation which have been mentioned, but as regards doctrinal modifications the test is thoroughly legitimate. Without going so far as to pronounce it absolutely impossible, we may say that _the theory, that doctrinal insertions may exist where the extant texts show no break, is improbable in the extreme_.

In taking this view, we are not confined to the plea of the early and abundant nature of textual evidence, or to the effect of controversy in preserving the purity of the text, though these are arguments of very great weight. A sufficiently decisive factor is _the character of the existing textual variants_.(48) If authentic items of Jesus‐tradition and “explanatory words and phrases” have not been able to enter the textual stream unnoticed, can we suppose that doctrinal modifications have breasted the waters without leaving so much as a ripple? If even an insertion like “Father, forgive them; for they know not what they do” has not been able successfully to conceal itself, can we believe Lk. i. 34 f. to have succeeded in doing this? Can we think that, like Melchizedek, the passage is without father, mother, genealogy and beginning of life? In asking these questions we need to recall the character of the section. It is such as radically to transform the standpoint of the chapters in which it occurs. It speaks of matters which, for a considerable time at least, were not known among the mass of Christian believers, and were never accepted by some. To suppose, then, that it is a non‐Lukan doctrinal interpolation, is a flight of faith, for which those who can make it should receive the credit that is due, but of which the present writer must confess that he is not capable.

While, however, we conclude that the theory we are discussing is manifestly improbable, we have admitted our inability to pronounce it impossible in any shape and form. Provided we agree that the Third Gospel never circulated without Lk. i. 34 f., there is one point where the passage might have entered as an insertion, and that is in the interval before circulation. But even here it is difficult to suppose that the passage was added by some one other than St. Luke himself. In our entire ignorance of the circumstances under which the Gospel came to have a wider circulation, we cannot say that this supposition is inadmissible. It has a bare possibility in its favour, but not more. If a linguistic examination of the passage gave a result unfavourable to Lukan authorship, the possibility would become more significant. But if the contrary proves to be the case, then it becomes so remote as to be unworthy of serious consideration. It is because of this position that we have described the present argument as being not completely conclusive in itself, and the one line of reasoning as complementary to the other. Quite apart, however, from the linguistic argument, the difficulties which the theory of non‐ Lukan interpolation has to face on textual grounds are formidable.

II. Linguistic and Stylistic Examination of Lk. i. 34 f

Our second task is to make _a linguistic and stylistic examination of Lk. i. 34 f._ At the beginning of the last chapter we drew attention to the importance of the test. It cannot be too strongly affirmed that any hypothesis of interpolation, which does not take account of the linguistic characteristics of the passage, is premature; indeed, it may easily turn out to be a rather glaring case of _non sequitur_.

It is precisely the linguistic test which we miss in the arguments of those who claim that Lk. i. 34 f. was not written by St. Luke. Usually it is thought enough to argue an incompatibility between this passage and its context, and straightway to assign the former to the pen of an unknown redactor. We may illustrate this method from the two articles in the _Encyclopaedia Biblica_ to which reference has been made. In the article on “Mary”, Schmiedel says (col. 2956): “It has to be pointed out that even in Lk. i only two verses—vv. 34 f.—contain the idea of the virgin birth clearly and effectively; and these disturb the connexion so manifestly that we are compelled to regard them as a later insertion”. The only argument of a linguistic character is the remark: “Note, further, that apart from i. 34 ἐπεί (‘since’) is not met with either in the third gospel or in Acts”. Usener writes (col. 3349): “To Joh. Hillmann (JPT. 17, 221 ff.) belongs the merit of having conclusively shown that the two verses in Lk. (i. 34 f.), the only verses in the Third Gospel in which the supernatural birth of Jesus of the Virgin Mary is stated, are incompatible with the entire representation of the rest of chaps, i and ii, and _thus must have been interpolated by a redactor_”.(49) It is theories of this kind that we have in view when we say (p. 47) that to state such a conclusion is to take two steps where there is ground for one only.

The importance of the linguistic argument is manifest in such works as Sir John C. Hawkins’s _Horae Synopticae_ (2nd ed., 1909) and Dr. W. K. Hobart’s _Medical Language of St. Luke_ (1882). It has also received great emphasis in the books in which Harnack has sought to prove the Lukan authorship and early date of the Acts, viz. _Luke the Physician_, _The Acts of __ the Apostles_, and _The Date of the Acts and of the Synoptic Gospels_.

It may not be without value to ask how far the linguistic argument can take us. We may certainly lay down the broad proposition that arguments in favour of an interpolation ought to be supported by the linguistic facts; provided, of course, that the suspected passage is susceptible of the linguistic test. We do not forget that a passage may be of such a neutral character as not to admit of that test. In that case we have to be content with other available arguments. Where, however, the linguistic test can be applied, and where the result is strongly favourable to the genuineness of the passage, that, assuredly, is a very serious objection for the theory of interpolation to face. It becomes especially formidable, if we can bring forward no evidence to prove an anachronism, or if we can allege no real textual objections. Under such circumstances, indeed, we may well adopt the rule that, in cases of this kind, we have not to do with the insertion of a redactor; unless, of course, we have good reason for saying that the interpolator has entered deeply into the original writer’s style. The view here taken does not mean that all objections to a passage are sufficiently met if we can state a strong linguistic case on the other side. We shall have reason to take up this point again (p. 69). For the present it is sufficient to say that each kind of argument must be given its own particular force. In the case of a passage where objections arising from context and subject‐matter cannot be gainsaid, we must conclude that the passage is of later date than its context, but not more. In a case where the facts of vocabulary, style, and subject‐matter are sufficiently favourable, and no textual difficulties forbid, we must ascribe the passage to the original writer. In a case, finally, where both kinds of conditions occur, we must suppose that the passage was afterwards inserted by the writer himself into the body of his own work. Clearly, then, the linguistic examination of a suspected passage is a matter of great importance. In the case of Lk. i. 34 f., it is not too much to say that it is a task as necessary as it is neglected.

It may be objected that the passage is one of two verses only, and that, in consequence, it is much too brief to allow of satisfactory results. On the other hand, it should be remembered that the thirty‐seven words of the section include several interesting phrases and points of construction, which are so important in matters of this kind. Moreover, in the case of St. Luke, we are dealing with a writer who has a very distinctive style.(50)

Harnack has recognized the force of the linguistic argument in the case of two verses (thirty‐one words). These are the last two verses of the Acts. After remarking that, so far as he knows, it has never been questioned that these words come from the author of the complete work, though they have the appearance of being a postscript, he continues: “Moreover, in content and in form they agree so closely with the Lukan style that from this point of view strong arguments can be produced in favour of their genuineness” (_Date of Acts_, &c., p. 94). In a footnote he adds the linguistic argument. This is quite enough for our purpose. It is true that the genuineness of Lk. i. 34 f. is questioned by many (on other than linguistic and textual grounds). Nevertheless, the field is open for inquiry as to whether “in content and form they agree so closely with the Lukan style that from this point of view strong arguments can be produced in favour of their genuineness”. After all, the length of the passage is not the vital consideration, but its _character_ (which may, or may not, be more striking than that of a much longer section); and this is something which can come out only after actual examination.

We turn, then, to the linguistic examination of Lk. i. 34 f. According to the Westcott and Hort text, the passage is as follows:

34. εἶπεν δὲ Μαριὰμ πρὸς τὸν ἄγγελον Πῶς ἔσται τοῦτο, ἐπεὶ ἄνδρα οὐ γινώσκω; 35. καὶ ἀποκριθεὶς ὁ ἄγγελος εἶπεν αὐτῇ Πνεῦμα ἅγιον ἐπελεύσεται ἐπὶ σέ, καὶ δύναμις Ὑψίστου ἐπισκιάσει σοι· διὸ καὶ τὸ γεννώμενον ᾿ἍΓΙΟΝ ΚΛΗΘΉΣΕΤΑΙ, υἱὸς θεοῦ.

In treating these words, we shall not follow the order in which they occur, but the order of their importance for our investigation.(51) It is clear that the words fall into different classes: (_a_) according as they are neutral in character, that is to say, of insufficient importance either way in deciding the question; (_b_) in so far as they create difficulty on the assumption of Lukan authorship, and, to that extent, support the theory of interpolation; (_c_) in so far as they give clear support in favour of Lukan origin.

a.

In the first class we may include the words: ἀνήρ, καὶ ἀποκριθείς, πῶς, ἄγγελος, δύναμις, ἅγιος, εἶπεν with dat., υἱὸσ θεοῦ, and perhaps even Πνεῦμα ἅγιον.

Every one of these words and phrases is well represented in the Lukan writings, and in the case of some of them we get, on investigation, remarkable results.(52)

Take the case of ἀνήρ. In the NT. it occurs 212 times, and of these no less than 125 appear in St. Luke’s works (26 in G. and 99 in Acts), i.e. 58 per cent. Still more remarkable is the result when we compare ἀνήρ and ἄνθρωπος. Whereas the other Evangelists use ἄνθρωπος very frequently indeed (218 times), they employ ἀνήρ only 20 times. St. Luke also (especially in the Gospel) uses ἄνθρωπος frequently (93 times), but he has ἀνήρ 26 times (cf. Mt. 8 times, Mk. 4 times, Jn. 8 times). If we take both Lukan writings, the usage of ἄνθρωπος and ἀνήρ is roughly equal, whereas in the rest of the NT. it is as 9 is to 2. We can say, therefore, that St. Luke shows a liking for ἀνήρ, whereas Mt. Mk. and Jn. markedly prefer ἄνθρωπος. However, the word is so common that we can lay no stress on the fact that it occurs in i. 34, where the connexion demands it. We can only note its congruity with a Lukan liking.

Καὶ ἀποκριθείς is also interesting, though not, of course, in any way decisive. In Lk. the phrase occurs 14 times; in Mt. it is found 6 times; in Mk. 8; never in the Fourth Gospel, and never in the Acts. It occurs, that is to say, in those parts of the New Testament in which sources, probably Aramaic,(53) are employed. This is in line with the view expressed by Moulton and Milligan with regard to the aorist passive forms of the verb.(54) They say that they incline to the opinion that ἀπεκρίθην “belongs only to early Hellenistic, whence it was taken by the LXX translators to render a common Hebrew phrase, passing thence into the narrative parts of NT. as a definite ‘Septuagintalism’ ”. It is in keeping with this view that καὶ ἀποκριθείς ... εἶπεν should appear in that part of St. Luke’s Gospel where most of all we have reason to posit Semitic sources, whether oral or documentary. As we have seen, half the record of this expression in the New Testament, apart from Lk. i. 35, is in the Third Gospel. The presence, then, of καὶ ἀποκριθείς in Lk. i. 35 is congruous with these facts; more, perhaps, we cannot say.

A word like πῶς has no bearing on our present investigation, and the same is true of ἄγγελος, δύναμις (otherwise, however, of δ. in combination with nouns, &c., in the gen.), ἅγιος (very frequently in Lk.), εἶπεν (with dat.),(55) and υἱὸς θεοῦ.

Μαριάμ (of the mother of Jesus) occurs more often in Lk. than in other NT. writers (9 times and probably 10 in the G., once in Acts); the form Μαρία appears but once (ii. 19 is doubtful). In Mk. Μαρία occurs once, Μαριάμ never; in Mt. we find Μαρία 3 times and Μαριάμ probably twice. The use of the form Μαριάμ in i. 34 is therefore in agreement with St. Luke’s usage, but of course this does not preclude the hand of an interpolator, since every instance of Μαριάμ (of the mother of Jesus) in the Third Gospel occurs in the first two chapters.

As is well known, the phrase Πνεῦμα ἅγιον is very frequently found in the Lukan writings. The percentage is as much as 60, and out of the instances in the NT., where the phrase is anarthrous, more than 50 per cent, are in St. Luke (G. and Acts). The phrase is therefore very strongly Lukan. But perhaps we ought not to include the phrase among those which tell strongly against the theory of interpolation, since a redactor would easily and naturally introduce it in the connexion in which it appears in i. 35. “The new view was not an intruder from the sphere of heathen mythology, but a logical conclusion from the belief that our Lord was _God’s Son by the operation of the Holy Spirit_” (Harnack’s _Date of Acts_,(56) p. 144). We can say therefore that Πνεῦμα ἅγιον is admirably in keeping with Lukan usage but hardly more. The case is quite otherwise with the whole phrase, Πνεῦμα ἅγ. ἐπελεύσεται ἐπὶ σέ, as we shall see.

Καλέω is also a word which might be considered here, for it is, of course, a very common word. Having regard, however, to the way in which it is used, it will be better to take it later.

Summing up our results thus far, we may say that we have found nothing that is out of accord with Lukan usage. On the other hand, indeed, every word and phrase we have examined is well represented in St. Luke’s writings. Nevertheless, the words are common elsewhere, and in no case do they tell decisively either way.

b.

_We now come to words which present difficulties, less or greater, on the assumption of Lukan authorship, and so far tell in favour of the theory of interpolation._ These are—ἐπεί, γινώσκω, and perhaps τὸ γεννώμενον.

1. We introduce τὸ γεννώμενον here, because the expression, as distinct from the construction, occurs nowhere else in Lk. As a matter of fact it occurs nowhere else in the New Testament in this form. The perfect passive participle, however, appears twice in the Johannine writings: τὸ γεγεννημένον ἐκ τ. σαρκὸς σάρξ ἐστιν (Jn. iii. 6), and ὅτι πᾶν γεγεννημένον ἐκ τ. θεοῦ νικᾷ τ. κόσμον (1 Jn. v. 4). What is more important is that there is a close parallel to τὸ γεννώμενον in Mt. i. 30, which reads, τὸ γὰρ ἐν αὐτῇ γεννηθὲν ἐκ πνεύματος ἐστιν ἁγίου. The complete clause in Lk. runs, διὸ καὶ τὸ γεννώμενον ἅγιον κληθήσεται, υἱὸς θεοῦ.

It is certainly open to any one to argue that the passage in Lk. is introduced by an interpolator who is under the influence of Mt. i. 20. Why, however, while under that influence, he should so far enter into Lukan usage as to introduce the Lukan διὸ καί, and κληθήσεται, to say nothing of putting Πνεῦμα ἅγιον into a different connexion in a characteristically Lukan phrase (Π. ἅγ. ἐπελεύσεται ἐπὶ σέ. Cf. Acts i. 8 and see later), are questions which it is not easy to answer. Assuredly there is not much here to support the hypothesis of interpolation, and when we consider the constructional use of the article with the participle, there is still less, if indeed anything at all. To consider τὸ γεννώμενον is rather a concession to carefulness than the acknowledgement of a real difficulty.

2. Γινώσκω must be examined, because in i. 34 it is used of knowledge in the way of marital relationship. The only parallel in the New Testament is Mt. i. 25, where, however, it is used of a man: καὶ οὐκ ἐγίνωσκεν αὐτὴν ἕως οὖ ἕτεκεν υἱόν. On the other hand, in other senses, γινώσκω occurs fairly frequently in Lk. It is, however, in no sense Lukan, being distributed evenly throughout the New Testament, except in the Johannine writings, where it is very common.

We cannot, therefore, produce evidence to show that elsewhere St. Luke uses γ. in the special sense of i. 34. Nevertheless, there is no reason why he should not have written γ. in that passage, and there are considerations which go to show how he could easily have used the word.

In i. 34 and also in Mt. i. 25 γινώσκω is by no means a “Hebraistic euphemism”,(57) yet it is probable that the influence of the Septuagint is to be found in both passages. In the LXX there are several instances of γ. used, as in i. 34, of a woman. It is so used in Gen. xix. 8 (of Lot’s daughters), in Judg. xi. 39 (of Jephthah’s daughter), and in Num. xxxi. 17 (of the women of Midian). If, then, we are right in tracing the influence of the LXX, in i. 34, we have ground for finding the hand of St. Luke in that passage, even though he never again uses γ. in that sense. For it is just in Lk. i, ii that the influence of the LXX is most marked.(58)

Even if we do not press LXX influence (for γ. in this special sense is found “in Greek writers from the Alexandrian age down”),(59) it is not at all apparent why St. Luke himself should not have used the word. And if the argument in favour of the theory of interpolation is to be sustained, it is scarcely enough to urge the bare fact that St. Luke does not use γ. as in i. 34 elsewhere. An idiom which occurs in Greek writers from the time of Menander(60) (B.C. 325) may well have been known to a writer like St. Luke, apart from its presence in the Septuagint. If verses 34, 35 are indeed Lukan, it is quite probable that in γ. we should find the influence of the Septuagint, but we are not at all shut up to Septuagint usage. In the connexion in which it occurs γινώσκω was a suitable word to employ, and its presence there is in no way incongruous with Lukan authorship.

3. In these verses the word which is of greatest difficulty is without doubt ἐπεί. In the rest of the New Testament it occurs 25 times. Of these 10 are found in the Pauline Epistles and 9 in the Epistle to the Hebrews. The remaining 6 appear in the Gospels; 3 in Mt., 1 in Mk., and 2 in Jn. Apart then from i. 34 ἐπεί occurs nowhere in St. Luke’s works.

There are, it is true, two Lukan passages, one in the Gospel (vii. 1) and the other in the Acts (xiii. 46), where ἐπεί δέ occurs in some MSS. The true reading, however, in both cases is probably ἐπειδή.(61) We have, therefore, to face the fact, that not only is ἐπεί found nowhere else in St. Luke’s works, but that elsewhere he seems to prefer ἐπειδή and ἐπειδήπερ (the latter in the Prologue to the Gospel, and the former five times out of the ten cases in which it occurs in the New Testament). Here is the strongest argument, which on linguistic grounds can be urged against the genuineness of i. 34 f. The richness of St. Luke’s vocabulary increases the difficulty.(62) Why, if he has used ἐπεί in i. 34, he should never employ it again, is a question which it is not easy to answer. If, in view of the evidence as a whole, the case for an interpolation fails, we shall have to content ourselves with the fact, however strange, that here and here only έπεί occurs in Lk. A writer indeed may use a word once and never again. Ἐπεί occurs but once in Mk. (xv. 42), and it may be so here. Assuredly, in a linguistic argument room must always be left for the occurrence of ἅπαξ λεγόμενα in an individual writer. The force of this contention is, however, somewhat weakened by the preference which St. Luke seems to show for ἐπειδή, and it must be allowed that the case for an interpolation does receive support from ἐπεί.

c.

We have now to consider the _third_ division of the linguistic evidence. It includes the following words and phrases:

τὸ γεννώμενον (the construction), κληθήσεται, δύναμις Ὑψίστου, διὸ καί, ἐπισκιάσει σοι, Πνεῦμα ἅγιον ἐπελεύσεται ἐπὶ σέ, εἶπεν δὲ ... πρὸς ...

1. We begin with τὸ γεννώμενον (the construction.) As is well known, the article with the participle is quite a characteristic of the Lukan writings. “Participles with the article often take the place of substantives”, writes Plummer (ICC., St. Lk., p. lxii). The instances given by Plummer are as follows:

ii. 27. κατὰ τὸ εἰθισμένον. (Here only in NT.)

iv. 16. κατὰ τό εἰωθός. (Here and Acts xvii, 2 only.)

viii. 34. ἰδόντες δὲ οἱ βόσκοντες τὸ γεγονὸς ἕφυγον. (Here and Mk. v. 14; Lk. [xxiv. 12]. Cf. also Acts iv. 21.)

xxii. 22. κατὰ τὸ ὡρισμένον. (Here only in NT. Cf. the parallel passages, Mt. xxvi. 24 and Mk. xiv. 21, where we find καθὼς γέγραπται περὶ αὐτοῦ.)

xxiv. 14. περὶ πάντων τῶν συμβεβηκὸτων τούτων. (Cf. Acts iii. 10.)

To these may be added xxi. 36, xxiii. 47, 48. The construction is clearly Lukan, without, of course, being exclusively Lukan, and though τὸ γεννώμενον does not occur elsewhere in St. Luke’s works, the verb is not uncommon (10 times out of 93 in the NT., of which 40 occur in the Genealogy in Mt.).

2. Κληθήσεται. In his _Date of Acts_ Harnack underlines this verb, as a Lukan trait, wherever it occurs in the “We” Sections, which he prints on pp. 4‐12. Out of the total number of cases in which it occurs in the New Testament, no less than 44 per cent. are found in the Lukan writings. In the Gospel it is present 41 times. It should also be noted that when we compare καὶ τὸ γ. ἅγιον κληθήσεται with the analogous phrase in Mt. 1. 20, τὸ γὰρ ἐν αὐτῇ γεννηθὲν ἐκ πνεύματός ἐστιν ἁγίου, in the latter the Lukan καλέω is absent. Of course καλέω is a common word, but St. Luke’s use of it is distinctive, and with this usage κληθήσεται in verse 35 agrees.

3. We have referred to δύναμις already,(63) and have said that while frequent in Lk., it is too common a word to be important for our present purpose. The case is otherwise with the phrase δύναμις Ὑψίστου. St. Luke is fond of using δ. in composition with other words in the genitive. In his Gospel, he employs it with τὸ πνεῦμα, ὁ θεός, Κύριος, οἱ οὐρανοί, and ὁ ἐχθρός. In the Acts (viii. 10) we have ἡ Δύναμις τ. θεοῦ ἡ καλουμένη Μεγάλη. In Mt. we find this usage twice; in Mk. once; in the main epistles of St. Paul it occurs 13 times; elsewhere in the New Testament 7 times. That is to say, out of 29 instances in the New Testament (other than i. 35),(64) St. Luke has 6 (or 20 per cent.). We may therefore say that this again is a marked characteristic of St. Luke’s usage, and though the phrase δ. Ὕ. does not occur again in Lk. (it occurs nowhere else in the NT.), it is thoroughly congruous with the Lukan style. We have also to note the word Ὕψιστος. Out of 12 instances in the New Testament St. Luke actually has 8, or 75 per cent. As, however, three of these occur in chaps. i and ii, it might be argued that the interpolator has introduced Ὕ. in verse 35 under the influence of these very chapters. That, however, he should combine it with δ. is interesting. Indeed, on the theory of interpolation, our interpolator has combined a distinctively Lukan word (Ὕψιστος) with another word (δύναμις) which St. Luke often uses (24 times), to produce a characteristic Lukan phrase (δ. in composition with a noun in the genitive)!

4. Διὸ καί. Elsewhere St. Luke uses διό 9 times (once in the Gospel and 8 times in the Acts). In this respect he may be compared with St. Paul, who uses the word 25 times, and the author of the Epistle to the Hebrews, who employs it 8 times. In the Catholic Epistles it appears 6 times. There is but one instance in Mt. and a doubtful case in Mk. The results are also interesting when we take διὸ καί. Out of 10 instances in the New Testament, St. Luke has 2 (Ac. x. 29 and xxiv. 26), St. Paul has 6, and Hebrews 2. There is not an instance in Mt. or Mk., or anywhere else in the New Testament. We are far from suggesting that no one else could use διὸ καί.(65) The point is that the supposed interpolator has introduced the phrase into the work of a writer who, with St. Paul and the author of Hebrews, alone among New Testament writers employs it!

5. Ἐπισκιάσει σοι. Ἐπισκιάζω appears in four other places in the New Testament. Of these, three are connected with the story of the Transfiguration (Mt. xvii. 5, Mk. ix. 7, Lk. ix. 34). That the remaining instance should be Acts v. 15 is, in connexion with our present problem, an interesting fact. Thayer‐Grimm remarks that the verb occurs in “profane” authors, “generally with an accusative of the object, and in the sense of obscuring”. In the Septuagint, however, it is used of the divine covering or overshadowing (cf. Ps. xc. (xci.) 4; Ps. cxxxix. (cxl.) 8; Ex. xl. 29 (35)). We have to ask whether these passages, especially the last, have influenced the writer of i. 35. We cannot assume the point, of course, but there is much to be said for it. The thought of the cloud of Yahweh overshadowing the tent of meeting may very well have shaped the thought and the phrasing of δ. Ὕψίστου ἐπισκιάσει σοι. If there is any weight in this suggestion (cf. Plummer, op. cit., p. 24), again it tells for Lukan authorship—so far, that is to say, as the undoubted fact that chaps. i and ii have a distinctly Old Testament atmosphere will take us. Apart, however, from such considerations it is a remarkable fact, on the theory of interpolation, that a word so rare in the New Testament, and one which St. Luke uses more than any one else, should appear in the suspected verses. Acts v. 15 (ἵνα ἐρχομένου Πέτρου κἂν ἡ σκιὰ ἐπισκιάσει τινὶ αὐτῶν) is enough in itself to raise the gravest doubt that we have here to do with an interpolator.

6. Πνεῦμα ἅγιον ἐπελεύσεται ἐπὶ σέ. Here we have first to call attention to the verb ἐπέρχομαι. Apart from Eph. ii. 7 and James v. 1, this verb is limited to the Lukan writings, where it occurs six times (i.e. besides i. 35). We have already spoken of Πνεῦμα ἅγιον and remarked that, while it is characteristic of St. Luke, we could not lay stress upon that fact, since even an interpolator would naturally introduce a reference to the Holy Spirit in such a connexion as i. 35. If, however, as now we take the whole phrase, we come to a very different conclusion. For in Acts i. 8 we have the significantly close parallel, ἐπελθόντος τ. ἁγίου πνεύματος ἐφ᾽ ὑμᾶς. The parallel speaks for itself!

7. We consider lastly, εἶπεν δὲ ... πρός. A comparison of passages in the four Gospels and the Acts gives the following results:

εἶπεν δέ: Jn. 1 (& 2?); Lk. (G.) 60; Acts 15; Lk. (G. & Ac.) 75 εἶπεν ... πρός: Mt. 1?; Mk. 2; Jn. 9; Lk. (G.) 79; Acts 26; Lk. (G. & Ac.) 105 εἶπεν δὲ ... πρός: Lk. (G.) 25; Acts 2; Lk. (G. & Ac.) 27

To the facts noted in the foregoing table we may add that εἶπεν πρός occurs nowhere else in the New Testament. St. Luke, therefore, has it 105 times out of 116. Still more is εἶπεν δὲ ... πρός limited to St. Luke. No other New Testament writer uses the phrase, and St. Luke has it 27 times.(66)

In his three books on the Acts, Harnack is fond of underlining Lukan characteristics in the “We” Sections, in order to show the linguistic identity which exists between these Sections and the rest of the work. Let us see how Lk. i. 34 f. appears, when treated in this way; not forgetting, of course, that we are dealing with two verses only. It is obviously impossible to indicate by this method the special significance of each word or phrase; this, however, has already been shown. Our results may be represented as follows: _εἶπεν δὲ_ Μαριὰμ _πρὸς τὸν_ ἄγγελον Πῶς ἔσται τοῦτο, ἐπεὶ ἄνδρα οὐ γινώσκω; καὶ ἀποκριθεὶς ὁ ἄγγελος εἷπεν αὐτῇ _Πνεῦμα ἅγιον ἐπελεύσεται ἐπὶ σέ_, καὶ _δύναμις Ὕψίστου ἐπισκιάσει σοι· διὸ καὶ τὸ γεννώμενον_ ἅγιον _κληθήσεται_, υἱὸς θεοῦ.

A possible reply to the linguistic argument presented above is that we may have to do with an interpolator who has thoroughly entered into the Lukan style. If our examination has shown anything at all, it has shown that Lk. i. 34. f. is very far from presenting neutral features: it is shot through and through with “Lukanisms”.(67) But, it may be asked, could not an interpolator, strongly influenced by the Lukan style, have penned these verses?

Let us see what, on that hypothesis, the interpolator has done. He has produced a passage of thirty‐seven words, in which there is not a construction, and only one word (ἐπεί), which is not well represented in the Lukan writings. He has used a word (γινώσκω) in a sense not elsewhere illustrated in those works, but a word which St. Luke would naturally employ in the connexion in which it occurs. He has employed words, phrases, and constructions for which St. Luke has a fondness, such as καλέω, δύναμις Ὕψίστου, διὸ καί, the article with the participle in place of a noun (τὸ γενν.).(68) He has used two verbs (ἐπισκιάζω and ἐπέρχομαι) which are rare in the New Testament, but which St. Luke uses more than once; the phrase Π. ἅ. ἐπελεύσεται ἐπὶ σέ, which is closely paralleled in Acts i. 8; and, above all, the markedly Lukan εἶπεν δὲ ... πρός.

This feat, it must be confessed, is a striking performance. If, indeed, it has been achieved, we must conclude that it has been carried out deliberately. We make every allowance for the possibility that a redactor may well enter into the style of an author. But to suppose that in so short a passage so many Lukan features have come together without premeditation or design is all but impossible. We make bold to say that, if we must admit such an undesigned collocation of “Lukanisms”, we can have little confidence in the linguistic argument anywhere.

But can we believe that the linguistic features of Lk. i. 34 f. have been _purposely_ introduced? Such a question is its own answer. No one, assuredly, would resort to the desperate expedient of supposing a redactor, who laboriously amasses Lukan characteristics, with the intention of passing off the very phraseology of his insertion as genuine. A modern interpolator might work along these lines, but not an ancient redactor. Interpolations are not forgeries. The thought of consciously reproducing stylistic features in an insertion would probably never have occurred to a redactor of the Gospels.(69)

So far then as linguistic considerations go, we must conclude that our unknown interpolator is a mythical personage. We do not forget the difficulty of ἐπεί, but if Lk. i. 34 f. is a non‐Lukan interpolation, we must have more support than this. Warp and woof are Lukan; only a single thread gives cause for hesitation. Must not this hesitation give way when we look at the facts as a whole? Can we strain out the gnat, and swallow the camel? Assuredly on linguistic grounds the most reasonable conclusion we can frame is that _Lk. i. 34 f. comes from the hand of St. Luke himself_.

III. Summary and Conclusion

We have now to co‐ordinate our results. However strong a linguistic argument may be, there is perhaps always room for the view that it is confirmatory rather than demonstrative. In the present case also, the shortness of the passage can be pleaded. In noticing this objection we urged that the character of the passage is the relevant consideration, and we think Lk. i. 34 f. meets this demand. But we have no need to press the linguistic argument to the extent we ourselves believe to be legitimate, when we find that both this argument and the textual argument point steadily in the same direction. It is this fact, that both arguments converge on the same point, which is the ultimate ground for our conclusion. Short of supplying a rigid demonstration, which should not be sought, it is sufficient to establish for us the Lukan authorship of Lk. i. 34 f.

This view carries with it at once the further conclusion that at some time or other St. Luke taught and believed in the Virgin Birth. But before we can rest satisfied with this result, we need to look more closely at _an alternative form of the interpolation‐hypothesis_, to which reference has already been made (p. 36). This is the view of Kattenbusch, Merx, Weinel, and J. M. Thompson (_Miracles in the New Testament_, p. 149).

According to this theory the interpolation consists in the phrase ἐπεὶ ἄνδρα οὐ γινώσκω, an insertion which, it is contended, has transformed the promise of a natural conception into the prophecy of a virgin birth. Mr. Thompson notices the two forms which the theory may assume. The insertion may be either “a modification of St. Luke’s source, introduced by the Evangelist himself, as editor”, or it may be “a later addition to the text of Lk. by some person or congregation who wished to make the miracle quite clear” (p. 149). It is obvious that, in its former shape, this hypothesis would not seriously affect our results reached thus far, provided we could agree that “verse 35 is not inconsistent with human parentage” (Thompson, p. 148), and is best interpreted in this way. As regards the second form of the theory, the case is different. If ἐπεὶ ἄνδρα οὐ γινώσκω is the addition of a later reader or congregation, it is much more difficult to think that St. Luke taught the Virgin Birth. It would not be impossible; but it would leave the whole problem to rest upon the interpretation of verse 35.

We are unable to accept the theory that ἐπεὶ ἄνδρα οὐ γινώσκω is an insertion of unknown origin, for the following reasons:

1. _On the whole, the more natural interpretation of verse 35 is that in itself it implies the Virgin Birth._ It is easier, on this view, to explain ἐπελεύσεται and ἐπισκιάσει followed by διὸ καί. (Cf. Schmiedel, col. 2957 n.; Plummer, _St. Lk._, p. 24f.; Lobstein, op. cit., p. 67.)

2. _No textual evidence can be cited in support of the theory._ This is frankly admitted by Mr. Thompson, and the insertion is explained as an editorial modification. We could regard this explanation as sufficient, if the “insertion” could be looked upon as an “explanatory phrase”, intended to sharpen a reference to the Virgin Birth, which had already been found in the context. On this reading of the problem, absence of textual variation might not be an insuperable difficulty. But if we must regard ἐπεὶ ἄνδρα οὐ γινώσκω as a doctrinal modification—an attempt on the part of an unknown editor to impose upon the narrative a sense quite different from that which previously it had been understood to bear—then the argument sketched in the first part of the present chapter is wholly against the theory. We cannot understand why no echoes of the earlier view have lingered.

3. _It is difficult to suppose that a later reader who sought to work up the original narrative in the interests of the Virgin Birth would have exercised such restraint._ To expand a narrative in the direction of the sense which it already bears is a conceivable suggestion. To transform it totally by merely adding four words is a theory which does not carry conviction. Was ever an interpolator so ingenious as this?

On the other side may be pleaded (1) the difficulty of ἐπεὶ, (2) many of the arguments we have sketched in Chapter II. The difficulty of ἐπεὶ we have to admit. As regards the second point, we believe that the theory we have yet to outline in the next chapter meets the case much better, without suffering from the special objections which can be brought against the view we have just discussed. For the reasons given we are unable to accept that view. We prefer to regard Lk. i. 34 f. as a unity, and to interpret both verses as implying the Virgin Birth. And as we have found sufficient reasons, both on textual and linguistic grounds, for ascribing the passage to St. Luke, we believe that he taught the Virgin Birth.