The Greek grammarian Nicanor of Alexandria, active during the reign of the Roman emperor Hadrian (a.d. 117–38), is known, among other things, as the author of a complex and somewhat idiosyncratic system of punctuation that consists of no fewer than eight different punctuation marks.Footnote ¹ Owing probably to its complexity, Nicanor's system largely remained an isolated phenomenon.Footnote ² For the scope of the present article, one punctuation mark is of particular importance. Its name is ἐνυπόκριτος ὑποστιγμή (literally ‘point below with special intonation’) and its function is to mark the pivot between a preceding subordinate clause and the subsequent main clause.Footnote ³ A good example is Paris’ indignant response to Antenor's suggestion that Helen be returned to the Greeks (Il. 7.359–60):

The ἐνυπόκριτος ὑποστιγμή after ἀγορεύεις indicates that the subordinate clause is coming to an end. This way of punctuating applies only to those subordinate clauses that precede the main clause, which, for Nicanor, is the standard sequence (called ὀρθὴ περίοδος, ‘straight/regular period’). In the case of an inverted period (ἀντεστραμμένη περίοδος), on the other hand, the preceding main clause is separated from the subsequent subordinate clause by means of a διαστολή (‘comma’).Footnote ⁵ As a matter of fact, the function that Nicanor assigns to the ἐνυπόκριτος ὑποστιγμή is not so different from the function that the ὑποστιγμή has in other ancient punctuation systems, most of which essentially operate with two marks, hard and soft. The grammar of Dionysius Thrax, for instance, sees the (soft) ὑποστιγμή as a ‘marker that the sense of the sentence is not yet complete but still defective’.Footnote ⁶ Nicanor's practice in putting the ἐνυπόκριτος ὑποστιγμή shows that he essentially agrees with this definition; for instance, the Homeric example cited above nicely illustrates the definition because the main clause is needed in order to complete the sentence.Footnote ⁷ The main difference between Dionysius Thrax and Nicanor is that the latter expressly wants his ὑποστιγμή to be accompanied by an appropriate intonation: it ought to be ἐνυπόκριτος (from ὑπόκρισις ‘delivery’).Footnote ⁸ This specification matters to Nicanor because his system has in fact two ὑποστιγμαί, one with special intonation (ἐνυπόκριτος), the other without (ἀνυπόκριτος). The function of the latter ὑποστιγμή is to mark the end of an insertion (called διὰ μέσου, literally ‘in-between’, that is, parenthesis broadly understood).Footnote ⁹ Nicanor has an obvious interest in making the two ὑποκριταί distinguishable.Footnote ¹⁰

The relevant sources do not explain which form this special intonation of the ἐνυπόκριτος ὑποστιγμή is to take. But it is a fair guess to assume, with Friedländer, that the reader raises the pitch of the voice towards the end of the subordinate clause, from where he or she then lets it descend into the main clause.Footnote ¹¹ The raised pitch of the voice indicates the incompleteness of the subordinate clause. A main clause is needed to round off the entire sentence. Raising the pitch of the voice towards the end is, of course, also a standard way of marking a question in oral discourse.Footnote ¹² Therefore it is not too difficult to see why someone might feel inclined to (mis)appropriate the ἐνυπόκριτος ὑποστιγμή as a question mark. For this is what in fact happened, as shall be demonstrated shortly. The argument of the present article does not hinge, however, on Friedländer's hypothesis of the raised pitch of the voice. As the texts quoted below shall show, Nicanor himself recognizes that the two fundamental phenomena that he urges to keep separate, preceding subordinate clauses and questions, do have similarities in delivery (ὑπόκρισις), whatever this delivery might have been like.

Nicanor criticizes the misappropriation of the ἐνυπόκριτος ὑποστιγμή on at least five occasions. The number is large enough to prove the point of the present article and, at the same time, small enough to allow reviewing all the instances.Footnote ¹³ Owing to the nature of the extant sources described in n. 1 above, the relevant notes all deal with passages from the Iliad.

In Iliad 3, Aphrodite rescues Paris from certain death at the hands of Menelaus and brings him back to the palace, where Helen addresses him a harsh speech. Its first line reads as follows (Il. 3.428, with Nicanor's punctuation):

Nicanor's note on the passage testifies to a scholarly disagreement over the illocutionary force and punctuation of the first half of the line:

The τελεία στιγμή (literally ‘complete point’) is Nicanor's hardest punctuation mark. It rounds off complete sentences that are followed by an asyndeton in the next sentence.Footnote ¹⁴ To punctuate Il. 3.428a with an ἐνυπόκριτος ὑποστιγμή is out of the question for him because the first three words of Helen's speech can in no way be analysed as a subordinate clause.Footnote ¹⁵ The misguided τινές might actually agree on this point. For their motive appears to be another one: they put an ἐνυπόκριτος ὑποστιγμή ‘owing to the delivery’ (διὰ τὴν ὑπόκρισιν). This seemingly vague expression must refer to the delivery as a question. The justification for this explanation is provided by none other than Nicanor himself when he expressly refers back to this comment in a note on a scene from Book 4 of the Iliad.Footnote ¹⁶

Instigated by Zeus, Athena arrives on the battlefield, illustrated by the ominous simile of a meteor, in order to renew the fighting. A τις-speech summarizes the anxiety felt among both armies alike (Il. 4.82–4, with Nicanor's accentuation and punctuation):

The first part of the speech (82–3a) is commented on in the following way (sch. A Il. 4.82–3a¹ Nic.):

The back reference to the previous discussion is unmistakable. As is the generalizing character of the note: questions must not be punctuated with an ἐνυπόκριτος ὑποστιγμή under any circumstances.Footnote ¹⁸ This general rule applies, even though, Nicanor admits, there is a noticeable comparability as far as the delivery (ὑπόκρισις) is concerned. In other words, with a view to delivery preceding subordinate clauses and independent questions do display similarities.Footnote ¹⁹ But the dissimilar syntactic status of a subordinate clause and a question—one is dependent, the other not—precludes that one and the same punctuation mark can apply to both because it would undermine the fundamental criterion of completeness.Footnote ²⁰ In all likelihood, Nicanor agrees with the τινές that lines 82–3a are in fact a question.Footnote ²¹

The τις-speech that has just been quoted is very similar to a section from Zeus's speech in the preceding divine assembly that results in the dispatch of Athena (Il. 4.14–16, with Nicanor's accentuation and punctuation):

Not unexpectedly, Nicanor argues along the same lines as in the preceding cases but adds a small point (sch. A Il. 4.14–16 Nic.):

Nicanor again criticizes unnamed critics for putting an ἐνυπόκριτος ὑποστιγμή because they are misled by the delivery as a question. He adds the observation that this delivery is common ‘also to dubitative questions’ (καὶ τοῖς δισταγμοῖς). What he means to say is that the relevant delivery does not only apply to simple yes-no-questions but also to dubitative questions (δισταγμοί).Footnote ²² In this particular case, Nicanor rejects not only the use of the ἐνυπόκριτος ὑποστιγμή to mark a question but also the analysis specifically of lines 15–16 as independent questions (hence no πρώτη ἄνω στιγμή after ὄρσομεν either). Instead, he reads them as indirect questions dependent on φραζώμεθ(α) (‘let us consider’), as his alliterating paraphrase shows.Footnote ²³

Another note, sch. T Il. 3.405 Nic., simply states the fact that the delivery as a question (here with an ironic undertone, ἠθικῶς) does not result in punctuating the line with an ἐνυπόκριτος ὑποστιγμή.Footnote ²⁴

The fifth and final case is again worth quoting, because it indirectly brings out a potential shortcoming of ancient systems of punctuation in general. Hera, fully dressed up and adorned with Aphrodite's girdle, arrives on Mt Ida in order to seduce her husband. He welcomes her with the following speech (Il. 14.298–9, with modern punctuation):

In his note on the speech's second line, Nicanor deliberates on how to take it (sch. A Il. 14.299a Nic.):

By now, the fundamental argument will sound familiar, irrespective of the open question whether someone actually put an ἐνυπόκριτος ὑποστιγμή at the end of line 299.Footnote ²⁵ More importantly, Nicanor expressly accepts the possibility of reading the line as a question.Footnote ²⁶ But his system of punctuation, in all its complexity, does not allow him to mark the text accordingly. The fact that line 299 can be delivered as a question must be spelled out in the commentary.Footnote ²⁷

This is where one might glimpse the rationale of the unnamed critics. They apparently felt the need to mark questions in the text itself and decided to fill what they considered a gap in the extant systems of punctuation. It seems unlikely that they themselves invented the ἐνυπόκριτος ὑποστιγμή to that purpose. More likely is the scenario that they appropriated an extant punctuation mark, the purpose of which was to signal a special intonation. This manoeuvre seemed to offer itself because the relevant intonation was in fact so similar. The corollary of the appropriation was, of course, that one and the same punctuation mark was to have multiple functions. This effect need not have bothered them too much. For a very similar appropriation happened more than a millennium later, albeit the other way around. In early modern Germany it became customary to mark some preceding subordinate clauses with a question mark.Footnote ²⁸ Likewise, the widespread habit of putting question marks after direct and indirect interrogative sentences alike was not reigned in until the middle of the eighteenth century in Germany and England. Moreover, the τινές might have felt that there was no risk that readers or listeners might truly mistake an independent question for a preceding subordinate clause. Or they simply focussed more on producing, as it were, a ‘score’ for the delivery of the text in question and less on the syntactical status of the individual sentence.

Be that as it may, Nicanor, for his part, did not like at all what they were doing. He forcefully argued against what he considered a fundamental misapplication of a particular punctuation mark because it blurred the crucial boundary between complete and incomplete sentences. Most interestingly, he does accept that the unnamed critics have a point when they align the delivery of questions with the delivery of preceding subordinate clauses. But he does not seem to have acknowledged that they equally have a point when they deplore the lack of a question mark and fill the gap. Like most ancient punctuators, he apparently saw no need for a question mark.Footnote ²⁹

Most—not all. Nicanor does not seem to have succeeded in completely eradicating the ‘bad habit’ of marking questions with a ὑποστιγμή. Four Church Fathers, at least, had no qualms about doing exactly that: Eusebius of Caesarea (c.260/4–339/40), Athanasius of Alexandria (c.300–73), Cyril of Alexandria (c.376–444) and Theodoretus of Cyrrhus (c.393–458/66).Footnote ³⁰ Their exegetical works on Biblical texts such as, for instance, the Psalms contain notes where the word ὑποστιγμή clearly betrays its use as a question mark.

Eusebius’ point of reference is a section from the beginning of Psalm 21, which reads as follows (verse 3, punctuated with modern marks as Eusebius understands the passage):

Eusebius opposes the view that the final part should be read as a statement. Quoting a passage from the Gospel of John in support (11.42–3: ‘I [sc. Jesus] knew that you always hear me’), he puts forward a different interpretation (Euseb. Demonstr. evang. 10.8.44):

Eusebius reads the passage as a question and therefore marks it with a ὑποστιγμή. Unsurprisingly, in his commentary on the Psalms he repeats essentially the same interpretation, but the point is less obvious because he gives a mere paraphrase of the positive implication that the leading question (introduced by οὐ) has (Euseb. PG 23.204.50–1):

For the purpose of the present article, the wording of this note is all the more revealing because Eusebius apparently expects his readers to understand without further explanation that he suggests putting what is in effect a question mark. This conciseness presupposes a certain familiarity with the concept, at least among the intended readers of his commentary.Footnote ³¹

Turning to Athanasius, his explanatory paraphrases leave no room for doubt either that he makes use of a question mark. In his case, a point of reference is the following passage from Psalm 9 (verse 25, punctuated with modern marks as Athanasius understands the passage):

Athanasius’ commentary reads as follows (starting with a paraphrase of the passage; PG 27.88.41–50):

Irrespective of whether this comment actually gives an accurate explanation of the difficult passage from Psalm 9, it demonstrates that Athanasius is prepared to use the ὑποστιγμή as a question mark. The phrase καθ’ ὑποστιγμὴν ἀναγνωστέον, which is even repeated in this note, and the subsequent paraphrases make it perfectly clear how he takes the passage and that he marks it accordingly.Footnote ³² This comment and those mentioned in n. 32 below deal with questions that are introduced by οὐ and thus suggest an affirmative answer.

The commentary on the same passage from Psalm 9 from the pen of Cyril of Alexandria (PG 69.780.44–59) is so similar, with plenty of verbatim repetition, that there is little point in quoting it too. The hypothesis that Cyril depends here on Athanasius (or on a common source) is confirmed by his use of the shorter phrase καθ᾿ ὑποστιγμήν in order to indicate a question mark. This is contrary to Cyril's regular practice, because elsewhere he tends to make things clearer by explicitly adding a form of the word ἐρώτησις—for instance, when he gives instructions on how to punctuate and deliver a passage from Exodus (Cyril. Comm. Ioann. 2.143.27–144.2 Pusey):

This combination of the terms ἐρώτησις and ὑποστιγμή is Cyril's standard way of dealing with passages that ought to be delivered and punctuated as questions. The fairly numerous examples discuss various yes-no-questions.Footnote ³³

Theodoretus of Cyrrhus, finally, provides examples in his commentary on the letters of Paul, more specifically, the letter to the Romans. For instance (Rom. 11.7, with modern punctuation):

Theodoretus tellingly comments (PG 82.173.12–14):

Theodoretus stands out because he does not deal with a yes-no-question but with a wh-question, which, however, is rhetorically emphatic and thus suggestive.Footnote ³⁴

Similar attestations of the word ὑποστιγμή (and cognates) can be found, for instance, in the various scholiastic corpora on Greek authors.Footnote ³⁵ The problem with these attestations is that it is virtually impossible to provide even an approximate date for them. One is nevertheless worth looking at because it has been linked to Nicanor.

The relevant Homeric scene is the questioning of the Trojan spy Dolon by Odysseus, who enquires about the exact position of the Trojan allies (Il. 10.424–5, with modern punctuation):

A scholium from MS A (Venetus A, tenth century) reads as follows (sch. A Il. 10.425a Nic.?):

Erbse tentatively attributes the scholium to Nicanor, but this attribution is unlikely to be correct. As Friedländer (n. 1), 71 n. 10 aptly remarks, the note contradicts Nicanor's doctrine both in letter and in spirit. In the present context, the salient point is that this critic suggests identifying a question by means of a punctuation mark that has, as seen, a different function for Nicanor, who does not mark questions in the first place.Footnote ³⁶ Unfortunately, both the identity and the date of this critic remain unknown, except that he must predate the writing down of the manuscript (tenth century), which is not particularly helpful. On the positive side, the striking instruction to mark the same word with two signs, a stroke (διαστολή) and a point (ὑποστιγμή), might help eventually to solve a question that remains elusive: how did the semicolon (;) become the standard question mark in Greek? A possible answer is, perhaps, to be sought in the following direction. The main argument of this paper presupposes that a single punctuation mark, the ὑποστιγμή, could be used for rather different purposes. Readers may have felt that the ambiguity of the ὑποστιγμή is unsatisfactory. As a remedy, scribes who intended to mark a question combined the ὑποστιγμή with another mark in order to make it unique and thus unambiguous. Such a combination of a point and a stroke would be particularly close to a semicolon if the prefix ὑπο- in ὑποστιγμή referred not to its position but to its weaker force.Footnote ³⁷ Needless to say, this hypothesis on the origin of the question mark in the form of a semicolon can be no more than a first attempt and needs further elaboration.

To sum up: the present article collects notes that testify to the use of the ὑποστιγμή as a question mark. These notes all predate the eighth or ninth century by several hundred years (except for the last, which cannot be dated with any certainty). In other words, they substantially predate the period which relevant scholarship almost universally credits with the ‘invention’ of the question mark both in the East and in the West.Footnote ³⁸ This communis opinio needs to be modified in the light of the present findings. On the one hand, the argument may well remain valid that physical evidence in the form of actual manuscripts which document the use of a question mark of sorts does not substantially predate the eighth and ninth centuries.Footnote ³⁹ On the other, the evidence collected in this paper clearly testifies to early attempts at identifying questions by means of a punctuation mark, the ὑποστιγμή. The most plausible conclusion is that contemporaries either largely ignored these early efforts or, as in Nicanor's case, expressly rejected them. The corollary is that these attempts remained more or less isolated and seem to have had no noticeable impact on general principles of punctuation. Apparently, the time was not yet ripe for a punctuation mark whose origins can now be traced back to the second century a.d.