ἦν, ἦ, ἤμην
(Phryn. Ecl. 123, Moer. η 2, Philemo [Vindob.] 394.19)

D. General commentary

Atticist lexicographers, including Phrynichus (A.1), Moeris (A.2), and Philemon (A.3), discuss the 1st-person singular of the imperfect of εἰμί ‘to be’; all agree in rejecting the innovative form ἤμην ‘I was’ in favour of the more conservative ἦν. Their preference is understandable given the two forms’ respective attestations, although, as Phrynichus acknowledges, ἤμην is not unattested in texts by classical authors. Indeed, ἤμην may have been even more common in the copies of classical texts circulating during the Atticists’ time.

Historically, ἦν is indeed the older form, although it first originated as an innovation. The inherited 1st-person singular form is ἦα, the regular reflex of PIE *h₁é-h₁es-m̥ (cf. Sanskrit ā́sam), attested 12x in Homer besides the variant ἔᾱ (4x), produced by quantitative metathesis, and the thematic by-form ἔον (3x; for the Homeric forms see in detail Chantraine 1958–1963 vol. 1, 287–90). In Attic, ἦα was contracted to ἦ, which was later recharacterised with the secondary ending -ν, making it homophonous with the 3rd-person singular ἦν. The latter is contracted from ἦεν < *h₁é-h₁s-ent (cf. Sanskrit ā́san), etymologically a 3rd-person plural which was reinterpreted as singular (see Willi 2018, 420 n. 11). In Ionic, it replaced the inherited ἦς (< *h₁é-h₁es-t), which was preserved in West Greek dialects. In turn, Attic-Ionic introduced the innovative form ἦσαν as a 3rd-person plural. In Homer, both the 3rd-person singular forms ἦεν (108x) and ἦν (88x) are well attested, together with ἔην (83x), which, since it is mainly used before consonants, may have replaced an earlier *ἔεν, the unaugmented counterpart to ἦεν.

The homophony between the two forms ἦν was probably an incentive for analogical innovation in order to better distinguish the 1st and 3rd persons of the singular. In later Greek, a morphologically middle-passive 1st-pers. sing. form ἤμην was created, which spearheaded the gradual creation of a whole middle paradigm for the verb ‘to be’. Cassio (forthcoming) also stresses the role of the present subjunctiveSubjunctive, which developed middle forms already in Hellenistic times, probably due to the functional overlap with the future ἔσομαι. The innovative middle subjunctives appear in the 2nd century BCE in Delphi (e.g. subj. pres. 3rd-pers. sing. ἦται in SGDI 1799.6 [173 BCE]; SGDI 1696.4 [150–140 BCE]) and later in Messenia (subj. pres. 3rd-pers. plur. ἦνται in the Sacred Law of Andania = IG 5,1.1390.83 [23/24 CE]). Eventually, a full middle present paradigm εἶμαι, εἶσαι, etc. was created (Horrocks 2010, 154; 303–4; for further developments in Medieval and Modern Greek, see E.).

On the basis of the above-mentioned geographical distribution of the earliest middle forms of εἰμί, Kretschmer (1901, 12–3) argued that this was an innovation of the North-West Greek dialects which was adopted by the koine. This hypothesis cannot be ruled out, since the few instances of ἤμην in Attic literature are either corrupt or late (see below), and the form is absent from Attic inscriptions, where ‘with scarcely any exceptions the attested forms of εἰμί are as would be expected on the basis of the literary traditions’ (Threatte 1996, 586; for examples of ἦν, 587–8). Whatever its origins may have been, ἤμην did indeed become the regular koine form. It is the norm in the Septuagint and in the New Testament, as well as in Ptolemaic papyriPapyri, where ἦν only occurs as the 3rd-person singular form (Mayser, Gramm. vol. 1,2.127), and in Roman and early Byzantine papyri (Gignac 1981, 402). Among authors of the Hellenistic and early imperial periods, ἤμην is used by Dionysius of Halicarnassus (3x), Philo (1x), Josephus (2x), and quite frequently by Plutarch (22x).

The scholarly debate about the various 1st-pers. sing. imperfect forms of εἰμί goes back at least to Herodian, who, according to Choeroboscus (B.2), wrote a Μονόβιβλον περὶ τοῦ ΗΝ in which he probably discussed all the homograph words spelled ΗΝ and their prosodies, including the verbal forms ἦν (1st/3rd sing. of εἰμί; 1st-pers. sing. of ἠμί), together with the pronoun ἥν (acc. sing. fem. of the relative pronoun or of the 3rd-person possessive), the conjunction ἤν (< ἐάν), and the interjection ην (see Sandri 2024, 134–5, who further argues that the ancient grammarians also distinguished a ‘deictic’ interjection ἤν/ἦν and an ‘affirmative’ interjection ἥν). Judging from B.11, Herodian probably also discussed the correctness of ἤμην in his treatise on the verbs in -μι (see Schmitz 1997, 76 n. 43), recognising its absence in Homer and its rarity in other classical authors.

The spread of the innovative form ἤμην in the koine easily explains why the Atticists rejected it, in keeping with their general opposition to analogical levelling in verb paradigms (see La Roi 2022, 208–16). More interesting, however, is their attitude towards the attestations of the proscribed form in the canonical authors, as implied by Phrynichus’ remark in A.1 that ἤμην is to be avoided ‘even though it is found in the ancient [authors]’. As Monaco (2024, 115) has recently observed, this is one of several cases in which Phrynichus in the Eclogue criticises authoritative sources for unapproved uses (see e.g. entries ἀλεκτορίς, Λάκαινα, Λακωνική and Phryn. Ecl. 64Phryn. Ecl. 64; 105Phryn. Ecl. 105; 297Phryn. Ecl. 297). What such cases have in common is that the rejected forms or constructions were (supposedly) typical of the koine: Phrynichus projected his linguistic ideology – based on a systematic opposition between Attic and the koine – onto classical authors, according to the principle ‘if you want to appear Attic, do not use what is common in the Koine! (even though it was used in Attic)’ (Monaco 2024, 128). In the case of ἤμην, it is doubtful where Phrynichus could have encountered it in classical authors: while there are a few attestations of ἤμην in classical Attic drama and prose, most of them are textually doubtful (see Rutherford 1881, 240–3; Kannicht 1969 vol. 2, 244–5; Carrara 2020, 27–9).

Looking first at poetic texts, ἤμην is transmitted as a variant reading in the direct tradition of Soph. Ai. 679, Tr. 24, Eur. Supp. 200, Tr. 474, although in these cases it was readily emended by the editors. A grammatical tradition represented by Choeroboscus (B.3) and the Etymologica (B.8) quotes an iambic line, allegedly from Euripides’ Demand for Helen (Ἑλένης ἀπαίτησις), as evidence that ἤμην was occasionally used by classical authors. Since Euripides is not known to have composed a play by this title, the quotation was long considered to be from Eur. Hel. 931 (C.2), a similar line where ἦν is transmitted and corrected to ἦ by the editors. Carrara (2020) has persuasively argued that the line more likely belongs to Sophocles’ securely attributed, though largely lost, Ἑλένης ἀπαίτησις, although in her opinion ἤμην is not the genuine reading – owing to its to foreignness to classical Attic – but an early corruption already widespread in the imperial period (see F.2). Carrara (2020, 23) even suggests that this line, in its corrupted form, was the locus classicus behind the Eclogue’s entry. Two other attestations of ἤμην occur in iambic adespota (C.8, C.9), where they may be textually sound, although both have been corrected to ἦμεν by some editors. However, neither line is very likely to come from a classical play. The former line (C.8) is quoted by Favorinus (Περὶ τύχης 4 = [D.Chr.] 64.4) as coming from a funerary inscription, while the latter (C.9) is transmitted by a cod. of the Etymologica (Vossianus Gr. 20 [12th century] = cod. V of the EM = cod. D of the Et.Sym., see the apparatus to B.8) as evidence that ‘some of the more recent authors’ (τινες … τῶν νεωτέρων) use ἤμην (see Carrara 2020, 27–8, for further references). The latter claim is ambiguous since the term νεώτεροι may refer to Post-classical Greek (in which case the doctrine would not contradict that of Phrynichus), or to the later stages of Attic itself (in which case the Etymologica would admit the coexistence of ἦν and ἤμην in Attic).

As far as Attic prose is concerned, there are only three occurrences of ἤμην in 4th-century BCE authors, of which Lys. Or. 7.34 (C.4) and X. Cyr. 6.1.9 (C.5) are generally regarded as corrupt, while the one in Hyperides’ oration Against Athenogenes (C.7), transmitted by a 2nd-century BCE papyrus (P.Louvre 9331+10438.col. xii.7 = TM 61288 [Panopolis]), stands a good chance of being genuine, and should be counted among the traits in Hyperides’ language that anticipate the koine (Carrara 2020, 29). Indeed, despite his place in the canon of the Ten Orators, Hyperides is generally rejected as a linguistic model in the Eclogue (cf. Phryn. Ecl. 309Phryn. Ecl. 309, 311Phryn. Ecl. 311 and see entry χρεολυτέω), making it possible that he was one of the authors Phrynichus had in mind when he warned against ἤμην. Lysias, too, is frequently criticised by Phrynichus for the colloquial traits of his language (Schmitz 1997, 76 n. 43 and 123; Monaco 2024, 114–5).

In conclusion, it is possible that Phrynichus found occurrences of ἤμην in contemporary copies of more than one classical author (as the plural παρὰ τοῖς ἀρχαίοις also suggests), but that he nevertheless proscribed the form, either because it occurred in the text of authors which he regarded with some suspicion (Lysias, Hyperides), or simply because of his general principle that koine forms should be avoided, even when they occur in Attic. The proscriptions against ἤμην (A.2, A.3) and its compound ἀπήμην ‘I was away’ (B.1, on which B.10 depends) in the other Atticist lexica, which are transmitted in a more abridged form, omit any mention of the classical attestations of ἤμην; the latter were, at any rate, known to the grammatical tradition, as evidenced by references in Choeroboscus and the Etymologica. In Byzantine lexicography (B.5, B.6, B.7) ἦ(ν) is presented as the classical equivalent of ἤμην, which was then the usual written form (see E.), without any discussion of the classical precedents of the latter form. It is instructive to compare the reception of ἤμην with that of the corresponding 1st-pers. plur. form ἤμεθα, which, despite a similar distribution in Post-classical Greek (LXX, NT, etc.), was completely ignored by the purists, owing to its lack of attestations – however marginal – in classical texts.

The more Atticising authors of the imperial period generally follow classical Attic usage in preferring ἦ(ν), although ἤμην is not avoided altogether by Dio of Prusa (Or. 44.6.9), Alciphron (Ep. 2.10.1, 18.3.5), and Lucian (DMort. 9.2.11, DMar. 2.2.14, Asin. 15.1; in VH 2.25.13 παρήμην has the v.l. παρῆν). These occurrences may be due to the influence of the koine (see Schmid, Atticismus vol. 4, 599; Deferrari 1916, 60), but in light of the above discussion it cannot be excluded that these authors felt that ἤμην was legitimised to some extent by its presence in contemporary copies of classical texts. A less rigorously Atticising author such as Achilles Tatius, by contrast, only uses ἤμην (3x), never ἦν (Gammage 2018, 242–3).

The scholarly tradition also attests to a debate as to whether ἦ or ἦν was the proper Attic form, showing an awareness that the former was chronologically earlier. A scholium (B.12), probably going back to Porphyry’s Homeric Questions (see Schrader 1880–1882 vol. 1, 82–3; Pontani 2020, 122–3), states that ἦ is the older form, ἦν the more recent one; similarly, Eustathius (B.9), quoting Aelius Dionysius, assigns ἦν to ‘middle Attic’Attic, middle (for a similar periodisation in Moer. χ 12Moer. χ 12, see entry χολάς, χόλιξ). See also B.4, which quotes Plato (C.6) to exemplify the Attic use of ἦ, and B.13, which reports that Didymus read ἦ in Eur. Hec. 13. Modern scholars tend to agree that ἦ was preferred by earlier Attic authors, although the exact distribution is difficult to reconstruct, since ἦν tended to replace the earlier form in the MSS, and ἦν is guaranteed only in metrical texts before a vowel. See Rutherford (1881, 242–4) and especially Harrison (1942), who noted that while it is always possible to read ἦ in Sophocles (14x) and Aristophanes (12x), with the exception of the late Wealth (ἦν 3x), including C.3 (which seems to have been the locus classicus for the Attic usage of ἦν, because the monosyllabic form is metrically guaranteed by Cario’s reply οἷδά τοι; cf. B.2, B.3, B.7 and see Willi 2003, 56), Euripides probably used both forms throughout his career, although ἦ is never metrically guaranteed. Barrett (1964, 292–3) argued even more strongly against reading ἦν in 5th-century authors (see also Kannicht 1969 vol. 2, 254). Somewhat unexpectedly, the prescriptive Atticist lexica do not deal with the choice between ἦ or ἦν, and indeed the form prescribed against ἤμην in A.1, A.2, and A.3 is ἦν, not ἦ. It is possible that both forms were considered equally acceptable because of their presence in the classical texts, especially if ἦν had already begun to replace ἦ in the MSS tradition.

E. Byzantine and Modern Greek commentary

The imperfect forms of εἰμί in the Byzantine period are treated in detail in AGP vol. 3, forthcoming. In Byzantine Greek the conservative form ἤμην, inherited from the koine, coexists with relics of the classical form ἦν and with the innovative variant ἤμουν (CGMEMG vol. 3, 1729–32). The latter, produced by the backing of /i/ to /u/ between two nasals, is abundantly attested since the early 12th century, but may have arisen even earlier. In turn, ἤμουν developed the variants ἤμου (attested since the 16th century) and ἤμουνε, ἤμουνα (attested since the 15th century). The rare forms ἤμη and ἤμον also occur. During this period, the exclusive use of ἤμην is restricted to the more archaising texts, while other sources mix the two forms; tellingly, Michael Glykas uses ἤμουν in his Verses Written in Prison, a poem with many vernacular traits, while using ἤμην (5x) in his other, more learned works. The more innovative form ήμουν has prevailed in Modern Greek, with the by-forms ήμουνε/-να still common in colloquial speech.

F. Commentary on individual texts and occurrences

(1)    Phryn. Ecl. 123 (A.1)

The innovative forms of the imperfect of εἰμί are also discussed by Phrynichus in Ecl. 118Phryn. Ecl. 118, where the 2nd-pers. sing. form ἦς is condemned in favour of ἦσθα (cf. Moer. η 4Moer. η 4, and see entry ἦς, ἦσθα, ἦσθας). However, whereas in A.1 ancient authors’ use of ἤμην is at least acknowledged (though this should not be taken as an endorsement of the form: see the discussion in D.), no such allowance is made for ἦς, which is bluntly rejected as a solecism, i.e. a syntactical error, because – as the text of the entry makes clear – it could be mistaken for the subjunctive ᾖς. Phrynichus’ less compromising attitude towards ἦς (noted by La Roi 2022, 211; Monaco 2024, 115 n. 23) is also justified by the fact that this form, unlike ἤμην, is not attested by canonical authors. By contrast, although the 2nd-pers. sing. form ἔφης ‘you say’ in place of ἔφησθα, from φημί, represents the same kind of innovation as ἦς, Phrynichus’ attitude towards it is again lenient – even ‘too lenient’, from Rutherford’s (1881, 225) neo-Atticist perspective – since the lexicographer admits that ἔφης is attested, albeit sparsely, in the ancient authors (ἔστι μὲν παρὰ τοῖς ἀρχαίοις, ἀλλ’ ὀλίγον; cf. Soph. fr. 314.444; X. Cyr. 4.1.23; Pl. Grg. 466e.6; [Pl.] Ep. 362e.4). See further entry οἶσθα, οἶσθας, οἶδας and AGP vol. 1, 306 and 308–10.

(2)    Soph. fr. dubium (C.1)

Choeroboscus (B.3) and the Etymologica (B.8) quote this line to exemplify the use of ἤμην, attributing it to a Euripidean play entitled Ἑλένης ἀπαίτησις. It has been traditionally thought that the quote refers to Eur. Hel. 931 (C.2), but this attribution is questionable for several reasons. Firstly, the lectio difficilior Ἑλένης ἀπαιτήσει would be an unlikely error for ἐν Ἑλένῃ, and no play with the former title is ever attributed to Euripides in ancient sources, while Sophocles is known to have written a Ἑλένης ἀπαίτησις, probably centred on the unsuccessful embassy of Menelaus and Odysseus to Troy to demand Helen’s return (cf. Hom. Il. 202–24), of which several fragments survive (Soph. frr. 176–180a; on the play’s place in Sophocles’ corpus, see Lupi 2023, 107–11). Moreover, the textual differences with C.2 are too great to assume that the source of B.3 and B.8 is simply misquoting Euripides’ line. The most likely solution, recently endorsed by Carrara (2020)    but already suggested by Welcker (1839, 120) and Kannicht (1969 vol. 1, 162), is to assume that the quoted title is correct, while the name of Sophocles has been replaced by that of Euripides, possibly because the latter’s Helen was better known. The declaration of innocence by the speaker in this fragment – arguably Helen herself – would not contradict the known fragments of Sophocles’ play, especially fr. *178, where the speaker laments being the target of δυσφημίαι (‘slanders’) from others. According to Carrara, although the grammatical tradition seized on C.1 as an example of the classical use of ἤμην, it can hardly have been the original reading, if the play was indeed by Sophocles or one of his contemporaries: the koine form ἤμην likely replaced a different construction, perhaps ἄρ’ ἦ (as in C.2).