The spread of Christianity among the educated classes of the Empire brought into sharp relief the formal inadequacy of Scripture when measured against the standards of classical literature, whose influence in cultural life was still pervasive in Late Antiquity. In particular, it became evident that the Book of Psalms – the most prominent example of ‘Christian’ poetry – could not rival pagan poetry in terms of literary form, at least in its Greek rendition. It is within this context that the composition of the Metaphrasis Psalmorum (hereafter M.) must be situated: a hexametric paraphrase of the Greek Psalter erroneously attributed to Apollinaris of Laodicea and dated to the 4th or 5th century. The work survives in 39 manuscripts. The earliest of these (V = Ottob.Gr. 59, Diktyon 65300) dates to the 13th century, while the majority stem from the 15th and 16th centuries. In several manuscripts, the paraphrase is preceded by the Protheoria, a programmatic prologue in which the poet outlines his motivations. According to this reflection, the Greek translation of the Psalms had compromised the original “grace of the metres”, prompting pagans to dismiss Christian poetry. Now that Christianity has triumphed throughout the world and across all languages since Pentecost, however, it is fitting that the Ionian language of epic poetry should also be employed to praise God. Though this argument may seem somewhat contrived – indeed, it is – it nonetheless reflects a broader cultural effort to reconcile Christianity with classical paideia, one of the most defining characteristics of late antique intellectual life.

However, the prestige of the biblical model was too great for the paraphrase to achieve a fully independent existence, and the textual tradition of M. continued to interact with that of the Psalter even after its composition. This interaction can be observed on two levels. The first concerns material transmission: some manuscripts of M. are also witnesses of the Psalter in the textual form of the Byzantine koine. In O (Barocci 25, Diktyon 47311 = Rahlfs 1704) and Barocci 48 (Diktyon 47334) the Psalm text is written in red ink within the line spacing of the M. text. In P (Par. Gr. 2743, Diktyon 52378 = Rahlfs 1742) it appears in the outer margins alongside with exegetical excerpts. In Q (Par. Gr. 2782^A, Diktyon 52419) and I₂ (Metora, Metamorph. Gr. 402, Diktyon 41812) only the beginning of each hemistich is transcribed on the right of the corresponding hexameter. In Laur. Plut. V 37 (Diktyon 15985 = Rahlfs 293), the text of M. (copied by L) and the text of the Psalms (copied by L², who also corrects L’s text) alternate folio by folio, mirroring each other. All these manuscripts appear to be stemmatically related with regard to the text of M.; however, this does not imply that their Psalm texts derive from a common sub-archetype. A brief comparison of variant readings (cf. e.g. Ps. 3.2_b: ἐπανίσταντο O; ἐπανίστανται PL²; ἐπανίσται ut vid. QI₂; Ps. 26.9_c μου PL²I₂: μοι O; Ps. 31.1 ψαλμὸς τῷ Δα(υὶδ) εἰς σύνεσιν L²I₂: ψ. τῷ Δ. συνέσεως P: non hab. O) rather suggests that it was inserted independently into several copies at different stages of transmission. Additionally, most witnesses of M. report the first hemistich of each psalm as an identifier preceding each paraphrase, suggesting that this feature likely dates back to the archetype. The coexistence of M. alongside its model within the manuscript tradition has fostered a more complex form of interaction, which leads us to the second point of our analysis. The Psalm text employed by the metaphrast is rather difficult to define but clearly does not coincide with the Byzantine koine. This discrepancy was noticed by readers of M., who sought to adapt the paraphrase to the version of the Psalter familiar to them. These interventions, unsurprisingly, are mostly attested in manuscripts that transmit M. alongside the Psalter and can be broadly classified into three types: (1) addition; (2) omission; (3) alteration. The third type is relatively rare, as it usually concerns specific details that could be overlooked, whether out of accuracy or negligence. One illustrative example occurs in M. 9.54 ῥιγίστοιο πόνου καὶ ἀρῆς στόμα τοῖο βέβυσται, where PR (with R = Par.Gr. 2868, Diktyon 52506) perhaps rightly conjecture δόλου, drawing on Ps. 9.28_a, οὗ ἀρᾶς τὸ στόμα αὐτοῦ γέμει καὶ πικρίας καὶ δόλου. An instance of omission concerns M. 13.8–15: this section is absent in QPRL²I₂, as it corresponds to Ps. 13.3_c–l, which is not found in the koine text. O, by contrast, takes a more conservative approach and preserves these hexametres; however, it mistakenly transcribes in their line spacing the text of Ps. 13.4–7, which was subsequently erased. The most frequent type of intervention is addition. In some cases, it serves solely to align the M. with the division of the verses in the koine. For example, L²I₂ interpolate a ‘filling-verse’ (εἰν βλεφάροισι τεοῖσιν ἅτ’ ἔργμασι παντοσσώοις) after M. 30.47, in order to reproduce the division of Ps. 30.21 into three hemistiches of their Psalm text (κατακρύψεις... σου | ἀπὸ... αὐτοὺς | ἐν... γλωσσῶν). Similarly, following M. 138.38, the verse καὶ γὰρ πάγχυ ἐρισταὶ τυγχάνετ’ εἰς ἐπινοίας is added by P. The purpose here is twofold: firstly, to restore the alignment with the Psalm text (since M. 138.39 condenses Ps. 138.20_a-b); second, to adapt the passage to the koine reading ἐρισταί ἐστε, whereas M. here presupposes ἐρεῖς. Elsewhere, additional verses are inserted to compensate for textual lacunae. M. 96.11–27, for example, appears only in the group of manuscripts that also transmit the Psalter (OL²I₂PR) and is undoubtedly spurious on stylistic and metrical grounds. These verses correspond to Ps. 96.6–12, and it is impossible to say whether they were absent from the metaphrast’s Vorlage or whether their paraphrase was lost – at least from the archetype. Similarly, in M. 43, no paraphrase is provided for Ps. 43.14. As a result, L²I₂ interpolate vv. 28–29 of Ludwich’s edition, while PR add only one verse (γείτοσιν ἄμμε ὄνειδος ἅπασι πέριξ τε γέλωτα). Such interventions have had a significant impact on the textual history of M., particularly since R was most likely used as Druckvorlage by A. Turnèbe for the editio princeps of the poem (Paris, 1552). Consequently, the interpolations found in R circulated as authentic until the publication of A. Ludwich’s critical edition (Leipzig, 1912), arousing perplexity among philologists.

The dating of these interventions remains uncertain, ranging from the time of the archetype (after the 10th century) to that of the extant written sources, most of which date to the 15th century. Notably, the desire to produce a text aligned with the koine Psalter often took precedence over the aim of transmitting an ‘authentic’ text. This approach should not be dismissed as mere falsification; rather, it is symptomatic of the late Byzantine reception of M., reflecting a weak conception of authorship. In this case, such fluidity was undoubtedly facilitated by the intrinsic authorial weakness of the paraphrastic genre and the prestige of the Vorlage. M. functioned as a supplement to the Psalter: for the Byzantine audience, this relationship carried far more weight than any connection to the individual author of the work.