In the verses before us, Venus presents the armour forged by Vulcan to Aeneas.Footnote ¹ The manuscripts and Macrobius (Sat. 5.8) give the following reading:

This might be translated: ‘Behold the gifts that have been finished by my husband's promised craft’; or: ‘Behold these gifts, made with my husband's promised craft’; or else, putting a comma after en: ‘Behold! These gifts have been finished by my husband's promised craft.’Footnote ²

It is strange, I think, that the scandal of these lines has offended so few of their readers. In particular, promissā … arte is a perplexing phrase.Footnote ³ One can promise labour, money, help, faith, or goods, but skill? The collocation is almost completely unparalleled, and perhaps rightfully so.Footnote ⁴ Indeed, when Venus first asks her husband Vulcan to make these very arms, he replies that he will undertake quidquid in arte mea possum promittere curae (8.401): ‘whatever pains in my craft I can promise you’. He does not promise ars itself, but his sedulous exercise thereof. Ovid makes a similar statement at Tr. 5.6.11–12: quem semel excepit numquam Podalirius aegro | promissam medicae non tulit artis opem. Podalirius does not promise his medical craft, but to help his patients with it.

The qualification of perfect[um] by ablative arte is also unique in the Aeneid. Virgil calls an object perfectum four times, but the accompanying ablative always expresses the material, not the means with which the thing was made: silver at 5.267, 9.263 and 9.357, and ivory at 6.895. (The construction perfectum arte, however, was not unknown to other Latin authors.)Footnote ⁵

Finally, though perfecta munera has an apparent analogue in perfecto munere diuae at 6.637, that phrase expresses the completion of a task, not the fashioning of any concrete object.Footnote ⁶ There is nothing necessarily wrong with perfecta munera in itself, but the phrase—if indeed it is Virgil's phrase—is unique in this sense.

I propose that two alliterative words have been transposed,Footnote ⁷ and that Virgil actually wrote:

For promissā arte, we have the far happier perfectā arte, a phrase that has numerous parallels in ancient usage.Footnote ⁸

The phrasing of the repaired verse is consistent with several other passages in the Aeneid. In the first place, for en promiss[um] + [np] there is the parallel en haec promissa fides est? (6.346) and the similarly enjambed cernes urbem et promissa Lauini | moenia (1.258–9). As for promissa munera itself, there are likewise passages of direct relevance: Sergestum Aeneas promisso munere donat (5.282) and sermonum memor et promissi muneris heros (8.464). The evidence of these two verses is indeed so strong as to render almost incredible the suggestion that promissa was brought into such close proximity to munera without being meant to modify it. Nor should we ignore the special significance of promissa in connection with Venus’ patronage of Aeneas.Footnote ⁹ Consider 8.530–1, less than a hundred lines above:

The promises made to Aeneas by his mother (not his stepfather!) are a repeating theme in Book 8, and at lines 612–13 we have their fulfilment in Venus’ delivery of the arms.

In defence of the received text, one might raise the possibility of a double adjective transference,Footnote ¹⁰ by which the sense of promissā can be attached to munera and that of perfecta to arte. If so, it would be analogous to the two other examples of the phenomenon in the poem: ibant obscuri sola sub nocte (6.268) and tepidaque recentem | caede locum (9.455–6). According to Servius, the latter passage was an example of hypallage, on account of which many manuscripts which he had seen contained the modified reading tepidumque recenti.Footnote ¹¹ Thus it was already debatable in Late Antiquity whether a difficulty like this should be solved by textual emendation or by explaining it away as a rhetorical figure.

Double transference was apparently first proposed for our own passage in the seventeenth century. This occurred in an informal setting: a private lesson given to Agosto Chigi (1662–1744), the future prince of Farnese. When his Piarist tutor pointed out that en perfecta mihi promissa coniugis arte | munera was hypallagic, the boy Agosto responded that, if Virgil had chosen not to employ such a device, he could alternatively have written en promissa mihi perfecta coniugis arte | munera.Footnote ¹² Apparently, however, it did not occur to either teacher or student that this hypothetical rearrangement might actually be the true reading.

In fact, there is a good reason to resist the explanation of 8.612–13 as an instance of hypallage—namely, it is an abstract proposition that relies for its whole meaning on exactness of wording. In this it can be distinguished sharply from the other examples of Virgilian double transference, whether the two mentioned above or a line like Iuppiter et laeto descendet plurimus imbri (Ecl. 7.60). These belong to scenic description, in which a certain looseness of diction can be tolerated, and in which the literal sense is so obviously backwards that the real, hypallagic meaning is immediately apparent. At 8.612–13, however, where the adjectives promissa and perfecta are not concrete but abstract, and where there is no visual imagery to serve as a guide to the poet's meaning, both the hypallagic and the literal interpretations are at least logically conceivable. An intentional double adjective transference is not at home in a setting like this, which is why I believe that the received word order is likelier an error than a poetic device.

Another potential objection is a parallel to the received word order in Allecto's declaration to Juno at 7.545: en perfecta tibi bello discordia tristi.Footnote ¹³ The similarity between en perfecta tibi and en perfecta mei is real, but I do not think that it is by any means decisive. It could very well be, for example, that contamination from 7.545 was the cause of the scribal lapse at 8.612. One could also be so bold as to substitute promissa for perfecta at 7.545. This would do away with the awkward ethical dative, and the verse would now have the satisfactory meaning: ‘behold in this grim war the strife I promised you’. There is perhaps some support for this emendation in the phrase promissi dea facta potens at 7.541 above.

One further observation: I think that it is plausible that Servius had the pristine form of 8.613 in front of him. His comment on the verse was restricted to one word in it:

The cross-reference seems actually to refer to the whole passage 8.530–6. Servius’ relevant note there belongs to the phrase in line 531 (which we have already cited above):

The implicit problem motivating both of these comments was the absence of any explicitly mentioned promise from Venus to Aeneas. But that was all right, Servius wrote: these verses were written κατὰ τὸ σιωπώμενον, that is, in reference to an element of the plot that happened offstage. By connecting 8.612 to Venus’ earlier promises to Aeneas, Servius implied that the promise mentioned now by Venus was likewise her own and not her husband's. In other words, he took promissa as a descriptor of Venus’ gifts, not of Vulcan's craft. Furthermore, his citation of the lone word promissa suggests that he read it as an accusative or nominative modifier of the distant munera. If it were an ablative that relied for its meaning on the neighbouring coniugis arte, it would surely have been more logical to comment on that phrase in its entirety, as with diuae promissa parentis at 8.534. Finally, if Servius was indeed looking at the words en promissa mei perfecta coniugis arte, then we also have our explanation for why he did not comment on the oddness of the phrase promissā arte in the first place.