Scholars have long recognized that Proba’s Cento Vergilianus contains a corruption at line 42. The transmitted text, as given by Fassina and Lucarini in their critical edition,Footnote ¹ is as follows:

Clearly, line 42 is corrupt on metrical grounds; semantically, it is also suspect (liquidi ignis umor does not make good sense). Thus, some editors print it as two lines, 42a and 42b, with an intervening lacuna:Footnote ²

An associated issue is that line 38 is syntactically (though not metrically) suspect, as there appears to be no verb upon which concreuerit could depend. However, this problem is readily dealt with by postulating (as, for example, Cullhed does) an ut in Proba’s original text,Footnote ³ modified back to the Virgilian reading of et by a later scribal error; such small changes to Virgilian material are common within Proba’s cento technique, and a scribe familiar with Virgil might well have reversed her change unconsciously.

A number of scholars have attempted a restoration of the lacuna, under the hypothesis that the original passage referenced the four classical elements;Footnote ⁴ with fire (et liquidi simul ignis) and water (et caeli mobilis umor) already mentioned, the missing material must by this reasoning reference earth and air. Such restorations also tend to feature a rearrangement of lines, on the assumption that the corruption extends beyond lines 42a–b (possibly encompassing the issue with line 38). Previous attempts to restore the passage include:

Fassina and Lucarini follow Schenkl and Pollman in their rearrangement of lines; Lucarini argues that, as the tener mundi orbis is formed out of the four elements, it should follow the summary of them. However, he also points out that these two proposals share a common flaw in that they place either et liquidi simul ignis or semina terrarumque in the second half of the hexameter; this appears to break Proba’s own compositional rule, whereby a half-verse extracted from Virgil is kept in the same metrical location in which it originated.Footnote ⁸ Though Lucarini does not discuss his reasons for rejecting Green, the difficulty is likely aeris incrementum, as aeris is nowhere found in a fourth-foot position in Virgil. (I would add that incrementum feels like a weak term for describing the creation of fire and air, and that the line as proposed by Green lacks Proba’s characteristic elegance.)

Faced with these difficulties, Lucarini comes to a rather pessimistic conclusion. Since diligent searches of the Virgilian corpus for material that satisfies the metrical and compositional requirements have repeatedly turned up empty-handed, he postulates that Proba, unable to find anything to fill the gap to her satisfaction, left lines 42a and b (et liquidi simul ignis and et caeli mobilis umor) as separate half-lines in the margin, perhaps planning to return to the passage later; these two partial lines were then combined and inserted after line 41 by a later editor. Thus, Fassina and Lucarini propose:Footnote ⁹

Although this scenario is logically possible, it seems rather inconsistent with the immense care that Proba has taken elsewhere in her poem, and with her clear wish, or even expectation, that her poetic fame among posterity should rest upon the Cento;Footnote ¹⁰ one would expect her to have been more painstaking with her magnum opus. In addition, the lines felix, qui potuit rerum cognoscere causas, | omnia et ipse tener mundi concreuerit orbis …, with lines 42a–b placed aside as marginalia, fail to hold together syntactically; one can of course speculate that Proba’s intention was to eventually bridge the syntactic gap as well as the semantic one.

I now propose a reconstruction along rather different lines. It is my contention that the central assumption of those who have previously tried to emend the passage—that is, that Proba’s original text referenced the four classical elements—is incorrect. Rather, I see in these lines a summary of the six days of Creation, but in reverse order:

If this hypothesis is correct, the missing material concerns Day 3 (the creation of dry land and of plants), with probably some additional material to fill out Day 4 (the creation of the sun, moon and stars). liquidi … ignis, the reading attested by most witnesses, would of course need to be completed by a nominative. Alternatively, a number of manuscripts read liquidi … ignes,Footnote ¹¹ which could either stand alone or take some supplementary material, such as a genitive; the latter is arguably more likely as liquidi ignes alone is a rather vague term for the heavenly bodies. I will propose one supplement for each of these two possibilities.

First, with liquidi … ignes:

This proposed reconstruction requires no rearrangement of lines; it fills the gap perfectly; and it is consistent with Proba’s compositional style. In addition, it meets the requirement that any material taken from Virgil should remain in its original metrical position. It is also not difficult to see how such a corruption might have happened. It would be natural for a copyist’s eye to jump from the et of line 42a to that of line 42b, leaving out the intervening material; a subsequent scribe, desperately seeking to amend the obviously unmetrical result (et liquidi simul ignes aetheris et caeli mobilis umor) might well have dropped out the word aetheris given that it is syntactically dispensable; and Proba’s ignes might easily have (in most manuscripts) mutated back to the original Virgilian ignis in a manner similar to the ut to et reversion discussed earlier.

Second, with liquidi … ignis:

There is some textual support for this latter reconstruction in that one manuscript (Ch) reads et liquidi fons ignis et caeli mobilis umor,Footnote ¹⁴ possibly a corrupted form of the original (assuming an intermediate version such as et liquidi simul ignis fontes et caeli mobilis umor, as for aetheris above). One potential objection against fontes is that it is rare in dactylic hexameter, and particularly in the Virgilian style, for a spondee to directly precede a bucolic diaeresis, particularly when there is a break in the sense as observed here;Footnote ¹⁵ another potential objection is that fontes is not found in a precise fourth-foot position anywhere in Virgil. However, Proba does occasionally allow some metrical flexibility within a composite excerpt (one formed by recombining two or more excerpts linked by one or more common keywords).Footnote ¹⁶ fontes occurs in close proximity to liquidi three times in Virgil (G. 2.200, 3.529, 4.18). G. 2.200 (non liquidi gregibus fontes, non gramina derunt) would require a shift of only half a foot to place fontes in the fourth-foot position, while G. 4.18 (at liquidi fontes) requires a larger metrical shift (1.5 feet), but could perhaps more intuitively be recombined with et liquidi simul ignis given the similarity of the introductory particle (at, et) and the lack of intervening text; either excerpt seems a plausible source for fontes in the composite passage.

Returning to the wider context, we might ask on what grounds we can assume that Proba intended this passage as a (reverse) summary of the six days of the Creation. In fact, I contend that this is more likely than the ‘four elements’ reading, for three reasons:

1. (1) Given that Proba is a Christian poet who is writing an explicitly Christian epic, the six days of the biblical Creation seem a more likely referent than the four elements of pagan Greek philosophy.

2. (2) A few lines earlier (in line 36), Musaeus is mentioned. As discussed in detail by Cullhed,Footnote ¹⁷ this is Proba’s Virgilian equivalent of the biblical Moses, and Proba is referring to the Pentateuch when she writes: Musaeum ante omnes uestrum cecinisse per orbem | quae sint, quae fuerint, quae mox uentura trahantur, 36–7. For her to continue this line of thought to reference the Mosaic account of Creation would seem only natural; the four Greek elements, on the other hand, are alien to the Mosaic worldview.

3. (3) The Cento contains a second, extended Creation account at lines 56–135. This account fails to strictly follow Genesis 1 in several respects: it displaces several events onto different days, condenses several days into one, and even leaves out material (such as the creation of the firmament). It also in places displays the influence of pagan thought regarding the created order; for example, the first line and a half, principio caelum ac terras camposque liquentes | lucentemque globum lunae, is taken directly from Anchises’ cosmological speech to Aeneas (Aen. 6.724–5). It is understandable that, preceding this longer account, Proba might wish to emphasize her underlying orthodoxy, assuring readers that, despite the idiosyncrasies of the Creation account to come, she remains a faithful Christian; and the definitive statement in lines 43–5 (haut aliter prima crescentis origine mundi | inluxisse dies aliumue habuisse tenorem | crediderim) suggests that this was indeed her goal in writing this first, shorter Creation account.

We might also ask why Proba should choose to arrange her (first) summary of the Creation in reverse order. Two reasons suggest themselves: first, as is made clear in her statement at Cento 23, she is committed to advancing the idea that Virgil’s works are prophetic of the Christian story (Vergilium cecinisse loquar pia munera Christi). The very fact that the lines hominum pecudumque genus uitaeque uolantum | et quae marmoreo fert monstra sub aequore pontus occur consecutively in Virgil (Aen. 6.728–9), appearing to summarize Days 5 and 6 of the Creation in a precise backwards sequence, might well have prompted her to continue the pattern through the remaining four days.

Second, looking at the introduction to the Cento (1–55) more broadly, we can observe an overall backwards temporal narrative: Proba reflects upon her own past as a poet (Cento 1–12); Virgil (12–23); the Incarnation (31–4); ueterum … monumenta uirorum (35); Musaeus/Moses’ writings (36–8); and finally turns to the first subject of her poem proper, the Creation narrative. This new section begins with the word principio (Cento 56); thus the ‘backwards Creation narrative’ functions as part of a larger movement leading the reader backwards in time to the principium, when the Creation story can begin.

To conclude, this article has presented two variants of a novel reconstruction of Proba’s Cento Vergilianus 42a–b based upon the six days of the biblical Creation, summarized in reverse. This reconstruction not only fills the textual gap more smoothly than previous attempts while meeting the compositional and metrical requirements for a Virgilian cento; it also fits more convincingly with the surrounding context and with Proba’s own Christian religious commitments, while providing a reasonable explanation as to how the textual corruption occurred. There are, then, multiple grounds on which to conclude that one of the reconstructions proposed here, or another produced along similar lines, likely represents the original text.