It is typically assumed that Greek readings take precedence over versional readings in a versional tradition, that in general, every opportunity is taken to correct the version within the availability of Greek texts and the skill to translate them. But in his commentary on the Old Latin text of Romans 5:14, Ambrosiaster does not follow the Greek reading simply because it is Greek, though he is well aware that it differs from his Old Latin text.
Ambrosiaster catches us by surprise in refusing to prefer Greek text forms as a matter of course. For Ambrosiaster, the fact that a reading is in Latin is not in itself reason to reject it as inferior. But how does he justify this preference for the reading of his translation over that of the original language? And what might this tell us about attitudes towards the Greek and Latin traditions in the West at the end of the fourth century?
The discussion begins with Ambrosiaster’s comment on the prevailing Greek reading of Romans 5:14 (with the negative particle):
“Some Greek manuscripts say that death reigned even in those who had not sinned in the way that Adam had.” (Comm. Rom 5:14) 
Ambrosiaster objects to the notion implied by the Greek text that death reigned over all human beings before Moses, whether or not they sinned in the same way as Adam, by turning away from the Creator (Rom 5:14). Ambrosiaster wants to insist that death did not reign over those who like Abraham, acknowledged the Creator before there was a written law that revealed the Creator.
To accomplish this, Ambrosiaster turns to the concept of “natural law,” which he has already noted in connection with the law of the Gentiles mentioned in Romans 2:14: 
“nature itself recognizes its Creator through its own capacity for discernment, not through the [Mosaic] law but rather through natural reason (per rationem naturae).” (Comm. Rom 2:14; CSEL, 81/1.75, translation mine) 
The significance of natural law for Ambrosiaster’s textual problem lies in its nature as an unwritten principle that precedes the written Law. If this natural law undergirds the written scriptures, Ambrosiaster believes he can appeal to its corollary, natural reason, to arbitrate between texts of uncertain provenance. Under this assumption of the conformity of scripture to natural law, it is simply understood that the true text will conform to reason, as Ambrosiaster argues elsewhere: 
“[t]he Scripture wants … many things to be implied, so that the sense gathered from the words should never be contrary to the reason of religion (rationi religionis).” (Quaest. 26) 
So for Ambrosiaster “religious reason” points to the true text in the same way that natural law anticipates the written law. It comes as little surprise then that Ambrosiaster includes “reason” (ratio) as the first of three qualities of “the true text”:
“I consider this to be the true text, when reason, history, and authority are all preserved.” (Comm. in Rom 5:14.5a, see my working translation)
So what are the text-historical implications of Ambrosiaster’s “criterion” of reason?
First, we might point out that while presented as a “criterion,” in practice Ambrosiaster’s appeal to natural reason serves a plainly apologetic function, namely, to defend Old Latin readings against the overwhelming claim to priority of the Greek tradition. Still Ambrosiaster’s appeal to natural reason reveals something of the tension that gripped one particular user of the Old Latin text in attempting to reconcile the clear discontinuity between his preferred text form and the “source of truth” in the Greek tradition. We see him grasping for justification for his preference of the secondary text form even as it stands in outright conflict with the original-language source.
Secondly, if the tension we find in Ambrosiaster is at all indicative of the contemporary situation, we might look for other cases of interaction between the Greek and Latin traditions that suggest a similar conflict. Of course, one obvious case is the Greco-Latin bilingual Codex Bezae. In light of Ambrosiaster, it suddenly begins to matter a great deal who transcribed Codex Bezae and where their loyalties might have been. Were they interested in mere transmission of an ancient text? Were they sympathetic to Ambrosiaster’s views? Were they more engaged in the conflicts of the times? If Bezae’s producers were active in the period between 385 and 415, it is hard to see how they could have been unaware of the Vulgate. But if this is so, the mere selection an Old Latin text already betrays something of their interests.
Finally, the case of Ambrosiaster is rather unsettling to the view that Bezae’s Greek text represents the stable culmination of a tradition of Old Latin exemplars. While not to suggest any personal involvement on his part, in his own words Ambrosiaster fits the profile of an individual who might consider correcting a Greek manuscript on the basis of Old Latin readings. So we have evidence that the necessary motive is attested at the time of Bezae’s production. But regardless of our conclusions, we might agree that Ambrosiaster introduces a certain unforeseen volatility to questions pertaining to the interaction of the Greek and Old Latin traditions in the last decades of the fourth century.
 Ambrosiaster. Commentaries on Romans and 1–2 Corinthians (ed. and trans. G. L. Bray; Downers Grove, Ill.: InterVarsity, 2009) 42, italics mine.
 On the role of natural law in Ambrosiaster’s thought, see S. Lunn-Rockliffe, Ambrosiaster’s Political Theology (Oxford: Oxford University Press, 2007) 50–52. The concept of natural law has a rich tradition in Roman juridical theory and early Christian apologetics, though it is seldom mentioned as explicitly as it is by Ambrosiaster. See A. J. Carlyle, A history of mediaeval political theory in the West, vol. 1, (New York, 1909) 104–105.
 Note that in the first edition (α) of his Commentary on Romans (10:8), Ambrosiaster seems to equate the two, referring to “the natural law, called reason” (lege naturae loquendi ratione) (CSEL, 81/1.347).
 L. Perrone, “Echi della polemica pagana sulla Bibbia negli scritti esegetici fra IV e V secolo,” Annali di storia dell’esegesi 11 (1994) 161–185 at 175.
 Translation by M.-P. Bussières, “Ambrosiaster’s Method of Interpretation in the Questions on the Old and New Testament” in Interpreting the Bible and Aristotle in Late Antiquity (ed. Josef Lösl; J. W. Watt; Burlington, Vt.: Ashgate, 2011) 49–65 at 53.