P46 is a manuscript copied from a Greek - Latin exemplar. This explains the many readings of P46 that are subsequently found in later Greek - Latin manuscripts and it explains the direct influence of Latin errors in the Greek.
[Don't quote me yet on this, please read on.]
In that doorstop of a study Scribal Habits, James Royse discusses the singular reading επευξ of P46 in Heb 5:6. This is almost certainly nonsense as it stands, having replaced the normal ιερευς ('priest'). Zuntz suggested that this reading is a Latin alphabet error, a replacement of Greek Rho ρ for a Latin 'p' which is then graphically represented as a Greek Pi, π. The Xi ξ is a 'simple' replacement of ς. Royse rejects this explanation ('there appears to be no other evidence that our scribe was in any way influenced by Latin') and treats the reading as an inexplicable error (though he treats the preceding ει as a separate variant).
First, I believe the whole variant is ειεπευξ for ιερευς. That is, the first syllable ει represents an itacistic reading of the initial iota of ιερευς.
Secondly, I may have found another instance of a Latin misreading in 2 Cor 10:12 where we have the puzzling νεκρουντες ('died') for μετρουντες.
The appearance of νεκρ- for μετρ- might be interference of the Latin stem MORT- as in MORTALES. Admittedly, a few other errors have to take place at the same time τρ - ρτ, but that is not an argument against this proposal. Τhe fact that we have a strange reading implies that a few errors have occurred—what is left for us is to reconstruct these errors.
Thirdly, I actually quite like Zuntz's suggestion, and would love to see it as the hypothesis for an extended study of P46. What if there is a large Latin element in this manuscript? Can other readings be explained by means of Latin influence? In addition, it would open up interesting questions on the 'how?' of the countless instances where P46 joins the Greek - Latin bilingual manuscripts virtually on its own.
So, this is the justification of my first paragraph. A theory that may be worthwhile to put to the test, either to reject or accept.