At Hebrews 5.6, in the citation from Ps 109.4(LXX), P46* clearly has the word EPEUX. The corrector has changed this, by marking the word with dots and writing IEREUS interlinearly.
This is an interesting reading because it raises a number of issues: on the one hand P46* represents 'the more difficult reading' (since this is clearly a rare word); while P46C here represents a reading harmonised towards the text of the LXX (an important general consideration in citations from the OT in the NT, and especially in Hebrews, as Cadwallader showed). Should we follow this reading as the earliest attested reading for this verse? (as indeed was argued by Hoskier, Commentary, 1938, 3-6)
The answer is I think not.
The crucial addition rule is (as always) Hort's dictum: "Knowledge of documents should precede final judgment upon Readings."
First, this is the only place in Hebrews in P46 where we have EPEUX for IEREUS. Every other citation or allusion to Ps 109.4(LXX) has IEREUS (e.g. 7.1, 17, 21).
Secondly, this reading is not supported by any other known witness. It is a singular reading to P46* and is corrected in the manuscript.
Thirdly, P46* has some form. Royse counts 16 singular readings in P46 which result in 'words unknown to grammar or lexicon' (Scribal Habits, 254).
Fourthly, this corrector, with the black ink and thick strokes is most probably the scriptorium corrector (the second corrector, after the scribe himself, so Zuntz, Text, 253, supported by Royse, Scribal Habits, 235) - his task is not comparing the scribe's production with an alternative exemplar, but to correct the worst errors of the scribe's production.
Fifthly, there is a visual similarity between EPEUX and IEREUS - many errors begin as simple mistakes of words that look somewhat similar (Zuntz suggested that there may have been Latin influence prompting the P - R confusion, although Royse finds no other examples, see pp. 260-262).
The answer would seem to be that sometimes the 'more difficult reading' is simply too difficult. The weight of all the evidence must also be considered. Especially in this example since:
Sixthly, this word, EPEUX, is beyond rare, it is attested nowhere (not in LSJ, nor BDAG, nor in Lampe's PGL; I suppose I could have checked TLG to make an even more definitive negative). Hoskier, presuming that it must mean something, thought it may have meant 'a precentor, leader of prayer and praise and song' (p. 6). He hypothesised that early Christians may have engaged in discussions about have to 'name' priests and come up with this word. (In this of course Hoskier has his own form, as Royse shows in discussion of Hoskier's discussion of Hebrew 9.3 at Scribal Habits, 262f.)