I have been reading Emanuel Tov's classic The Text-Critical Use of the Septuagint in Biblical Research (1981) and came across an interesting discussion he has on proposed tendentious paleographical exegesis in the Septuagint. The theory is that at times translators made paleographical decisions on how to read graphically similar letters based on exegetical concerns. Tov is rightly skeptical of such proposals. Two examples he uses will serve to illustrate the phenomenon:
MT: אבדת רשע "you destroyed the wicked"
LXX: απωλετο ο ασεβης = אבד הרשע "the wicked perished"
Explanation: It has been proposed that the translator of the LXX of Psalms chose to read the text in this way because he wanted to avoid making God himself actively destroy the wicked. The graphic similarity of the letters ת/ה gave the freedom for the translator to divide the words differently and read the letter of his preferred reading. But elsewhere the LXX of Psalms has no problem with the idea of God personally destroying the wicked (Ps 5:6-7; 143:12), so Tov rightly concludes that it is much more likely that this variant was extant in the Hebrew Vorlage of the LXX, or at least was an honest misreading by the translator, rather than an intentional rereading of the text for theological reasons.
Gen. 8:21 (and 3:17)
MT: בעבור האדם "because of man"
LXX: δια τα εργα των ανθρωπων = בעבוד(ת) האדם "because of the works of man"
Explanation: According to Tov, G. Bertram in TDNT argues that "the negative attitude of Hellen. Judaism to work decisively affects the text" in that the LXX deliberately read a ד instead of a ר to add an anti-work polemic into the text. Thus, the translator's background and exegesis influenced his choice on a difficult paleographical question to depart from his Vorlage. Tov again rightly rejects this proposal, insisting that it is far more likely to be an error in a Hebrew manuscript, or at worst an accidental misreading by the translator, possibly influenced by the context of the tilling of the ground and man's evil works before the Flood (both of which share the same root as the LXX here).
I would agree with Tov on both accounts. While I cannot completely rule out the possibility that the LXX translators would engage in some sort of intentional midrashic rereading of the text, I will remain skeptical of such proposals unless supported with strong circumstantial evidence. My experience with the LXX of Genesis has left me suspicious of suggestions that the LXX would engage in such an uncharacteristic method, especially when simpler explanations are readily available. More consistent patterns would have to be established to substantiate such theories, and though perhaps these can be discerned in other books, I find little evidence of it in Genesis. At this point, I would be inclined to say that theories of tendentious paleographical exegesis are the least likely of the three possibilities. Slightly more probable is the possibility that the translators may have accidentally misread their texts, though the meticulous care apparent in the versions and the likely necessity of double-checking one's translation still leaves this unlikely. I am inclined rather in the main to attribute variants easily explained on the basis of Hebrew paleography to variant Hebrew manuscripts.