ὧν δὲ δὴ μάλιστα δοκῶ ζῴων, ἔφη, ἐζηλωκέναι ἱπποκενταύρους,
This makes no sense, Loeb has ὃ.
If you read ὧν it would be a connecting relative governed by ἐζηλωκέναι (not agreeing with ζῴων, partitive gen), as in the construction ζηλῶ σε τῆς ἀρετῆς. I.e., ἐκείνων δοκῶ ἐζηλωκέναι τοὺς ἱπποκενταύρους. Make sense?
yes, it looks persuasive but still Loeb and the others prefer a different reading, why?
One correction to what I said above. I hadn’t looked far down enough after the clause you quoted. With ὧν, τhe construction with ζηλῶ is as I said, but ὧν isn’t a connecting relative, just a regular relative pronoun picked up, after the intervening subordinate clauses, by πάντα…ταῦτα. ὧν is Marchant’s emendation. Some mss have ὅ, which is preferred by other editors, including Miller’s Loeb. You could conceivably understand ὅ in at least two ways, either with ἱπποκενταύρους in apposition, or a double accusative construction (cp ζηλῶ σε τὴν ἀρετήν). The former (apposition) is far more likely. The latter is rare and poetic; I’m not sure if there are prose instances.
ὅ is odd; Marchant’s ὧν is appealing and neatly corrects the syntax. Still, I think ὅ is more likely right. It’s another instance of maundering anacoluthon. Not strictly grammatical, but nevertheless stylistically appropriate. “The animal I believe I’ve especially envied, namely centaurs, if they [really] existed in such a way that they deliberate with human intelligence, etc etc…well, don’t I certainly unite all these qualities in myself?” Most of the other variants reported by Marchant (eg ζῷον) are clearly later attempts at correction.
If you don’t have a copy of Marchant’s OCT (with the apparatus) it should be possible to find a scan online.
of course, it is in apposition (as my commentary says).
Yes, ὅ would have ἱπποκενταύρους in apposition; ὧν would be a construction of the type ζηλῶ σε τῆς ἀρετῆς.