I'm sorry for replying after so much time, but I had some problems with my browser. Grammar is a help, not the key to translation. I wouldn't say that 'Smith's' text is not translatable, but that one needs ways to protect at least common sense. Therefore, if we had only that text, we could perhaps consider φημί as implied.

Since ἐκράνθη is third person singular indicative, not in indirect statement, would the φημί only be implied for the ἀθανάτων δ᾽ ἀπέχειν χέρας?

No -> I say <- this is not true. The same happens here; indirect statement is not needed, if we suppose phemi cutting the sentence.

So "phemi" would apply to both clauses; it would cut the first one and govern indirect statement in ἀθανάτων δ᾽ ἀπέχειν χέρας?

This would be a possible way, if we had to ignore Wilamovitz and Murray, but not common sense.