Basil of Caesarea, on Ps 95:5

“For all the gods of the Gentiles are devils: but the Lord made the heavens.”

“But, if he is of too little importance, according to their reasoning, to be capable of the partnership of the term "Godhead" with the Father and the Son, he is not worthy of sharing any other of the terms proper to God. For, if the terms are considered and compared with each other according to the significance observed in each, none will be found to be inferior to the title of "God." And a proof of this is that many inferior things also are called by this name. Moreover, the divine Scripture does not refrain from using this equivocal term, not even in inconsistent matters, as when it calls images by the name of "God." "For," it says, "let the gods who did not make the heaven and the earth be destroyed and be cast under the earth." It also says, "All the gods of the Gentiles are devils." And the witch with her magic arts summoning up the souls sought by Saul said that she saw gods. Furthermore, even Balaam, a certain diviner and soothsayer, who bore his oracles in his hand, as the Scripture says, and who successfully procured for himself instruction from the demons through his divining trickery, is related by the Scripture to have taken counsel with God. And, it is possible, collecting many such passages from the divine Scriptures, to allege that this name has no precedence above the other appellations proper to God, since, as it has been said, we even find it used equivocally in incongruous matters. But the name of holiness, and of eternity, and of righteousness and of goodness, we are taught by the Scriptures, is nowhere communicated to things that are unfit. Therefore, if they do not deny that the Holy Spirit shares with the Son and the Father in the names piously used exclusively in the case of the divine nature alone, what reason is there to try to make out that he has no partnership in this one alone that both evil spirits and idols have been shown to share through a certain equivocal use?”

Imported from an open dataset — not yet checked against the printed edition.