Declinatione nominum latinorum a nominibus graecis quae -ης finiuntur perturbor.
Dictionarium L&S exhibet "Arsăces, is m." a nomine graeco Ἀρσάκης. (E in syllaba ultima de "Arsăces" longum esse praesumo.)
Nomen graecum primae est declinationis, sed nomen latinum tertiae est. Et dictionarium Wiktionary exhibet "Gōtarzēs, Gōtarzis" a nomine Graeco Γωτάρζης.
Cur dicimus "Arsacis" et "Gotarzis" et non "Arsacae" et "Gotarzae"?
(Latine scribere volui, sed male scribere timeo. Anglice scribam.)
I'm feeling a bit uncertain right now about how nouns taken from Greek nouns ending in -ης are declined. According to Lewis and Short, "Arsăces, is m." is the Latin name corresponding to the Greek name Ἀρσάκης. Although L&S do not mark the e in the last syllable with a macron, I assume it is long (-ēs).
The Greek name seems to be first declension, but the Latin name is given as a third-declension noun. Similarly, the Wiktionary entry for the name Gotarzes (from Greek Γωτάρζης) marks it as a third-declension noun with the forms Gōtarzēs, Gōtarzis.
I'm confused because I expected that Greek first-declension nouns ending in -ης would be taken into Latin as first-declension nouns of the "Greek" sub-type, with nominative singular forms in -ēs and genitive singular forms in -ae (as described in Joonas's answer here).
When I do a Google search for "Arsacis" or "Gotarzis", I do get many results. I tried searching for "Gotarzae" and found an example in this document: Flavii Iosephi Antiqvitatvm Ivdaicarvm libri XX. Should this be categorized as an erroneous form? I also found results for "Arsacen" and "Gotarzen".
(I realize that names like this aren't necessarily attested in many places, so maybe there is no good way to judge which forms aside from the nominative are "correct" in Latin. If the situation is like that, an explanation of this would also count as an answer.)