It is well established that the Latin word circus 'circle' is a loanword from Greek κίρκος kírkos 'circle, ring'. But it seems that κίρκος is of uncertain origin. One possibility is that κίρκος would be a loanword or an irregular development related to κύκλος kúklos 'circle' (< Proto-Indo-European *kʷkʷlos 'circle'). What about the posibility that in some Indo-European neighboring Greek (Old Macedonian, Peonian, Thracian, ...) we would have had the development:
(1) pIE *kʷekʷlos > *kʷekʷros > *kʷerkʷos -> κίρκος
Do you see any reason why this highly speculative conjecture could be incorrect at any point?