U+00B3

Wait, in UTF-8:
xe4 (which is ä? No, ä is ä which is xc3xa4 in UTF-8, but in Latin-1 it is xe4).
If the text was encoded in UTF-8, then converted to Latin-1/CP1252, and then those bytes were written as HTML entities:
xe3x80x8c (「) -> xe3 (ã), x80 (€ or €), x8c (Œ or Œ).
Let’s check: 【 -> ã is ã = xe3, € = x80,  = x90.
xe3x80x90 is 【! Yes, indeed.
And 〗 -> ã = xe3, € = x80, — = x97.
`xe3x80x