ENAs is well known, two different 3 sg. endings can be reconstructed for the Indo-European middle: *-to, ubiquitous in all languages that preserve the middle voice (Hitt, iya-tta(ri) “marches”, Ved. niri-yd- te “dies”, Lat. -tur, Gk. -rat, etc.), and a rarer variant *-o and/or *-e (Hitt, kis-a “becomes”, Ved. say-e “lies”). The same presence of two structurally different endings occurs in tire 3 pl. (Hitt, -anta(ri), Ved. -ate, -ate, Lat. -utur, Gk. -vrat, etc. < *-nto beside Ved. se-re “they lie”, YAv. aijhdire “they sit” < *-ro, *-ero) and the 2 sg. (Ved. -se, Lat. -re, Gk. -oat, Go. -za < *-so beside Hitt, -tta(ri), Ved. sec. -thah, Toch. A -tar, OIr. -ther < ^-thie). For tire 1 sg., on the other hand, only a perfect-like ending *-/12(e) can be reconstr ucted (Hitt, -lilia(ri), Ved. -e, Lat. -or, etc.). The 1st and 2nd plural endings, which present an entirely different structure, and the even more problematic dual endings will be left out of consideration hr what follows.