Punctuation before a Japanese repeat mark sorts one way after a Latin letter, the other after Han

FN-OP_LESS_THAN-001: Punctuation before a Japanese repeat mark sorts one way after a Latin letter, the other after Han

Quirk · measured · recorded 2026-11-04

Comparing a string with punctuation before a repeat or prolonged-sound mark against the same string without the punctuation, the order depends on the character at the start. After a Latin letter the punctuated string is smaller: "a-ゝx" < "aゝx". After a Han character it is larger: "時-ゝx" > "時ゝx". This held for every pair read, with the marks ゝ, 々 and ー and with hyphen-minus, U+2212 and U+2019 as the punctuation. In all 66 readings the two strings are never equal, and a string with the mark added sorts after the string without it, after Latin and after Han alike: "a−ゝ" > "a−".

Basis: 66 recorded readings of =, < and > on 22 pairs of strings, 11 pairs after 時 and 11 after a.

Open: The collation rule behind this is not identified, and none is claimed here.