白鳥 (2026), slide 44: netopa 'body/torso' < *ni-ato-pa '*胴-†脇-頭' (北海道・カラフト, 知里 人間篇:115; 千歳 netopa, -ke, 中川1995:308). *ni 'tree-trunk' extends to '*torso/body' (parallels: French tronc, English trunk 'trunk/torso'; Japanese 幹 analysed as containing 身). *ato '†脇 side/flank' is the bound front element of atopok 'armpit' (美幌・屈斜路, 人間篇:441), atopok = ato + pok '下/under'; cf. 幌別 yarpok 'armpit' = *i-ar-pok 'its-one.side-under' (人間篇:440). 白鳥 supports ato='脇' via a proposed calque atopok ≈ Middle Japanese wakinosita '脇の下' (何溥嘉 2026 p.c.). *pa = 'head' (頭). netopa is synchronically unsegmentable. Open question (白鳥 p.c.): the Sakhalin (鵜城) topa 'body' is hard to derive from *ni-ato-pa. Dialect tables and sources: morpheme_db/notes/netopa-flank-hypothesis.md.