In PE16:83 the editors have some trouble interpreting the Oilima
Markirya line _valka wilwarindon_ (OM2a) or _valkane wilwarindon_
(OM2) 'vague as a butterfly', since Q. _valka_ 'cruel, bitter' <
VL.KL. (PE12:101) and later Q. _nwalka_ 'cruel'< NGWAL- (V:377) don't
seem very fitting at the first glance.
However, in PE17 we find (p. 154):
GWAL- 'be stirred, excited, &c.', Q. _walya_ 'be excited (moved)',
_walta_ 'to excite, rouse, stir up', _walda_ 'excited, wild', _walme_
'excitement' and at last untranslated _walka_
This last form is fairly exceptional, since _-ka_ is not a very common
Quenya suffix and usually appears as an adjectival ending after
diphthongs: _tiuka_ 'thick, fat' < TIW- (V:394), _faika_
'contemptible, mean' < SPAY- (V:387), _fauka_ 'open-mouthed, thirsty,
parched' < PHAU- (V:381), _lauka 'warm' < LAW- (V:368).
In Sindarin the derivatives of GWAL died out due to coalescence with
BAL 'have power' except for _balch_ 'fierce, ferocious', cognate of
_walka_. Noldorin derives _balch_ 'cruel' from NGWAL- (V:377).
'Excited, stirring butterfly' indeed suits the simile of a 'rapid and
erratic motion of a butterfly that prevents it being distinctly
perceived' (wherefore 'vague'). So maybe this root was already
conceived at the time of OM2 (probably in a different incarnation,
like *WAL or *WL.K) or at least the idea that 'fierceness, cruelty'
derives from 'stirring, excitement', perhaps even by a blend with BAL-
or Q. _val-_.