=91. ILLA QVIES ANIMO.= _Animo_ is locative; or perhaps _in_ should be supplied from the following line: for the joining of a noun with a following preposition already with a complement, see Clausen on Persius I 131 'abaco numeros et secto in puluere metas'. I read _animo_ (found in one of Heinsius' Vatican manuscripts) because of the parallel structure it gives with the following _in ore_, but ANIMI (_BCMFHILT_) is possible enough: _OLD quies_ 7 cites _quies animi_ at Celsus III 18 5.
=91. QVAM TV LAVDARE SOLEBAS.= The same phrase at _Her_ XV 193 'haec sunt illa [_sc_ pectora], Phaon, _quae tu laudare solebas_'. For the persistence of Ovid's old habits, compare _EP_ I x 29-30 (he remains a moderate drinker, as formerly).
=93-94. SIC EGO SVM LONGE, SIC HIC, VBI BARBARVS HOSTIS / VT FERA PLVS VALEANT LEGIBVS ARMA, FACIT= is clearly corrupt, as will be seen from Wheeler's 'Such is my bearing in this far land, where the barbarian foe causes cruel arms to have more power than law' and André's 'Je vis au loin, ici, où un ennemi barbare donne aux armes cruelles plus de force qu'aux lois'. Merkel ejected the distich, which seems the best solution; it is not necessary to the poem's structure, and the iterated _facit ut_ in unrelated clauses at 94 and 97 is suspicious. Also, as Professor R. J. Tarrant notes, the _ut_ in 94 makes one expect that _ut_ in 95 will be correlative, when it in fact continues the thought of 93 (or rather of 91-92, after 93-94 are excised).
Heinsius thought 93 alone to be suspect; if so, the meaning lying behind the text is probably something like 'What I once was at Rome, I still am here'.
=93-94. HIC, VBI BARBARVS HOSTIS, / VT FERA PLVS VALEANT LEGIBVS ARMA FACIT.= Similar statements at _Tr_ V vii 47-48 'non metuunt leges, sed cedit uiribus aequum, / uictaque pugnaci iura sub ense iacent' and _Tr_ V x 43-44; see also Otto _lex_ 3.
=93. BARBARVS HOSTIS.= The same phrase at _Tr_ III x 54, _Tr_ IV i 82, and _EP_ II vii 70.
=95. RE ... NVLLA= _MHIL_ REM NVLLAM _BCFT_. The verb _queri_ can take a direct object, or be constructed with _de_ + ablative, but not both; this would in effect give the verb two objects. _Re ... nulla_ removes this difficulty and is obviously prone to corruption, the true object _de nobis_ being postponed to the following line.