[67] Qua temptet &c. depend on the preceding words, which imply, though they do not express, a state of doubt. Ribbeck, following Peerlkamp, throws this and the following line into the form of a question, comparing vv. 399 foll.; but this would not be so good. ‘Quae via’ Pal. originally, Rom., ‘qua via’ fragm. Vat., Med., Pal. corrected, Gud. &c. Serv. has both. Heyne and Wagn. rightly prefer the former, which is really a poetical variety for the abl. “qua via.” Those who introduced ‘qua’ may have mistaken the nom. for the abl., as Serv. finds it necessary to remark that ‘via’ abl. would be unmetrical. Ribbeck reads ‘qua vi’ from his own conj., which would introduce a rhythm seldom employed by Virg., and without justification here. ‘Via’ of a method 12. 405.