On pg. 393 I discuss the evidence for an Old Latin 3rd sg. perf. ending -īt, but I didn't say anything about an interesting phenomenon in Classical Latin poetry which has recently been studied by Marina del Castillo Herrera. In Ovid there are 42 examples of 3rd sg. perfects ending in -iit before a vowel which scan as -iīt and no examples of the prevocalic scansion -iĭt. This contrasts with the treatment in Vergil where 20 examples of -iĭt and one (maybe two) of the scansion -iīt (Alcides subiit, haec illum regia cepit Aen. 8.363). The probable explanation is that the one example in Vergil is a metrical lengthening on the Ennian and Homeric model whereas Ovid's usage suggest a real (i.e. not metrically lengthened) long vowel. The probable explanation for Ovid's -iīt is that -iit contracted to -īt, a form attested already in Plautus, and that -iīt is a conflation of the contracted and uncontracted forms.
See del Castillo Herrera, Marina. 2009. "Las formas de perfecto de indicativo en -iit." Cuad. Fil. Clás. Estud. Lat. 29.2:5-20.