In inscriptions dating to the early Roman Empire, it is used frequently but inconsistently to transcribe the long vowel /iː/. In Gordon’s 1957 study of inscriptions, it represented this vowel approximately 4% of the time in the 1st century CE, then 22.6% in the 2nd century, 11% in the 3rd, and not at all from the 4th century onward, reflecting a loss of phonemic vowel length by this time (one of the phonological changes from Classical Latin to Proto-Romance). In this role it is equivalent to the (also inconsistently-used) apex, which can appear on any long vowel: ⟨
á é í ó v́⟩/aː eː iː oː uː/. An example would be ⟨fIliI⟩, which is generally spelled fīliī today, using macrons rather than apices to indicate long vowels. On rare occasions, an apex could combine with long i to form ⟨Í⟩, e.g. ⟨dÍs·mánibus⟩.
The long i could also be used to indicate the semivowel [j], e.g. ⟨IVSTVS⟩ or ⟨CVIIVS⟩, the latter also ⟨CVIVS⟩, pronounced [ˈjʊstʊs, ˈkʊjːʊs]. It was also used to write a close allophone[i] of the short i phoneme, used before another vowel, as in ⟨CLAVDIO⟩, representing [ˈklau̯.di.oː].
Later on in the late Empire and afterwards, in some forms of New Roman cursive, as well as pre-Carolingian scripts of the Early Middle Ages such as Visigothic or Merovingian, it came to stand for the vowel ⟨i⟩ in word-initial position. For example, ⟨iNponunt in umeroſ⟩, which would be inpōnunt in umerōs in modern spelling.