On pg. 66 I note that Latin had distinctive long or geminate consonants and give some minimal pairs, but I should have added some significant restrictions:
First, the only geminates that occur morpheme-internally are the voiceless obstruents /p, t, k, s/ and the sonorants /m, n, l, r, j/. Voiced geminate stops do occur, but only at a morpheme boundary (agger 'earthwork' < *ad-ger, ad-dō 'I add', ab-bibō ~ ad-bibō) and in loanwords (abbas 'abbot' post-classical from Greek from Aramaic, addax (a type of gazelle), an African word according to Plin. 11.124). Geminate f has a similar limited distribution (of-ferō, suf(f)es 'chief magistrate of Carthage' from Punic), but there is offa 'small lump' whose origin is unknown.
Second, in Classical Latin geminates are limited to medial intervocalic position with the exception of the neuter hocc < *hod-ke, which is always a heavy syllable in Latin poetry unless shortened by IS. Priscian (Keil 1.592) claims et sic in antiquissimis codicibus inuenitur bis c scriptum, but Velius Longus explicitly contradicts this (Keil 7.54: unum c scribimus et duo audimus). There don't seem to be any certain transmitted literary examples of this spelling in absolute final position but OCC is found on CIL 8.17938 from Thamugadas, present-day Timgad, Algeria. In any case hocc is at most a partial exception since it formed a phonological phrase with the following word. In Old Latin there were more final geminates to judge from Plautine scansions like miless, ess, corr, etc. See Questa 2007:20 for examples.