Why isn't the 'p' in "Colapis" (ancient name for the Kupa river) and "Serapia" (ancient name for the Bednja river) geminated, since 'a' was short?

Question

As far as I can see, all or nearly all scholars who study Croatian toponyms agree on those two things:

1. In the Illyrian language, consonants after a short vowel geminated (like in Middle English, more or less). That’s why Pannonia, from *pen (marsh), has a double n. For example, Šimunovíc, ‘Predantički toponimi u današnjoj (i povijesnoj) Hrvatskoj’ (p. 189) says this (emphasis mine):

Etimolozi polaze od ie. korijena *pen- ‘vlažan, mokar’, koji se nalazi u indo-europskoj imenici *pŏni-i̯o ‘vlažno područje’, ‘močvara’. Ie. *pŏnii̯o razvio se glasovnim zakonom u panonski *pana ‘močvarno područje’. Od skupa *‑ni̯‑ (păni̯o) nastaje reduplikacijom ‑nn‑ zbog kratkoga /ă/, te se oblikovao apelativ *pann-ōn- ‘vlažno, močvarno zemljište’, a onda konverzijom *Pann-ōn-ii̯a ‘močvarni predio’ te plemensko ime *Pann-ōn-es ‘stanovnici na tom zemljištu’. Budući da je prvotno ie. *ŏ (*pŏnii̯o) postalo panonsko ă, a dugo ie. *ō (*pănn-ōn-) zadržano, oblikovan je horonim *Pănnōnīa (hrv. Panonija).

My translation:

Etymologists start from the IE root *pen- (wet), that existed in the Indo-European noun *pŏni-i̯o (marsh). Indo-European *pŏnii̯o developed by regular sound changes into Pannonian (a dialect of Illyrian) *pana (marshy land). From the consonant cluster *‑ni̯‑ (păni̯o) forms the geminate ‑nn‑ because of the short /ă/, so the appelative was formed *pann-ōn- 'marshy land', and then by conversion *Pann-ōn-ii̯a 'marshy land', and the name of a tribe *Pann-ōn-es (people living on that land). Since the short Indo-European *ŏ (*pŏnii̯o) changed to Pannonian ă, and long Indo-European *ō (*pănn-ōn-) was conserved, a horonym *Pănnōnīa was formed (Croatian "Panonija").

2. The ‑ap‑ in the Croatian river names Col-ap-is (ancient name for Kupa) and Ser-ap-ia (ancient name for Bednja) is the Illyrian word for ‘water’, coming from the Indo-European root *h₂ep-.

Now, how do they reconcile those two things? The a in ‑ap‑, if it indeed comes from *h₂ep-, must have been short, so why wasn’t the p geminated?

Answer 1

That quote is not making a general statement that consonants after short vowels are geminated. It is instead saying only that *‑ni̯‑ > geminate -nn- after short vowels.

It makes no claim about the evolution of other consonants or consonant clusters. Strictly speaking it doesn't even make such a claim about short vowels other than *ă.

Note also that in the example given in the quote, there is no gemination of a single consonant, instead a cluster becomes a geminate.

Additionally, whilst it is common for languages to develop a restriction that all syllables be heavy (i.e. short vowels must be followed by a coda consonant), it's more typical for this to be done by lengthening vowels in open syllables than by geminating a following consonant (likewise it's more typical for a restriction against ultraheavy syllables to be realised by shortening long vowels in closed syllables than by losing coda consonants after long vowels).

It's actually this change (open syllable lengthening) and not the change you suggest (post-short vowel gemination) that occurred in Middle English.