This is what I know.
1) Unlike, say, -p, the final -t of Chinese readings was maintained in careful speech at least as long as the Christian resources of the XVIth century (Frellesvig 2010: 316-7). At least for some, there is a confirmed orthographical confirmation that there was no borrowing with epenthetic vowel.
2) Most of these occur in words that, according to actual spelling, are hidden in kanji, making any kind of ‘tradition’ improbable.
3) The overwhelming majority of those that survivd to modern jōyō list now close it with -ツ.
4) To the moment when the vowel was needed, there was no hearable Middle Chinese to check any feature of readings.
Question: why were some of the syllables fixed in -チ-readings? These are the ones officially declared:
- セチ 節 ‘node’ (Go-on)
- ニチ 日 ‘day’ (Go-on)
- バチ 罰 ‘punish’ (Kan’yō-on, for non-Jōyō ぼち)
- リチ 律 ‘law’ (Go-on)
- イチ 一 ‘one’, 壱 ‘judicial one’ (Go-on)
- キチ 吉 ‘good luck’ (Go-on)
- シチ 七 ‘seven’, 質 ‘content’ (also チ、Go-on)
- ハチ 八 ‘eight’, 鉢 ‘bowl’ (Go-on)
Most of these are 1) unique or occurring at most twice; 2) an astounding number are numerals. What is going on in the assigning of Go-on readings?