Katakana were developed as a reading help for Chinese texts imported into Japan, by taking parts of kanji with the desired reading. I found a list of these kanji in the book The Key to Kanji and figured that it might help to remember the reading of a bunch of kanji, so enjoy. [For some kana it is not so obvious which parts were taken, though.]
It’s the first two strokes. I guess the font makes it unclear, but the first stroke of 良 doesn’t have to be a straight down stroke, it can be a bit angled.
Ok, I see it now. Same is true for 介. I was somehow expecting the kanji chosen to represent a kana were the “main” reading, maybe things changed over time.
As far as I understand, the kana weren’t taken directly from kanji, but from man’yōgana, which was basically a way of using kanji for phonetic readings (and how the man’yōshū was written, hence the name). Basically a whole system of ateji. So the assigned mora wouldn’t depend on how the kanji is read, but on how the man’yōgana is read.
I’m pretty sure manyogana were themselves chosen merely for their phonetics… Which makes sense. But it’s possible for those phonetics to change in that time, given how long it’s been.
Nanori readings are basically a collection of archaic readings. But also the mapping of sounds wasn’t always straightforward and occurred in a variety of ways, using both On and Kun readings.
It’s still somewhat arbitrary as Manyogana weren’t particularly standardized. 阿,安,英,足, and 鞅 for example all were used for the あ-sound, there’s no particularly reason to pick 阿 over the others. But it’s also worth noting there are a bunch of Hentaigana which meant one sound could be written in a number of ways up until they were officially made obsolete in 1900.