I'm not sure if I've mentioned this here before, but I guess I'll do so now in depth. In my spare time these past few weeks, I've been transcribing an 18th-century Japanese text into modern Japanese. Specifically, the Tale of Moonlight and Rain (雨月物語; Ugetsu Monogatari). I'm pretty sure it's an original, or at least, one of the original copies, given both the grammar and the calligraphy. (ie, it was entirely brush-written; not typed or printed. On traditional rice-paper, and bound by string.)
Actually, my family has a few of those kind of books too from pre-communist China, in my grandmother's collection. It would probably explain my relative aptitude for Asian calligraphy, given my virtually non-existent exposure to Asian society.
Anyway, I first chanced upon Ugetsu while browsing through the Waseda online database of classics. I was attracted by the poetic title, and therefore knew NOTHING about its contents. So, I started blindly transcribing what could have been anything from a dull diary of daily drudgeries to an erotic fantasy. Like all classical compositions, the opening was straight-forward, stating the location of the tale. "In country X, in the city of Y, lived a man named K."
Orthographically, the text was fascinating, as it gave me an accurate depiction of how Japanese looked like 300 years ago. Of course, since it was only one sample, it would be impossible for me to distinguish between characteristics of the writing system in general, and the characteristics of the author's personal penmanship. So, I went back to the database and dug up the other two big classics in Japanese literature: the Tale of Genji, and the Tale of the Heike.
Sure enough, the penmanship styles were different enough that I couldn't read them with the same ease that I had acquired to read Ueda Akinari's Ugetsu. So now I had my rough sample set. But one thing I noticed was that looking at the page as a whole had a different aesthetic value than looking at each character individually.
Which, oddly enough, is pretty similar to my own handwriting, when I write in English. I'll look at a page of my writing and think, "yah, it's neat enough," but when I actually come and edit my own text, I find that individual letters or words look off or wrong. Very strange.
Anyway, what I personally found fascinating was the lack of standard for spelling. It'd be tough to find an English example, but let's say that at one point in history, English was contesting between the older, Latin letters versus the newer reformed letters. Then some writers would appreciate the aesthetic value of certain letters in each set, and combine them in some way. That's sorta basically what I was looking at in Ugetsu; there would be two or three (sometimes more) distinct variants of the same character. In other cases, there was no evidence of the modern letter at all; only the older variant seemed to be in use.
Actually, a good example would be the letters I and J in English, as well as U, V, and W. (I and J came from the same letter, as did U, V and W.) So in older English texts, sometimes it's more useful to substitute the J with an I, or vice versa (eg. the word "jmagery" would really be "imagery". Or "wniuersity" for "university".) The difference here, is that all these letters exist in English now, whereas the hiragana variants simply disappeared in Japanese.
Imagine how much fun we could have if English had two or three alternatives to the already existent 26 standard letters?? (Come to think of it, I suppose technically we could treat minuscule and majuscule letters as being variants of the same letter, and it's only by convention that write majuscules larger... hrmm...)
But so far, I haven't been able to discern any pattern concerning the choice of one hiragana varient over another. Maybe it's purely random/aesthetic, but maybe not...
It's no small wonder that people were slower on the uptake to literacy back then, with their uneven and standardless writing system...