What is "won't" a contraction of?

"Don't", "wouldn't", "couldn't" and "isn't" are all contractions of "do not", "would not", "could not" and "is not"... So what's "won't" a contraction of?

It appears to be "will not", but if so, why isn't it "willn't"? (And if there's no good reason, has it ever been "willn't" at any point in history?)

Wiktionary says:

Abbreviation of wollnot or woll + not, negations of archaic form of will.

The Concise Oxford Dictionary of English Etymology agrees:

XVII. contr. of wonnot, assim. of wol not

As to other forms, Etymonline only mentions wynnot:

first recorded mid-15c. as wynnot, later wonnot (1580s) before the modern form [won't] emerged 1660s.

Won’t actually has a pretty interesting and complex history. Ultimately it does come from a contraction of will and not, but it all happened in a rather roundabout way.

It all started off with the Old English verb willan/wyllan, meaning to will, wish, or want. Even in Old English it was used occasionally to denote a future intent. “Ic wille gan” could mean “I want to go” or “I will go”, depending on context.

Now, the thing about negatives in Old English is that they were often reduced:

na(w)ðer = nahwæðer = ne + hwæðer
neither = not + whether

næfre = ne + æfre
never = not + ever

nabbað = ne + habbað
haven’t = have + not

We nabbað naðor ne hlaf ne wæter.
We have neither bread nor water.

Not comes from naht via noht. Related to nawiht meaning naught, it originally meant in no way, but came to be used as an emphatic form of ne. Subsequently it became unstressed and supplanted ne altogether. This is an example of Jespersen’s Cycle.

All these things combined led to a new negative form of willan, wynnot. The past forms of willan began with wold-, which is where we get would. Under the influence of these forms and the related verb wol, wynnot became wonnot by the late 1500s.

Finally, the modern form won’t emerged by the 1660s as a result of reducing the final vowel in wonnot. It appears to be the first word so contracted; most of the other -n’t contractions we use today (can’t, couldn’t, shouldn’t, &c.) arose in the 1700s, modelled after won’t. In modern English, cannot is the only uncontracted -not compound that survives.

As for the other contractions such as -’ll and -’ve, their history is just as long, though perhaps slightly less convoluted. But that’s a story for a different question. ;)

Also, remember that spelling in Old English was less standardised than in modern English. There were often several equally valid ways to spell the same word, especially when you took different accents and dialects into account. So sometimes it’s difficult to get a good historical account of pronunciation and usage changes. Still, as far as I can tell, this is basically how it went down.

Source: The Online Etymology Dictionary.

It's a contraction of woll not.

A friend found this explanation of the etymology of "will" via Google Books. I found it interesting.

Will is from the Gothic wilgan, Saxon willan, Dutch willen, German wollen. Among the various writers cited, I find it written wil, wyl, wyll, wrill, wol, woll. In Scotland it is often, to this day, pronounced wull, and in some parts of England as if it were written wool! The past tense was variously written when used as an auxiliary, wold, wolde, wouldin, wulde, would. The German has present will, imperfect wolte. Now, as wol and will seem originally to have been the same in meaning, if not in every thing but spelling, and that difference perhaps merely arising from ignorance or conforming to a vicious pronunciation, it does not seem amiss to consider would as derived from will; it certainly is derived from wol, and is as certainly the past tense of will, the auxiliary, to which wol is equivalent, and with which it was interchangeably used.