Here's a neat feature of the online OED. (Alas! You can only connect if you or your institution has a subscription. So if you can't get there, and if you don't have the paper copy, you'll just have to take my word for it. Well, for the online interface stuff I'm talking about, even the paper copy doesn't work....)
As you look at the definition of a word, the word also appears in an alphabetical list in a frame in the left margin. For instance, for "cert (n.)," you can see "cert" in the middle of a list stretching from "cerosin," "ceroso-," "cerote," etc., down to "certes," "certie," "certifiable." This is good for checking similar words or correcting your own spelling, and also simulates to a small extent the experience of being seduced by random words you might find while flipping through a dictionary.
But you can switch from "List by entry" to "List by date"; then, instead of seeing your word in alphabetical context, you can see it in chronological context, where each word is indexed by the earliest listed occurrence (over the entire entry). "Cert" (in the "dead cert" meaning) is listed as appearing in 1889, so the words in the left margin run from "1889 catalytical," "1889 cataphoresis," "1889 catatonia," etc., down to "1889 chemiluminescence," "1889 chicle," "1889 chit-chatty."
There are a lot of 1889 words, but you can also go up or down a screen to see alphabetically prior or subsequent words in the same year, and further to see words from different years. This is fun because, if you're interested in a period, seeing what words first showed up in the language in that period gives you a nice sense of what was going on at the time in society and culture.
For example, "lonely" was invented by Shakespeare, who used it in Coriolanus ("I go alone Like to a lonely Dragon, that his Fenne Makes fear'd, and talk'd of more then seene.") in 1607 — did you know that "Alpine," "archaeology," "birthplace," "bloodshot," "exasperated," "Machiavellianism," "maestro," and "procrastinator" also appeared (as far as the OED knows) in that year?
Now I know what you're thinking: "O.K., Volokh, enough with the 1889 and 1607, what are the earliest known English words?"
We can ignore the first two words on the list. The first is "chiule" (c. 550), which is just a technical term for an old English or Norse ship of war. We can ignore the second, "alder" (a600) [I take it this means "before 600"], for two reasons.
First, in this meaning (not the tree), it's marked as obsolete. In its meaning of "parent, ancestor," it's been superseded by "elder," and in its meaning of "the head of a family or clan; a patriarch, chief, prince, or ruler," it's not used today either.
Second, this occurrence is from Beowulf. Beowulf words are inconsistently dated in the OED — you can find "c1000," "a1000," "a800," "a600," and without a date at all. This is probably an oversight — not so important in the paper edition, but more important if you want to see the chronological list of words like I described above. What's the "correct date" for Beowulf? This is complicated — see, for instance, the Wikipedia entry, which places the manuscript date at roughly 1000 and the "date of composition" at 650-800. But for OED purposes, it's arguably the manuscript date that should count for oral traditional stuff, since some words may have been introduced at the time, or shortly before, the manuscript was created. So calling Beowulf "c. 1000" throughout seems defensible.
O.K., so we ignore "chiule" and "alder," and what do we have left? Finally, we have three words marked as "601." Well, one is marked "?601," but all three refer to the Laws of Ethelbert of Kent, which they estimate as dating from 601-04.
And what are those three words? The three nouns town, yield, and priest.
Come again? The Laws of Ethelbert (Æthelberht for the sticklers — we have a cat named Æthelwulf, which we spell Ethelwolf) only have three words??? Let's just consider the evidence quoted for those three words (the modern text in brackets has the translation from a version I have, with that version's section numbers, which may differ from what the OED has):
601-4 Laws Ethelbert c. 17 Ȝif man in mannes tun ærest ȝeirneþ, vi scillingum ȝebete; se þe æfter irneþ, iii scillingas. [22. If a person breaks [as the] first into someone's dwelling, let him pay with 6 shillings. He who breaks in next, 3 shillings.]
601-4 Laws of Ethelbert xxviii, Ȝif man inne feoh ȝenimeþ, se man III ȝelde ȝebete. [28.1. If a person takes property therein, let than man pay 3[-fold] as compensation.]
?601-4 (MS. c 1120) Laws of Æthelberht c. 1 Biscopes feoh xi ȝylde. Preostes feoh ix ȝylde. Diacones feoh vi ȝylde. Cleroces feoh iii ȝylde. [2. A bishop's property [is to be compensated] with 11[-fold] compensation. 3. A priest's property [is to be compensated] with 9[-fold] compensation. 4. A deacon's property [is to be compensated] with 6[-fold] compensation. 5. A cleric's property [is to be compensated] with 3[-fold] compensation.]
Let's forgive the inconsistent citation style. Now note this "MS. c 1120" business. For the same
Beowulf-related reasons I listed above, perhaps we shouldn't even count Ethelbert's code? That would be a shame, since I think it's the first known work of written English prose or something like that. But on the other hand, perhaps a legal code is less problematic than an epic, since the original manuscript of the legal code was probably written down at the time, and the 1120 manuscript of the legal code was probably created by successive recopyings, while
Beowulf was constantly changing as time went by and may not have been written down until fairly late. Henceforth, I ignore this issue.
Consider "town" ("tun") — this is the specialized, obsolete sense of "the enclosed land surrounding or belonging to a single dwelling." The first word of that sentence is "Ȝif," meaning "if." (The first letter in that word is a yogh.) In case you want to complain about the spelling, the OED doesn't mind, and indeed, it currently lists the word "if" with the year 805, with the same spelling. Next, note the words "scillingum" and "scillingas"; these are just forms of the word "shilling," which the OED lists as "c900" with spellings like "scylling," "ssillinges," etc.
Plus, the "town" sentence also includes "after."
In the "yield" sentence, we have "fee" (in its obsolete meaning of livestock, movable property, money, etc., currently listed as "c870" with spellings like "fioh," "fe," and this spelling "feoh") and "inne" (listed as obsolete and meaning "therein"). And the "priest" sentence also contains "bishop," "deacon," and "cleric." Consider also the very first clause of the Laws of Ethelbert — "Godes feoh & ciricean XII gylde" ("God's property and the church's with 12 compensation") — and we get the words "God" and "church."
What's the moral of all this? The OED's chronological listing feature, in principle, rocks. But, alas, you can't take it too seriously. Those seventh-century guys weren't just going around saying "town yield church, yield yield church town!" with others answering, "Alder chiule." Sometimes they were also saying, "God bishop deacon cleric if after fee. Inne." To say nothing of "freeman king slay maiden third churl wife other ear mouth eye nose chinbone."
UPDATE: Pat at Stubborn Facts responds. Also, note a useful comment by David Leon Gil in the comments.