When the Oxford English Dictionary began to appear, Thomas Hardy—the English poet and novelist—found words said to be archaic or obsolete and stuck them into his literary works. Finally, James Murray, the editor of the OED, wrote him a letter and asked him not to do that.

We tend to be egocentric about words: if we don't know a word we encounter, we suspect it isn't really a word or maybe it's a dead word. But old words never die; they just end up in the dictionary.

Writers have often given mouth-to-mouth resuscitation to breathless words. When the Oxford English Dictionary began to appear, Thomas Hardy—the English poet and novelist—found words said to be archaic or obsolete and stuck them into his literary works. Finally, James Murray, the editor of the OED, wrote him a letter and asked him not to do that. The one example of acoast ("ashore") that Murray had found was from 1579; Hardy used it in a poem in 1898, so it wasn't obsolete any more. This seemed to Murray not to be playing by the rules. But who made the rules, and who enforced them? Nobody.

It's easier than ever to locate the first uses of words. So far, the earliest example of OK in the record is from March 1839. Lots of people have looked for earlier ones, but so far without success.

But how do we discover when words stop being used?

At last we have a way of getting evidence. At Brigham Young University, Mark Davies has created an on-line file of 450 million running words of American English from the 1810s to the 2000s. You can sign on to it, though after ten searches he will ask you for information about yourself. After that, you can use it to your heart's content.

Take the word galluses ("suspenders"). It first turns up in the 1880s and is in limited use right down to the present day. The high-water mark for galluses was in the 1920s, but there are only fifty examples in all that great body of American English.

How about serensified ("satisfied"): "I have had a sufficiency and am fully serensified," says the diner pushing back from the table. Sounds like a blend of serene and satisfied, doesn't it? But it doesn't turn up in Davies' big collection—though Google finds it in song lyrics from 2007. (Thirty years ago, Frederic Cassidy, editor of the Dictionary of American Regional English, used that very sentence at my dinner table. He related it to the fondness of Americans for high-sounding words like sockdolager, bodacious, and rambunctious.)

Twenty-three skidoo ("scram") seems to be gasping for air. It hasn't been used lately, and I suspect anybody using it would suffer from a queer and puzzled look. "Let's blow this popsicle stand" means just about the same thing, and I know someone who says that fairly regularly.

There is a mathematical model for the phenomenon we're discussing: Zipf's law. That law states that the rank of a word in a large body of text is inversely proportional to its frequency. Thus, the most frequent word in a large collection of English is the; the second most frequent is of; and the third most frequent is and. The occurs twice as often as of and three times as often as and, and so on.

Natural languages, including English, have distributions with very long tails: by far the most words will appear just once, and, if we enlarge the size of the text collection, more words will appear twice but even more will appear once.

The size of the English vocabulary is thus infinite. If Professor Davies's collection doubled (or tripled) in size, the frequencies would follow Zipf's law, and he would capture even more outliers.

Frumicate ("to put on airs") probably came to New England with the Pilgrims, but we don't have any evidence for it. How many words would we have to search to find it? And even if it never turned up, we couldn't be sure that it never existed.

Contest: Can someone tell me the meaning of the rare word in this sentence?

I'm hoping to hire a frumberdling to rake my leaves.

Put your answer in the comments section below.