>I found several sites that gave... as the definition.

this is potentially too germane a topic to leave lying doggo in an obscure word thread, in that it speaks to the whole issue of what makes it into the English lexicon as delimited by OED, W3, et al.

OED2 is loaded with obsolete words, nonce-words, and even words for which there is no other extant recorded evidence other than inclusion in some other old, obscure dictionary. In light of this historical (in the extreme) nature of OED, I asked the current U.S. editor if we should expect to see (for instance) Menckenisms such as 'bootician' and 'ombibulous' in the forthcoming OED3.

Jesse Sheidlower's reply:

In general, for a word to be included, we'd need some indication of wider use. For Menckenisms, _ecdysiast_ and _booboisie_ are both coinages that successfully made it into broader currency and will be included. The two you mention seem not to have done so, though we do have some examples from Mencken.

not content to let the matter drop, I pushed on:

Okay, this puzzles me a bit. I have had it pounded into my head that the OED is (or was) descriptivist and historical in nature. OED2 is loaded with nonce words from all sorts of obscure writers (Lytton leaps to mind)--so either things have changed, or Mencken is chopped liver, which?

Jesse's reply:

Well, a few things, including that Lytton was not an obscure writer when the OED was first being edited, and that our standards have changed so that we'd be much less likely to include a nonce-word now. But there can be a lot of reasons why one would include one--for example, the word is the sole example of a particular part of speech, but other parts of speech exist and the nonce example supports or sheds light on it in an interesting way.
 [E.A.]