1. Export your database as a text file using Refer format.
2. Read the exported database into Word (or other word-processor).
3. Spell-check.
4. Import back into EndNote as a new database. Warning! You will lose your existing record numbers. Workaround: keep a copy of the existing database. If necessary, you'll be able to use it to reformat bibliographies in papers that reference the original record numbers. If this seems too formidable, just write yourself a new EndNote style that includes whatever field(s) you would like to spellcheck. Do a formatted copy (Ctrl-k) on the whole database, and paste the result into Word. Perform the spellcheck. Then in EndNote manually correct any spelling errors you find. Tip: EndNote's built-in "terms lists" are very useful for identifying multiple spellings of publishers, journal names, author names and suchlike. For all except the title field, I prefer scanning terms lists to spell-checking Word. It's not too difficult to write a macro that "hides" all but the title field prior to performing the spellcheck. Request: The more people who pester Niles about the need for a built-in spell checker, the sooner they will include one!

To create the subset library, the natural instinct would be to select the desired references and cut and paste them to a new library. The problem with this method is that all the references will be assigned new record numbers and will no longer work with the manuscript. To get around this problem, make a copy of the full library file, then select and delete from the copy file the records that you DO NOT want in the subset library. (On the PC, for instance, select the menu item Edit / Select All [or press Control-A]; then "unselect" the proper references from your library by holding down the control key while selecting with the mouse.) This will leave only the desired records with their reference numbers unchanged.
--Contributed by Drew Tagliabue.