Harry Binswanger says: -> Someone here, I think Carl Distefano, mentioned that the -> best way to spellcheck is to use the command spell file1,file2 -> where "file1" is the file to be checked and "file2" -> is the name of the file Xy will create that contains -> just the questionable words. ... I don't say it's the "best" way; it happens to be my preference. I find it quicker to scan down a list of words and just fix what needs fixing; easier than answering all those prompts. (And it eliminates the annoying pause on double words.) But there are problems. There's a bug whereby SPELL file1 file2, whether executed directly from the command line or via an XPL routine, corrupts the contents of one or more "permanent" Save/Gets! In my case, I find that even a single use of SPELL file1 file2 almost invariably corrupts the contents of Save/Get 623, which holds the command history in Holmgren's STACK.PM. Maybe others notice it; I'd be curious to know. Still, the convenience of SPELLing to file outweighs the grief, especially in a multitasking environment, where you can perform the spell-check in a separate session. -> Then you can write a little XPL code to process the errors ... Actually you shouldn't have to. CORRECT file1 file2 is supposed to perform batch corrections (after you type the corrections alongside the appropriate "questionable" words). Unfortunately, it doesn't work. The corrections don't "take". So, as I mentioned a while back, I do have an XPL routine that automates SPELL to file and word-by-word corrections using either a CVA or FUNC SO procedure (user's choice), which addresses the labour/labor pitfall. I haven't gone so far as Nathan, however, and implemented a batch correction routine -- only because I haven't felt a pressing need (a word here and there is enough for me). TTG should revisit the SPELL/CORRECT duo and make 'em sing in key. -------------- Carl Distefano 70154.3452@xxxxxxxx