Hey, wait a minute.  None of the people providing (or trying to provide)
help in these forums work for Microsoft.  So don't tell me what to do.

If I had a printer capable of handling continuous feed labels I would try
and replicate the problem.  However I do not have such a printer so I cannot
investigate your problem.

I am in the same situation as Doug, i.e. no suitable continuous-feed printer
(the ancient Canon BJ300 I just dragged from my cellar just fizzled out, but
I had
no suitable consumables either). So it's impossible to experiment with the
real thing. However, it is possible to examine what Word is outputting,
so...

There are certainly no (obvious) settings to set up the /printer/ a certian
way, but you can select the "print tray" as tractor and the paper size as
8.5in by 12 in. on an FX850 as far as I can see.

I had a look at the ESC/P sequences Word generates when you set up Word 2003
to produce Avery 4013 labels. First of all, the page setup Word creates
seems OK - it uses a 12in form length and generates the cells necessary for
labels 15/16in high each.

I set up the driver to use tracter feed 8.5 x 12in continuous feed paper and
sent Word's output to a file.

Word generates the following ESP/P commands:
Header:
- printer reset
- carriage return (notice that that should not normally affect vertical
position
- set PC437 character set
- set characters 128-159 to be printable
- set USA character set
- set NLQ mode
- set 10cpi font
- set line spacing to 1/6in
- set page length to 84 lines
- carriage return

It then essentially repeats the following for each page:
- carriage return
- move down a specified amount
- move right a specified amount
- send a line of 9-pin graphics
- form feed

At the very end, it outputs
- carriage return

Now the really weird thing about that is that 84 lines at 1/6in spacing
isn't 12in but 14in, so if I were printing that stuff on an LX850, I would
expect to see a 2in gap at the bottom of each page. That's not what you're
reporting, but it is almost precisely what Devon Stout reported in the
original post in this thread. However, at this point
a. there are too many variables to explore withut a real printer with which
to experiment
b. I've begun to doubt my own findings
so I wonder if anyone else is in a position to look at this stuff?

If you, "CMLCBigGuy", would like to do a similar experiment, all you really
need do is set up the FX850 driver to output to the FILE: device instead of
(say) LPT1: and generate 2 to 3 pages of labels. If you know how to examine
the ESC/P codes yourself, fine, but if you don't, by all means despam my
e-mail address and send me the output for 2-3 pages of labels and I'll
compare it with the above findings.

FWIW the circumstances in which I can imagine Word behaving the way /you/
(CMLCBigGuy) report are:
a. Word never outputs FF, and miscalculates the page length by one line per
page
b. Word sets page length to be one more line than it is
c. Word does FF, then outputs a line, then sets page length, (which, for
Epson printers, resets the top of page).

However, the ESC/P sequences I see do not seem to be doing that...

Peter Jamieson