This is more or less exactly what we did for the Colonial and State Records of North Carolina (, with additional metadata records in MODS in the dmdSec.  It seemed to work well for representing all the information we had.


On Nov 10, 2009, at 12:27 PM, Rick Beaubien wrote:

I need to state right off that I am no expert in TEI.  However, if I understand correctly what you are trying to do, I would recommend using the mets/structmap/div/fptr/area element.  The area element is specifically set up to allow you to point to specific areas within an integral, typically external content file.  In a case where the external content file is a structured text file, such as an xml file, the relevant attributes on the area element would be:

FILEID -- reference the ID value of the <file> element within the METS document's fileSec that represents the integral TEI file.
BEGIN -- reference the ID value of the first relevant element in the external TEI content file
END -- reference the ID value of the last relevant element in the external TEI content file.  But note, that if all of the pertinent content is contained within a single element, you only need to include the BEGIN attribute.
BETYPE--set the value of this to "IDREF"

Hope this helps.

Rick Beaubien

Stefan Krause wrote:
[log in to unmask]" type="cite">
Hash: SHA1


we've made some attempts with a larger TEI file which contains the text
of some issues of an journal. Structural units like issues, articles,
and subdivisions are annotated with xml:ids. ), still I would think the way to go
for content of scale would be to keep TEI external and point to its
structures out from the METS structMap, as you would point to other

We included the TEI file in a mets:dmdSec. So we can link from the
logical structmap to the xml:ids via mets:[log in to unmask] This is pretty
easy, but I'm not shure if this is the METS way.

Stefan Krause
- ----------------------------------------------
Editura Gesellschaft für Verlagsdienstleistungen mbH
Tempelhofer Damm 2
12101 Berlin
Telefon +49 (30) 789 59 49-66
Telefax +49 (30) 789 59 49-90
E-Mail [log in to unmask]

Eingetragen im Handelsregister Berlin-Charlottenburg
Registernummer: HR B 81823

Ralf Szymanski

USt.Id. DE217180548

Stephen Simon schrieb:

I would tend to agree with the approach suggested by David Lowe here. I
would guess that you can make it work either way but the incorporate by
reference approach seems like it would be simpler and less likely to cause
headaches. Easier to troubleshoot, anyway.



On Mon, Nov 9, 2009 at 8:25 PM, David Lowe <[log in to unmask]> wrote:

I've thought about this, too, and although I have no experiential
background to share, and although the P5 version of TEI is supposed to
be designed to allow embedding within METS (see, for example ), still I would think the way to go
for content of scale would be to keep TEI external and point to its
structures out from the METS structMap, as you would point to other
representations of the information object's content, like page images or
plain OCR'd text.

Does this jive with anyone else's practice or at least thinking?  (I
think in terms of monographs, but there might be great reasons for doing
things another way.)

-----Original Message-----
From: Metadata Encoding and Transmission Standard [mailto:[log in to unmask]]
On Behalf Of Schwartz, Christine
Sent: Tuesday, November 03, 2009 2:10 PM
To: [log in to unmask]
Subject: [METS] METS and TEI?


I'm looking for advice on the integration of METS and TEI documents. Is
it best to integrate TEI documents within a METS document, or keep the
TEI external and just link out to it?

We are at the beginning stages of planning METS/TEI, so there are no
legacy practices that we have to consider. Right now we use METS, Simple
DC in the mdWrap, and MIX.



Christine Schwartz
Metadata Librarian
Princeton Theological Seminary Libraries

Version: GnuPG v2.0.12 (MingW32)
Comment: Using GnuPG with Mozilla -


Rick Beaubien
Pro bono consultant
U.C. Berkeley Library

Contact information:

88 Herrada Rd
Santa Fe, NM 87508