This is the fifth in a series of emails in which I will highlight upcoming changes to EAD.  As I previously wrote, I hope this stimulates discussion and comment, and serves as inspiration for the EAD community to engage with the beta release of EAD3.  More information on the revision and the beta comment form are available at  

As a reminder, the deadline for comments on the Beta version of EAD3 is October 1st.  So please submit your comments in the next two weeks!

Structured <did> Elements

We have updated a number of <did> elements to improve their data models, making them more structured and hopefully more consistent.  

<origination> and <repository>

In EAD 2002, both of these elements accept mixed content, both text and various elements.  In EAD3 they will not contain mixed content, rather a choice of elements.  They will be able to contain one or more of <corpname>, <famname>, <name>, or <persname>.  <repository> will have an optional (zero or one) <address> element. Both will have an optional <depscriptivenote> element.  Because <corpanme>, <famname>, <name>, and <persname> now all consist of one or more <part> lements, the <subarea> element in <repository> and <corpname> has been deprecated.

<unitdatestructured> and <unitdate>

In order to better model the complexity of dates, including single dates, date ranges, and combinations thereof, we have created a new element called <unitdatestructured>.  This elelement will allow one of the following: <datesingle> (for single dates), <daterange> (for date ranges), or <dateset> (for some combination of two or more <datesingle> and <daterange> elements.  This will allow for much more precise normalization and processing of dates.

However, in order to provide a migration path for existing dates which do not follow the model provided in <unitdatestructured>, we will be keeping the <unitdate> element as an unstructured alternative.


In EAD 2002, <langmaterial> is a mixed content element where one can include a <langugae> element, though none is required.  In EAD3, <langmaterial> will have a more tightly structured model.  It will contain one or more of the following: <language> (for recording the language used in the materials being described), or <languageset>, which wraps both <language> and <script>.  In EAD 2002, script is an attribute on <language>.  Moving <script> to an element allows us to capture instances where one language is written in mutlipe scripts and where multiple languages are written in the same script.  The <langmaterial> element will also have an optional <descriptivenote> element for textual notes regarding languages.

The final structured <did> element is <physdescstructured>, which I will describe in my next - and final! - EAD revision highlights email.

For now, remember to submit your comments!


Mike Rush
TS-EAD co-chair