I agree with Jane here - although legacy data can of course be an issue.
In an ISAD(G) context there can be a tension in applying the multi level
description rules. That for 'non-repetition of information' (2.4) is
often cited as meaning that, in our example, the word 'Correspondence'
at the series level should not be repeated at the file level. This is
too take the rule too literally and see it has somehow overriding the
other rules. In order to properly describe the file then I would repeat
the word 'Correspondence' as the title (with more detail specific to it
if appropriate). The rules for multi-level description should then be
viewed as a whole in order to produce clear and accurate descriptions at
all levels. As Jane suggests, I think such complete description is even
more important as we move from paper to digital: what was implicit on
the page of the paper finding aid, will not be useful in contexts where
descriptions at all levels are initially found and viewed online on
their own, even if they are (as they should be) linked to their
hierarchical parents, children and siblings.
S&C Cataloguing Systems and Processing Manager
Tel: (0)20 7412 7188
Please think before you print!
From: Encoded Archival Description List [mailto:[log in to unmask]] On
Behalf Of Jane Stevenson
Sent: 28 March 2012 08:26
To: [log in to unmask]
Subject: Re: dates wrapped in title field in box list
I think a title should always be supplied. There are rules...and then
there is what is sensible on the Web. If there is no title, then
searches outside of the context of the collection might just throw up
dates. We shouldn't only think about our context, but other possible
contexts for the data.
I've found that when we've tried to really think about what can be
inherited at lower levels, its less than we thought. I just think that
rules still seem to be based on the old paper-based perspective, where
the description of components is always within the context of the
On 28 Mar 2012, at 01:44, Mark Carlson wrote:
> I don't think anyone would argue that encoding dates in a separate
<unitdate> element is not a good idea. My point was that in this one
instance, within the description of subordinate components, when you
have a redundant title that you don't wish to repeat, how should one
handle or account for the lack of a <unitittle> at the component level?
If the <unitittle> is lacking, does that automatically mean that we are
to assume that it is to be inferred from its parent? (Nothing/lack of
something means something). Or did someone just make a mistake? Can a
sole <unitdate> within a component lend enough context? What if we
choose to digitize a portion of this component? The only contextual
data we have is a date. What if we want to re-purpose the data? How do
we map the title? (Of course, we would have to grab the parent title by
assuming the lack of a <unittitle> means to do so). And, as Michael
pointed out in his example, we are to assume the title "Correspondence"
is there but just not repeated. The only reason I bring this up is
because we made the mistake, early on, of assuming a lack of something
had some sort of "default" meaning. That caused problems down the
road. This is why we always require <unittitle> at the component level
and if one does not wish to repeat it, we place <unitdate> inside an
empty <unittitle>, NOT to mean that the dates are to be treated as
titles, but that the title is to be inferred from its parent (i.e. An
empty <unittitle> with <unitdate> inside has a meaning because of what
IS encoded not what is left out) and to avoid an empty <untititle> on
its own (which my scripts would strip out).
> In other words:
> can also be encoded as:
> <unittitle>[Correspondence inferred]</unittitle> <-- Empty element
> I guess my ultimate point is that nothing should be inferred from the
lack of something which could just be the result of poor encoding or a
mistake. It's best to state (or encode) exactly what something means.
But, I'm in the minority, I know.
> On 3/27/2012 12:17 AM, Jane Stevenson wrote:
>> Hi there,
>>> 4. Encoding dates as parts of titles in EAD is not desirable.
>> I agree with Michael on this. At the Archives Hub we don't do the
cataloguing ourselves, but we do provide tools for cataloguing, and
training in how to catalogue. As far as I'm concerned the data is a
separate entity from the title. This is not about display, as it is
possible to display the title and date how you wish, but it is about
what is semantically correct.
>> If we did think of the date as part of the title, we would have some
confusing titles, as some contributors just use a creator name as the
>> - the title given is just 'Rastrick, Henry', which isn't ideal, but
we can't control the way people create titles. Adding the dates as an
integral part of this would be confusing because it would not be clear
what the dates refer to.
>> Unfortunately, we do have some descriptions where the life dates have
been given in the title, which is not ideal, eg.
>> is 'PAPERS OF J HENRY LLOYD (1883 - ?1966)' This points to one good
reason to keep all dates separate and then apply styling to display them
>> It seems to me that keeping parts of the finding aid as discreet
entities is more likely to be appropriate for potential future
developments, and following consistent practice will help us with data
>> Jane Stevenson
>> The Archives Hub
>> Mimas, The University of Manchester
>> Devonshire House, Oxford Road
>> Manchester M13 9QH
>> email:[log in to unmask]
>> tel: 0161 275 6055
>> website: archiveshub.ac.uk
>> blog: archiveshub.ac.uk/blog
>> twitter: twitter.com/archiveshub
>> On 26 Mar 2012, at 22:11, Michael Fox wrote:
>>> Several interesting points have been raised in this discussion that
bear a more extended consideration.
>>> 1. In descriptive practice, we are guided by two types of
>>> * Descriptive rules that define the elements to be recorded and
the accompanying thesauri and authority files that specify how we
populate certain elements. In a US context these are ISAD(G) and DACS.
>>> * Electronic communication standards for data interchange.
Again, in a US context these are MARC and EAD.
>>> The answer to the question as to whether the dates of archival
materials are part of the title of those records or completely separate
is the business of the descriptive rules, not our communication
standards. MARC and EAD are not normative in this case. EAD was
written to enable a wide range of international practice, created
according to different conventions including national standards and
local ad hoc practices that exist either because of institutional
preference or a lack of national standards. It is not the place one
ought to go for direction on this question.
>>> 2. Our practices confuse the issue at stake.
>>> One encounters many finding aids that look like this.
>>> Example A:
>>> What is really meant here is
>>> Example B:
>>> Correspondence, 1919-1925
>>> Correspondence, 1926-1950
>>> Correspondence, 1951-1965
>>> In example A, the dates given for the contents of these three files
are the dates and not the title proper of the material being described,
even if one considers the date to be a segment of the title.
>>> DACS countenances this through the long-standing principle of
non-repetition in multi-level description that is expressed in Principle
>>> Non-repetition can be a useful concept. After all, who would wish
to repeat the name of the repository at every level of description?
However, it can also be confusing and is troubling for the future.
>>> The non-repetition of the title proper in Example A is really a
holdover from the days when we considered a finding aid to be a
narrative description of the collection to be read serially on typed
pages rather than as data the describes units of archival materials.
>>> An early question in this thread was whether or nor the nesting of
unitdate within unittitle would be confusing to databases. I don't
think that relational databases are the cause of the problem but rather
that they expose the priority that we have given to presentation on the
printed page, rather than taking a more data-centric view of the
information we record. Is this where we want to be?
>>> 3. Is the date then, looking at descriptive rules, a segment of
the title element or something completely separate and independent?
>>> a. As has been pointed out, this appears to be strictly a US
problem. Or more properly, a problem for some US institutions. No
other cataloging practice with which I am familiar treats unit dates as
anything but a separate element.
>>> b. ISAD(G) treats it as separate.
>>> c. DACS treats it as separate element. It identifies three
possible segments that might be included in a title. Date is not one
of them. In discussions at the time DACS was being drafted, the authors
were very clear that dates were not a part of the title element.
>>> This is reenforced in the Introduction to Describing Archival
Materials (p. 3) by the statement that data elements are mutually
exclusive. They can go in one place or the other.
>>> d. The practice of including dates as a segment of the title in
MARC comes from APPM which in turn reflected in some examples an older
manuscript cataloging practice for constructing title. Consider this
example of a title from the first edition.
>>> ALS, [ca. 1898 Jan. 1], Worcester Park Surrey to George Gissing,
>>> But even then in 1983, this practice was clearly antique and
inconsistent with newer formats for constructing titles
>>> 4. Encoding dates as parts of titles in EAD is not desirable. For
a standard created to facilitate the international exchange of data,
having most archivists worldwide following one practice and some in the
US another only confounds interchange and complicates training and
>>> 5. An archival description that treats unitdates as part of the
unittitle is simply not DACS compliant.
>>> Michael Fox
>>> On Mon, Mar 26, 2012 at 10:16 AM, Prom, Christopher John
>>> <[log in to unmask]>
>>> Thanks for laying this out in such detail. I just wanted to point
out that although Archon currently wraps unitdate w/in unittitle, I
actually think it best if in were NOT wrapped, in compliance w/ ISAD(G)
and DACS. I didn't give much thought to it at the time we programmed
the first EAD export in Archon, and we haven't reviewed this coding
since the original development of the application. Moving it outside
would be a trivial matter, and I'll take care of that in one of our next
>>> Christopher J. Prom, PhD
>>> Assistant University Archivist and Associate Professor
>>> University of Illinois Archives
>>> 19 Library
>>> 1408 W. Gregory Dr.
>>> Urbana, IL 61801
>>> [log in to unmask]
>>> +1 217 333 0798
>>> On Mar 24, 2012, at 10:12 AM, Cory Nimer wrote:
>>> While there is what you might call a growing consensus that
<unitdate> should not be encoded within the <unittitle>, there is still
disagreement within the community. This can be seen in comparing some of
the different best practices documents, management tools, and content
standards available. In the best practice guidelines listed on the EAD
>>> Northwest Digital Archives -- Outside of <unittitle>
>>> "To insure compliance with ISAD(G), do not nest <unitdate> inside
<unittitle>." -- Online table
>>> Online Archive of California -- Outside of <unittitle>
>>> "The <unitdate> should be encoded outside of <unittitle>." -- p. 12
>>> Library of Congress -- Either inside or outside of <unittitle>
>>> "Following/within <unittitle> and before <unitid> in Collection
Summary." -- Online table
>>> RLG Best Practice Guidelines -- Either inside or outside of
>>> "US repositories following APPM practice normally include <unitdate>
as part of <unittitle>, whereas British and Canadian practice, following
ISAD(G)v2 use <unitdate> at the same level as <unittitle>. Given the
likelihood of further international standardization, separate title and
date is preferred but both practices are permitted. Repeat <unitdate> if
both inclusive and bulk dates are given. This element is considered an
essential element for data exchange by ISAD(G)v2." -- p. 12
>>> NCEAD Best Practice Guidelines -- Inside of <unittitle>
>>> "Always place the date within a <unittitle>, as the date is part of
the title of the unit as well as a date." -- p. 40
>>> There is also some variety in date handling between the different
community-developed archival management systems:
>>> Archivists' Toolkit -- Outside of <unittitle>
>>> ICA-AtoM -- Outside of <unittitle>
>>> Archon -- Inside <unittitle>
>>> In terms of content standards, because of DACS's foundations in
ISAD(G) date is clearly a separate element from the title. While this
might be an argument for not wrapping the <unitdate> in <unittitle>, the
examples provided in the text include instances of both including
<unitdate> in <unittitle> and keeping it separate. The understanding
that <unitdate> should be wrapped in <unittitle> is reinforced, I think,
by the long-term practice of including dates as part of the title in
MARC, as established in APPM. All but one of the MARC examples in DACS
are marked up in this fashion. However, it is worth noting that the MARC
mapping provided in RDA no longer places dates of production of archival
materials in the 245$f, but in the 260$c. In order to make this clearer
in MARC, the 264 field was also recently approved to make the role of
the date entry clearer.
>>> Ultimately, this might be considered to be a legacy practice, but
one that is still widely implemented. Perhaps this is something that the
current EAD and DACS revisions will clarify for the community. In the
meantime, I personally believe that the RLG Guidelines are correct that
it is preferable to keep <unitdate> outside of <unittitle>.
>>> Cory Nimer
>>> Manuscripts Cataloger/Metadata Specialist
>>> Brigham Young University
>>> From: Encoded Archival Description List [
>>> [log in to unmask]] on behalf of Elizabeth H Dow [[log in to unmask]
>>> Sent: Saturday, March 24, 2012 6:07 AM
>>> [log in to unmask]
>>> Subject: Re: dates wrapped in title field in box list
>>> Where to place <unitdate> raises the question of whether a date a is
an integral part of a name/title, as it is in MARC, or if it's a
separate data element on its own. About 12-15 years ago, the question
was a point of disagreement between the US and UK EAD communities. Both
are legal in EAD; has a general consensus appeared about which is the
better way to look at dates?
>>> Elizabeth H. Dow, PhD.
>>> School of Library and Information Science
>>> Louisiana State University
>>> From: Encoded Archival Description List [
>>> [log in to unmask]] on behalf of MicheleR [[log in to unmask]
>>> Sent: Friday, March 23, 2012 7:29 PM
>>> [log in to unmask]
>>> Subject: Re: dates wrapped in title field in box list
>>> We do this all the time. Instead of <date> though you'd want to use
>>> The <date> element is just for any old date info, including
>>> narrative text, whereas <unitdate> is specifically for the date(s)
>>> archival unit being described -- see
>>> So like this:
>>> I have no idea whether it would cause problems, or what sort, in a
>>> (depends on how the database is set up, I guess?) but in terms of
>>> perfectly legal and logical. It doesn't affect searching for us
>>> index and search the EAD directly.
>>> On 3/23/2012 5:42 PM, Ashley Knox wrote:
>>>> What is the consensus on date fields being wrapped inside of title
>>>> especially when the title of a folder IS a date? (specifically in a
>>>> <unittitle> Bills and Receipts,
>>>> <date normal="1850/1859" type="inclusive">1850-1859</date>
>>>> <date calendar="gregorian" normal="1980/1985"
>>>> All of this being related to the effect on searching in a database
(built in a
>>>> mysql database). I've been told that this will hinder searching and
>>>> needs to change, that title and date need to separate into their
>>>> fields. I'm working in AT. Would I just not have folder titles, but
>>>> I would love to hear opinions on the pros and cons of mysql
databases vs. xslt
>>>> styling as well, if anyone would like to comment.
>>>> Thanks for any and all comments!
>>>> Ashley Knox
>>>> Project Manager
>>>> USC Digital Collections
>>> America may be unique in being a country
>>> which has leapt from barbarism to decadence
>>> without touching civilization.
>>> -- John O'Hara
>>> Michael Fox