Introduction
Between 2008 and 2012 the Bentley Historical Library at the University of Michigan
undertook a major project geared toward developing a reliable mechanism for identifying
and preserving e-mail correspondence of long-term value created by university leaders.1
The first phase of this effort involved attempting to identify an off-the-shelf software tool
that would work well with a range of e-mail servers and services which was also
affordable and easy to implement. Several vendors considered to be major industry
leaders were worthy of consideration but in the end, the University Archives retreated
from this option. The second phase focused on a low-tech and completely voluntary
approach of capturing e-mail correspondence we call the “Copy Method.” This paper
briefly describes the project background, what the “Copy Method” is, and lessonslearned. For more details on our work in the area of e-mail correspondence visit the
Society of American Archivists Case Studies website.2
Background
Archivists at the Bentley Historical Library who contributed to this project
included Nancy Bartlett, Francis X. Blouin, Nancy Deromedi, Gregory Kinney,
Michael Shallcross, William Wallach, and Brian Williams.
2 McKay, A. “Partnering with IT to Identify a Commercial Tool for Capturing Archival
E-mail of University Executives at the University of Michigan,” and “Will They
Populate the Boxes? Piloting a Low-Tech Method for Capturing Executive E-Mail
and a Workflow for Preserving It at the University of Michigan,” Society of American
Archivists Campus Case Studies (2013).
http://www2.archivists.org/publications/epubs/Campus-Case-Studies
1
1
The decentralized culture of the University of Michigan presented several challenges for
us, one of which was the lack of a single, centralized e-mail service at the University.
When our project began, some schools and colleges on campus ran their own Exchange
servers. These servers were traditionally independent of each other and on different
upgrade schedules, and therefore ran different versions of Exchange. During 2009 and
2010 (as we were conducting our work) several of the Exchange “forests” were combined
and upgraded. In addition to Outlook/Exchange, the university also offered a generic
IMAP service, and supported a variety of desktop e-mail applications to access it,
including Mulberry, Apple Mail, Pine, Thunderbird, Entourage, and two home-grown
webmail applications. Users could choose one desktop or web-based e-mail application
for an office computer, and a different one for a home computer or PDA. In addition,
there was no policy to prevent employees from forwarding their e-mail to a third-party
cloud service such as Gmail, and many employees did choose to do so.
In early 2010, as the second phase of this project began, there was a general
acknowledgment that the multiplicity of e-mail tools was inefficient and inhibited
interdisciplinary collaboration. While the University leadership strategized about
whether to move e-mail services to a cloud vendor, the project team moved forward with
a plan to test whether or not well-motivated executives and faculty members would
nominate archival-quality messages on-the-fly and to create policies, workflow and tools
to store and provide access for the captured e-mail messages.
2
The project team identified 50 executives, primarily deans and executive officers who we
believed could fill important gaps in the correspondence series coming to the archives if
we could capture the archival-quality e-mail. We understood that our approach would
need to be highly customized for each executive and the process needed to be simple. We
would need the conscious cooperation of the executives to declare archival records from
their e-mail correspondence as they actively worked with it. Experience had shown us
that at the end of their tenures, most officials do not have the interest, time or energy to
search through accumulated e-mail to decide what should be turned over to the archives
or to a successor.
We resolved to conduct a series of pilots with several officers and faculty members that
had deep understanding of the value of archives and the historical record to determine
whether they would participate in records declaration. After we had successfully
launched the pilots, we needed to turn to the task of figuring out how we would transfer,
process, describe, store and provide access to the captured messages. The third phase of
the project overlapped with the second phase and involved making decisions on how to
manage the acquired correspondence. This work is primarily policy and procedure
development work, is ongoing and is not discussed in this paper.
Introducing the “Copy Method”
During preliminary user interviews we asked whether participants would be willing to
forward particular e-mail messages to the archives for management. A few participants
3
thought that this might work. We called this method of transfer the “Copy Method”
because all administrators had to do was Cc or forward pertinent messages to this other
account. To facilitate the process, a personal e-mail account was set up for each
individual whose e-mail was collected. This account we named “bhl-“ (for Bentley
Historical Library) prefixed to the person’s existing e-mail address. Thus, Francis
Blouin’s e-mail address, fblouin@umich.edu produced an archival account that was
called “bhl-fblouin@umich.edu.” Both the executive and an archivist had read and delete
access to the messages deposited in the archival account, though the archivist promised
not to read or download the messages until given explicit permission from the executive.
It was not possible to send mail from the archival accounts; they could only receive
inbound mail.
Most e-mail messaging programs can display a second e-mail account as if it were a
folder in the primary account. This allowed the administrators to drag and drop messages
into the archival account directly from their regular e-mail reader, without needing to
forward messages. The administrator also had the ability to set up folders within the
archival account, if so desired, so as to further differentiate the mail intended for the
archives. Unfortunately, the setup for this functionality differed among the various email applications and was not straightforward. Therefore we went to the executives
offices and manipulated the settings to create the appropriate display.
We planned to allow mail to accumulate in the accounts for a time, after which the
archivist would secure permission to download the messages from the administrator and
4
preserve them permanently. Usually, mail accounts at the university must be affiliated
with real people as a matter of policy. Since the archival accounts were primarily
affiliated with a particular person, and since they were not allowed to send mail, but only
to receive it, we relied on the support of our e-mail administrators to execute our pilot
project.
Since this was a pilot, and we were unable to assure the preservation of e-mail transferred
to the account (since the mechanisms were being simultaneously developed), we decided
to ask participants to transfer copies of messages, instead of “original” messages to the
archival account.
Testing the “Copy Method”
Since employees of the University of Michigan could use a variety of e-mail applications,
before we presented this method to participants, we needed to test it in all applications.
Testing fell into two major groups: IMAP and Exchange.
We developed a protocol that tested setup, cc’ing, forwarding, search, deletion, draggingand-dropping originals and copies, adding assistants, setting up mailbox rules so mail
would be sorted into the account automatically, taking screenshots of the appearance and
functionality, and reflecting on usability. (See, for example the completed test protocols
for Apple Mail and for Outlook.)
5
To test the Copy Method we set up test accounts for Bentley archivists using Pine, Apple
Mail, Mulberry, UM Webmail (Blue version), UM Webmail (Maize version) and
Thunderbird. The Copy Method worked on a fundamental level with all of the
applications although it was not easy to use in Pine and Webmail Maize. There were
important differences in how the different applications worked. For example, to move
copies of messages rather than originals, different key combinations were necessary and
the setup procedure varied with each application meaning it would be necessary to
conduct setup in person on each computer that the mailbox owner used to read his or her
mail.
Results
The pilot did not produce as many archival e-mail messages as we anticipated. In total,
from 11 positions over a period of nearly 18 months, we collected 3,171 messages. Four
of the participating positions contributed less than 25 messages and one position
contributed 2,475 messages. Despite these low numbers we gained a great deal of insight
for our future efforts. Participants had several general reactions to their experience:

Difficult to predict the significance of a message at the time of creation;

Third-party privacy and confidentiality;

Plans to thoroughly review legacy e-mail for transfer but never found time;

Much less business of significance transacted over e-mail than they thought;

System worked exactly as needed; and

System itself was not too difficult to manage for anyone.
6
One executive did not use the “Copy Method” to transfer messages of significance, but
did use it for a delivery method for other university records. In this way, the archival email box acted as a digital drop box.
Lessons Learned
This project taught several useful lessons. First, we learned that the changing
technological landscapes required constant vigilance. Despite its simplicity, our Copy
Method required a lot of desktop support and was not very scalable.
Second, we learned that populating the boxes with e-mail was difficult to do. Despite the
best of intentions, participants would not reliably target messages for the archives on-thefly. We agreed that any future system would need to capture larger chunks of data
(defined by folders or time periods), and require less effort for the executives. This result
emphasized what we already knew: we needed better incentives, most significantly,
university policy that would be clear about mail that should be retained long-term.
Finally, we learned that administrators were concerned about the total information flow
within their offices and therefore they observed that important e-mail derived from a
number of contexts and not simply from their own desktops. The project team concluded
that we would have to move beyond our specific strategy to recover e-mail toward a more
comprehensive approach to the capture all types of born digital records.
7
8