The Marshall Papers Project

The Marshall Papers project (http://www.marshallfoundation.org/Database.htm) is a subcollection of five volumes drawn from the large manuscript collection of General George C. Marshall's papers at the Marshall Foundation's Archive in Lexington, Virginia. According to the website, "the published volumes span the years 1900 to 1946 and comprise some 3,000 pages" that include "letters, telegrams, memos, reports, and speeches."

Selection Decisions: The George C. Marshall Foundation's website has many interesting and attractive features. Overall, it appears to be a very well-designed website. Unfortunately, however, the digital library collection of the Marshall Papers appears to be an anomaly on this otherwise current website. The selection of the five volumes covers a chronological period, but the website acknowledges that it is incomplete, as it is missing the important period from 1947 to 1959 when General Marshall served as the Secretary of State and then later as the Secretary of Defense during the Korean War. Other than the brief one paragraph "About this collection" page, I could not find any other explicit collection policy. Also, the collection does not appear to be actively curated -- it contains an invitation to researchers to "visit often in 2007."

Metadata: The digital library is very straightforward and simple, but it is surprisingly not user-friendly and not aesthetically pleasing to the eye. There are three ways to search the collection: (i) by searching for words in particular fields (anywhere, title, text, keywords, or date), (ii) by using the word search tool that produces an unattractive drop-down menu of browsing options of documents with number identifiers that are not very helpful, and (iii) by browsing documents in their entirety. There is sufficient metadata to produce results for each of these results and to allow users to select their own presentation preferences: interface language (numerous languages to choose from), Encoding (default is Unicode UTF-8), Interface format (graphical or textual). Users can also select search preferences: query mode (simple or advanced), query style, case differences, word endings, search history, and how many search hits to display per page. On the positive side, word search brings up text documents with yellow highlighting on the search terms. On the negative side, if you click on Volume 5 in the browse contents feature, there are some pages that do not have titles, dates or other basic metadata.

Object Characteristics: This may be the only digital library collections that I have encountered this semester that does not provide images in PDF format. The only options to view the objects are in HTML text or in Microsoft Word, which is odd since the Marshall Foundation's other digital collections seem to provide at least some PDF images. I could not locate a finding aid for the collections, which makes identifying numbers, such as #1-001, largely superfluous. Objects are presented as search results in long lists that require a lot of vertical scrolling. When one clicks on an object, the metadata is relatively sparse in that it only includes: the document number, the publisher, the date of the document, copyright information, and a very brief description of the object.

Intended Audience: The intended audience is probably students, historians, and other researchers. One indication of this is the recommended citation form that the digital library provides for each digital object, which can be very helpful to researchers. Nonetheless, the digital library has largely limited its usefulness to today's researchers by not including PDF images of the original document pages. Relying on OCR'd text in HTML may have been acceptable in the early days of the web, but in my experience, it is not relied upon, and certainly not preferred, by historical researchers today.