Executive  summaryChapter1Chapter2Chapter3



The Fowler reported that data extraction from their collection management system took less than one hour, while reformatting the data to fit the MESL field structure, editing, and testing constituted the bulk of the time spent. The MESL project prompted the Fowler to undertake a major internal effort to input data to describe images for distribution and sharing. The added benefit to this was improved local access. They report much time spent editing and inputting changes to text, including a summer intern spending 160 hours inputting editing changes for African objects in the first year. Also in the first year, Peruvian ceramics data was distilled from a museum exhibition catalog and entered manually into MESL records. Because of the nature of the Fowler Museum's objects (ethnographic and archaeological artifacts), they had a particularly difficult time mapping their data to the MESL fields. For example, the "creator" person, culture, and creation place had not been separated, and it was difficult and time-consuming to manually do so. In their report, the Fowler outlines a 15-step process that they used for putting their data in the MESL format.

George Eastman House

GEH reports that they routinely left nine fields empty, but felt that their data mapped easily to the MESL data dictionary.


Harvard routinely left six MESL fields blank because of the lack of data available for those fields. Like GEH, they also found the mapping from their data to be fairly simple. Harvard reports no problems with mapping their database to the MESL requirements, but ends up with more hours spent on data transmission to Michigan due to incompatibility and file naming problems. Their fields did not match, and they also had problems with the export routine adding delimiters and random characters. Since this type of process was new to them, the Information Systems staff received assistance from someone at the University of Virginia with Perl scripting for the first year, and from a couple of technical people from Michigan to help map their data in the second year.


Houston lists seven steps they used to put their data into MESL format. They report that their "Quixis" system field names did not match the fields for the MESL specifications. Another problem was that certain special characters did not convert (È, Á, etc.), which caused more work for the museum.

National Gallery of Art

The National Gallery of Art created a special program to draw MESL data from their collection information system. They routinely left five fields blank, but included unstructured text in an accompanying document.

National Museum of American Art

The National Museum of American Art established special query/export routines in the first year that were reused in the second year. The NMAA database administrator spent approximately two and a half days on it, and was assisted by a programmer. They list only three steps in the extraction and conversion of data: writing a macro to extract data, running an executable file to format and add field delimiters, and downloading data into FileMaker Pro, where it was merged with its associated image file data. The FileMaker Pro database was used after they unsuccessfully attempted to use SGML as a transfer format. The museum listed eleven MESL fields that they left empty.

Library of Congress

While they did not provide specific numbers, the Library of Congress reported that reworking their data to fit the MESL categories was very time-consuming, and that adding or deleting data to meet museum-oriented practices (as opposed to the Library's book-oriented practices) was a complex exercise in data mapping. The Library used a succession of macros and manual editing to map the data. They also provided additional unstructured prose texts to accompany the collections.

Back to Chapter 2


The Cost of Digital Image Distribution:
The Social and Economic Implications of
the Production, Distribution, and Usage of Image Data

By Howard Besser & Robert Yamashita