Vol. XVI, No. 2
CSA Newsletter Logo
Fall, 2003

ARTstor, a Web-based Image Collection*

Susan C. Jones


ARTstor is the Andrew W. Mellon Foundation's initiative to bring digital images of objects and structures to the academic internet community. Its stated purpose is to "to use digital technology to enhance scholarship, teaching, and learning in the arts and associated fields. As part of a community-wide effort to further this goal, the project will build, continually develop, and distribute a library of digital images and related information that serves the needs of researchers, teachers, and students."1 This fall it is beginning a test phase at a limited number of non-profit institutions with access to seven different collections of images. The institutions for this phase include colleges, universities and museums.

ARTstor is based on the model of the foundation's highly successful JSTOR initiative, which provides on-line access to articles in major scholarly journals to participating institutions. (JSTOR is now independent of the Mellon foundation.) As JSTOR provides internet access to journals, ARTstor will provide its subscribers with access to on-line collections of digital images. Differences between collections of digital images and the contents of published journals, however, dictate somewhat different approaches to the management of ARTstor and JSTOR. Some of theses differences are examined below, along with some of the philosophy behind ARTstor. (Note that some issues involved in providing access to and preserving imagery have yet to be discovered at this stage in the development of web-based academic research.)

Current Content

ARTstor currently has static images of the world's art and architecture, combining both digitized photographic prints or slides and images that are created as digital ones. Seven collections, each quite different from the others in many ways, were selected as the beginning core of the collection to help ARTstor understand the range and interest of its potential users and contributors. There has been no attempt at this point to include digital video (originating as film, video, or DAD). Presumably as video and interactive visual media (VR reconstructions, computer animations, performance art using digital formats) become the subject of scholarly attention, their unique problems will be addressed, either by ARTstor or by yet another initiative.2

Current content includes:
  1. Art History Survey Collection. A basic collection of images for art history survey classes, the Academic Image Cooperative's four thousand images were assembled based on a concordance of monuments found in ten common art history survey texts.
  2. The Image Gallery. This collection of approximately 200,000 images is intended to support pedagogy in various fields of art history; it is derived in large part from the slide library at the University of California, San Diego.
  3. A selection from The Huntington Archive of Asian Art from Ohio State University. This archive consists of digitized images of black and white photographs and color slides taken by art historians John and Susan Huntington during their travels.
  4. The Mellon International Dunhuang Archive. This archive will ultimately be a combination of digital photographs of the current restoration/conservation of the caves; digitized historic photographic prints of the caves from the Lo Archive of Princeton, New Jersey, taken during the 1940's; and digital images of the artifacts from the caves now distributed among various museums and libraries around the world. This is an on-going project with some preliminary material currently available on ARTstor.
  5. Carnegie Arts of the United States Collection. These images are the digitized versions of the archival, large-format color photographs from this collection which have ceased to be available in a 35 mm slide format.
  6. The Museum of Modern Art (MoMA) Architecture and Design Collection. Since many of these objects are very large and fragile; they are not generally on view and some have never been published. Thus it is difficult for scholars and student to gain access to them. ARTstor will provide the primary access to these images.
  7. The Illustrated Bartsch Collection. This collection is based on the authoritative 96 volume publication of Old Master European prints. ARTstor includes digitized images of the published plates, cataloging data, and the scholarly commentary available in the printed version.

Copyright Issues

Copyright issues are particularly difficult. ARTstor's management hopes that traditional academic fair-use principles can be applied and enforced by limiting subscribers to nonprofit institutions.3 The law firm of Cowan, Liebowitz & Latman, PC, has been retained to address complex intellectual property issues.4 Specifically, copyright to each collection will remain with the provider; ARTstor needs and wants non-exclusive, ongoing, and royality-free license to integrate each collection into its on-line database and make all available to non-commercial users. Obviously ARTstor also has no need for exclusive use of the material, but the need for perpetual and royality-free access is fundamental for both research and educational use. The management sees ARTstor as an intermediary between the copyright holders and the users, seeking to balance interests of a broad arts community. The broad range of ARTstor's current collections (in terms of content, source, source photography, and related intellectual property issues) was a conscious choice, order to further the exploration of the range of issues involved in this kind of initiative.

To illustrate the type of complex issues that ARTstor collections might raise, consider the collections that contain images generated from books and other copyrighted material such as the Image Gallery and the Illustrated Bartsch. The copyright may not be held by the image supplier; many images are probably taken directly from book illustrations or part of commercially purchased slide sets. If problems arise with these images, ARTstor personnel expect that they can mediate between the various parties based on fair-use standards. An entirely different set of issues are raised when the owner of the copyright is identical to the supplier of images to ARTstor.

Image Format and Access

ARTstor will provide centralized control in the key areas of standardized file formats and on-line access. The flexibility to combine ARTstor's images with those from other sources is seen as key to using it in both pedagogy and research. To this end, ARTstor's guiding principles are to remain compatible with as many different online access systems as possible and not to require the user to purchase/lease/download specialized software. These principles are seen in the use of existing file formats and in the development of an XML- and web-based search to access its images.

ARTstor typically digitizes images contributed to the service, returning digital versions to the source institution or individual, generating TIFF and JPEG formats. The TIFF formats will not be directly available to the users, they will have access only to several resolutions of JPEG images, including a high resolution one.5 Any web browser will be able to access the JPEG images, so the user's ability to access them along with other other digitized images from different sources will depend upon the user and her/his software's sophistication.

Access to the images is through a simple, self-contained, web-based search engine, and not through any proprietary package. This eliminates most concern about cross-platform compatibility. Obviously, the Linux, MAC, and PC worlds all have versions of web browsers, and currently all browsers share fundamental features, like processing basic XML. The database behind this search engine is under the direct administration of ARTstor and will employ the use of several Getty standard "vocabularies" (the Getty Thesaurus of Geographic Names, the Union List of Artists, etc.). This may not be the ideal vocabulary or access method for a particular collection, but it will be a good starting place. Textual descriptions of any image are problematic, and for the near term, ARTstor personnel are scrutinizing individual entries and attempting to standardize them. The data about the images currently complies with VRA (Visual Resources Association) core categories,6 but ARTstor experts are exploring several such standards.

Redundancy and Quality Control

Any combination of image collections will probably contain multiple shots of the same object or structure. Individuals at ARTstor are examining each image to eliminate redundancy and still provide multiple views where appropriate. For example, images of of the same structure which vary in time, viewpoint, lighting, etc. will be allowed, as will images that highlight different details.

ARTstor personnel are also examining the entries for fundamental quality, correcting those that do not meet its standards. This quality assessment procedure includes everything from standardizing the textual information to color-correcting the digitized images where feasible and appropriate.

The needs of the academic community require records on the provenience of materials used in research and the tracing of any restoration/preservation work done on that material. This would include at least some of the work done by ARTstor's staff to standardize image presentation, description, and access. Even such a simple thing as standardizing a transliteration of non-Latin names into a Latin alphabet might obscure data vital to a researcher. Preserving a copy of the data as originally presented to ARTstor should alleviate this problem.

Archiving Issues

An on-line collection of images must be permanent to be of value to the academic community. ARTstor's management is committed to providing such a collection and to the archiving implicit in such a commitment. Since its image collections vary greatly in content and source, the management does not propose a unique solution to the community-wide challenge of funding digital archiving and preservation. It assumes that each collection will require an individual archiving arrangement between ARTstor and the providers. For those collections that are generated specifically under its auspices, such as the Dunhuang Project, it will be the "archive of record," providing for the preservation of those digital images. For other types of collections, it will leave archival procedures to their provider, and finally, for some images, such as those in the Image Gallery which are taken directly from printed images, archiving is less an issue. (If necessary, the original image can be redigitized.)

This individualized approach to archives begs the overall question of cost and responsibility.

Cost Issues

The ARTstor initiative is relying on the JSTOR model to handle the cost of digitizing collections and adding them to its corpus. Under JSTOR, initial costs to bring the collections into the project are the responsibility of the donors; the cost of accessing the images and maintaining that access falls on the site licensees; and archival costs will be handled by a separate endowment. Funding for this endowment is problematic at this point in time.

Like JSTOR, ARTstor will ultimately depend upon institutional license fees to support its ongoing operations. These fees will not support development of collections, for which other sources of funding will be sought, nor will they support archiving costs. As we have noted elsewhere in the Newsletter, archival costs are only just coming to the attention of the academic community.7

Concluding Remarks

As currently conceived, ARTstor is both a pedagogical and research tool. It will provide fundamental on-line access to image collections. Through its use, the academic community should also gain important insights into how such a resource could affect pedagogical and research methodologies.

The underlying assumptions that pedagogy and research have equal priorities in academia and that academia can resolve resulting conflicts of interest is another fundamental debate in academic circles. As in the case of for-profit education, such issues are outside the scope of the CSA Newsletter.

ARTstor's great strength at this early stage comes from its commitment to developing a workable database of images. Its designers are aware of its position as a front-runner and its ability to shape future web-based image collections. ARTstor's personnel listen closely to the problems and concerns of their clientele and are committed to addressing them. The interactions between ARTstor's designers and its earliest users have included modifications to ARTstor's fundamental approach to accessing, storing, and presenting images. Therefore, its current state must be viewed as the first steps of an evolutionary process. ARTstor clearly understands this and remains very responsive to its users' concerns.

-- Susan C. Jones

To send comments or questions to the author, please see our email contacts page.


Acknowledgements: Many of the issues involved in providing access to imagery and preserving the data behind it have yet to be discovered, let alone addressed, at this stage in the development of web-based academic research. I wish to express my appreciation to Mr. Max Marmor, Director of Collection Development at ARTstor, for discussing the current state of ARTstor with me. I also thank Professor Marc Simpson of Williams College, who used ARTstor last year in an art history class on Winslow Homer, for sharing his class's experience; and Professor Jeffrey A. Cohen of Bryn Mawr College, for sharing his concerns as a potential user. Return to article.

1. http://www.artstor.org/, accessed 22 Sept 03. This site contains descriptions of the collections, a listing of participates in current tests, a FAQ page, and background on the project and its personnel. Return to body.

2. JSTOR is only beginning to address problems of archiving the increasing amount of original e-publications, material that does not fit into a static, linear digitized print model. To some extent, it has passed the initiative to yet another Mellon project, the Electronic Journal Archiving Program. JSTOR personnel concluded "that a separate organization dedicated to the task of e-archiving journal literature is needed to maximize the probability for success." http://www.jstor.org/news/2002.09/EarchivingBornDigitalContent.html, accessed 22 Aug 03. No. 6, Issue 2, JSTORNEWS, October 2002 "E-Archiving Born Digital Content" (http://www.diglib.org/preserve/ejp.htm). Return to body.

3. Restricting site licenses to non-profit institutions raises an interesting issue that is outside the scope of this Newsletter, but which must be raised. The traditional equation, educational-institute = not-for-profit-institute, can no longer be taken as given in the United States; the University of Phoenix and Edison Schools come immediately to mind. In the current philosophical climate of replacing "inefficient," not-for-profit sectors of our economy with "efficient," for-profit ones, the trend towards for-profit educational institutions can only grow. How do not-for-profit, fair-use issues fit into this new educational model? Return to body.

4. http://lists.village.virginia.edu/lists_archive/Humanist/v14/0691.html, accessed 20 Aug 03. Return to body.

5. There will be Web access to, in addition to JPEG images, compressed versions of some images, using a wavelet compression algorithm. The exact format is from TrueSpectra. http://www.truespectra.com/, accesses 26 Sep 03. Return to body.

6. Visual Resources Association is an international organization that has set up a widely used standard to describe images similar to the Dublin Core project for textual data. http://www.vraweb.org/, accessed 26 Sep 03. Return to body.

7. See "The Archaeological Data Archive Project Ceases Operation," by Harrison Eiteljorg, II, in Vol. XV, no. 2, Fall 2002, issues of the CSA Newsletter, http://csanet.org/newsletter/fall02/nlf0201.html. Note also this statement from a JSTOR newsletter: "Any long-term archive must have a business model in place to fund the ongoing investments that will be required. In our view, this is a major challenge facing the higher education community. Although there is unanimous agreement among constituents that e-archiving is important, this enthusiasm is not yet matched by budget lines that support the objective." (emphasis added) "JSTOR Takes on the Challenge of Journals 'Born Digital,'" No. 6, Issue 1, JSTORNEWS, March 2002 http://www.jstor.org/news/2002.03/EarchivingMarch2002.html, accessed on 22 Aug 03. Return to body.


For other Newsletter articles concerning on electronic publishing or the use of electronic media in the humanities, consult the Subject index.

Next Article: Virtual Reality Guide to Good Practice

Table of Contents for the Fall, 2003 issue of the CSA Newsletter (Vol. XVI, no. 2)

Master Index Table of Contents for all CSA Newsletter issues on the Web

CSA Home Page