[Mirrored from: ]

What follows is the Narrative submitted as part of a grant proposal to the National Endowment for the Humanities, an independent agency of the federal government of the United States of America. Roger Bagnall prepared the final version of the proposal; We at Duke have modified it slightly to conform to the requirements for placing it on the World Wide Web. Note that the appendices mentioned throughout this Narrative are not included here.

Questions and comments can be addressed to joates@acpub.duke.edu or to Roger Bagnall at bagnall@columbia.edu.

Advanced Papyrological Information System

A Joint Project of

Columbia University

Duke University

Princeton University

The University of California, Berkeley

The University of Michigan

Université Libre de Bruxelles

Yale University


1. Introduction

This application requests support for both preservation of a large and important body of ancient manuscript material, through conservation and imaging, and improvement of intellectual access to this material, through cataloguing and an innovative and experimental electronic system linking cataloguing with images, bibliography, text, and published literature. The Advanced Papyrological Information System (APIS) project, although leading to a permanent "product" and conceived of as a continuing cooperative venture among six major American universities and one European institution, is presented here as a four-year project, for the first two years of which the present application seeks Endowment support.

Preservation activities generally stand today in the position of the Roman god Janus, looking both forward to the astonishing capabilities of new technology and back to traditional methods of recording and preserving the intellectual and physical heritage of earlier millennia. The APIS project shares this bidirectional character. The larger part of the activities for which support is requested in this grant period are established, well tested, and traditional: physical conservation of ancient artifacts with writing, cataloguing in standard library records (the US-MARC AMC format), and recording of the images of these unique objects to reduce the wear and tear of use of the originals and preserve their intellectual contents against the possibility of catastrophe. The value of these activities is well recognized, and the importance of the collections in question will be described in section 2. But this project is something more than an assemblage of separate proposals for preservation.

What is distinctive about this project is exactly the reason that it is not a batch of unrelated applications for these preservation purposes: the institutions involved will adopt collectively a set of standards for imaging, for the formats of the electronic data generated, and for the linking of the various sets of electronic data. The entire project will thus be carried out with a view to the creation of an integrated information system, available over the Internet. The cooperative aspect of the proposal is thus central to its existence, for it will replace the prospect of a world of incompatible, separate systems, each with its own standards, with that of a single, seamless system that will be readily usable not only by papyrologists but by scholars and students in other fields. By beginning APIS now, we will be able to lead to the adoption of these standards not only in North America but worldwide and set an example to other disciplines of what is possible.

APIS will therefore, in time, not only entirely transform instruction and research in papyrology but also, for the first time, make papyrological materials readily accessible to nonspecialists. This will in fact be its most central outcome. The vast resources of the papyri have until now been relatively little used either by scholars of most fields concerned with antiquity (literature, history, philosophy, religion, archaeology) or by a broader educated public. In large part this is the result of the extreme difficulty of access to the material. APIS will make it possible to change that situation. These are large claims, but they are fully justified, we believe, in what follows.

Finally, APIS will serve as a model, both in its collaborative creation of field-wide standards and in its integration of different types of information resources, of what is possible for a wide variety of fields in humanistic studies. The range of languages recorded in the papyri will stretch the capabilities of information technology in a fashion certain to be useful to other fields.

2. Significance of the Papyri

Papyrus was the most important writing material of the ancient world and perhaps ancient Egypt's most important legacy; alongside it were used other (often cheaper) materials, like wood and clay (broken pottery sherds with writing are called ostraca). On these materials were recorded everything from high literature to the myriad of documents and other communications of daily life. About one in ten of those studied to date is a fragment of literature, either a far more ancient witness to a work known otherwise from medieval manuscripts or a text hitherto lost in antiquity. From the literary papyri the modern scholar learns about the state of literary texts in antiquity before errors were compounded in the manuscript tradition of the Middle Ages. From among these papyri the modern world has recovered such important lost works as the lyrics of Sappho and the Paeans of Pindar, the comedies of Menander, the Mimes of Herodas, the orations of Hypereides, the Constitution of the Athenians by Aristotle, and early Christian and Gnostic works which once competed with the New Testament.

Nine of ten published texts are private letters or documents of every conceivable sort - legal and business papers, government regulations, property records and transactions, petitions to high officials, tax and rent receipts, bank deposits and payments, and farm and crop reports. As such, these documentary papyri differ little from modern archival material; except for their usually fragmentary nature and extreme antiquity, they reflect the quotidian affairs of government, commerce, and personal life in much the same way that modern records do.

From the documentary papyri were born the new fields of social, economic, and administrative history, which have all but displaced the older histories of kings and battles. The papyri (using the term to encompass the other materials) are thus the source of a large part of what we know about many aspects of antiquity, particularly those concerned with economic life, social relations, cultural interaction in a pluralistic society, and daily life.

Until now relatively little classical material available to a general audience has been available in electronic form, and that (mainly through the Perseus project) has primarily been concentrated on the archetypal canonical period, Periclean Athens.[1] APIS, designed to be usable by nonspecialists, can open up material outside the canon and allow the full diversity of a multilingual and multicultural ancient society to be visible both in text and in images to students. There is no other body of ancient material with such dramatic potential for broadening the ability of students to grasp the reality of a world in which not everyone was a Greek or a Roman, not all activities were the sole province of men, and not everyone was rich.

This material is self-evidently of central importance for classical history and literature, but it is also of immense importance for other areas. For example, the papyri have transformed our understanding of the development of the Greek and Latin languages in everyday use, a matter of importance not only for historical linguistics but for the way scholars read Jewish and Christian sacred texts. From the papyri, moreover, have come abundant new works of religious literature not only for Judaism and Christianity but also for traditional Greek and Roman cults, for Manicheism, and for the early history of Islam. An active project in Italy collecting the philosophical papyri bears witness to the importance of the papyri for the history of philosophy. The papyri are also our most important source for the actual working of law in ancient societies and help make it possible to test the theoretical doctrines derived from jurisprudential literature. And the papyri are (along with archaeology) the main source of raw data from antiquity capable of allowing the insights of the quantitative social sciences to be applied to antiquity.

The American collections involved in the first phase of this project are the six most substantial in size and scholarly value of contents in the country. They also include those with the most active current programs of graduate instruction in papyrology in the United States (Michigan, Duke, and Columbia). These collections have distinctive histories and strengths which we set out briefly here; collectively they dominate American holdings in the area.

Berkeley's collection stems from a single excavation, that carried out by B. P. Grenfell and A. S. Hunt at Tebtunis (southern Fayum) in 1899/1900 with financing from Phoebe Apperson Hearst. The excavation yielded three large lots of material: (1) cartonnage from the mummification of crocodiles, mainly from the late second and early first centuries B.C.; (2) cartonnage from human mummies of the third and second centuries B.C.; and (3) papyri found in the ruins of the town, from the first three centuries of our era. Only a small part of this collection has ever been given careful study; the total number of fragments exceeds 21,000, although many of these are extremely small. The bulk of the collection is in Greek, but there is also a significant body of material in Demotic Egyptian.

Michigan's, the largest American collection if small fragments are discounted, was the earliest to begin large-scale collecting in 1920, with continuous subsequent purchases until 1943, and then again in the 1980s. In the early years, the papyri were acquired through a "cartel" that was comprised of the British Museum, several American universities (Columbia, Michigan, Yale, Princeton, etc.) and a number of European universities. Michigan's holdings grew further from an unprecedented excavation over 11 consecutive seasons (1924-1935) in the ancient Egyptian town of Karanis. Most of the papyri from this expedition were returned to the Egyptian government in 1954 as part of the original agreement, but approximately 1,000 individual papyrus fragments, along with Polaroid pictures of many of the returned papyri, remained at the University of Michigan. Today, the Michigan Papyrus Collection is among the largest worldwide. It contains over 7,000 inventory numbers and has more than 10,000 individual fragments (many of the inventory numbers include multiple fragments). The papyri in the Michigan collection cover almost two millennia of history, ranging from ca. 1000 BC to 1000 AD, with the majority from the third century BC to the seventh century AD. The majority of the papyri are in Greek, but there are large groups of papyri in all languages of the various ethnicities that once lived in Egypt, in particular Hieratic Egyptian, Demotic, Coptic, Arabic, Latin, and Aramaic. In addition to the papyri, the Michigan collection contains other writing surfaces that were in use in the ancient world, such as ostraca (pot shards), lead, wax and wooden tablets, parchment, and rarely, paper.

Columbia's papyrus collection began with distributions from the Egypt Exploration Fund in the early years of this century, but serious purchasing started in 1923. Most of the collection was acquired through the "cartel" mentioned earlier. It was supplemented with a sizable collection of ostraca purchased in 1964 and 1965 from the Metropolitan Museum of Art. It contains today about 800 papyrus inventory numbers (again, many of these covering multiple fragments) and about 3,600 ostraca. The papyri are mainly in Greek, but with a range similar to that of Michigan. The ostraca are mainly in Coptic but include some in Hieratic, Demotic, and Greek.

The Princeton papyrus collection also began with Egypt Exploration Society distributions, but most of its holdings were acquired through the same cartel or in separate but related purchases in the 1920s. Part of this lot was acquired by Robert Garrett for his own collection but deposited in the Princeton University Library; these papyri were donated to Princeton in 1942. In addition, some papyri have come into the collection through a variety of gifts and purchases since that time. It totals approximately 1,250 pieces. The two-thirds of the collection that has received publication or preliminary inventorying is mainly in Greek, but Latin, hieroglyphic, Hieratic, Demotic, Coptic, Syriac, and Arabic are also represented.

Yale's papyrus collection began in 1889 with a gift of papyri from the excavations of W. M. Flinders Petrie at Hawara. Additions were made in the first decade of the 20th century through distributions of the Egypt Exploration Fund, in the 1920s through the cartel mentioned above, and, after 1925, from the purchases made in Europe and the Middle East by Michael Rostovtzeff and other members of the Yale faculty. The Yale excavations in the 1920s and 1930s at Dura-Europos yielded several hundred further items, including a large number of Latin texts. Large groups of material were acquired in the 1930s and again in the 1960s, including a sizable number of Arabic documents. Items have been added sporadically since the 1960s. The collection as of December, 1994, has over 3,800 inventoried items and many times that number uninventoried. The inventoried level represents ca. 4,800 pieces (many small fragments are catalogued and mounted under a single number), of which ca. 3,800 are fully catalogued, conserved, and mounted in plexiglass, and ca. 1,000 are catalogued but not yet conserved (mounted in glass). Additionally, there is a much larger number of pieces (many of them small fragments) not yet catalogued or numbered. There are a small number of wooden and wax tablets, parchment, palm leaf, and paper items. The date of the material ranges from ca. 1700 B.C. to the 12th century of our era, with the majority of the items falling in the Roman and Byzantine periods. Although most items in the collection are Greek, the number of Latin and Arabic papyri is uncommonly high. The range of languages correponds to that in the Michigan collection but also includes Pahlavi, Syriac, and Hebrew.

Duke began to acquire papyri seriously only in 1969, but has emerged as a major collection in the quarter-century since that time, with total holdings of about 1200 pieces, mostly in Greek but some in Latin, Demotic, Coptic, and Arabic.

These collections are all of great international importance, and collectively they are the core of the American holdings of such documents, with approximately 45,000 items. Nearly forty volumes of texts from them have been published or are in press, and active work is in progress on all of the collections. Because several important European collections have never disclosed information about the number of papyri they possess, it is impossible to provide an accurate world context for these numbers, but at a rough guess the six institutions possess a tenth of the total.

3. Background and previous history of this project

Like classical studies generally, papyrology has been ahead of most humanistic disciplines in applying information technology to the management of information and the support of research. In this respect it has shown the electronic equivalent of the leadership role that it has played in conventional research tools since the early part of this century.

Today, for example, we have in electronic form 100 percent of the published texts of Greek and Latin documentary papyri and ostraca through the Duke Data Bank of Documentary Papyri (DDBDP); the bulk of recent bibliography through the files for 1960-1968 and 1976-1994 of the Bibliographie Papyrologique; a well-advanced project in the area of cataloguing of collections, using a standard library catalogue record type, with extended subject access (at Duke) and developing projects in image capture (at Michigan and Duke). These resources are described in more detail later in this application.

There are also other relevant undertakings. The Chicago Demotic Dictionary project has both text and imagery for cursive late Egyptian in Macintosh word processing format. Prosopographical projects are underway in Louvain (converting the Prosopographia Ptolemaica to electronic form) and London (creating a prosopography of Roman Egypt). In addition, a typological electronic catalogue is under development in Heidelberg.[2]

These developments, though relatively recent, have together already transformed scholarship in the field and continue to do so. They are far from realizing the full potential for changing scholarship and education in this discipline, however, and already it can be recognized that for the most part these projects represent a technology of the 1970s and 1980s - database files, whether text or structured, with separate access systems. Some are available on the network, some on floppy disk, some on CD-ROM, and some not at all outside the place of creation. Like most projects of that era in the humanities, these are rapidly being left behind as user interface software moves forward at astonishing speed. Proposals that left most people gasping with disbelief in 1992[3] are everyday phenomena in 1995. Moreover, the present array of unintegrated tools has done little to lower barriers to the study of papyri by nonspecialists, including students. It is time to build an Advanced Papyrological Information System to fill in the gaps in our digitized resources, take advantage of current access technology, and allow later developments to be added as they become available.

There is another grave difficulty in the present state of things. Everything said above about research tools concerns only the published papyri and ostraca, a sizable body of material (worldwide, about 50,000 documentary texts, to which must be added perhaps 5,000 literary texts) but only a fraction of what has survived. Although many surviving papyri are fragmentary, these modern research tools provide scholars with the capability of extracting information from them to a degree hardly possible before, and even of combining fragments scattered in different collections with an ease never before known. No one knows how many unpublished pieces lie in the world's collections, but it is unlikely to be less than five times the number of published texts. Some of these enjoy excellent conditions of conservation and storage, but others have scarcely been touched in three-quarters of a century or more since their acquisition; this is true even of major collections with resident papyrological experts, and far more true of small, isolated collections.

Much work has already been carried out by the partner institutions in laying the foundations for APIS:

(1) Duke. The Endowment has supported a project at Duke University for conservation, cataloguing (to US-MARC AMC standards), and imaging (by color scanning) of the entire collection, work carried out by Peter van Minnen and Suzanne Corr. Duke will be, at the conclusion of this project in December, 1995, one of the few collections in the world to be in such good order. Already Duke has established a home page, the Duke Papyrus Archive (URL: http://scriptorium.lib.duke.edu/papyrus/), which contains 200 images of papyri and the full catalogue records of 1000 texts. By the end of this year more than 2000 images of papyri and their catalogue records linked by hypertext markers will be available at this source.

(2) Michigan. The papyrus collection at the University of Michigan began to experiment with capturing digital images of its papyrus holdings as early as 1991. At the same time, the first elements of an electronic catalog (especially of the published and assigned holdings) were placed into an electronic database. The Provost of the University funded a half-time position (occupied since July, 1991 by Dr. Traianos Gagos) in order to digitize the entire collection and create an electronic database that would be available online. The goals of his position were inter alia to start bringing the Michigan papyrus collection under physical and intellectual control. This involved converting old hand or typewritten notes into electronic form and devising a database structure to accommodate bibliographic information of the published materials.

When the Michigan project started in 1991, the papyri were scanned on a flatbed grayscale scanner. Approximately 500 images were scanned with the use of this technology at a standard resolution of 300 dpi. This result, given space limitations, was considered to be satisfactory. It was discovered in the experimental phase that the use of imaging technology is a very realistic alternative to traditional teaching based on the study of photographs. The dramatic improvement of technology in the meantime has led to the use of color scanners and electronic cameras; multispectral cameras - used primarily for military purposes until a year ago - are now available for general research purposes. The availability of alternative media for image capture and the desire to create a high resolution electronic image of archival quality led Michigan in summer, 1994 to a period of extensive experimentation. Papyri were scanned, put on the WWW for access via Mosaic, and made available on the Internet since August, 1994. Scholars in the United States and abroad have had the chance to look at the scanned images and respond to a questionnaire to help in determining the standards and equipment that are best for image capture and electronic distribution. (See Appendix 8 for a brief report on this experiment.) This study led to a further investigation of imaging by the partner institutions in APIS, under a contract from the Commission on Preservation and Access. The report of this investigation (Appendix 8) has served as the basis for the imaging methodology adopted in the current project.

(3) Columbia. The preliminary work for APIS at Columbia has taken two forms. First, it has since 1988 funded student research assistance for the conversion of the retrospective Bibliographie Papyrologique (1960-1992) into electronic form, a project that at present has reached the 75 percent mark. Second, in 1994 a part-time assistant curator, Dr. Raffaella Cribiore, began work on the least studied part of the collection, the Coptic ostraca. She has begun the creation of an electronic catalogue readily convertible to US-MARC AMC standard.

(4) Yale. Beginning in 1983, Yale undertook at its own expense a conservation program for the papyrus collection. All items are now individually sorted and housed in acid-free paper, and the process of conserving, cataloguing, and mounting the items in plexiglass, starting with the older acquisitions and continuing through the most recent, is well under way. Approximately 1,000 items were mounted in glass at the time the project began, and these will be rehoused in plexiglass after conservation. With the exception of a small number of severely damaged pieces, conservation and mounting of the entire collection should be completed by spring, 1997. A cataloguing project was also initiated at Yale in conjunction with conservation. Records are created in an inhouse database using the program Inmagic and are then mounted as a text file on the Yale gopher, making them accessible worldwide over the Internet. (Correspondence from around the world indicates that the files are frequently consulted.) The cataloguing follows standard papyrological conventions. Each item is described physically and intellectually and equipped with a full bibliography as well as an account of the acquisition and publication history of the item. The catalogue is updated at regular intervals during the year with newly catalogued items (ca. 600 additions in 1994) as well as corrections and additions to the earlier entries. The catalogue of the entire collection should be completed by the beginning of 1997.

(5) Berkeley. The absence of a papyrologist on the Berkeley faculty for decades led to very little activity in the collection. The International Photographic Archive of Papyri, funded by the Endowment, photographed about 6,500 pieces and put more than 20,000 fragments into acid-free folders. More recently, a Dutch doctoral student in papyrology, Arthur Verhoogt, visited Berkeley and drew up a comprehensive condition report on the collection at the request of Anthony Bliss, Curator of Manuscripts and Rare Books in the Bancroft Library. This body of this report is attached as Appendix 10.

(6) Princeton. Princeton also for many years lacked a faculty papyrologist. In 1993, however, it appointed a demoticist, Joseph Manning, to the faculty of the Department of Classics. It has also begun the creation of a basic finding aid for the papyrus collection, in the form of a Preliminary Checklist of the Princeton University Collections of Papyri (1995). Dr. Traianos Gagos, of the University of Michigan, was brought in as a consultant to this effort in May, 1994.

4. Condition of the Materials

Papyrus is a remarkably durable material, far more permanent than the acidic paper of the period since 1850 and on the whole even more than the rag paper in use before then. But it is of course much older than most paper manuscripts, and most papyri are torn on several, if not all, sides. They usually emerge dirty, crumpled, and twisted, unless they have been preserved in a box or jar (as occasionally happens). Ostraca are often broken, and sometimes have significant salt in the fabric if they have lain in land reached by the Nile's waters. Some preliminary conservation is generally done by dealers or in the field, but usually full cleaning and straightening is left for "laboratory" work in the library, which has often meant never.

In most papyrus collections conservation work is carried out by papyrologists, that is, by scholars, rather than by professional conservators. Reasons for this state of affairs include the relatively simple character of much of the work to be done and the absence in many places of any institutional support for conservation work. In a few of the larger collections, like the Austrian National Library in Vienna, full-time conservators are employed. There is fairly widespread agreement in the field on the main techniques used; these are described in a book by Michael Fackelmann, former conservator in Vienna. [4] Special techniques for working with carbonized rolls have been developed by the International Center for the Study of the Herculaneum Papyri (Naples) and by Jaakko FrîsÇn (Helsinki), the latter of whom has been retained for work on the carbonized rolls found in recent excavations at Petra.[5]

The University of Michigan has created a brief guide to current best practice in papyrus conservation, which has been developed to its present form in collaboration with conservator staff at the other APIS partner institutions. This is attached as Appendix 7. In order to provide a fairly uniform level of conservation expertise for this project, the partner institutions will conduct a one-week training seminar at the beginning of the grant period. All project staff who will be carrying out conservation, including both papyrologists and conservators, will attend. The seminar will be led by staff from Duke, Michigan, and Yale involved in work carried out in the last few years, plus a professional European papyrus conservator (we will invite Andrea Donau, the present conservator in Vienna, to serve in this role).

As already indicated, the Duke collection is now, thanks to Endowment support, thoroughly cleaned, straightened, and mounted in suitable glass. They are stored in acid-free boxes in the vaults of the Special Collections Library.

Conditions at Columbia and Michigan are very different and not easy to describe concisely, because different batches of papyri have not received uniform treatment. Many of Columbia's papyri were glassed in the 1960s, but in sandwiches of excessively heavy glass and with unstable tape that now oozes everywhere. Its ostraca, acquired in the 1960s from the Metropolitan Museum of Art, have not until recently even been systematically examined, let alone given any care. In general, only those texts already published or currently being studied for publication have been photographed, and those almost always in standard black and white.

A condition survey of published documents on papyrus and other substrates in the Michigan Papyrology Collection was conducted in June-July, 1994. The survey report (see Appendix 11) provides detailed information about the present condition of materials that will receive conservation treatment during this project. Over the last five years Michigan has refoldered the bulk of the collection and systematically checked the pieces framed in glass; a total of some 6,000 papyri has been thus given preservation attention.

As described above, the bulk of Berkeley's collection was put in acid-free folders by the International Photographic Archive of Papyri staff in 1979. Much of the rest, however, is either in Vinylite mounts of the 1930s, which need replacement, or still in need of initial conservation treatment and kept in tin boxes from the time of the excavation (see Appendix 10 for details).

Recent work on the Yale collection is described above in section 3. It is expected that the entire collection will have been given conservation treatment and rehoused by early 1997.

Princeton's collection has had little conservation attention for some years, and conservation will be a large part of its activities in Phase 1. On completion of this phase, the papyri will be relocated to the climate-controlled chamber described below in section 14 and Appendix 12.

5. Access to the Collections and Bibliographic Control

Once again, there are large differences among the partners in this project. All of the Duke papyri will, by the end of 1995, have been catalogued in US-MARC AMC records that are part of the Duke Library catalogue and also added to RLIN and OCLC. It has, thus, a mode of access that for the combination of accessibility and quality is unparalleled in the world.

At Michigan, between 1920 and 1991, most of the information on the studied and published papyri was kept manually and did not meet modern standards. The information had not been updated for several years. Beginning in 1991 Michigan undertook to experiment with and create a detailed electronic catalogue both for the published and also for the unpublished papyri, ostraca and other writing surfaces at the University of Michigan. Such a catalog was also to include papyri that once belonged to other collections (e.g., the Cornell collection, part of the Michigan collections since the 1970s) and to be uploaded ultimately to one of the standard national databases. After extensive internal and external consultation, a detailed plan of fields for this database was developed. The first step was to capture information about materials that had already been studied in order to avoid duplication of efforts by scholars. This database, created with a program called 4th Dimension, gives detailed information on the first and later editions, journals or series in which the papyri and ostraca appeared, plates in the edition, and subsequent corrections cited in the Berichtigungsliste. Because of the magnitude of the collection and the variety of ways in which papyri are being stored (in regular boxes; in oversize boxes; sandwiched between glass; in lockers provided for the use of scholars; in lockers for "problematic" pieces), the database that was created at the time included a special field that marked the exact location of each published papyrus. Thus, the database secured also physical control over the collection (see Appendix 13 for a sample from this database).

At Columbia there was, as of 1994, only an index-card file, with typed entries of the briefest sort, including only some of the papyri and none of the ostraca, and kept in the curator's office. In summer, 1994, work began on creating records for the ostraca. Work is beginning in summer, 1995, on entering these into an electronic catalogue designed to allow easy uploading into US-MARC AMC-compatible records.

Yale has, as indicated above in section 3, created its own electronic database of the papyri in its collection, using the program Inmagic. This catalogue should be complete by early 1997.

Princeton's Preliminary Checklist has been mentioned above; this is the principal current means of access to the collection.

As we have indicated above, so far electronic technology has made some headway in making papyrological material available to scholars in other areas of the study of antiquity, but hardly beyond that domain and certainly not to a broader educated public. We believe that it is now possible to cross that boundary and devise a system in which the high technical threshold of serious research is no longer a barrier to wider educational use of the material. The core of this breakthrough is the combination of the subject cataloguing and descriptive text in the cataloguing records (as developed in the Duke catalogue project) with contemporary knowledge-base navigation tools and images deliverable over the Internet. These can together provide sufficient access information to allow students and people with general background in ancient studies to search for data on subjects of interest to them. A critical element of the current proposal is the unification of all of the existing catalogues by converting them for public consultation into a single format (US-MARC AMC).

APIS will thus be part of the growing riches available over the information superhighway to all those connected to it. Because it will draw on existing standards rather than creating idiosyncratic data structures or access methods, it will have maximum transparency for the user. Its methodology is described further below.

6. Selection for Preservation

Because of the uniqueness of the materials involved, the entireties of these collections are obvious candidates for a preservation and access project. This project envisages ultimately preserving, recording, and cataloguing all of the pertinent materials in these collections. In the case of Duke, this preservation work has already been carried out with NEH support. Yale plans to complete conservation work by early 1997. At Columbia, the entirety of the collection is targeted for work during a four-year period, of which the present application covers the first half.

With the much larger collection of Michigan, the initial two-year period is planned to encompass about 2,800 published and 100 unpublished objects, the remainder been reserved for future periods. These will be drawn in the main from those published so far, because these fragments, by virtue of having been studied, are all represented in the other resources that form part of APIS and will thus offer the most immediate gains for access. (In addition provision will be made for the stable housing and storage of all original documents of this project in order to keep them at Michigan as records of preservation and of ultimate authority for scholars in years to come.)

The focus at Berkeley in the initial phase will be on the roughly 1535 papyri currently housed in Vinylite, which are the most at risk of the entire collection, as well as being the most heavily used (because many of them are published). Work on these items will be accompanied at the start of the project by a comprehensive condition survey of the entire collection, to provide an in-depth analysis of the condition of the rest of the material and to allow intelligent planning for Phase 2. A detailed examination of the material in tin boxes will be carried out as part of this survey, giving for the first time an accurate estimate of the amount of material there and the work required to put it into satisfactory condition.

Princeton: All of Princeton's collection will be conserved, catalogued, and imaged in Phase 1.

7. The elements of APIS

Before describing the specific tasks planned in the grant period for which this application is submitted, we give a general account of the main components of the larger system that they are part of. A key characteristic of APIS is its modular, scalable character. It will be designed so that both new types of information and greater quantities of the existing types can be added. Beginning with a set of existing data resources for three collections, it will in its first pilot phase (1996-1998) begin to fill in the gaps for these six collections to create a fully operative system embracing the principal centers of papyrology in North America. These six have altogether about 45,000 items (including papyri, ostraca, and other ancient items with writing in ink), of which about one-eighth have been published to date. The first phase will include a pilot system available over the Internet to begin to get user response. See section 10 for further details.

The second phase of two years (1998-2000) will aim to complete these tasks, to make the pilot system with a large amount of data available for user testing locally and over the Internet, and perhaps to bring additional institutions into APIS. It will then, after the four year period, be ready to expand to embrace the entire field, by adding collections one at a time or types of material in bulk. It aims to set a standard for worldwide cooperation.

The initial model of APIS includes the following elements:

(a) Catalogue of papyri and ostraca, in a format acceptable nationally and internationally to bibliographic utilities. (These records will be accessible through those utilities as well as through APIS.) This is the heart of the system. The base format will be the US-MARC record type in its specific version for manuscript collections (AMC), and it will use standard Library of Congress Subject Headings and Art and Architecture Thesaurus index terms. (This is no small task, for these headings do not correspond to traditional papyrological classifications, but it is critical to the openness of the system to wide use. See Peter van Minnen's article in Bulletin of the American Society of Papyrologists 31 [1994] 159-170 for a full description.) The project will result in substantial additions to these standard repertories of headings. The experience gained by Duke will allow the other APIS partners to use these subject headings rather than develop their own. Examples of records from the Duke project are included as Appendix 16.

Although the AMC record type was originally designed for collections rather than individual items, it has been successfully adapted at Duke for individual papyri with significant content. For very small fragments which preserve little information, the time and cost involved in creating individual MARC records is not appropriate, and such records would amount to little more than debris in the databases. The institutions will for these instead create standard MARC group records, which will then be linked to item-level SGML-coded text documents to function as finding aids, drawing on the experience of Berkeley in its current Title II-C Finding Aids project on the creation of such aids and using the standards emerging from that project. We believe that this combination of cataloguing forms will apply the most appropriate form of access to different categories of material.

* Duke's catalogue will be virtually completed before the project start date. All of these records will be in standard US-MARC format. Duke will provide lists of the subject headings it has used to the other APIS partners.

* Michigan has about 3,000 bibliographic records in a relational database that will be adapted for project needs. This database has been closely mapped to US-MARC fields and can be converted automatically.

* The catalogue of papyri must be created out of existing manual records at Columbia; considerable intellectual effort will be required to enhance these records. Work on cataloguing the Columbia ostraca into an electronic database is beginning in 1995. These records will be mapped into standard US-MARC AMC records; these will then need to be enhanced with subject headings but will not have to be created afresh.

* Yale's in-house electronic catalogue will contain about 4500 records by early 1997. It will also be converted by Academic Information Systems staff at Columbia into US-MARC format.

* At Princeton and Berkeley, as at Columbia, work will begin from existing manual catalogues and lists, which contain only minimal information.

Both published and unpublished objects will be included in all of the completed catalogues, as already with the Duke collection, a major breakthrough in access for a field in which unpublished objects have generally been wholly unknown outside the home institution. This process will, naturally, take considerable time, especially in the case of the unpublished material.

The Catalogue will, by virtue of its combination of library standard subject headings with normal papyrological information, be the prime access point to the collections for a wide range of users. Full-text searching will be provided. Individuals with subject interests and librarians with no subject specialization in antiquity will be able to use the standardized subject headings, while (at the other end of the spectrum) scholars already equipped with specific papyrological references will be able to go straight to them. Most of the activity of Phase 1 is connected with the Catalogue and the underlying conservation work. More extensive detail about the individual institutional plans is given in the appendixes.

(b) Texts of the documents, drawn from the Duke Data Bank of Documentary Papyri. All published texts from these collections, except for a few in recent journal articles, are now included in the DDBDP and can readily be extracted. The texts will be marked up using SGML in accordance with the standards of the Text Encoding Initiative. The Greek is represented in a standard character coding format referred to as Beta Code, which was developed originally in connection with the Thesaurus Linguae Graecae. Michigan has already begun work on conversion of the DDBDP into SGML coding.

It is our present intention eventually to transfer all of these texts into Unicode (or its 32-bit counterpart ISO 10646, depending on the software available) at such time as appropriate indexing, searching and word processing software for Unicode text becomes available. At the moment, there is little such software on the market, and Unicode is unusable for most software. It is impossible to predict a timetable for such availability. We plan to retain Beta Code or some derivative of it until such time as it is possible to convert the texts to Unicode. The problem is not merely one of having Greek that is stored in the central database displayed as Greek for the user,[6 ]but rather of the user's ability to use Greek characters to describe a search which can be executed on the central database and its results retrieved for the user in Greek.

"Subliterary" and literary papyri are not yet systematically included; the project will add these texts for the three collections (about 900 pieces in all). Earlier projects at the University of Michigan have already captured several genres of such materials (oracles, prophecies, dramatic hypotheses, gem inscriptions, and others). Adding languages other than Greek and Latin will present technical problems that will be resolved in Phase 1; relatively few of these texts (Arabic, Coptic, and Demotic, in the main) have so far been published.

(c) Images of the documents. The APIS partners have as part of their preparations for this application conducted a study of digital imaging technology, funded by the Commission on Preservation and Access, the report of which is included here as Appendix 8. The conclusions of this study are that true 600 dpi imaging through digital cameras (not flatbed scanners) is an appropriate minimum standard for images. For views of the entirety of large documents, 300 dpi images will be included, but 600 dpi or higher images of portions of the original. Papyrologists using these images, with the advantage of modern computer tools for text reconstruction, have consistently found them much more valuable than traditional photographs, either in black and white or in color, and 600 dpi provides sufficient resolution and color accuracy for almost all purposes. Some papyri or parts of papyri with exceptionally dense information will be captured at higher resolution. In Appendix 8 we detail the standards for file format and other pertinent aspects of the imaging that we will adopt. These standards, which are in line with industry standards and projects like the Vatican Library Project, will make images fully interchangeable among the institutions involved and allow ready consultation over the Internet.

In the case of relatively small papyri, our study concluded that 600 dpi imaging was feasible with existing digital cameras, and that such cameras would produce imagery that would never need replacing. For larger papyri, however, it is not clear that future advances in imaging technology will not bring considerable advantages. For papyri over the size of 17 x 25 cm, therefore, we will also produce 4x5 color photographic transparencies. Experience in reproducing paintings and other fine arts has shown that these are sufficient to allow retaking of digital imagery as the capabilities of the latter improve.

APIS will also eventually make use of multi-spectral imaging (MSI), a technology developed at the Jet Propulsion Laboratory (Pasadena) and applied to some Dead Sea Scrolls fragments with considerable success. MSI uses bandwidths in the extreme infrared to bring out writing by identifying what part of its spectral signature is unique to it and extracting that bandwidth. This technology also uses a digital camera, but with a special lens designed for sampling the spectrum at specified intervals of bandwidth. At present, MSI is still a rapidly developing technology, and the cameras that use it are not interchangeable with those supporting the standard color imaging already described. It is also useful only for that minority of papyri (and, importantly, ostraca, where contrast is often poor) where there are significant difficulties in bringing the writing out from the background. APIS plans to bring MSI into use in Phase 2, after the materials on which it will be used have been identified; at present it is impossible to estimate on what part of the collections it would be applicable.[7]

All of the collections have existing bodies of photographic negatives made for various purposes in the past, and these will be scanned at an early stage. Michigan also has many photographs of texts once there but now returned to the Egyptian Museum in Cairo, and these photographs can also be digitized and included. Ultimately the vast resources of the International Photographic Archive of Papyri (centers in Brussels, Cologne, Oxford, and Urbana) can be drawn on as APIS grows beyond the original six collections. The project will need to resolve intellectual property issues connected with images when a wider range of repositories is included, particularly because it will not be possible to impose any standard policies about matters like publication permission on foreign institutions that adopt our standards.

(d) Bibliography, from the Bibliographie Papyrologique project at Columbia, which has converted an index-card bibliography published in Brussels since 1932. This database exists in Pro-Cite at present, and covers the period 1960-1994 except for a gap in 1969-1975, on which work is in progress; some six years of cards will need to be entered during the project. The Centre de Papyrologie et de l'êpigraphie Grecque of the UniversitÇ Libre de Bruxelles has committed itself to preparing these cards for data entry, and this will be the major participation of Brussels in Phase 1. It would ultimately be desirable for the cards for the period 1932-1959, at least for the participating collections, to be entered. These were, however, not provided with subject indexing when originally issued, and their availability with such indexing is thus much more difficult. We may look for other ways of covering the pre-1960 bibliography.

(e) Corrections and republications of texts (including redatings), as noted in the Berichtigungsliste griechischen Papyrusurkunden aus égypten (1913- ) and elsewhere. This eight-volume set (with one index volume; vol. 9 is ready for press) lists the textual history of papyri and ostraca after their publication.[8] Unique among the documentary disciplines concerned with antiquity, it has been a major element in the progress of papyrology. At present it exists only in book form; the digitization of its information will be required. This will not mean the entry of the BL itself as a separate element, however, nor even the entry of the information in the form presented there. Rather, published texts in the DDBDP will be corrected and other information added to the bibliographic citations in the BP, (d) above. See Appendix 3 for detailed discussion.

(f) Printed editions of the texts, including usually introduction, text, commentary, and translation. Most of these are in printed volumes, but some are in journal articles. These will be scanned using the techniques developed for digital library projects (see below) and included. In this way the actual image of the publication can be called up, while the ASCII text of the edition will be searchable. Where translations were not included in these editions (a tiny fraction of the cases), they will be created by the papyrologists on the staff and added as ASCII files in order to enhance access for students and scholars who do not know the original languages. (The intellectual rights to the existing volumes of papyri are held by the institutions holding the papyri; volumes are copyrighted, if at all, either by the universities or by their publishing agents. The texts themselves are viewed in papyrology as being in the public domain once they are first published, and no collection in the world requires scholars to get permission for reprinting a text.)

(g) Checklist of Editions: The printed editions, along with the other main components of the papyrological literature, are listed, along with the abbreviations commonly used for them, in the Checklist of Editions of Greek and Latin Papyri, Ostraca and Tablets, 4th ed. (1992), created by two of the developers of APIS (Bagnall and Oates) with other collaborators. This tool is available on disk and will be included in APIS as a pull-down tool to help users identify abbreviations in the scholarly literature.

8. APIS and digital libraries

APIS obviously does not exist in isolation. The side of APIS aimed at integrated electronic access, which is a pilot project at this point and will come to full fruition only in Phase 2 and beyond, has a context. There are many initiatives underway to digitize information and provide sophisticated access to it. Two of these are underway at partner institutions of APIS, Michigan and Columbia. These are described in Appendixes 14 and 15. These initiatives have developed sophisticated interfaces and extensive experience in dealing with scanned text that has passed through optical character recognition software. Both institutions are committed to using the expertise developed for these larger projects to assist in the work of integrating the components of APIS, and in both institutions APIS will actually be treated as a part of the larger digital library. The present application seeks support only for work that supplements or modifies what is being created as a standard part of these digital library initiatives.

9. Data structure and the user interface

APIS will be built essentially as a hypertext system, with built-in dynamic links among the various components. In this way users will be able to move easily from a text to the image of the original, to the first publication, to bibliography about it, or to its formal catalogue description. A substantial part of the work involved in creation of the system is the development of these hypertext links. Catalogue records will use the Z39.50 protocol.

To provide hypertext links, APIS will adopt the HTML (hypertext markup language) standard, which is developed from the industry standard SGML with the addition of a facility for indicating hypertext links. In this way, the data files will be readily adaptable to new and more flexible interfaces as they are created. The rapid development of WWW access tools like Mosaic and Netscape has made ready access to such hypertext-linked data much easier, and we expect that there will be many new developments in such access tools before APIS is online. APIS may therefore be accessible through more than one software package. The distinction between the data structures and the access software is critical here. The data files will continue to be usable by many different means; for example, separate distribution of the DDBDP will continue for those who find this more convenient, catalogue records will exist in other databases, and so forth.

A major question yet to be resolved is the level of detail to be tagged. It is a relatively simple matter to tag at the document level, so that clicking on a papyrus brings up its image or text. But optimal convenience for scholarly use would provide line, word, or even character-level linking. It is evident that such tagging done manually would be prohibitively expensive, but development work is underway in various places to automate such middle- and lower-level tagging. We expect to benefit from the expertise of the Center for Electronic Texts in the Humanities (Rutgers/Princeton) and from the experience of the Vatican Library Project in this area.[9]

For the purposes of APIS, the major deficiency of the existing tools is their inability to work in the rich range of alphabets represented in the papyri. It will be necessary to add the tools to enhance network navigation tools for Greek, Coptic, and other languages. These tools should be of much wider benefit than APIS, of course; at the moment it is impossible to consult any data base using Greek over the network and be confident that Greek will be understood in the same way by client and server and that it will be properly displayed to the user. Current browsers are also deficient in global boolean searches. For both purposes, project staff at Columbia and Michigan will develop specifications and look for commercial software capable of meeting them. For any need for which no outside source can be found, staff will write the necessary programs. The staff will also develop an APIS router for use on servers. The software developed for APIS should be generally useful for similar projects.

APIS will also enrich this basic software with additional tools as they become available. For example, a program for converting dates in the papyri into julian dates has been created for the Macintosh by Willy Clarysse (Leuven); this could be a tool button in APIS, allowing the user to call it up whenever a date appears and an equivalent is needed. Professor Clarysse has given permission for this use of his work (letter attached as Appendix 18).

10. Stages of Making APIS available

The Duke Papyrus Archive has already brought online part of its catalogue records and images as a prototype of what part of APIS will look like. (Instructions for using this and the Michigan online images can be found in Appendix 17.) Duke plans to continue to enhance this resource (for example, with the Checklist of Editions) while work on APIS is underway.

We anticipate that the first experimental APIS interface will be brought up around the middle of Phase 1, in order to allow participants in the project and other users to comment on its design as this evolves. Because of the uncertainties about the swiftness with which necessary software will become available, it is impossible to predict exactly what will be available when; this is particularly true for non-roman alphabets and for word-level HTML tagging. But by the end of Phase 1 there should be a pilot system containing at a minimum thousands of catalogue records and images plus bibliographic resources; we hope that the full texts in Greek will also be online by that point.

By the end of Phase 2, the full system will be operational, with data being constantly added as they pass through quality control checks and are ready for public release.

11. Plan of Work in Phase 1

The project is conceived as a four-year effort, with an initial application to cover the first two years. In this two-year period, the project will focus on the following:

Year 1: Enhancement of the DDBDP at Duke; cataloguing, conservation and imaging at Columbia, Michigan, Berkeley, and Princeton; completion of cataloguing and conservation at Yale; completion of the BP for 1969-1975; start of work on APIS system software and interface.

Year 2: Continuation of cataloguing, conservation, and imaging work; beginning of imaging work at Yale; first version of interface ready; experimental set of data converted into common format.

The project will be operated under a grant to Columbia University, which will, apart from the work on its own papyrus collection, provide programming and management services for the electronic systems aspects of the project. The work at Duke, Michigan, Berkeley, and Princeton will be carried out by subcontracts to those institutions from Columbia. Yale is funding its component internally. The work necessary at Brussels, the seventh sponsoring institution, will also be funded locally. The details of what will be done at each U.S. institution are described in the plans of work in Appendixes 1-6.[10]

A central concern of any project like this must be quality control. Each component of APIS brings somewhat different issues of quality control. Our general approach has been to try to insure initial quality as far as possible, but we recognize that no system is proof against slips. Details of how each institution will carry out checks are provided in their work plans, but a few general principles can be set out here.

Text: The DDBDP has been proofread by papyrologists at Michigan, volume by volume, separately from the data entry at Duke. Usage has shown that the error rate in the corrected texts is extremely low.

Bibliography: The BP was proofread at Columbia after data entry, and a second reading has been given it at Brussels, source of the original cards. The Brussels office is responsible for continuing file maintenance and correction, taking advantage also of any corrections reported by users. It will periodically provide corrected versions to APIS.

Images: The principles of quality control for electronic imaging are discussed in Appendix 8. These are in part mechanical and in part a matter of checking of the images by papyrologists for correct identity, tagging, and fidelity to the original shortly after their creation.

Catalogue records: The papyrologists in charge at each institution will check the papyrological content of each record for any evident problems after it is created by the staff papyrologist; it will be checked for conformity to cataloguing rules by the library cataloguer responsible for adding subject headings and completing the US-MARC record.

12. Plan of Work in Phase 2 (Summary)

The second phase will focus on completing the cataloguing, conservation, and imaging at Berkeley, Columbia, and Michigan, continuing imaging at Yale, and testing and refining the user interface. The work required in the papyrus collections is readily visible from the description of Phase 1 in the work plans (Appendixes 1-6); it is the balance of each of these activities, with major effort devoted to cataloguing. The work necessary on the electronic portion of the project is much more difficult to assess at this stage, because it depends on both the progress of Phase 1 and user response to the trial interface. An increasing amount of the work in Phase 2 will probably be involved in responding to user comments and in evaluation.

It is very probable that at Berkeley and Michigan considerable work will remain at the end of Phase 2, because of the exceptionally large size of the collections. After the pilot two-year period, Michigan plans to start cataloging and digitizing gradually the entire corpus of unstudied papyri, ostraca, and other writing surfaces that form part of its holdings. It is hoped that the 100 unstudied papyri included in the two-year pilot, together with the experience gained at partner institutions, will generate enough experience to guarantee successful work on the remaining materials. Because of the bulk of the Michigan collection, it is hard to estimate at this stage a date of completion. It is estimated that the remaining unstudied materials at the University of Michigan are in the range of approximately 7,000 pieces. Phase 2 may deal with about half of this body of material, leaving still a substantial amount for years after 2000.

Phase 2 will also probably see the expansion of APIS beyond the original collections. APIS will seek to enlist additional partners who will add the data about their collections into the system, looking over a period of time to add all appropriate artifacts and texts to the database. Many papyrologists elsewhere have expressed strong interest in APIS, and we believe that the six-collection beginning will provide a decisive impetus for others to adopt the data standards and open systems that we are planning. We are hoping in particular that several European collections can begin adding records and images to the system during Phase 2, and negotiations to this end will be starting in the coming year.

13. Availability of Preserved Materials

Two quite separate, though related, questions must be addressed here. The first is the availability of the original papyri etc. Most collections have been extremely protective of the originals, and the APIS partners are among them. In general, access is restricted to those with appropriate professional training and sufficient knowledge of the materials to use them without damaging them. Library and faculty curatorial staff generally share the responsibility of controlling access. These policies are not likely to change, although proper glassing of unglassed material should make future use of the papyri by students and others less damaging to the original artifacts. In any case, what is absolutely certain is that different institutions will have different policies about access. Current policies will themselves no doubt evolve, but as additional institutions become part of APIS their policies will also have to be respected. It is no part of our intention to impose one particular policy on the entire world of papyrus collections. The system will therefore build in the capability of providing different levels of access on a document-by-document basis.

Access to the images of the papyri, however, is a different matter. Up to now, most collections have tended to provide photographs of unpublished papyri only to those to whom they would give permission to use the originals, and only ability to be physically present at a collection site has distinguished these groups of users. The electronic image availability over the Internet will dramatically change the situation for both published and unpublished texts. Published texts will now be freely accessible via image, far more freely than constraints of physical safety of the objects would allow the originals to be. Unpublished texts, which have for the most part been both inaccessible and even unknown to all except collection staff, will now be readily available as well. Although individual collections may for a time wish to reserve publication rights to particular texts for students, faculty, or other scholars, the vast bulk - and in some cases all - of these images will nonetheless be available to all who wish to consult them. In this way, APIS will bring about a dramatic transformation of the accessibilty of these manuscripts.

Most existing electronic tools in classical studies are distributed on CD-ROMs or floppy disks and used on local microcomputers. The scale of APIS is unlikely to permit this mode of access in the near term for the entire system, particularly given the volume of stored images and the patchwork of access regulations governing them. The primary method of access will therefore be over the Internet. As with various recently-developed WWW tools, usage will require client software on the user's workstation for some operations, particularly in manipulating images. Users will need workstations of significant power, but these will be widely available at reasonable prices before APIS is operational. As with other tools, versions of the software will be made available for the Macintosh (in its PowerMac version), X-Windows workstations, and high-end Windows systems. Full use of APIS is likely in the near term to require direct network connection by the users, but software tools for access to the WWW are changing so rapidly that specifying any particular one here would be pointless. Just during the development of this proposal the range of tools available has changed dramatically. As the ability to transfer large volumes of data over telephone or cable lines (now uncommon and expensive) grows and becomes more affordable, home access to the full system should become a reality.

A distributed form of use will also be readily possible because of the capabilities of the newer WWW tools to extract subsets of the database for use on local machines. A researcher going into the field with an archaeological expedition, for example, could choose from the database all papyri and ostraca of the third and fourth centuries, storing images only of a sample useful for palaeographical analysis, thus creating a personalized database suited to the most immediate needs while away from network access.

Broadening public and educational access to the papyri will require some specific steps beyond those that will provide availability to scholars. We will create a home page designed for general users, to provide a less technical route into the resources of APIS. In its first stages such a home page will of necessity be experimental, for none of us has much experience with what is required and what is possible in this domain. All usage on this home page will be logged and studied (cf. below, section 15) to help guide planning for Phase 2 and beyond.

14. Storage of Preserved Materials

Here again there is a dual question: the artifacts and their intellectual contents. Where photographic negatives are created (of larger pieces), these will be placed in the National Underground Storage facility;[11] copy negatives will be maintained in the institutions. For the rest, the contents will be electronic files of considerable bulk. These will be backed up in the originating institution using standard academic computing procedures and in the same manner that the institution backs up similar materials like the library catalogue. Further protection will be provided by distribution of copies of the data to each partner institution (initially on tape, but a new medium may be adopted by unanimous agreement as technology develops).

The partners in APIS are indeed aware of the need to be certain that these large electronic files are not left behind when new generations of mass storage devices are adopted. It is impossible to predict such developments with any accuracy. Rather, the six institutions will provide for this migration through two mechanisms: first, treating the contents of APIS like other critical components of the institution's digital existence; and second, continuing consultation among the partners (under the aegis of the American Society of Papyrologists) to ensure compatibility among the institutions.

The original papyri will remain in the pertinent sections of the six university libraries.

(a) Michigan has recently built a highly sophisticated separate storage chamber for the papyrus collection. This is described in Appendix 12.

(b) Duke's papyri are kept in the vaults of the Special Collections Library.

(c) Columbia's papyri are kept in the vaults of the Special Collections Library. The ostraca are kept in drawers of metal cabinets and in boxes. The areas in question will be air-conditioned as part of the planned renovation of Butler Library, the first phases of which will begin in summer, 1995.

(d) Yale's papyrus collection is housed in a climate-controlled vault in the Beinecke Library with humidity at 50 percent and temperature at 20 degrees C (both considered the ideal levels for papyrus) and protected by a non-aqueous fire suppression system. The mounted items are shelved horizontally in acid-free folders within archival boxes.

(e) Berkeley: The Bancroft Library is in the process of creating new vault space within its present building. Accommodation will be made in the most secure area of the vault for housing the papyri. Details of the plans for the use of this space are given in the Berkeley work plan (Appendix 1).

(f) Princeton has a climate-controlled storage chamber in Firestone Library, described in Appendix 12, in which the papyri will be stored after conservation work is completed.

15. Research

Although APIS is primarily concerned with carrying out actual work in preservation and access, it will also have a research component, which will mostly develop in Phase 2. There is, after all, much we do not know as we embark on APIS. We have no idea, for example, how much electronic access in a hypertext system will change user demand for particular types of information. We cannot at this point tell what curve of user demand over the network to anticipate. A particular point of interest will be the structure of individual work sessions, which will tell us much about how patterns of scholarly investigation are changed by the availability of APIS and similar systems. APIS will benefit from the research and evaluation capability established for the digital libraries to assess these questions. The results of this evaluation will then help provide more efficient access to APIS in the future.

A special part of this effort will be the logging and study of public usage of the generalist home page briefly described in section 13 above. We hope from this to learn what types of access work best for the non-specialist user of the papyri and how to broaden our efforts to make the material more available outside scholarly circles.

16. Project Management and Staffing

The project was initially developed by a group of scholars convened by the Project Director in his role as President of the American Society of Papyrologists. Ludwig Koenen, the Herbert C. Youtie Professor of Papyrology and University Professor at Michigan, and John F. Oates, Professor of Classical Studies at Duke, are the other two senior members of the team. In addition the ASP group included Peter van Minnen (Duke) and Traianos Gagos (Michigan), expert and widely-published papyrologists serving in staff positions with the papyrus collections at their institutions. Subsequent phases brought a wide range of rare book and manuscript curators, preservation librarians, library systems staff, academic information systems specialists, library management, and conservators into the discussions. The present application reflects this broad involvement of the various parties involved; the work plans in particular list these personnel and describe their duties.

We are under no illusion that managing a six-institution project will be simple, particularly without a high-overhead central institutional management. Columbia University will act as project manager, with the funds for the other institutions provided through subcontracts. Direct central intervention in local operations will thus be minimized. This practice is in accordance with the fundamental philosophy of APIS, which is to ensure compatibility and interoperability of results but not to micromanage local routes to achieving these. The elements of our consortial arrangements are as follows:

(1) A Steering Committee composed of two representatives from each partner institution. This body will communicate in the main by electronic means, but the budget provides for a meeting twice each year. Our experience in developing the project is that such face-to-face meetings are necessary to deal with common issues. A list of its members is given in Appendix 19.

(2) An electronic list of all parties involved in the project, by means of which communications can reach not only those on the Steering Committee but all others concerned. This list is already in existence (l-apis@umich.edu) and has facilitated the study on imaging and the preparation of this proposal greatly. It is our main means of internal communication on matters of general interest.

(3) A Steering Committee in each institution, responsible for coordinating work locally. These will be responsible, working with the local project director, for ensuring that the individual subcontracts are carried out.

(4) An Advisory Committee of persons not otherwise connected to the project but whose expertise in areas central to its work will be called upon, mainly by electronic consultations. A preliminary list of some persons we will invite to be part of the Advisory Committee is in Appendix 19, but we expect to expand it as work proceeds.

Details of staffing at the individual institutions are given in the work plans (Appendixes 1-6).