A Collaborative Workflow for theDigitization of Unique Materialsby Gretchen Gueguen and Ann M. HanlonAvailable online 18 August 2009

This paper examines the experience of oneinstitution, the University of Maryland

Libraries, as it made organizational effortsto harness existing workflows and to

capture digitization done in the course ofresponding to patron requests. By

examining the way this organizationadjusted its existing workflows to put in

place more systematic methods for digitalcapture of unique collections, the authorshope to provide insight into the benefitsand pitfalls of one model for scaling up

digitization.

Gretchen Gueguen, Head, Digital Collections,East Carolina University, Greenville,

North Carolina, USA<guegueng@ecu.edu>;

Ann M. Hanlon, Digital Projects Librarian, Marquette University,Milwaukee, Wisconsin 53233, USA

<ann.hanlon@marquette.edu>.

468 The Journal of Academic Librarianship Volume 35, Number 5, pages

INTRODUCTIONA number of major research universities have under-taken mass digitization of their book collections,including efforts associated with such well-knownprojects as the Google Book Search Project and theOpen Content Alliance. So it is not surprising that callsfor the “mass” digitization of our special collectionsmaterials have followed. Indeed, prominent players inthe library world, including OCLC and the Council onLibrary and Information Research (CLIR), have arguedfor the need to scale up digitization efforts in order tomove from project-based digitization to more systema-tic programs focused on enabling deep research ofheretofore hidden or geographically inaccessible (forsome) collections.1,2 But libraries face the basic chal-lenge of how to scale up in themidst of already strappedbudgets and overburdened organizations. Given ourexisting workforce and workflows, how canwe begin tomake our unique materials more systematically avail-able online? This paper will examine the experience ofone institution, the University of Maryland Libraries, asit made organizational efforts to harness existingworkflows and to capture digitization done in thecourse of responding to patron requests. By examiningthe way this organization adjusted its existing work-flows to put in place more systematic methods fordigital capture of unique collections, the authors hopeto provide insight into the benefits and pitfalls of onemodel for scaling up digitization.

“prominent players in the library world...have argued for the need to scale up

digitization efforts in order to move fromproject-based digitization to more

systematic programs focused on enablingdeep research of heretofore hidden

or geographically inaccessible(for some) collections.”

468–474

LITERATURE REVIEW

Several recent articles have suggested methods to scaleup digitization. Much of the focus in this literature hasbeen centered on “mass digitization” projects such asGoogle Book and the Open Content Alliance. KarenCoyle, in a 2006 overview of such projects, points outthat “mass” projects have different qualities fromprevious “large scale” projects. The uniformity of thebook format has made it possible to automate much ofthe digitization, increasing the scale to that of entirecollections. However, Coyle wisely points out that thereare two fallacies in themass digitizationmodel. The firstis the assumption that all books are suited to thismethod, no matter how fragile or uniquely formatted.The second is the assumption that time and money willbe saved by digitizing materials only once and makingthe subsequent digital copy universally accessible. Theuniversal accessibility of the digital copy, particularlywith regard to fragile materials, would also be a boon tonon-book materials. However, the fragile and idiosyn-cratic nature of special collections and archivalmaterialsmake automation much more problematic. Humanintervention is likely to be necessary at the item levelin nearly every case, making it difficult to move beyond“boutique” digitization projects.3

This boutique model, and the hurdle it presents tothe systematic digitization of special collections mate-rials, is likewise favored by funding models based ongrants that focus on digitizing a specific body ofmaterials selected to meet grant guidelines. Howeverwell-designed these guidelines might have been forselection, the limits they implymean that only a portionof any collection can be digitized in this manner withina grant's timeframe and budget. Thus, while grantfunding can be an excellent means to establishimportant digital collections, it cannot be a fundamen-tal part of a digitization program. As Daniel Greensteinand Suzanne E. Thorin (2002) write:

Many believe that as the digital library becomes libraryinfrastructure, the financial resources needed to maintain itwill come from numerous budget lines rather than from oneline that is earmarked for digitization. In the adult digitallibrary, electronic resources will be acquired from generalcollection budgets, and digital preservation activities will besupported with general preservation funds.4

Laurie Lopatin (2006) notes that while movementtoward sustainable funding has been seen in somequarters (she cites a 2005 survey of libraries in the NewYork City area inwhich 51% of respondents reported theirbudgets were funded internally), many others reported acontinual search for new funding to begin and sustainprojects. The high profile of themass digitizationprojectsalready noted further muddies the waters.5 As NicholasJoint (2008) points out, when Google Books sounds like “‘a 110 million pound scholarly digital library available forfree,’ administrators may think: why ever spend anotherpenny on your local library?”6 While Joint is primarilyconcerned with scholarly open access projects, the fightto gain recognition for the extraordinary effort put intodigital library development remains the same.

Along with inconsistent funding, systematic digitiza-tion initiatives may be harmed by a lack of internalorganizational support. Boock and Vondracek (2006)conducted a survey of 40 ARL libraries and found that38 of them (95%) had engaged in digitization. Of these,84% were found to rely on “cross-departmental projectgroups” in these efforts. That is, the bulk of libraries thatare making digitization happen are those that are ableto leverage the expertise of their larger institution.7

Although new units and new positions may be createdin support of these initiatives, using the existingstrengths of the organization appears to be the mostviable strategy to adapt to changing needs.

With specific regard to special collections, RickyErway and Jennifer Schaffner's 2007 report for OCLCPrograms and Research attempts to distill the senti-ments and discussions of the “Digitization Matters”forum attended by two hundred directors, adminis-trators and curators of special collections in libraries,archives and museums. The report argues “that largequantities of digitized special collections materials willbetter serve our users,” and that we should therefore“optimize procedures primarily for access.” The reportdoes not call for librarians to abandon standards andbest practices for digitization. But it does call for betterdecision-making. Erway and Schaffner ask whether thisis a viable standard for special collections — does everyitem we digitize need to be treated as though it cannotor will never be digitized again? Is it possible to digitizefor access and assume that the opportunity to digitizefor preservation still lies ahead? And as for descriptionof special collections materials, a major impediment tothe mass digitization of those materials, might there beroom in the item-level world of special collectionsdigitization for group-level description and collection-level decision-making?8

Finally, a recent CLIR report on “Reconceiving ResearchLibraries for the 21st Century” (2008) calls “for moreaggressive intervention to better structure and managethe challengeswe face.”Drawingupon theproceedingsofa symposium featuring leaders in the field of digitallibraries, the report argues for rethinking what weconventionally consider to be “fringe activities,” such asmetadata building and digitization, and reprioritizingsuch activities as core investments that we need to makein order to “make material available to the scholarlycommunity in a systematic way.”9 Shifting our basicorientation in this way is no small task. But the authors ofthe current case study hope our efforts serve as oneexample of the processes by which libraries might beginto organize for the systematic digitization of uniquematerials in special collections and other holdings.

The University of Maryland digitization programsheds some light on how nearly all of these suggestionsmight be applied. The project systematized digitizationby implementing a policy to deposit all digitization donefor patron requests into a newly created digital reposi-tory. This policy had far-reaching implications. And itprovoked many new questions: how would this newpolicy affect the daily digitization workflow; how couldthe scope of the digital collections be defined if we werecollecting the arbitrary digitization requests of patrons;

September 2009 469

and how could we adapt our standards and bestpractices to accommodate this new workflow withoutoverburdening the special collections and digital collec-tions staff affected? The following case study exploreshow some of these questions were answered and howthe University of Maryland, as an organization, adaptedworkflow and policy to meet the goal of capturing thisexisting digitization workflow in order to implementmore systematic digitization efforts.

CASE STUDY: THE UNIVERSITY OF MARYLAND'SDIGITIZATION WORKFLOW

In December 2004, the UM Libraries established theOffice of Digital Collections and Research (DCR), tocoordinate and plan for digital initiatives, and todevelop and manage a central digital repository (usingthe Fedora architecture) to house digitized objects fromacross the UM Libraries' departments. The repositorywould limit the re-scanning of frequently requestedmaterial and at the same time repurpose those scans foronline digital collections. The repository was to bepopulated with materials created from patron requests,particularly those generated by the Department ofArchives and Manuscripts, along with any materialsdigitized as part of other digitization projects.

“the [University of Maryland] Librariesestablished the Office of Digital Collectionsand Research... to develop and manage acentral digital repository... populated withmaterials created from patron requests.”

As DCR began the task of coordinating efforts to createthe repository, the patron scanningworkflow in Archivesand Manuscripts was growing, particularly due to effortsto document the University's history for its 150thbirthday celebration. With the increase in patron andexhibit scanning, joined with the significant timerequired to program, design, and develop the Fedora-based repository, an urgent need emerged to create atleast a stop-gap measure to capture and track the scansbeing created. In response, a Project Archivist hired toassist with digital image management and the Curatorfor Historical Manuscripts, working in close consultationwith DCR, developed a “scanning database.” This was aMicrosoft Access database with fields that, whencompleted by staff and students in the course of scanningmaterials, wouldmap directly into the repository's newlydeveloped XML metadata schema. Scanning would bedone according to specifications published by DCR. A file-naming scheme was added, and a dedicated directorywas created for saving new digital images. The databaseof metadata, along with the directory of images, was tobe migrated over when the repository infrastructure wasfinished. At that point, a web-based administrativeinterface would be launched, giving staff and studentsthe ability to upload objects and metadata records

470 The Journal of Academic Librarianship

directly with sophisticated tools for handling metadataand rights management.

Stumbling BlocksAs with any digital initiative, the goals and aims of the

repository project changed over time. In some cases thislearning process required going back to the drawingboard and starting over. But in the case of UM'sdigitization workflow project, staff continued to add tothe scanning database in anticipation of the completeddigital repository. Thus, the digitization program alreadyunderway had to be robust enough to adapt to changingpolicies and the repository had to be flexible enough toaccommodate legacy data. These issues necessitatedanswering the following questions: how to create qualitydigital objects, how to handle the scale of the operation,and how to present this mixture of materials online in away that made sense to a diverse audience.

Quality Digital ObjectsBuilding a repository while simultaneously populat-

ing it led to certain obvious difficulties. First, changing astandard — for example, requiring images to be createdwith a 24-bit color profile rather than a 48-bit one —meant rendering potentially “unacceptable” thousandsof images created up to that point, not to mentionthousands of work hours. Second, given the distributedworkforce, day-to-day decisions about standards andpractices were not easily communicated or implemen-ted. The range of archivists, curators, librarians, andstudent assistants participating in this project shared anuneven awareness of current digital standards andtechnology. While many were willing to learn, accu-rately communicating a message to a diverse anddistributed group was a difficult challenge. Finally, thedigitization program was not necessarily the toppriority of archives and special collections departmentsdealing with the more immediate pressures of dailypatron requests, reference questions, processing back-logs, exhibit building and fundraising.

Several methods were employed to resolve some ofthe inherent complications of the project. The first wasthe creation of in-house documentation and standards.This documentation ranged from statements of missionand responsibility and simple guidelines codifyingbenchmarks for digital output, to more in-depthexplanations of “input referred” scanning techniquesand step-by-step instructions for typical practices inwhich staff might need to engage, such as resizing animage, using a histogram to evaluate target aim points,or sharpening an image.

These efforts were supported by a series of work-shops and trainings organized to increase personalcommunication among staff in Archives and Manu-scripts and DCR. In addition, quality control procedureswere devised to balance responsibilities among thestaff. Curators would be responsible for regularlyreviewing the metadata records created by graduateassistants to insure against items piling up. Curatorswould ensure a consistent level of quality control bychecking to see that item records were completely filled

out and the information was basically correct. DCR staffwere to follow up by checking a statistically significantportion of these records for style and consistency inmetadata and image quality. This system played into thestrengths of those involved: the curatorial staff's abilityto verify the correctness of the information, the DCRstaff's familiarity with technical standards. Despite aninitial reluctance to interfere with existing scanningoperations in the Archives and Manuscripts depart-ments, DCR eventually stepped in to fill the role oftrainer and project manager.

The creation of documentation and standards pro-vided much needed limitations for the image creationprocess. The standards removed the necessity forindividual decision-making about digitizing items,synthesizing the wide range of possible color profiles,resolutions, and post-processes into amoremanageablerange of “if–then” scenarios. By choosing a standardthat would be acceptable in most cases (such as arelatively high spatial resolution, or an RGB colorprofile), context became irrelevant and the workflowprocess was streamlined. In addition, the organizationagreed early on to accept into the repository legacy datathat did not meet the current standards. This meantthat the Library had to accept the possibility thatimages might be rescanned in the future if a higherquality version was needed. However, we were guidedby the growing realization in the field that accesstrumps preservation and “digitize once” may not be afoolproof plan.

Scale and PresentationPerhaps the largest hurdle for the project was the

scale of the endeavor. Although close to 3000 imageshave been entered into the repository through thismethod as of January 2009, more than 4000 are stillbacklogged, many with only preliminary metadatarecords. That may seem a small number relative toother digital initiatives; but is significant given thatthese materials were all “captured” from existingwork — a repository built in the interstices betweenmeetings, processing, desk work, and the day-to-dayactivities of a typical librarian or archivist. Mostimpressive, these 7000 images represent an archivethat no one person had curated, collected, or plannedfor; a wildly diverse collection that was, in a sense,found on the doorstep.

If the repository had been finished, and the web-based administrative interface made available, it ispossible that many of those images in the backlogwould now be online. However, a relatively robustmetadata standard, designed so that records could beeasily repurposed and shared, added a significantburden to the existing scanning workflow. Added tothe robustness of the metadata was the volume ofscanning requests, often so dense that there was littletime left over for metadata — staff were more inclinedto be preparing their next item for scanning, rather thancreating a metadata record.

To address the problems of scale, one solution mighthave been to divide the labor for creating individualmetadata records by assigning initial basic descriptive

information (a title, a creator if applicable, and adescription) to an image when it was scanned. Oncethe imagewas digitized, the “stub record”would go intothe repository with the image. Then catalogers from theTechnical Services department would go through andaugment these records with more detail and controlledsubject headings. In this way, items would not sit in abacklog far from the public view, but would be availablewith some basic metadata even before a fuller descrip-tion could be created. In addition, the curator's knowl-edge of the collection could be harnessed, but withoutasking that curator to acquire the skills of a catalogeralong with those of an archivist. Although suchprocesses were not part of the original plans forworkflow at the UM Libraries, one of the authors hassuccessfully implemented such a workflow for digitalcollections at East Carolina University.

Aside from scale, issues of collection-scope proved aconcern with this project, and ultimately provokedamong administrators a desire for stricter guidelinesconcerning what was to be captured and stored in therepository. Patrons tend to request a predominance ofimages of sports events and sports figures — certainly apart of the University of Maryland's history, but not theonly part that should be highlighted. It might be arguedthat the unplanned bias of this collection accuratelyrepresents the most-used parts of our archive; it couldalso be argued that the Libraries have a responsibility toprovide materials for all forms of research and inquiry,not just those that present themselves most often.While the project had originally been designed inresponse to the need for an image management systemand a hope to avoid the repeated scanning of the mostpopular requests that might come from restrictingcontent, administrators also argued for the benefit ofhaving the organization spend its time and resources ongetting the best materials online first.

Added to the concern about sports-centric contentwasa concern that the lack of an overall selection focus for thethousands of captured images represented a problematicdeparture from the way that other digital collectionswere created at UM. Indeed, the original concept behindDCR was that digitized objects would be created in“collections.” As with traditional archival arrangement,these collections of similar material would be presentedtogether for researchers to examine as a group. But itemsscanned as a result of a patron request belong to nosingle collection. Presenting this vast sampling of ourholdings online and through an interface that wouldgive users some context was a challenge.

“But items scanned as a result of a patronrequest belong to no single collection.Presenting this vast sampling of our

holdings online and through an interfacethat would give users some context

was a challenge.”

September 2009 471

It might be argued that the conventional idea of“digital collections” is itself inherently limiting andpotentially outdated. Relying on the “first order ofinformation” concept described by Michael Weinbergerin Everything is Miscellaneous, the traditional under-standing of the “collection” relies on the idea (and evennecessity) that things belong in one particular place andone place only.10 But the realities of digital access makethat unnecessary. Moreover, it is unlikely that manyusers are arriving at our digital libraries through the“front door” and browsing through our carefully craftedcollections as we intend. Instead, they are findingindividual objects through search engines. As internetsearching statistics show, and numerous usabilitystudies and library web analytics confirm, users lookfor information using search engines. The Pew Internetand American Life project reported in 2008 that thenumber of individuals using a search engine daily is justunder 50% and is above 60% for certain demographicslike college graduates.11

Given these statistics, it might well be asked: whyput digital objects into collections at all? In answer, itcan be noted that, even if most users find content on theweb through search engines they might still find usefulinformation in the relationships between objects thatcollections can provide. Taking that notion further, itmight be argued that objects may be part of manydifferent “collections” based on their diverse qualities.For example, a 19th century work on agriculturepublished at the University might belong to collectionson the history of agriculture, the history of theUniversity, the bookshelf of a noted agrarian, or anumber of other topics.

So the problem with the UM project, then, was notthat the digitized material collected in response to userrequests fit into no collection, but that with itemsselected from across the institution's holdings possiblecollections were too numerous to define. With morematerials added every day, the difficulty was in trying tologically group items when there was no idea if whatwas added in the next day, week, month or year wouldchange the scope of the online materials.

“In response to the problems with scale andpresentation of the materials, a collectiondevelopment policy of sorts was created in

late 2007 requiring that all content fitinto one of 18 broad and browse-able

subject categories.”

In response to the problems with scale and presenta-tion of the materials, a collection development policy ofsorts was created in late 2007 requiring that all contentfit into one of 18 broad and browse-able subjectcategories. This policy was developed by a representa-tive team of staff members from Archives and Manu-scripts and DCR. The subject categories would not limitthe creation of new collections should they arise in

472 The Journal of Academic Librarianship

response to other needs. But curators were asked tokeep these collecting areas in mind when adding digitalobjects to the repository. When materials fit the guide-lines of the policy, a metadata record was to be createdand the object added to the online collection. If the itemfell outside of the guidelines, it could be simply scannedand deleted.

Abandoning the idea of attaching every digitizedobject to a unique collection was a move towards whatWeinberger has described as the “third order ofinformation,” in which materials are not grouped atall, but retain multiple, not pre-determined qualities.12

These qualities, like the broad subjects, can be searchedand aggregated into groups of relevant results and canpromote serendipitous discovery. For example, a searchfor images on “kindergarten” might lead a user back toan “Education” collection, which could potentially leadto many more relevant images. With the rich assort-ment of potentially useful subjects, themes and inter-esting content hidden beneath the repository interface,other discovery methods were discussed that couldutilize emergent web 2.0 and data visualization techni-ques, such as tag clouds and hyperlinked terms inmetadata records.

OutcomesAs a result of this approach more than 7000 images

have been created and either ingested into therepository or await ingest in the scanning database.Out of that 7000, 1200 items have been selected forinclusion in two thematic collections. The single biggestbeneficiary of this approach, in terms of sheer numbers,was the University Archives. That department, whichnormally receives the most requests for scans ofmaterials, was also in the midst of the publicitycampaign for the University's 150th anniversary. Aglossy coffee table book and a full-length documentarywere two of the major projects undertaken, and bothrelied heavily on scans of images and documents fromthe University Archives. With the addition of imagesthat had been scanned and saved prior to the beginningof this project, approximately 2000 images weredocumented in this manner and were ultimatelyadded to the digital repository to form a still-growingcollection called University AlbUM bhttp://www.lib.umd.edu/digital/album.jspN.

Another important set of images captured in thismanner was a collection of postcards held by theNational Trust for Historic Preservation Library Collec-tion housed at University of Maryland. Thousands ofpostcards documenting historic buildings, destinations,and important architectural styles proved to be popularrequests from patrons. In addition to capturing theserequests, the librarian in charge of this collection decidedto fill in some of the gaps. She set out to systematicallydigitize the collection and to use the scanning databaseto capture metadata for future ingest into the digitalrepository. That effort is now publicly available as theNational Trust Library Historic Postcard Collectionbhttp://www.lib.umd.edu/digital/ntlpostcards.jspN.Although much smaller than the University AlbUM, theonline collection represents only a small portion of the

digitized postcards, which will be added regularly to theonline collection.

DISCUSSION: WHAT CAN BE LEARNED FROM THIS CASE STUDYLibrarianship, and certainly curatorship, does notnaturally gravitate toward ceding control over anyaspect of collections. However, giving up some controlover digital selection at the University of MarylandLibraries created amore efficient path to building digitalcollections by capturing and supplementing an existingworkflow. By involving people across the organizationand not just those identified as part of the “digital”department, production increased. By distributing the“burdens” (and the satisfactions that come frombuilding a publicly accessible collection), a digitalcollection was created that was larger and more diversethan one requiring the careful selection of eachdigitized item. Capturing the existing workflow frompatron requests meant building an ostensibly neutralcollection. Nothing is ever really without interpretationor bias, of course; and nothing could highlight that factmore clearly than the very pronounced bias towardsports in the University AlbUM collection. But thisconcept and practice of “neutral collection-building,” asopposed to building a collection based on curatorselection, enables a collection to capture items thathave built-in value for users and potential users. In theend, that collectionwill reflect the everyday and heavilyused holdings rather than the jewels in the crown. Thedevelopment of digital collections at the University ofMaryland became less of a “trophy” service and, instead,began to build toward the critical mass of online,original research content that will enable our digitalcollections to be a truly valuable part of how research isdone in the 21st century.

“giving up some control over digitalselection at the University of Maryland

Libraries created a more efficient path tobuilding digital collections by capturing

and supplementing an existing workflow.”

This neutral collection-building requires a differentfocus of concentration, however. As the Marylandexample shows, rethinking our current paradigms forpackaging and presenting information is key to thesuccess of initiatives like this one. The inheritedmuseum model of the earlier part of this new centuryrelied on creating “exhibit-style” digital collections thatprovided large amounts of context to guide usersthrough the carefully shaped narrative of a givencollection. The University of Maryland's intention fromthe beginning of the repository project was to breakthat mold and focus instead on access to many moreimages, in many more ways, in order to allow theresearcher to build their own context and connections,just as they do in their current research in the library's

archives and special collections. Truly providing accessat this level requires trying new methods to bridge thegap between repository and user. It seems counter tothis line of thinking to insist that this type of under-taking also requires the creation of clear policies aboutwhat will and will not be done, but the Marylandinitiative might have been buried under a mountain ofunreasonable demands if limitations were not devel-oped. These limitations turned out to be advantageousas they offered the opportunity, once again, to thinkabout presentation and collection-building.

Finally, it's worth noting that not only digitization,but problem-solving was distributed in the Universityof Marylandmodel. Many of the ideas to solve particularworkflow problems — such as stub records, minimizingcollections and using a broad vocabulary of subjects —these ideas came about because the “problems”weren'tjust owned by DCR, but rather by everyone involved inthe project. The meeting of minds between archivistsand digital collection librarians is a good example of theways that digitization can benefit from the input andstrategic planning of the entire institution.

By focusing on ways to streamline the process ofbuilding digital collections, and building upon theexisting workflows and expertise of the organizationas is possible and effective, digital collection-buildingcan become a core function of the library, and digitalcollections can begin to build to a critical mass, so thatresearchers can come to the web to conduct systematicoriginal research using digitized primary sources.

Perhaps the overarching challenge in this endeavor isthat digitization is still not considered a core function ofmost libraries' missions. Even though it may be stated innewmission statements, very little has really been donein most libraries to organize around digitization. But inorder to open up our collections to new and excitingforms of scholarship, the digitization of our uniquematerials must become more central to library opera-tions. The model the authors pursued at the Universityof Maryland Libraries may point toward at least onemethod for moving digitization to the core of Libraryoperations by tapping into existing resources. It shouldnot be the final step, but it can be the first.

NOTES AND REFERENCES

1. Ricky Erway and Jennifer Schaffner. Shifting Gears: GearingUp to Get into the Flow. Online. Report produced by OCLCPrograms and Research. 2007. Available: http://www.oclc.org/programs/publications/reports/2007-02.pdf (December17, 2008).

2. Council on Library and Information Resources (CLIR), NoBrief Candle: Reconceiving Research Libraries for the 21stCentury. Report produced by the Council on Library andInformation Resources, 2008. Available: http://www.clir.org/pubs/reports/pub142/pub142.pdf (February 5, 2009).

3. Karen Coyle, “Mass Digitization of Books,” The Journal ofAcademic Librarianship 32 (November 2006):641–645.

4. Daniel Greenstein and Susan E. Thorin, The Digital Library: ABiography. Online. Report produced by the Council onLibrary and Information Resources, 2002. Available: http://www.clir.org/pubs/reports/pub109/contents.html (Febru-ary 5, 2009).

September 2009 473

5. Laurie Lopatin, “Library Digitization Projects, Issues andGuidelines: A Survey of the Literature,” Library Hi Tech 24(2006): 273–289.

6. Nicholas Joint, “It Is Not All Free on the Web: Advocacy forLibrary Funding in the Digital Age,” Library Review 57(2008): 270–275.

7.M. Boock and Ruth Vondracek, “Organizing for Digitization:A Survey,” Portal 6 (April 2006): 197–217.

474 The Journal of Academic Librarianship

8. Erway and Schaffner, “Shifting Gears”.9. CLIR, “No Brief Candle”.10. David Weinberger, Everything is Miscellaneous (New York:

Times Books, 2007), p. 17–18.11. Deborah Fallows, Pew Internet and American Life Project:

Search Engine Use. Online. Available: http://www.pewinternet.org/pdfs/PIP_Search_Aug08.pdf (February 5, 2009).

12.Weinberger, Everything is Miscellaneous, p. 19.