Building Preservation 211

Building Preservation: The University of Illinois at Urbana-Champaign’s Stacks Assessment

Thomas H. Teper and Stephanie S. Atkins

This article discusses the results of two collection assessments conducted at the University of Illinois at Urbana-Champaign. The library’s Preserva- tion Committee completed the first assessment and reported its results in College & Research Libraries in 1989. The second assessment was com- pleted in 2002 and accompanies the institution’s initiative to establish the library’s first centralized preservation and conservation program. Both as- sessments focused on the central stacks collection, a repository collection serving the institution’s forty-two branch and departmental libraries. Although a reanalysis of the first assessment’s data was impossible, the authors attempted to draw comparisons between the two assessments’ results. Af- ter thirteen years without a preservation program and without any signifi- cant facilities improvements, the results provide insight into the results of deferred collections care and facilities maintenance and offer guidance for conducting similar studies with other research library collections.

ince the publication of the sur- than fifty years.2 Written thirty years af- vey results from Stanford ter Barrow published his research, the University’s Green Library in authors in Illinois indicated that the real- 1982, the library preservation ity of this prediction was becoming “pain- community has come to rely on collection fully obvious” to librarians and other cul- assessments as a means of learning about tural resource managers.3 Indeed, the the physical state of library and archival 1989 survey’s results indicated that 37 collections.1 In 1989, the University of Il- percent of the collection suffered from linois at Urbana-Champaign (UIUC) pub- serious deterioration, 33.6 percent from lished the results of its own assessment— moderate deterioration, and 29.4 percent Library Collection Deterioration: A Study of from no significant deterioration.4 the University of Illinois at Urbana- Although an assessment does not pro- Champaign. In the introduction, the au- vide concrete data about each and every thors cited William Barrow’s assertion volume, a properly planned and con- that many of the twentieth century’s pub- ducted assessment provides a preserva- lished books may have a shelf life of less tion administrator with the means to de-

Thomas H. Teper is the Head of Preservation at the University of Illinois at Urbana-Champaign; e-mail: [email protected]. Stephanie S. Atkins is the Assistant Circulation and Bookstacks Librarian at the Uni- versity of Illinois at Urbana-Champaign; e-mail: [email protected]. The authors wish to thank and ac- knowledge the Research and Publication Committee of the University of Illinois at Urbana-Champaign Library, which provided support for the completion of this research.

211 212 College & Research Libraries May 2003 velop a better understanding of an reality that became painfully obvious to institution’s collections. Indeed, operat- many of the institution’s librarians in 1989 ing under the assumption that preserva- has been compounded by thirteen years tion funds must be targeted at the areas of additional acquisition and deteriora- of greatest need, the value of an assess- tion. At the time of the 1989 assessment, ment in the initial planning of preserva- the central stacks collection held roughly tion activities and those expenditures is 5.3 million volumes. Thirteen years and still of paramount importance for maxi- almost 500,000 additional volumes later, mizing expenditures. Brian J. Baird and the current assessment evaluates possible colleagues recently reiterated this point comparisons between the two data pools in a report on surveys conducted at the and quantifies the collection’s physical University of Kansas: condition as the library launches a new preservation and conservation program. One of the greatest challenges fac- ing academic and research libraries UIUC Library and Its Collections today is the preservation of collec- The University of Illinois at Urbana- tions. The majority of materials Champaign library is nationally and in- housed in research libraries are ternationally renowned for the depth and unique and irreplaceable. Unfortu- breadth of its collections. Its specialized nately, the number of these items collections are dispersed among forty-two needing treatment far exceeds the departmental libraries, more than twenty availability of resources. Therefore, of which are located within the main li- libraries must carefully evaluate brary building. Currently, the library’s their collections and use that infor- total collections number nearly 10 million mation to develop proactive preser- volumes and 22 million pieces. vation plans.5 With approximately 5.8 million vol- umes, the central stacks collection is the Spurred by a growth in professional single largest repository in the campus interest in preservation, the University of library system. As collections of currently Illinois began to investigate various acquired material grow, departmental li- means of addressing preservation needs braries transfer older and seldom-used throughout the library in the late 1970s titles into the stacks. As a result, the stacks and completed a report in 1980. Another contain the most diversified and oldest body of individuals completed a state- circulating materials in the library. wide analysis project in 1986.6 Despite this Unlike many of its peer institutions, preparatory work, preservation within UIUC has maintained a closed stack sys- both the state of Illinois and the Univer- tem, granting access only to faculty, sity of Illinois at Urbana-Champaign’s li- graduate students, and a limited number brary failed to make the headway that it of undergraduates. The result is a circu- did at other institutions, headway that can lation model in which many patrons re- be seen in the publication of reports from quest items, while staff retrieve and re- many institutions completing the Asso- shelve them. ciation of Research Libraries’ Preservation Planning Program and the continued de- The Central Stacks Collection and velopment of preservation programs at Collection Maintenance many of these institutions.7 For many years, the library’s administra- Although preservation programs be- tion focused primarily on amassing its came almost commonplace in many uni- unique and valuable collection more than versities and, to a more limited extent, on the collection’s condition or mainte- colleges, UIUC’s library was unable to nance. With the notable exception of li- develop and maintain a comprehensive brary binding, rare books and special col- preservation program. As a result, the lections received most available Building Preservation 213 preservation and conservation funds. tiple stacks leaks in 2001–2002. In addi- Browsing through the stacks, it is evident tion, poor air circulation, accumulated that the collection is showing signs of sig- dust, and insect infestations have all left nificant deterioration. The stacks office their marks on various parts of the col- staff stabilized materials, when possible, lections. using boxes and pamphlet binders; and The collection’s impressive growth in the library’s small book repair and pam- the second half of the twentieth century phlet-binding unit followed suit when- has created another significant problem— ever requested. However, the number of overcrowding. In many areas, the items needing attention exceeded the collection’s growth has exceeded its staff’s ability to keep pace with the level shelving capacity to such an extent that of need. books are stacked on the floor in aisles. The staff also faces the downside of Presently, the library is in the process of such comprehensive collection building; building a high-density shelving facility the depth and breadth result in large col- to alleviate the overcrowding. In prepa- lections of non-Western material and sig- ration for this, the library’s administra- nificant collections of semi-ephemeral tion imposed a moratorium on all depart- material. Consequently, the stacks hold mental library transfers into the central the bulk of the library’s aging collections stacks collection, a move that alleviated and the most significant holdings of ma- some pressure on the stacks but added terials that continue to be produced on significant pressure to the various branch acidic paper—Eastern European, African, and departmental libraries. and Latin American materials. The stacks’ poor storage conditions The Stacks Collection and the exacerbate the problem of aging collec- Assessment tions and acidic paper.8 Built in 1926, the The central stacks collection consists of original building and most of its subse- monographs, bound periodicals, and a quent additions lack basic air-condition- collection of 82,000 microforms. The mi- ing and humidity controls. In air-condi- croform collection resides in an environ- tioned portions of the stacks, tempera- mentally controlled room and has re- tures generally remain relatively stable cently received some collection despite the lack of operational reheat sys- maintenance by re-housing a significant tems. However, those portions not air- portion of the collection. Special formats conditioned often experience extreme such as audiovisual materials and digital fluctuations in temperature and humid- media are not material types represented ity, with ranges from around 60° to well in the collection. over 80°F and humidity fluctuations be- The assessment’s focus was guided by tween the 30 percent range to over 70 per- that of the 1989 assessment. In keeping cent in various parts of the building. The with the 1989 assessment, specialized col- combination of high ambient temperature lections located within the stacks’ physi- and humidity has resulted in periodic cal space were excluded. For example, the mold outbreaks. Significant, documented stacks’ microform collection was not in- mold blooms affected the central stacks cluded nor were the government docu- collection in 1993 and the Rare Book ments collections or the Asian library. Room’s collection from 1981 to 1987. Similarly, the assessment excluded un- The collection also faces the results of bound periodical titles. This is in keep- decades of campuswide deferred facili- ing with the protocol established by other ties maintenance. Periodic roof and foun- assessments.9 Finally, the assessment ex- dation leaks led to limited flooding and cluded a brittle book backlog of roughly subsequent mold outbreaks, including a 6,000 British and American periodical significant mold bloom in the Rare Book volumes and monographs dating back to Room during the spring of 2001 and mul- the early nineteenth century. 214 College & Research Libraries May 2003

Why an Assessment? stitution with no opportunity to reevalu- Beginning with the publication of ate the original sample. For instance, the Stanford University’s assessment of 1989 assessment defined only three levels Green Library in 1982 and proceeding of condition—good, moderate, and poor— until the present, preservation assess- for the paper, binding, and boards (table ments proved themselves a valuable 1). The new assessment enabled the library method of informing preservation admin- to document the condition of the collec- istrators and collection managers about tion and to begin the process of thoroughly their collections’ condition and enabled analyzing needs. them to plan their preservation program’s development.10 An assessment permits an Project Hypothesis institution to identify needs and priori- To ensure the maximum level of objectiv- ties and provides justification for re- ity, the assistant circulation and sources. bookstacks librarian and the head of pres- The authors chose an assessment us- ervation discussed any preconceptions ing random sampling for four primary they held about the collection based on reasons. First, this enabled some level of their knowledge of the stacks and their comparison between the results of the understanding of the information gath- 1989 assessment and data gathered thir- ered by the 1989 assessment. As with the teen years later. Second, because the cen- Stanford University assessment pub- tral stacks collection represents the larg- lished in 1982, the data from UIUC’s ini- est and oldest circulating collection in the tial assessment reported in College & Re- library system, assessing this population search Libraries employed a system of offered the opportunity to learn about a weighing certain elements of an item’s significant portion of the library’s collec- condition in an effort to score overall item tions. condition.11 Although this model was ef- Third, the imminent construction of a fective for gathering the general collec- high-density shelving facility on campus tion-wide data needed at the time, the influenced the decision to assess the cen- sample-level data from 1989 were lost. tral stacks collection. The facility’s first Because a direct comparison with the phase will house two million volumes, 1989 UIUC study was impossible, the 100,000 of which are initially being se- authors designed the new assessment lected from the stacks. This process means with the purpose of gathering data that that the stacks will no longer exist as cur- were more specific. However, the 1989 rently conceived; in essence, the stacks’ assessment did lead the authors to hy- current repository function will diminish. pothesize that roughly one third of the Consequently, the authors perceived this collection would exhibit signs of signifi- to be the last opportunity to conduct such cant damage, one third would exhibit an assessment. Moreover, assessing the signs of mild to moderate damage, and central stacks collection will give collec- tion managers and administrators an idea of overall condition before relocating the TABLE 1 materials. Paper, Binding, And Board and Finally, the authors strongly believed Cover Condition as Reported in the that a second, more thorough stacks assess- 1989 UIUC Assessment ment would be invaluable for long-range preservation planning. Although the 1989 Paper Binding Board assessment produced useful data about the and Cover overall state of the central stacks collection, Good 32.0% 70.8% 49.7% no data remained pertaining to the indi- Moderate 31.0% 24.5% 41.9% vidual pieces surveyed. Instead, only the Poor 37.0% 4.7% 8.3% compiled data remained, leaving the in- Building Preservation 215 the remainder would show no significant estimate of the population was not nec- signs of damage. Although this is their essary before beginning the assessment. hypothesis, the authors also understood Consequently, the next step in this project that multiple factors could have changed was to determine the sample size required this: (1) an increase in the percentage of to provide the desired level of accuracy. material published on acid-free paper, (2) Rather than using statistical sampling patrons’ increased usage of electronic re- tables, the AZPlanSite provided a calcu- sources, (3) the creation of a separate lator that automatically calculated sample brittle book backlog, and (4) the impact sizes based on collection size, maximum of air-conditioning on a significant por- acceptable margin of error, and degree of tion of the collection. confidence desired.12 At the recommendation of a statistician Statistical Methodology with the campus’s Survey Research Lab, Confidence and Tolerance the authors chose to ensure that each deck Using proper sampling methodology, a within the stacks had thirty items ran- sample of the collection will reasonably domly chosen for the study. This sugges- estimate the characteristics of any popu- tion ensured an adequate representation lation within a certain margin of error. A of the discrete populations within the col- confidence level of 95 percent and a tol- lection, a situation that stems from the erance level of ±5 percent were acceptable central stacks collection’s physical ar- for this study. This requires a sample size rangement into seventeen distinct decks. of 385 items. This means, for example, that As mentioned earlier, some decks hold if 32 percent of the books are brittle, the collections that are administratively sepa- authors are 95 percent certain that the ac- rate from the central stacks collections, tual percentage of brittle books is between resulting in the exclusion of three decks 27 and 37 percent. The earlier survey of from the sample. Two half-decks were the stacks used the same confidence and combined into one, as they contained ex- tolerance levels in its assessment, provid- actly half the number of ranges as the full ing a basis for comparison between the decks. Thus, the sample assessment con- results of the two surveys. Moreover, one sisted of thirteen decks. downside of a higher level of confidence Initially, the authors wanted to base the is the significant increase in the number number of samples per deck on the deck’s of items sampled. For example, an in- approximate collection size. Unfortu- crease from 95 to 99 percent requires a nately, determining the exact number of nearly 72 percent increase in the number items for each deck was impossible. The of items sampled—a significant expense stacks have undergone neither a complete relative to the data’s potential use for gen- inventory nor a full retrospective conver- erating broad-based information. De- sion. Moreover, overcrowding is so severe creasing the tolerance level by one per- in some spaces that books are stacked on cent increments also would require large the floor. Consequently, it is difficult to increases in the sample size: 56 percent estimate accurately the number of items increase for ±4 percent; 177 percent in- located on each deck. With all these limi- crease for ±3 percent; and 524 percent in- tations known, the statistician recom- crease for ±2 percent. This will greatly mended that selecting a certain number increase the precision of the results, but, of items per deck would suffice. Thirty once again, the expense would exceed items per deck would be analyzed, what additional information the authors achieving a sample size of 390 items. This can gain by a lower tolerance level. sample size provides a confidence and tolerance of 95 percent ±5 percent. Sampling Method To locate items within the collection, the As sample sizes do not increase for popu- authors used Microsoft Excel’s random lations over 50,000, a completely accurate number generator to provide a single num- 216 College & Research Libraries May 2003 ber that corresponded to each range within have been made in this area, these condi- the stacks. Three numbers were generated tions could make a difference in the col- to locate each column and each shelf. Six lections housed here. numbers were generated for each book. The number of columns and Variations in the number of columns, shelves varies depending on book size. shelves, and books per shelf necessitated The folio cabinets have fewer columns the generation of multiple numbers. These and shelves than standard regular ranges. provided a backup for locating items when Thus, they likely constitute a higher num- there was no corresponding column, shelf, ber of rejections. The original assessment or book. Student employees received fifty used a similar method to generate ran- samples each and instructions to stop dom numbers, and the authors chose to when they had analyzed thirty books on duplicate the process, recognizing that the each deck. These additional samples en- results may not reflect exactly how many sured that rejects due to insufficient col- items in the collection are folios. umns, shelves, and books per shelf would not interfere with sampling the necessary Work Flow, Labor Management, and number of items per deck. Instruction The assistant circulation and bookstacks Biases librarian hired five student employees to Although the authors made every effort conduct the study. At the time, it was not to include as much of the collection as feasible to reallocate central circulation possible within the sample, the necessary and preservation staff to work on the exclusion of certain collection materials study. Instead, funding opportunities means that some biases exist in the study. were available from sources within the A 6,000-volume brittle book back- library to use student employees for re- log was not considered within the assess- search projects, and the authors received ment. a grant to hire students for the study. No satisfactory manner was de- In preparation for the project, the stu- vised to include the books currently dents were required to attend a single stacked on the floor. The authors gave sig- training program consisting of an over- nificant consideration to these view of the project’s goals, an introduction “nonrespondents.” Consultation with a to book construction, and a discussion of statistician produced no better solution assessment techniques and methods. than including these materials within a Through a Microsoft PowerPoint presen- broad category of biases.13 tation, images illustrated a variety of dif- The number of shelving units var- ferent types of cover and paper damage, ied depending on the deck, ranging from as well as some basic binding information. 229 to 387. Because the same number of Each student received a copy of the samples was set for each deck, the mate- PowerPoint presentation for reference. rials on the decks with the lowest num- This portion of the workshop concluded ber of ranges had a greater probability of with the students analyzing books and selection than did materials on decks with presenting their findings to the other stu- the highest number of ranges. However, dents. This hands-on exercise provided an this method guaranteed sampling from opportunity to test what they learned and all portions of the collection. enabled the authors to emphasize and Most of the collections reside in clarify certain details. portions of the stacks without air-condi- The workshop included a tour of the tioning. However, the newest, high-den- stacks to familiarize the students with their sity addition to the stacks does receive arrangement and the method for locating central air. Nearly 45 percent of the items materials using maps and the random sampled came from the decks in this ad- numbers. At the end of the tour, the assis- dition. Although shifting and additions tant circulation and bookstacks librarian Building Preservation 217 gave each student a set of random num- The Assessment Results bers and a map and instructed the students In the most basic terms, the results from to find the correct book on the deck. Only the first assessment corroborated those one student out of five had trouble find- provided by the second assessment. In ing the correct book on the shelf, and after terms of brittleness, 35.64 percent of the a thorough explanation, the librarian was pieces sampled in 2002 exhibited some satisfied that the students had a good grasp degree of embrittlement, as opposed to of how to use the random number sets and 37 percent in 1989. Similarly, 29.2 percent the maps. On the random sample form, the of those pieces sampled in 1989 suffered students noted the reason why a random from moderate or poor binding condi- number set was rejected (e.g., insufficient tions.14 In 2002, those with detached number of books on the shelf). The librar- boards, loose hinges, tears, or missing ian reviewed the reasons for each rejection covers accounted for 24.88 percent of the to ensure that no sets were discarded in collection. Given the tolerance of ±5%, the error. data initially suggested no significant dif- Instruction did not end with the work- ferences between the populations. shop and tour. The students regularly As noted earlier, the current assess- communicated with the authors and ment was divided into four basic sections: asked a variety of questions, ranging from a brief examination of publication infor- finding the date of publication to identi- mation, an examination of the binding, fying types of damage. The authors also an assessment of any enclosures, and an reviewed completed forms and discov- assessment of the text block that focused ered that the students had a good eye for on damage to the pages. detail. However, some students were ini- tially too quick to note damage when Publication Information none was present. The authors spoke to To develop a better understanding of the the students to clarify differences (i.e., collection, the assessment gathered data yellowed paper versus the natural color about the publications’ bibliographic his- of the paper) and the students corrected tory. This included data on the relative their forms. In total, the students took size, date, and place of publication. eighty-three hours to analyze 390 books, averaging thirteen minutes per book. For Size books that the students required consul- Of the 390 pieces examined, 79.23 percent tation to complete, the assessment process of the collection was of a standard size, took longer than thirteen minutes, but for measuring less than ten and a half inches books with minimal or no damage, stu- and over six inches in height. This is equal dents needed only a few minutes to fill to nearly 4.6 million volumes from the out the form. One of the reasons why the entire collection of nearly 5.8 million students were efficient in assessing the pieces. Undersized material equaled 2.82 books was that all of their samples were percent of the collection, and oversized on two to four decks. The students be- material equaled 15.9 percent. Another came intimately familiar with the layout 2.05 percent of the pieces reflected folios and peculiarities of their assigned decks requiring flat shelving.15 and were able to navigate with ease. When all forms were completed, a stu- Date of Publication dent entered the data from the assessment As one would expect with such a repre- into a relational database. The preserva- sentative sampling method, a year-by- tion librarian examined the forms for dis- year analysis of publication dates yields crepancies or typos before the data were little substantive data about the entire inputted into the database. After being collection. However, analyzing the data entered, the data again were reviewed for by decade of publication does present an possible errors or omissions. interesting illustration of both the 218 College & Research Libraries May 2003 collection’s development and composi- tion. Although a separate project began TABLE 2 isolating those items in the central stacks Date of Publication as Reported collection published before 1800, the as- in the 2002 Assessment sessment indicates that nearly 74,202 vol- umes were published before the begin- Category % of Total ning of the nineteenth century. As part of Pre- 1800 1.28 the selection for high-density shelving, 1801–1810 0.26 rare book and special collections librar- 1811–1820 0.51 ian is reviewing early imprints for trans- 1821–1830 0.26 fer to the Rare Book Room. 1831–1840 0.51 The collection’s growth from 1870 on- 1841–1850 0.26 ward appears to be almost unchecked 1851–1860 1.03 (table 2). Until 1980, its rate of growth in- 1861–1870 0.77 creased nearly every decade with the ex- 1871–1880 1.28 ceptions of the decades 1881–1890 and 1881–1890 1.03 1941–1950. The first of these decades fol- 1891–1900 2.05 lows a period of economic depression in 1901–1910 3.85 the early 1870s and actually precedes the 1911–1920 4.87 depression that accompanied discussions 1921–1930 5.13 about abandoning the gold standard in 1931–1940 8.46 1890. The second decade encompasses the 1941–1950 8.21 Second World War. Whether either of these 1951–1960 10.51 periods of diminished growth can be di- 1961–1970 13.85 rectly attributed to diminishing resources 1971–1980 16.15 or decreasing book production is unlikely. 1981–1990 9.23 However, it is likely that both played a role. 1991–2000 8.21 From 1981 to the present, the rate of 2001–Present 0.51 growth appears to have declined. How- Unknown 1.79 ever, the system comprises forty-two sub- ject and departmental libraries. With the Place of Publication exception of Slavic and Eastern European, Predictably, North American publica- Latin American, and European blanket tions constitute the largest single por- orders, many core titles and the most cur- tion of the collection, as shown in table rent acquisitions are not sent to the cen- 3. At 42.82 percent, this represents tral stacks collection. Consequently, the nearly 2.5 million volumes. This is vast bulk of volumes acquired since 1981 closely followed by the combined are still housed in departmental libraries, weight of roughly two million Western meaning that the significant drop in col- European volumes and nearly three- lection breadth during this period does quarters of a million Eastern European not necessarily indicate a decline in the volumes. The most surprising result is breadth of the overall collection. that 4.1 percent of the sampled volumes The collection of material published in trace their origin to India or other parts the decades before the university’s incor- of Asia. The Asian library is one of the poration in 1867 is more difficult to gener- few libraries that does not transfer col- alize. Lower numbers of these items within lections into the stacks; however, many the stacks resulted in lower populations for other libraries purchase materials pub- each of these decades. For example, imprint lished in these areas and include them dates from the first three-quarters of the within their collections. Moreover, nineteenth century indicate that the collec- many of these items are received tion ranges from lows of 14,864 volumes to through the PL480 program and the highs of 74,321 volumes per decade. Farmington Plan. Building Preservation 219

Binding Information TABLE 4 Binding Style Examination of the binding styles focused Binding Style as Reported in on determining which of eight choices 2002 Assessment best characterized the type of binding rep- resented by each piece. In general, the Category Number % of Total definitions were clear to most of the stu- Publishers’ cloth 123 31.54 dent employees. The one area that re- Library binding 169 43.33 quired some additional definition was the Softcover 47 12.05 relationship between library binding and Velobind/comb 3 0.77 leather binding. For the purposes of this Spiral 0 0.00 assessment, a library binding consisted of Pamphlet binder 34 8.72 any binding performed by a binder after Other binding 8 2.05 the library had purchased the piece. These Vellum binding 0 0.00 items ranged from early quarter- and half- Leather 5 1.28 bound pieces with marbled boards to con- Unbound 1 0.26 temporary library bindings. A leather- Total 390 100.00 bound volume was a volume bound entirely in leather presumed to have been sold in that binding and generally mono- lection exhibited some signs of damage, graphic in nature. With 43.33 percent of including one or more of the following its collection library bound, 1.28 percent conditions: detached boards (3.33%), of its material in full leather, and less than loose hinges (13.33%), tears (9.23%), and a third of the collection in an original no boards or covers whatsoever (2.56%). publisher’s cloth binding, these results In terms of pure numbers, this means that suggest that the library has relied heavily approximately 341,447 items suffer from on library binding as both a management either missing or detached boards. and maintenance tool (table 4). When compared to the 29.2 percent of items indicated as exhibiting poor or External Cover Damage: Hinges and moderate binding condition in the 1989 Mechanical Construction assessment, it is interesting to note that An examination of the mechanical con- the items currently suffering from de- struction of the volumes’ covers indicated tached and missing boards vary from the that 24.88 percent ±5 percent of the col- original data by only 4.32 percent, within the margin of error outlined by the ear- lier assessment. Similarly, the combined TABLE 3 percentages of those covers with tears and Place of Publication as Reported loose hinges equal 22.56 percent of the in the 2002 Assessment collection. In the 1989 assessment, 24.5 percent of the bindings exhibited “mod- Category % of Total Total erate damage.”16 Pieces N. Amer. 42.82 2,482,305 External Cover Damage: Other Damage L. Amer. 6.67 386,665 Another component of the assessment W. Europe 33.59 1,947,236 was damage to the cover of a piece. Any E. Europe 12.56 728,112 item with visible cover damage, such as India 1.79 103,768 water damage, warped and misshapen Africa 0.26 15,072 boards, light bleaching, staining, and in- Aus/NZ 0.00 0 sect damage, was counted as one with Asia 2.31 133,912 external cover damage. The total percent- Total 100.00 5,797,070 age of sampled items without any visible external damage of this type is 43.59 per- 220 College & Research Libraries May 2003 cent. Examination of the other types of indicates that roughly 16.15 percent of the visible damage indicates that there are collection is misshapen. This translates some significant problems with materi- into 936,227 pieces permanently marred als and their storage environment. Two through mishandling, poor stewardship, areas of particular concern are materials and overcrowding. When simultaneously suffering from combinations of visible calculated with those items that are both water damage, staining, and misshapen scratched and misshapen, 4.1 percent of covers and materials exhibiting signifi- the collection, or approximately 237,680 cant abrasion. pieces, exhibits this type of compound In general, the quantity of items with damage. both water damage and staining is high. Approximately 3.59 percent of the mate- Extraneous Material rials sampled are water damaged or The amount of extraneous material asso- stained. Although this percentage does ciated with the covers of sampled items not sound excessive, it does indicate that is relatively low. In general, it is reassur- roughly 208,115 pieces have been exposed ing to see that few pieces received im- to leaks, spills, accidents, or other mis- proper repairs using adhesive or cloth haps in their history. This damage may tape. In this case, the lack of regular care have happened while the pieces were in has proved itself a benefit in terms of departmental libraries or in circulation or long-term care. The total number of items before they became a part of the library’s with tape or other extraneous material collections; however, it does indicate that applied to the cover total only 4.36 per- facilities problems and care and handling cent. issues are taking their toll. Although multiple mold outbreaks Enclosure Information have occurred within the stacks, some In general, the assessment produced very affecting several deck levels simulta- little information about the enclosures neously, none of the sampled materials present throughout the collection. Indeed, exhibited visible evidence of external only one piece of the 390 sampled had any mold damage. There are a couple of pos- type of enclosure. It was a phase box, and sible explanations for this potential dis- it, incidentally, suffered from some minor crepancy. Cleaning projects followed sev- damage itself. This figure represents 0.26 eral of the mold outbreaks. Moreover, percent of the collection, or 15,072 vol- some pieces may have been damaged to umes. such an extent that they were rebound, At first glance, this low number may resulting in no visible signs of damage to indicate that few pieces actually need this the cover. Another explanation is that the type of enclosure; however, the small students did not assess the damage cor- number of enclosures also can reflect two rectly and mistook dormant mold for dirt very different conclusions. As mentioned or other debris. Because no systematic earlier, the “storage collection” was not cleaning of the stacks has taken place, included in the sampled pieces. It repre- items with mold may not be easily iden- sents a backlog of severely embrittled tifiable from other dusty volumes. Finally, pieces. This collection ceased growing in some items marked as stained may, in the mid-1990s. Consequently, it is likely fact, be showing signs of mold damage. that a significant portion of the collection The other area of concern is the quan- that should have been boxed was put into tity of misshapen pieces. Although evi- this holding area. dent when walking through the stacks, The second conclusion that one may the results of improper shelving, the im- draw from the low number of enclosures proper use of bookends, and overly tight is the relatively low level of past preser- shelving are more striking when the dam- vation efforts. With about 3.33 percent of age is quantifiable. The 2002 assessment the collection suffering from detached Building Preservation 221 boards and over 17 percent of the collec- TABLE 5 tion brittle to the point that the paper can- not withstand a single double-fold (table Acidity to Breakability as 5), there is likely a significant backlog of Reported in 2002 Assessment material that will, at minimum, likely Category % of Total benefit from an enclosure. Acidic/One Fold 17.18 Internal Pages Acidic/Two Folds 8.72 Leaf Damage Acidic/Three Folds 4.87 Actual paper damage is best represented Acidic/Four Folds 4.87 by noting that 159 of the 390 sampled Acidic/Not Brittle 54.62 items exhibited no signs of damage. This Non-acidic/One Fold 0.26 represents 40.77 percent of the collection, Non-acidic/Two Folds 0.26 or about 2,363,465 volumes with no in- Non-acidic/Three Folds 0.00 ternal damage. Of those exhibiting one or Non-acidic/Four Folds 0.00 more categories of leaf damage, the rea- Non-acidic/Non-Brittle 9.23 sons are likely varied; however, staining and water damage often result from the than 4 percent of the collection had adhe- same event. sive tape affixed to leaves, likely in an Perhaps the most striking concern attempt to repair damaged pages. In raised is the number of volumes indicat- terms of patron-deposited materials, 6.41 ing evidence of marking (as characterized percent had paper slips inserted into text by highlighting and underlining) and/or blocks, 1.28 percent had paperclips on staining. Nearly 11 percent of the sampled leaves, and another 1.03 percent had self- items were obviously marked. The library adhesive notes applied to pages. Approxi- has a long-standing history of encourag- mately 3.33 percent of the collection had ing access both on campus and off, a tra- other extraneous material inserted within dition that is valued and unlikely to the text blocks. Unfortunately, the assess- change in the near future. However, the ment did not include provision to gather high incidence of marking is indicative of data on the types of extraneous material users’ poor care and handling practices. found. Ten percent of the sampled pieces suffered from some sort of staining. The staining Paper Acidity was the result of a variety of sources, but As the repository for much of the only a small portion of the stained items library’s older materials, it stands to rea- also visibly suffered from other related son that the stacks collection would suf- forms of damage, including water, visible fer from a higher-than-expected level of mold, and insect damage. One conclusion paper acidity. Indeed, that is the case. Of is that many of these other stains resulted the 390 items sampled using an Abbey from user-related damage. pH pen, 352 tested acidic on the exterior Of additional concern is the 3.08 per- margin of the last page of text. Even cent of volumes (or 178,550) that exhibit though this measures edge acidity and some sign of pest-related damage to the not core acidity, the result indicates that text block. Although this may include 90.26 percent of the central stacks collec- damage that transpired before their acces- tion is acidic. In terms of actual numbers, sion by UIUC, this evidence indicates that this translates into approximately some contemporary damage is occurring. 5,232,435 pieces from a population total- ing 5,797,070. Extraneous Material As would be expected, a number of vol- Paper Acidity as Related to Brittleness umes suffered from the presence of vari- As was demonstrated in the earlier subsec- ous extraneous materials. Slightly less tion on date of publication, approximately 222 College & Research Libraries May 2003

TABLE 6 Paper Acidity to Brittleness as Reported in 2002 Assessment

Category Number % of Acidic Books % of Total Collection Acidic & Not Brittle 213 60.51 54.62 Acidic & Brittle 139 39.49 35.64 Total 352 100.00 90.26

80.26 percent of the central stacks collection tion maintenance activities. Of the 139 was published before 1980. Nearly all of pieces that were both acidic and brittle, the these were published on acidic paper. As paper’s functionality varies greatly. A to- noted earlier, only a portion of the central tal of 17.18 percent of the sampled pieces stacks is air-conditioned. Indeed, nearly 55 were both acidic and embrittled to such a percent of this collection resides in space point that the paper could not withstand completely governed by the ambient air one double-fold without breaking (table 5). temperature. Consequently, the collection Acidic pieces that could withstand two would be expected to suffer from a high folds accounted for 8.72 percent of the en- level of embrittlement. tire collection, and acidic pieces that could Of those portions of the collection withstand three to four folds accounted for printed on acidic paper, a significant per- 9.74 percent of the collection. centage suffers from some level of Of additional interest is the fact that embrittlement. As shown in table 6, 60.51 the library completed an air-conditioned percent of the acidic books are not yet stacks addition in 1984.17 Those items in brittle whereas 39.49 percent are suffer- the new addition were located in climate- ing from some level of embrittlement. In controlled environment for only five terms of the entire collection, these fig- years before the 1989 survey was com- ures translate into 54.62 and 35.64 percent, pleted. At the time of the original survey, respectively, of the entire collection. That the change in the collection’s environ- means that the stacks collection is com- ment had only a marginal impact on the posed of nearly 2,066,076 volumes that are state of the collection. However, because acidic and brittle and another 3,166,360 eighteen years have passed since the new volumes that are acidic, but not yet addition was completed, the authors con- embrittled. Although some older volumes ducted a preliminary investigation to de- constructed of higher-quality papers will termine the effects of superior environ- remain pliable despite high levels of acid, mental conditions on the state of the col- the vast majority of the volumes that are lection in the new addition. acidic, but not yet brittle, will likely con- Of those items sampled in the addition, tinue to degrade and become increasingly the number published on acidic paper fragile. nearly equaled the results for the entire sample population. However, the num- Paper Acidity as Related to Breakability ber of items that were both acidic and Even though the assessment determined brittle is significantly reduced. For ex- that 35.64 percent of the entire collection ample, only 11 percent of this subpopu- is both acidic and brittle, planning for lation broke on one double fold, com- long-term collection maintenance activi- pared to 17 percent from the entire sample ties requires a more detailed analysis of population. Similarly, only 4.3 percent of the pieces themselves and their ability to those items located in better environmen- withstand other forms of treatment. tal conditions broke on two double folds, In terms of treating this collection, the compared to 8.72 percent from the entire relation of acidity to embrittlement level population. This trend continues is crucial in planning for long-term collec- throughout the entire brittle component Building Preservation 223 of this population. Some of the difference In addition, the library hired a preser- is certainly due to the acquisition of vation librarian and a conservator to be- nonacidic materials, but the possibility gin the planning and implementation of exists that this may be, in part, from eigh- a preservation program. Coupled with a teen years in better storage conditions. significant grant from the Andrew F. However, more comprehensive research Mellon Foundation to outfit and staff a is needed to determine this conclusively. conservation lab in its high-density shelv- ing facility, the program is making great Acidity to Decade of Production strides in meeting the needs of such a The assessment tool used to manage the large and varied collection. sample also permitted the authors to gen- However, this does not address how erate data relating the decade of produc- this assessment will shape the preserva- tion to a publication’s acidity. From ana- tion and conservation program’s overall lyzing these data, it is possible to see the development. Broadly viewed, the data distribution of acidic to nonacidic books generated can be grouped into three areas by decade of publication. As data from the of general need and/or utility: education current assessment demonstrate, the per- and outreach; collection repair and main- centage of acidic volumes begins to drop tenance; and administrative development. off in 1981–1990, the decade that includes the widespread introduction of acid-free Education and Outreach paper (table 7). However, it is not until the As evidenced by Jeanne M. Drewes and next decade that the number of nonacidic Julie A. Page’s Promoting Preservation publications begins to have a significant impact on the relative number of acidic TABLE 7 publications in the collection. Acidity to Date of Production as With the current assessment in hand, the library now has a unique opportunity Reported in 2002 Assessment to begin to address its history of benign Decade % Acidic neglect. Pre- 1800 80.00 Assessment Conclusions and 1801–1810 100.00 Programmatic Development 1811–1820 50.00 Recognizing that the collection has suf- 1821–1830 100.00 fered from the lack of regular mainte- 1831–1840 100.00 nance, the University of Illinois at Ur- 1841–1850 100.00 bana-Champaign seriously began to 1851–1860 100.00 address preservation concerns in the past 1861–1870 100.00 few years. Two years ago, UIUC hired a 1871–1880 100.00 consultant to complete a needs assess- 1881–1890 100.00 1891–1900 87.50 ment. Primarily focusing on the facilities, 1901–1910 100.00 the consultant’s report gave the institu- 1911–1920 100.00 tion the ability to tangibly discuss need. 1921–1930 100.00 For example, in addition to highlighting 1931–1940 100.00 the deficiencies caused by deferred main- 1941–1950 93.75 tenance, the assessment provided the li- 1951–1960 95.12 brary with an estimate of the collection’s 1961–1970 98.15 replacement value.18 Although this type 1971–1980 92.06 of information does not immediately ad- 1981–1990 77.78 dress concerns, it has been valuable in 1991–2000 53.13 lobbying the administration and increas- 2001–Present 50.00 ing awareness of the library’s value to the Unknown 85.71 campus. 224 College & Research Libraries May 2003

Awareness in Libraries: A Sourcebook for In addition, the head of preservation Academic, Public, School, and Special Col- initiated a silent campaign of raising pres- lections, education and outreach have a ervation awareness among both employ- long tradition within the preservation ees and users. Several months ago, the program arsenal.19 This education gener- program began producing a periodic ally takes multiple forms, targeting both newsletter that provides basic informa- employees and users. Based on the sig- tion on the preservation program and its nificant number of volumes suffering offerings, how to take better care of col- from highlighting, underlining, and other lections, and preservation in general. Dis- markings (637,678) or from the use of in- tributed electronically, the newsletter also appropriate bookmarks such as includes a What’s This? section with a paperclips and self-adhesive notes (74,202 picture highlighting either examples of and 59,710, respectively), neither the preservation do’s and don’ts or items of library’s users nor its employees have interest such as images of high-density been educated about the damaging effects storage facilities in operation. of such items or the importance of remov- Efforts to reach the library’s user popu- ing them when the volumes are returned. lation are still developing. However, Moreover, the fact that significant por- through a collaborative project with a tions of the collection’s external cases are member of the university’s faculty and a scratched (1,233,616) and misshapen senior graphics design class, the preser- (936,227) indicates that significant care and vation and conservation program now handling issues have not been addressed. has designs for posters, bookmarks, and This may include poor handling by users, screensavers to communicate the preser- but it also likely includes the deleterious vation message. effects of overly tight shelving. Even though it is expected that the library’s Collection Maintenance and Repair planned high-density shelving facility will Although the library’s bookstacks office resolve much of this issue, the fact remains and book repair unit have traditionally that there are serious deficiencies in basic completed repairs, establishing the pres- care and handling education. ervation and conservation program has Finally, a significant portion of the col- focused heavily on developing these ca- lection suffers from water damage and pacities to better serve the institution’s staining. Although it is likely that leaks needs. The program’s first year and a half and other faults in the building structure focused primarily on the improvement have caused much of this damage, some and rationalization of existing services. of it is the result of careless handling or For example, before September 2001, there untreated damage. was no formal relationship between li- To address some of these issues, the brary binding and book repair. Units with preservation and conservation program damaged materials sent them to either has taken a three-pronged approach. In- book repair or binding, generally with- ternally, the head of preservation, the con- out instruction and frequently without servation librarian, the associate univer- knowledge of actual need. Consequently, sity librarian for collections, and a mem- the library outsourced to a library binder ber of the repair unit have begun to de- many repairs that could have been done velop and hold a series of educational by existing internal staff and completed workshops for employees. Thus far, the repairs on many items that should have program consists of a preservation orien- been rebound. tation, training in library binding, and Similarly, many items did not receive training in basic book repair. Additional the more in-depth treatment their dam- workshops on disaster response and pres- age warranted. As evidenced by the ervation decision making for collection library’s long-neglected backlog of brittle managers are planned for the future. materials, the treatment of brittle materi- Building Preservation 225 als was scattered and infrequent. Some tion with concrete data for administrative items received phase boxes and others development. As mentioned earlier in this various inappropriate or outdated re- article, the campus has followed a long- pairs, many of which caused permanent term policy of deferred facilities mainte- damage to the original items. To address nance. The site survey performed by a this issue in a more comprehensive man- consultant in 2001 provided the campus ner, the library’s administration autho- and the library’s administration with a rized the new position of a brittle books common language when discussing the coordinator. Working with the head of collection—its monetary value to the cam- preservation, the brittle books coordina- pus. This assessment follows suit by pro- tor is establishing the institution’s first viding the library’s administration with comprehensive replacement and refor- the ability to justify improvements based matting program, both digging through on specific needs. the backlog and dealing with those ma- For example, the 1989 assessment con- terials that are now being routed to the cluded that 37 percent of the collection’s preservation and conservation unit. paper quality was poor.20 At the conclu- In addition to tackling damaged ma- sion of the assessment described in this terials, the preservation and conservation article, the authors can assert that slightly program began to systematically address over 35 percent of the collection is com- the issue of binding paperback acquisi- posed of paper suffering some level of tions. The assessment indicates that, of the embrittlement. Although this is generally entire sample population, forty-eight in keeping with the 1989 assessment, items, or 12 percent of the sampled items, there is both good news and bad. The appeared in soft covers. Excluding those good news is that although 35 percent of that the library had bound at some point, the collection is embrittled, only 25.9 per- nearly 39 percent of the sampled items cent is on acidic paper that fractures after produced since 1989 are paperbacks. In one or two double-folds. Another 0.52 the long run, this indicates that the percent of the sampled items are printed library’s collection patterns are changing on nonacidic paper that is likely coated significantly and that its binding patterns and breaks after two or fewer folds. The need to follow suit. Assessment results bad news is that the combination of these indicate that 4.2 percent of all paperbacks numbers indicates that the institution is sampled suffer from damaged or mended in possession of roughly 1,531,586 ex- leaves and that an additional 4.2 percent tremely brittle volumes within this single have covers mended with either cloth or collection. adhesive tape. Even though this equals Despite this significant corpus of only one percent of the entire sample, it embrittled materials, the more shocking implies that roughly 58,000 paperbacks figure is that a further 54.62 percent of the have received or need repairs. In addition, collection is printed on acidic paper that eleven of the paperbacks sampled were has not yet become embrittled. Conse- misshapen. This only equals 2.8 percent quently, the library is in possession of a of the entire collection but amounts to a collection composed of 90.26 percent startling 23.4 percent of all paperbacks acidic paper in a facility with extremely sampled. Using these data, the preserva- fluid environmental conditions with an- tion and conservation program has begun other 0.52 percent that, although acid free to implement a systematic means of ad- on the surface, is very fragile. These data dressing the binding needs of its paper- illustrate that, untreated, the rate of dete- back collections. rioration is having an impact that pres- ently outpaces the rate of acid-free trans- Administrative Development fers into the collection. Despite the sig- The assessment’s most far-reaching out- nificant increase in the acquisition of acid- come is the ability to provide the institu- free pieces in the past two decades, the 226 College & Research Libraries May 2003 percentage of embrittled pieces is remain- embrittlement. Although the cost of ing relatively stable. This could stem from proper HVAC controls is staggering, it is a number of reasons, including the con- dwarfed by the costs of replacement, re- tinued rate of deterioration in an unstable formatting, and deacidification. If the environment, an increased rate of acqui- possibility exists that improved condi- sition of older materials through the tions did slow the rate of embrittlement, institution’s robust gifts program, or the this assessment’s results provide the delay in the transfer of more current acid- library’s administration with a significant free acquisitions into the central stacks. tool in its quest to improve the facility’s Regardless of the cause for such contin- conditions. ued deterioration, the data in table 7 in- Within the library, the data provided by dicate that the production of acid-free this assessment are guiding programmatic paper will have a significant impact on development in some significant ways. As the long-term health of our collection. noted earlier, the preservation and conser- However, the continued deterioration of vation program has been making strides our older collection items indicates that in improving and expanding services. This the long-term effects of acid-free paper are development also has included the devel- not likely to be seen for some time when opment of policies and procedures that considering larger collection-wide pres- seek to address internal deficiencies. Ex- ervation issues at UIUC. amples include the development and The ability to contrast the replacement phased implementation of uniform bind- and reformatting or deacidification needs ing criteria and efforts to increase aware- of this collection against the significant ness among individual employees about costs of introducing environmental con- their stewardship role. Although these ef- trols is valuable for the library’s admin- forts are meeting varying levels of success istration. Despite the addition of nearly in such a large system, they are steps in 250,000 acid-free publications to this col- the library’s long process of developing a lection since 1989, the institution is just modern and responsive preservation and keeping pace with the overall rate of conservation program.

Notes 1. Sara Buchanan and Sandra Coleman, “Deterioration Survey of the Stanford University Libraries Green Library Stack Collection,” in Preservation Planning Program Resource Notebook, comp. Pamela W. Darling (Washington, D.C.: Association of Research Libraries, 1987), 159–230. 2. Tina Chrzastowski, et al., “Library Collection Deterioration: A Study at the University of Illinois at Urbana-Champaign,” College & Research Libraries 50 (Sept. 1989): 577–84. Barrow’s as- sertion is available at W. J. Barrow, Deterioration of Book Stock, Causes and Remedies, ed. Randolph W. Church (Richmond, Va.: The Virginia State Library, 1959), 15. 3. Ibid., 577. 4. Ibid., 579. 5. Brian J. Baird, et al., “Findings from the Condition Surveys Conducted by the University of Kansas Libraries,” College & Research Libraries 58 (Mar. 1997): 115. 6. Task Force on Preservation, “University of Illinois at Urbana-Champaign, Collection Analy- sis Project: Report of the Task Force on Preservation” (Urbana: Univ. of Illinois at Urbana- Champaign, 1980, photocopy) and Illinois Library Materials Preservation Task Force, “The Pres- ervation of Library and Archival Resources in Illinois: A Five-year Plan of Action” (Carbondale, Ill.: Illinois Library Materials Preservation Task Force, 1986). 7. First published in 1982, the Association of Research Libraries’ (ARL) Preservation Planning Program: An Assisted Self-study Manual for Libraries encouraged libraries to systematically study their collections and facilities through a process of assessment. In 1987, Pamela W. Darling ex- panded the publication and Wesley Boomgaarden revised the resource notebook to include sig- nificant selections from the burgeoning preservation literature. In 1986–1987, Northwestern Uni- versity, the Colorado State University Libraries, Ohio State University, and many other institu- tions published these reports. Building Preservation 227

8. Bradley L. Schaffner and Brian J. Baird, “Into the Dustbin of History? The Evaluation and Preservation of Slavic Materials,” College & Research Libraries 60 (Mar. 1999): 144. 9. The 1989 UIUC assessment and the 1982 Stanford assessment upon which it was based both excluded unbound serials. 10. Buchanan and Coleman, “Deterioration Survey of the Stanford University Libraries Green Library Stack Collection.” 11. The UIUC 1989 assessment evaluated books according to these criteria: condition of the paper, condition of the binding, and condition of the external covers. The books received a score for each category, and the overall score for each book was determined based on the combination of the three scores. Before the final score was assessed, the surveyors were instructed to give paper condition twice the weight of binding or external covers. The rationale for this decision was that paper represents the intellectual content of the book and thus the authors surmised that the library would be most interested in its condition. 12. AZPlanSite is located at http://www.azplansite.com/samplesize.htm and was last up- dated on June 6, 2002. It includes the online sample calculator and a brief explanation of its equations. For further elaboration on determining sample size, please refer to Ronald Czaja and Johnny Blair, Designing Surveys: A Guide to Decisions and Procedures (Thousand Oaks, Calif.: Pine Forge Pr., 1996), 126–33. 13. John Dean, “Managing Collection Information for Preservation Planning,” in Advances in Preservation and Access, vol. 2, ed. Barbara Higgenbotham (Medford, N.J.: Learned Information, Inc., 1995), 206. Although John Dean’s article emphasized the significant issues with excluding such nonrespondents from surveys and assessments, our statistician’s assessment was that an adequate sample size was generally based on respondents in relation to the population, not the population itself. 14. Chrzastowski, et al., “Library Collection Deterioration,” 580. 15. Oversized pieces are between ten and a half and seventeen inches. Folios are pieces that are seventeen inches or higher. Undersized pieces measure less than six inches. 16. Chrzastowski, et al., “Library Collection Deterioration,” 580. 17. The call numbers in the West stacks in 2002 are the same call numbers moved in 1984. Thus, books in the collection during the original move have benefited from the climate-con- trolled environment for eighteen years. The move is described in the library’s newsletter: Ruth McBride, “Central Bookstacks: Progress Report,” Library Office Notes, no. 30 (July 1984). 18. John DePew, A Library, Media, and Archival Preservation Handbook (Santa Barbara, Calif.: ABC-CLIO, 1991): 298–99. The figures provided by the library’s consultant were taken from those cited in DePew’s volume and adjusted for 2000 dollars. 19. Jeanne M. Drewes and Julie A. Page, Promoting Preservation Awareness in Libraries: A Sourcebook for Academic, Public, School, and Special Collections (Westport, Conn.: Greenwood Pr., 1997). 20. Chrzastowski, et al., “Library Collection Deterioration,” 580.