By: Carole L. Palmer, Lauren C. Teffeau Carrie and Mr. Pirmann for OCLC Research

Introduction
Research libraries exist to support scholarly work. In recent years, the scholarly literature on practices and information use has been growing, and research libraries should be Prospering from this increased base of knowledge. Unfortunately, the profession has no effective means for Systematically monitoring or synthesizing the published results. This review begins to address the problem by reporting on the state of knowledge on scholarly information behavior, focusing on the information activities involved in the research process and how they differ among disciplines. It provides an empirical basis for identifying promising directions and setting priorities for development of digital information services to support and advance scholarship. Preparing this report required the project team to make decisions about what publications to cover, what results to extract, and how to integrate and present the many valuable purpose often incongruous Findings on scholarly information behavior. Across studies there is considerable variation in how the object of study is defined and in how data is collected and analyzed. The variety of approaches is a natural outcome of the increase in number and sophistication of studies in recent decades and the Complexity of the processes under investigation. The challenges of designing and conducting a solid scholarly study of information behavior are many, but they are rarely apparent when reading the published reports.

Scholars and scientists carry out layers of physical and intellectual activity through a complicated mix of mundane and seemingly idiosyncratic tasks that result in a range of immediate and long-term outcomes. It is difficult to collect data that captures these socio-cognitive processes, and interpreting that data in ways that advance our understanding is even more challenging. The value and uses of information, individual items or entire genres-can change over time, at a micro level as Scholars gather, evaluate, analyze, assimilate and write, and at a more macro level as their ideas evolve, projects move forward and careers unfold. In our studies of scholarly information work, we have seen how a highly influential text a scholar can move into a new research project but then become overshadowed in the course of inquiry, with no trace left in the final, tangible scholarly product. On the other hand, we have also seen how scholars will reference materials from a diverse range of subject areas that reaches far beyond what they can readily recall as part of their directory information. These twists, turns, perceptions and practices are part of the intricate constellation of information activities that generate new scholarship and that we Strive to document and understand through empirical studies of scholarly information behavior. The term “information behavior “has become the field’s preferred term for studies of information needs and uses, but here, and in previous related papers, we use the terms “information practices” and “information work” since they we believe they are a better representation of the social aspects of scholarly activities and the purposeful, Workaday nature of how scholars spend their time.

Scope of the Literature
The scholarly literature on information behavior dates back at least to the reports from the 1948 Royal Society Scientific Information Conference and the 1952 symposium on Chicago School specialized information (Egan, 1954). Since that time, there has been a stream Increasing Steadily of research that has moved beyond the science to address the range of disciplines. By the 1980s,
user research studies Had taken hold as a significant subfield in DSL, and much of the research on scholarly groups produced over the decades is still highly relevant. For example, in the early digital era, RLG released a series of studies that provided a broad examination of information needs across the humanities, social sciences and sciences. The three reports covered a total of 20 disciplines profiling the kind of information, its uses and sources in each field (Gould, 1988 Gould & Handler 1989, Gould & Pearce, 1991). Around the same time, a book-length study of the work processes of Art historians was published by the J. Paul Getty Trust and the Institute for Research in Information and Scholarship at Brown University (Bakewell, Beeman, & Reese, 1988). These two contrasting approaches stand as exemplars of the broad and deep analysis needed to understand how to Develop effective information resources and tools for scholars. The results from these and other earlier works still hold implications for contemporary research libraries. In addition to supplying benchmarks from the past for Assessing Consistency and Change Over time, they contain important insights on information work before it was influenced by current technologies. Some information practices have not been altered in any fundamental way in the digital environment, but many may be advanced or enhanced with new information resources and tools. For example, the table below presents a distillation of results from the reports on the RLG types of information sources found to be of importance across the various disciplines studied. What has changed in the digital environment is not the value of these kinds of sources rather how to they are searched, accessed and used in the scholarly process. Therefore, while recent literature is Emphasized in the review, selected earlier studies have been consulted dating back to 1962.

As suggested above, it is complicated to integrate gold make close comparisons among different kinds of studies. Quantitative and qualitative approaches and make separate contributions together Provide complimentary perspectives and results. For example, ethnographic data provide richer and more nuanced analysis of research as it happens, while quantitative surveys produce more general results on patterns and trends in information behavior. In recent years, qualitative studies have become more common and are strongly represented Therefore in this review. Coverage of bibliometric studies is limited, since they tend to tell us more about the structure and flow of information than the actual work practices of scholars. Across studies the groups of scholars and scientists studied have been scoped in different ways, ranging from very broad classes (eg, science, humanities) to more narrowly defined disciplines (eg, Jewish studies, literary criticism, genomics) and mixed groups of interdisciplinary scholars working in many different research areas. Since it was not possible to align and integrate the results by population studied, our approach was to bring
findings together information on scholarly activities, associating studies of similar domains and Providing loose comparisons where possible.

Most of the literature covered is from newspaper publications in library and information science (LIS) and a number of important books and professional reports have also been included. Selected items from cognate fields in the social sciences and other information science domains, such as computer supported cooperative work (CSCW), are also discussed. Other research areas such as human – computer and internet interaction studies have not been included, partly due to space constraints Because their goal also applying to information services is often less direct. Because of our focus on disciplinary practices, we have not included research from LIS or cognate areas that examines Information behavior Primarily from the perspective of the individual. Many such studies have been Demonstrating how important in the local context information influences behavior, Although the definition of context has been applied Debated and Inconsistently (Courtright, 2007; Talja, Keso, & Pietilainen, 1999). Our focus in gathering literature for review was on more socio-cultural approaches that interpret information behavior as practiced within a discipline or field of study. As suggested by Hjørland and Albrechtsen (1995), since domain based studies of information seeking and use aim to represent consensus among communities of scholars, they are potentially more Those explanatory than that analyze the behavior of discrete individuals.

Table 1: Source Materials by subject reported in RLG reports

Analytical framework
To synthesize the large and diverse body of literature on disciplinary information practices, we have Relating results are concentrated on specific information activities provided by each study. We began by drawing from our previous review of scholarship and disciplinary practices (Palmer & Cragin, 2008), incorporating additional library oriented literature and covering a full range of information activities involved in the process of research. Adapting Unsworth’s (2000) concept of “scholarly primitive, “we derived a framework of scholarly information and primitive activities to guide discussion and to serve as points for comparison across domains. Our aim has been to advance understanding of the information work of scholarly communities, not the behaviors of individual researchers. Scholarship is a dynamic enterprise, however, and scholarly communities can be defined in many ways.

Many studies are designed to investigate standard academic structures and disciplinary categories. However, we know from our previous work on interdisciplinary research processes that it can be counterproductive to assume that formal academic departments are true representations of scholarly affinities (Palmer 1996, 1999a, 2001a, 2005, Palmer & Neumann, 2002). Scholars
regularly cross disciplinary boundaries in their work information, and recent studies have verified The importance and prominence of interdisciplinarity at research universities (eg University of Minnesota Libraries, 2006). Under these conditions, studies of scholarly groups need to take into Account of the “trading zones” that emerge as researchers exchange their expertise and products to solve research problems (Galison, 1996).

The basis of reviewed literature may not Adequately reflect the dynamics of how researchers interact with information and people across fields and specialties, but our activity-based framework is useful foregrounding for information work processes rather than a priori assumptions about disciplinary structures that may be built into the design of individual studies. At the same time, in gathering and discussing the research related to each activity, disciplinary differences across studies can easily be discerned. In addition, the review incorporates the growing body of information on work practices of interdisciplinary scholars and cross-disciplinary collaborations. As a whole, the activity centric narrative offers a landscape view of the many interdependent processes involved in information scholarly work.

Scholarly information activities and primitive
As discussed by Unsworth (2000) Scholarly primitives are basic functions common to scholarly activity across disciplines. He clarified the concept with a list of primitive-discovering, Annotating, comparing, referring, sampling, illustrating and representing-and provided further explanation with examples from humanities computing projects. Our concept of scholarly information activities is related goal emphasizes the explicit role of information in the conduct of research and production of scholarship (Palmer & Cragin, 2008). The concept of the primitive is distinct in that it is meant to refer to activities that are common across disciplines, at least within the humanities where the concept was originally developed and applied, and the examples provided by Unsworth suggest that the activities are relatively discrete in nature. In our implementation, we refine the concept further by Emphasizing a sense of the primitive have something at the base or beginning of a larger process. For example, in our framework, searching for information is interpreted as a scholarly information activity, while the more granular activities of chaining and browsing that contribute to the larger search and discovery process are considered primitive. What qualified strictly as an information activity or a primitive form stands as a basic information science research question in need of further empirical investigation. For example, the
University of Minnesota Libraries (2006) took a different approach in their project to Develop a framework for assessing support for scholarship on their campuses. They identified four general primitive-discover, gather, create and share-that “described the range of Activities Undertaken by scholars throughout the research process “(p. 38). Thus, while firm Criteria for Determining primitives have yet to be developed, the concept has proven valuable and intuitive in both the digital humanities and in DSL. For our purposes, the distinction between more general information primitive and associated activities has been helpful for structuring the array of findings on scholarly information that work currently exist in the literature. Both activities and primitive, we believe, tends to be common across disciplines and integral to how scholars create new works.

Additionally, they can happen at any stage of research, within data collection, analysis, and dissemination processes, or more during the formative stages of a research project. Our activity centric approach is reminiscent of some existing models of information seeking and use in LIS. One of the most well-known, developed by Ellis and colleagues, identified six common processes-starting, chaining, browsing, differentiating, monitoring and extracting, based on qualitative comparative analysis of information seeking in the social sciences, physical sciences, and literature (Ellis 1989, 1993, Ellis, Cox, & Hall, 1993). Competing process models exist (eg,
Foster, 2004; Kuhlthau, 1991), and revisions have been proposed based on new research. In particular, Meho and Tibbo (2003) suggested adding a number of activities to the Ellis model, such ‘re accessing, networking, and verifying, drawing on data from e-mail interviews with social scientists working in the research area of stateless nations. In a multi-method qualitative study, Palmer and Neumann (2002) showed that for interdisciplinary humanities scholars such a model should include exploring and translating the activities involved in working with information and colleagues in outside domains. As the case of Ellis’s model demonstrates, our understanding of the information activities that make up scholarly work, and how they map to scholarly communities, remains questionable and incomplete.

The sections that follow present the literature on scholarly information work framed around five core scholarly activities: searching, collecting, reading, writing and collaborating, with two or more primitive distinguished for each activity. Four cross-cutting primitives that are associated with more than one activity are also covered and serve as an important indicator of how the processes involved in the conduct of research and scholarship overlap and interact. Additionally, in the electronic environment “the flow of seeking, using, and creating information is becoming seamless” and new tools and resources are altering how scholars interact with information (Borgman, 2000). The activity / initial framework Allows us to see the components of this set of fluid Increasingly processes and how they may vary in application by researchers working in different fields. This report does not offer a comprehensive account of all possible primitive activities and involved in scholarly information work. The scheme, outlined below, was derived from the literature and enumerates Those With a basis of findings from empirical studies across multiple disciplines.

Searching
Searching Involves Deciding where and how to look for information. It may be performed with a single query and may result in the retrieval of one bibliographic record or one fact, but more often it is a complex and iterative process. It is A particularly important aspect of the “starting” or “opening” Stages of research identified in process models of information seeking (eg, Ellis 1993, Ellis et al.
1993 Foster, 2004; Meho & Tibbo, 2003). The searching activity in itself may open in a number of ways-with references to a current article or book, advice from a colleague, or keywords Aimed at finding specifics or for exploring a new area. The primitives associated with the activity-searching direct searching, chaining, browsing, probing, accessing, and distinctions are that apply in both the analog and digital information environment. Recent studies have tended to investigate on the Web searching, and searching for information Clearly online is a widespread practice across fields, but the actual impact of the shift to digital search systems remains difficult to Assess in any comprehensive way. One recent campus-wide survey reported that researchers in science and medicine almost exclusively preferred to search using electronic resources (Hemminger, Lu, Vaughan, & Adams, 2007). Approximately half of the interdisciplinary scientists surveyed in another study reported that their clothes were searching
markedly different from five years earlier due to the availability of digital resources (Murphy, 2003). As would be expected, studies of scholars have consistently found high use of search engines, especially Google, Which allow rival search across a wide and diverse array of sources. Has recent study examined decades of citations and the impact of online availability, showing that patterns have shifted as print browsing has become Eclipsed by searching online. It guarantees that scholars appear to be avoiding older, but falling, and literature to be focusing only on previously cited sources, Resulting in the use of a narrower, more homogenized range of literature (Evans, 2008).

Direct searching
Direct searching occurs when a scholar has a well-defined goal. For example, they may be looking for information on a particular chemical compound or trying to find a particular journal article. In most cases, direct searching is conducted with familiar keywords, names or other known terms in databases, online catalogs, search engines and online journals (Foster, 2004). Keyword searching predominates in digital resources, as seen in a survey of users of a Finnish national digital resource that showed a high level of keyword searches in both databases newspaper (63%) and reference databases (53%) (Vakkari & Talja, 2006). Keywords have been found to be important to historians for locating items known to be in an archive (Duff & Johnson, 2002) and for humanities scholars who frequently search for names, places, titles of works, and other proper nouns associated with familiar materials (Bates 1994, Bates 1996a, Bates, 1996b; Bates, Wilde, & Siegfried, 1995 Siegfried, Bates, & Wilde, 1993; Wiberley, 2003). In the humanities, keywords that represent subject – specific terms and other indicators of domain knowledge have also been shown to be important in
Formulating effective search strategies (Buchanan, Cunningham, Blandford, Rimmer, & Warwick, 2005).

Over the course of a research project, scholars use direct searching to confirm and verify their ideas facts. This kind of confirmation searching is used by humanities scholars to solidify their ideas and assist in recall of previous work done in a particular area. Studies of scientists, however, have shown that direct their searching tend to be Aimed at specific issues or a problem at hand when
conducting an experiment or writing up results (Palmer, 2005), or for checking the accuracy of information in hand (Ellis, 1993). Searching to check for accuracy of quotes and references has been Observed in the searching practices of interdisciplinary scholars (Foster, 2004; Meho & Tibbo, 2003) and for women’s studies scholars, identifying gaps in the literature was found to be Particularly
important at the end of a project (Westbrook, 2003). More uniquely, since interdisciplinary researchers often identify information that is “intellectually remote gold from unknown sources,” another layer of confirmation searching may be required “to interpret, verify and anchor the new material “(Palmer, 2005, p. 1144).

In the online environment, searchers tend to work more quickly and less deeply. A search pattern documented in one study of neuroscientists using PubMed is typical. Searchers faced with large retrieval sets only selected items from the first few pages of results, Although some did export result sets into bibliographic software for further review at a later time. Few searchers changed their queries or used the advanced search modes to get better results (Vibert, Rouet, Ros, Ramond, & Deshoullieres, 2007). It is common knowledge that search queries that contain imprecise terminology may result in very large retrieval sets, and misspellings may result in no results at all, yet many databases and online catalogs do not offer users suggestions for improving a query. Humanities scholars have noted their continued appreciation of library card catalogs for being more forgiving of minor terminology issues (Brockman, Neumann, Palmer, & Tidline, 2001). Search system recommendations, based on an evaluation of the University of California’s library services, have included incorporation of multi-lingual spell-checking, increased sensitivity to obscure scholarly terms, and presentation of options for alternative and related terms and topics. Provision of search expansions into other catalogs, like WorldCat, Amazon and other search engines, was also suggested, as well as access to librarian assistance via chat or e-mail reference (Bibliographic Services Task Force, 2005).

Chaining
Scholars depend on bibliographic references found in scholarly books, journal papers and Web sites to identify items to consult or read. This practice of backward chaining, or footnote chasing, has been confirmed as a distinct and prominent searching technique used across scholarly groups, ranging from humanities graduate students to Jewish studies scholars, and from sociologists and
computer scientists to researchers in economics and engineering (Barrett, 2005; Bronstein, 2007; Buchanan et al., 2005; Covi, 1999; Ileperuma, 2002; Vakkari & Talja, 2006; Westbrook, 2003). By following references, scholars are able to trace previous relevant publications. Forward chaining, or citation searching, is the correlate practice for finding subsequent relevant publications.
Bibliographic connections of this kind have been used for centuries to identify relationships among texts, but with networked information technologies, this kind of linking has exploded as scholars more readily chain through digital content (Bates, 2002).
For humanities scholars, chaining provides an important path to secondary materials via books, articles and reviews (Bates, 1994; Bates, 1996a; Bates, 1996b; Bates, Wilde, & Siegfried, 1995; Siegfried, Bates, & Wilde, 1993). But, other kinds of value are inherent in the practice of “mining” the expert bibliographies complied by others. Chaining works to build an understanding of the landscape of a field, shortens research time on a project, and helps in identifying the most important works on a topic (Brockman et al., 2001). One study found that the “seed documents” used for chaining by humanities scholars were particularly valuable for identifying sources not listed in standard indexes (Green, 2000). Among scientists, chaining has been shown as a key strategy for identifying older information for use in both teaching and research (CM Brown, 1999). The notion of chaining can be extended to include sources discovered through personal contacts, where the “link” takes the form of a suggestion from a colleague or collaborator (Meho & Tibbo, 2003). Evidence of this type of interpersonal chaining is found in studies across disciplines, and it is also discussed below as a kind of “consulting,” a primitive associated with collaborating activities. Women’s studies scholars, library and information scientists, graduate students in the humanities, sociologists, interdisciplinary scholars and astronomers have all been shown to rely on their colleagues for recommending relevant materials (Barrett, 2005; Covi, 1999; Spanner, 2001; Tenopir,
King, Boyce, Grayson, & Paulson, 2005; Westbrook, 2003; Zhang, 2001). A recent study reported that “using colleagues as information sources for journals was more typical in humanities than in other fields,” even though the act of chaining in general was a significantly more important approach in economics and engineering compared to humanities and medicine (Vakkari & Talja,
2006). Computer scientists have been relying on e-mail discussion lists to solicit references for some time (Covi, 1999). The practice may be more irregular with interdisciplinary scientists, but it can produce very high quality information. One study documented a case where a scholar sent a “cold contact” e-mail to a high-profile expert in an outside field and received a lengthy and highly
valuable bibliography in response (Palmer, 2005).

Chaining on the Web has been referred to as “quasi-footnote chasing,” since it usually combines linking among various kinds of digital content with search engine queries (Brockman et al., 2001). Studies have shown how geographers move between digital and print resources while chaining, performing iterative cycles of searching and working by following links in Web sites and online
journal articles, tracking citations in book reviews and in print publications, and searching library catalogs (Borgman et al., 2005). A study of how scholars use e-texts suggested that the difference between traditional chaining and browsing disappears in the online environment and is replaced with the practice of “netchaining,” which establishes and shapes “online information chains that
link sources and people” (Sukovic, 2008, p. 274-275).

Browsing
Browsing has long been recognized as an important and widely practiced information behavior (Bates, 2007; Chang & Rice, 1993; O’Connor, 1993). Unlike directed searching and chaining, browsing tends to be open ended with the searcher looking through a body of assembled or accessible information. As with other types of searching, the Web has had a tremendous impact on what and how scholars browse, and on the rate at which they can move through digital material from a diverse array of sources. For example, a study combining deep log analysis and surveys of Web usage found that users often engage in “bouncing” or “flicking,” moving rapidly from site to site and only occasionally returning to explore material in more depth (Nicholas, Huntington, Williams, &
Dobrowolski, 2004). Other studies confirm that print browsing continues to be of value to scholars. One large survey of scholars in the humanities and social sciences reported that approximately 80% considered browsing the library shelves to be an important, although infrequent, activity (University of Minnesota, 2006). Among studies of scholarly information use more generally, browsing has been strongly associated with the humanities and interdisciplinary fields (Bronstein, 2007; Ellis & Oldman, 2005; Meho & Tibbo, 2003). Nearly all the respondents in one small study of interdisciplinary humanities and social science scholars engaged in some kind of browsing as part of their research process (Spanner, 2001). In the sciences, there appears to be more variation. For example, surveys have demonstrated that physicists, chemists and biologists identify up to half their reading material by browsing, but browsing among astronomers was considerably lower, at 20%, possibly due to the more comprehensive and integrated online information systems in the field (Tenopir et al., 2005).

Collections of various kinds lend themselves to browsing. Studies of humanities scholars and geographers have shown that library shelves are valued as a browsing environment, especially sections devoted to new books and journals (Borgman et al., 2005; Brockman et al., 2001). Women’s studies scholars have relied heavily on browsing both publisher catalogs and bookstores, due in part to the less developed base of indexing and reference sources in newer interdisciplinary fields (Westbrook, 2003). Case studies of neuroscience research have demonstrated how the speed and flexibility of digital browsing can encourage review of material that might otherwise be ignored, such as compilations of conference poster abstracts that cover a multitude of very current research
projects (Palmer, Cragin, & Hogan, 2007). Web browsing can also lead scholars to more conventional library resources that might not have been pursued through a library portal or gateway (Zainab, Huzaimah, & Ang, 2007). Moreover, table of contents browsing in journals has been readily adopted in the electronic environment, in conjunction with the follow-on activities of accessing and
assessing, discussed further below (eg, Eason & Harker, 2000; Eason, Yu, & Harker, 2000). More specifically, studies have suggested that social scientists retrieve recent articles of interest through vertical chaining, moving from table of contents, to abstract, to full-text, while scientists browse journal titles and then perform vertical leaping, going directly to the full-text (Tenopir, 2003).

Among searching techniques, browsing is notable for its potential to result in serendipitous discovery. Because browsing tends to be broad and flexible, scholars encounter materials that would not be found through searching or chaining, and the new information may stimulate unexpected and fortuitous intellectual connections. A large survey of scholars in the United Kingdom showed that few scientists valued print collections for serendipitous browsing, but scholars in the arts, humanities, area studies and languages were twice as likely to consider it an essential aspect of browsing (Education for Change, 2002, p. 25). Interdisciplinary researchers have reported that physical libraries are more conducive to serendipitous discovery than digital libraries, and other scholars have reported that the ability to browse is a distinct benefit to having a library in close proximity (Borgman et al., 2005; Engel & Antell, 2004; Foster, 2004). In fact, some humanities scholars have reported that browsing is “difficult” to perform in the electronic environment (Buchanan et al., 2005).

Probing
For scholars seeking information across multiple domains, standard searching and browsing approaches can be inadequate due to the scatter of information and the disparity in vocabularies across fields (Mote, 1962; Weisgerber, 1993; White, 1996). Probing is an exploratory strategy used by interdisciplinary researchers to find relevant information that falls outside their discipline or area of expertise (Palmer, 2001b; Palmer & Neumann, 2002). Similar to browsing, probing may be loosely directed across a topic area, but it is distinct in its investigative nature and aim of identifying information in unfamiliar domains. It encompasses breadth exploration, identified by Foster (2004)—a “deliberate expansion of information horizons to bring within range different information types, sources, concepts, and disciplines” (p. 233). Not all probing is aimed at broadening or expanding the search scope, however. While researchers frequently probe into peripheral areas to increase their breadth of perspective and generate new ideas, probing in an outside field can also be deep and directed at solving a particular problem or locating a particular piece of missing information (Palmer, 1999b).

Cross-disciplinary searching and probing are not practiced exclusively by interdisciplinary researchers. Studies have documented the practice in various academic disciplines. Examples include molecular biologists conducting broad-based searches in databases to learn about unfamiliar areas and to stay current (Covi, 1999), and historians and music scholars searching for materials across a wide range of fields, including philosophy, anthropology, art history, literature, statistics, sociology, criminology and geography (CD Brown, 2002; Case, 1991). Domain-specific searching can also be “probing” in nature, especially when a research focus is not yet well defined or understood. Studies of how domain knowledge is applied by users, or represented in search tools, have implications for the support of scholarly probing. For example, one study of humanities scholars suggested that domain knowledge improved success with conceptual searches, but subject classifications were not useful due to their misalignment with how scholars conceptualized their fields (Buchanan et al., 2005).

While much of the documented probing activity has been associated with exploratory database searching, interesting examples can also be found in studies of archival information work. A study of the research process of historians showed how they probe archival finding aids to discover unknown source material and to identify new keywords for expanding their base of searching (Duff &
Johnson, 2002). This study also gave an account of interpersonal probing, in which the historian engaged the archivist in conversation and deliberately avoided the use of known keywords to better elicit additional terms. Probing far afield to identify keywords or find new information from an outside discipline can, of course, introduce terminology problems. “Translating,” to construct search queries or to interpret content from an outside domain, is one of the cross-cutting primitives discussed below.

Accessing
Information service providers understand that providing “discovery is not enough;” scholars want direct access to the materials they identify (Bibliographic Services Task Force, 2005; Research Information Network, 2006). This expectation may be reflected in a recent report showing that over a three-year period scholars were increasingly less likely to view libraries as a gateway to information, with the trend markedly greater for the sciences and the social sciences compared to the humanities (Houseright & Schonfeld, 2008). Another suggestion of access needs can be seen in a study that showed how chemists at one institution “created and relied upon their own list of relevant e- journals” rather than using those created by the library, showing that personal tools “increase efficient connections to what they consider to be their core literature” (Davis, 2004, p. 331). When full-text digital library content is available, however, scholars across fields have exploited its efficiency and convenience (Barrett, 2005; Brockman et al., 2001; Eason, Richardson, & Yu, 2000; Hallmark, 2004).

The lower levels of production and distribution of digital sources is an important factor in lagging adoption in the humanities. In one large-scale survey, arts and humanities scholars were found to be three times as likely as medical, biological, and physical scientists to consider physical access to library collections essential to their research, and twice as likely as social scientists (Education for
Change, 2002). Studies of humanities scholars have also demonstrated a continued reliance on primary materials held in special collections, archives and museums, coupled with regular travel to work on site with physical resources (Brockman et al., 2001; Case, 1991; Palmer, 2005; Palmer & Neumann, 2002; University of Minnesota Libraries, 2006; Wiberley, 2003).

In the sciences, e-journal use is dominant and strongly preferred over print (Hemminger et al., 2007; Tenopir et al., 2005). Some researchers save electronic copies for later access, but many studies have shown that scholars tend to print out articles or other sources for later reading (CM Brown, 1999; Eason & Harker, 2000; Eason, Richardson, & Yu, 2000; Murphy, 2003; Tenopir et al., 2005). For example, even though astronomers accessed about 80% of their readings from electronic sources, they printed out more than half on paper before reading, and less than one-fourth were read on a computer screen (Tenopir et al., 2005). That study found that HTML was favored for reading online and PDF format was preferred for printing. A “deep log analysis” of scholarly databases indicated that Blackwell Synergy users accessed documents in PDF twice as often as HTML, and Emerald Insight users accessed PDFs 56% of the time (Nicholas, Huntington, Jamali, & Watkinson, 2006). Another study reported that 70% of scholars surveyed preferred PDFs to HTML (Zainab et al., 2007). Print is still considered by many to have a distinct portability advantage, and
improvements are needed in e-text functionality for backward and forward movement through pages within a document and between different documents (Institute for the Future, 2002).

Collecting
As researchers search for and access information, they build personal collections that support their current and long-term research. Gathering and organizing are the primitives associated with scholarly collecting of research materials. Unfortunately, only a few early studies have specifically examined the personal collecting behavior or scholars, and therefore there is limited understanding
of patterns in content and use. The practice seems to be continuing over time with some variation due to e-resource availability. For example, studies ten years apart reported similar use of personal journal collections by medical and biological scientists for locating articles and keeping up with research in their field (Curtis, Weller, & Hurd, 1997; Kuruppu & Gruber, 2006). On the other hand,
longitudinal surveys have tracked a decline in scientists’ annual personal subscriptions, from an average of 5.8 titles in 1977 to 2.2 titles in 2002, with reliance on library collections increasing with the availability of e-journals (Tenopir et al., 2003; Tenopir et al., 2005).

As will be discussed briefly here and again more thoroughly in a later section on data sharing, reuse of data collections is a topic of great interest in LIS and is considered to be particularly important for data-intensive fields in the sciences. Personal scholarly collections, consisting primarily of documents rather than raw data, have also been perceived to have potential value for other users (Spanner, 2001), but there is little evidence of actual sharing practices. One study reported that one-third of fine arts respondents referred their students to resources in their own collections, because of its superiority and relevance over their university’s collection (Reed & Tanner, 2001). Nearly half of the scholars in another survey felt that their personal collections would be of value to other researchers (University of Minnesota Libraries, 2006).

Gathering
Collections may be built due to a need for long-term accessibility, convenience or to support sustained work with a set of resources. An early, broad survey of humanities scholars, social scientists and scientists indicated that some sort of personal collection, consisting mostly of monographs and journals, was maintained by most respondents (Soper, 1976). A later study revealed similar gathering patterns, but showed that social scientists’ collections were more likely to also include items such as bound reports, manuals and loose leaf materials (Case, 1986). The RLG studies reported on collections developed by academic departments tailored to disciplinary needs of their faculty that included materials such as monographs, journals, reports, prepublication papers, maps and photographs. More specifically, engineering labs often collected technical reports and linguistics departments collected significant dissertations in the field (Gould, 1988; Gould & Handler, 1989; Gould & Pierce, 1991).

Among humanities scholars and social scientists participating in a recent study, 37% claimed to have gathered unique research collections, and 56% reported engaging in “personal archiving activities” (University of Minnesota Libraries, 2006). Humanities scholars, in particular, have been shown to cultivate and take pride in their personal collections of books and other print sources
(Brockman et al., 2001). Their collections “are a necessity since rereading is a significant part of their interpretive work. Any number of texts may require periodic or systematic reading, and some may be ‘read’ for years or decades” (Palmer, 2005, p. 1144). Scientists also gather collections of literature in the form of journals, conference proceedings, and individual photocopied and digital papers. Approximately 85 to 95% of the scientists surveyed in one study maintained collections of reprints and article copies, and 63% of chemists reported personal collections of over 500 reprints (CM Brown, 1999). In a more recent survey, 70% of faculty in science and medicine kept both print and electronic article collections (Hemminger et al., 2007). Scientists have reported the high value
of being able to gather together large quantities of digital papers and have them mobile on their laptops (Palmer, 2005). Scientists also collect the data they generate through experimentation and field studies, as well as data produced by other researchers for modeling purposes. In geography, for instance, collections of field notes are maintained, maps are collected in paper and digital forms and serve as both primary and reference resources, and “by mid-career, many have built substantial image collections of their own” (Borgman et al., 2005). With data becoming increasingly digital and more easily mobilized for other purposes, some researchers are becoming involved in data repository and federation efforts (Palmer, 2005), and some agencies are moving toward requirements for long-term data management plans for projects they fund.

Primary materials in the humanities are not generally generated by the researcher, but collected from a range of sources in the form of texts, images, facsimiles and artifacts. Other primary sources frequently collected include manuscripts, letters, plays and photographs (University of Minnesota Libraries, 2006). The qualitative study of art historians, mentioned in the introduction, examined how researchers developed their own collections of art reproductions to “compensate for inaccessible or deficient institutional collections,” and because they “have the advantage of being personally selected and indexed” (Bakewell et al., 1988, p. 19). Fine arts faculty collect books, videos, plays and musical scores (Reed & Tanner, 2001), and literary theorists prefer to purchase the books they analyze, rarely relying on borrowed copies (Covi, 1999). Historians’ collections may be among the most diverse, since they rely heavily on their own data recorded in personal notebooks filled with annotations, facts and references, and collected cultural artifacts, such as items found in junk shops that relate to the time, place, or object of study (Case, 1991).

Organizing
Over time, personal collections become larger and more complex assemblages, and scholars devise organizational systems and tools for storing and managing the content. Accordingly, management of information was one of the activities proposed in Meho and Tibbo’s (2003) extension of Ellis’s model of scholarly information seeking discussed in the introduction. Most of the studies that have examined the organization of personal collections have been focused on the humanities or the social sciences. For example, despite the prevalence of digital materials, 98% of humanities and social science faculty in one university study reported keeping hard copies of print materials because of fear of computer failure, lack of technological skills, and computer storage space limitations (University of Minnesota Libraries, 2006). Much less is currently known about scientists’ approaches to managing their collections of literature and data, possibly because organization is less of a concern or assumed to be more straightforward in the sciences.
Humanities scholars often develop personalized organizational systems for their collections. Arrangement and storage of materials may vary from piles on the floor to structured file folder systems and elaborate databases (Palmer & Neumann, 2002). As described in one study, a historical biographer captured each moment of the life of a person on individual 3×5 cards (Case, 1991). Art historians in another study developed custom approaches to organizing materials in accordance with their needs for both teaching and research, and expressed a need to develop more detailed cataloging systems (Bakewell et al., 1988). Reports have also demonstrated that scholars from various fields have been adopting citation management tools to assist with organizing digital
content (Borgman et al., 2005; Brockman et al., 2001; CM Brown, 1999). At this point in time, scholars have recognized that the materials they collect have potential value to other researchers, but they consider their idiosyncratic organizational systems, as well as copyright restrictions, to be barriers to sharing (University of Minnesota Libraries, 2006). On the other hand, some scholars are
creating highly sophisticated digital collections around their personal scholarly interests that are meant to be shared. For example, the thematic research collections being developed by humanities scholars are scholarly products that that bring together specialized source material, tools and expertise to support inquiry in a specific research area (Palmer, 2004, 2005).

Articles, Library management, Research
cultural policies, discourse analysis, public library