College and Research Libraries 50th Anniversary Feature- Old Forms, New Forms: The Challenge of Collection Development Ross Atkinson Collection development is intended primarily to improve the academic library's ability to fulfill competing information responsibilities with chronically inadequate resources. In order to meet this challenge, collection development has sought to create a system out of the processes which are already endemic to selection. If this system is to progress, if it is to adapt to rapidly chang- ing technical and economic conditions, it must have the capacity to exert greater control over scholarly and educational information. Steps toward this increased control should include the categorization of sources and access by function rather than merely by subject, the ongoing definition of a title-specific core, and the development of prescriptive access and collection poli- cies. t is testimony both to the per- ceived significance of collection development and to the status of the library in the academy that the primary responsibility for the se- lection of library materials has passed from faculty users to academic library staff. The transfer of that responsibility, which began in the 1960s, has still not run its course. 1 Most larger academic libraries have by now assumed full responsibility for selection, although even in some of these larger institutions the transfer of au- thority has occurred quite recently. 2 The development of collections has, to be sure, always been a basic concern of all types of libraries, but what we today understand as academic library collection develop- ment is to a great extent the ongoing sys- tematization and professionalization of collection building and management which has evolved both as a product of and as a rationale for this transfer of the se- lection effort from faculty users to library staff. The reasons the academic library needed to assume responsibility for selec- tion have been frequently discussed. 3 The most important of these were probably (a) a rapid increase in funding and research, supported mainly by federal subsidies, 1 and (b) the increasing realization, which began at least as far back as the 1936 com- parative study by Douglas Waples and Harold Lasswell, that superior research collections could be built by professional bibliographers.5 From a more general per- spective, the transfer of selection respon- sibilities was intended to create a mecha- nism to improve the academic library's ability to respond rapidly and rationally to the manifold information needs of its us- ers. How collection development has sought to achieve that objective, and what further actions need to be taken in order to refine that ability, will be the subject of this paper. THE RECONCILIATION OF LIBRARY FUNCTIONS The academic library has neither a single mission nor a homogenous constituency, Ross Atkinson is Assistant University Librarian for Collection Development and Preservation at Cornell Uni- versity, Ithaca, New York 14853-5301 . 507 508 College & Research Libraries but rather is obliged to respond to a multi- plicity of academic needs and interest groups. Although there are many schemes which could be used to catego- rize these responsibilities, let us posit for the purposes of this discussion five essen- tial functions which the academic library attempts and is expected to fulfill. 1. The notification function. The aca- demic library continues to serve as the principal (although never exclusive) means by which scholars communicate the results of their research to each other across space and time. 2. The documentation function. The ac- ademic library maintains the essential raw data upon which many disciplines base their research. 3. The historical function. To all li- braries, but to the academic library espe- cially, falls the responsibility for maintain- ing the records of civilization, without which the future will be denied access to the past. 4. The instructional function. The stu- dents, whose education is after all the pri- mary purpose of all academic institutions, depend upon the library as a means to supplement and enrich their learning. 5. The bibliographical metafunction. In order to achieve the preceding four func- tions, the library must promote and facili- tate access to information sources. We must note at once that these func- tions have very different characteristics. The first four functions are direct re- sponses to user needs, while the biblio- graphic metafunction drives and regulates the other functions. The historical func- tion exists to ensure that records which one day may be needed will still be avail- able. It is closely connected to, but must be distinguished from the documentation function, which is not a long-range curato- rial responsibility; the documentation function is rather intended to provide ac- cess to information presently needed, es- pecially for the humanities and social sci- ences. To respond to the current needs of historians is to fulfill the documentary function. To maintain materials or access to databases for future generations, on the other hand, is to respond to the historical function. Thus while the historical func- September 1989 tion is intended to serve future scholars (or at least all future scholars concerned in any way with history), the documentation function responds directly to the needs and interests of contemporary clientele. All academic libraries normally serve all five functions to varying degrees, depend- ing upon available resources. Those resources-funding, staffing, space-are always and have always beeri limited. The five functions are therefore in a state of perpetual competition for inevitably in- adequate resources. The fundamental re- sponsibility of academic library collection development (although it may not always have been viewed in these terms in the · course of its evolution) has been and re- mains the reconciliation of such compet- ing library functions. The balancing of competing responsibil- ities is, of course, necessary for all library operations, but collection development has been, as we shall see, especially well designed to achieve such a purpose. This capacity will doubtless become increas- ingly evident (and, one hopes, effective) as more sources of information become available in electronic format. Although electronic publication is not proceeding nearly as rapidly as was once expected, there can be little doubt that many paper publications will eventually be replaced by sources in electronic form. While this will alter the nature of collections signifi- cantly, it is unlikely that it will induce changes in the fundamental purposes of collection development, because the cost of meeting all information needs for in- struction and research will very likely con- tinue to exceed available resources. Medi- ating among those competing needs, reconciling divergent academic library functions with conspicuously inadequate resources, will remain the fundamental responsibility of collection development, regardless of the formats in which schol- arly and instructional information is pub- lished.6 THE FOUR CONTACT GROUPS In her frequently cited 1973 dissertation, Elaine Sloan characterized collection de- velopment as a boundary spanning activ- ity: Collection development is viewed as an activity which is at the boundary of the organization and which also engages in extensive intra- organizational transactions. Those who are re- sponsible for developing collections will be re- quired to interact with users of the collections, who are outside of the boundaries of the uni- versity. Within the boundaries of the library, those responsible for collection development may interact with public service librarians who are in contact with users and with technical ser- vice librarians who are in contact with dealers and publishers. Those responsible for develop- ing collections will therefore be required to co- ordinate their activities with many other orga- nizational units.7 The competing functions which the aca- demic library must fulfill are not only ab- stractions. Those functions are also pow- erfully represented by interest groups of varying authority with which collection development librarians maintain routine contact. We can distinguish these primary contact groups according to their relation- ships with the local institution and/or the library profession (see figure 1). Providing an acceptable but practicable response to the disparate needs, demands, aspira- tions, and biases of these four groups is the activity in which many collection de- velopment librarians are engaged much of the time. Two initial assumptions relating to this scheme should be noted. First, the most immediately perceivable political influ- ence clearly originates from above the hor- izontal line, i.e., from local institutional forces. When there is clear competition for resources between institutional and non- institutional contact groups, the former groups will usually prevail. Second, most of the economic power lies outside of the Professional Old Forms, New Forms 509 profession, to the right of the vertical line. One reason, for example, that the achieve- ments of cooperative collection develop- ment have been relatively modest is that the main proponents or representatives of cooperative activities are often collection development staff at other institutions, · and they have neither the political author- ity nor the economic influence to compete with the demands represented by other contact groups. We should also note parenthetically that, while boundary spanning is indeed an accurate and insightful description, there are also aspects of collection devel- opment which have necessarily been at the same time boundary-defining. A link- age can only be achieved if the linking agent becomes a true third component, distinguished from those elements on ei- ther side of the boundary. Collection de- velopment, in order to establish its own identity, has been compelled to disengage itself from the two key contact groups at the institutional level, other library staff (usually in the acquisitions department) and faculty users. It is in fact very difficult to establish a collection development pro- gram without temporarily weakening the connection between the emerging pro- gram and those two institutional groups from which the program is assuming its responsibility and authority. Once the col- lection development program is in place, however, and its legitimacy is no longer suspect, a primary objective must be tore- establish and to reenforce those local con- tacts as rapidly as possible. Let us now consider the relationship of collection development to each of its four main contact groups. Nonprofessional Institutional Local library staff outside of collection development Users Noninstitutional Collection development librarians elsewhere FIGURE 1 Collection Development Contact Groups Publishers 510 College & Research Libraries ''During periods of austerity, faculty users often lobby to protect acquisi- tions at the expense of other critical library operations.'' Library Staff Outside of Collection Development Of all the diverse functions for which a library is responsible, it is the key biblio- graphical metafunction which remains the most obscure to users. This is especially the case with fundamental processing ser- vices. Because the main political and eco- nomic authority in the academy resides with the users, the bibliographical meta- function can become vulnerable. During periods of austerity, faculty users often lobby to protect acquisitions at the ex- pense of other critical library operations. ''Faculty members will accept many radi- cal changes as long as funds are available with which to buy essential material. This is an area where miscalculation can bring disaster; if allowed to grow haphazardly, this budget [i.e., for acquisitions] will de- vour other funds.' ' 8 Because collection de- velopment has become the library's pri- mary link with faculty, and because faculty generally recognize collection de- velopment librarians as the representa- tives of the collections, it becomes an es- sential responsibility of collection officers to educate faculty as to the dependence of the collection on quality processing and staff. Collection development is therefore in a special position to protect the biblio- graphical metafunction by translating the values and concerns of the library into those of the faculty users. Library Users Local library users are clearly the most prominent and influential contact group for any collection development operation. Much effort has been devoted to the de- sign of surveys and other mechanisms to identify user needs and attitudes. 9 Re- cently the competing needs of current fac- ulty users have become especially prob- lematic as a result of the escalating prices September 1989 of materials in the sciences. 10 We have now become very sensitive to the fact that such prices have driven the cost of fulfill- ing the notification function in the sci- ences many times higher than the cost of fulfilling that same function in the human- ities and social sciences. Selection respon- sibility has been assumed by the library in order to ensure, among other things, that the basic information (notification, docu- mentation) needs of all faculty users are being met as consistently as possible within the confines of available resources. In a very real sense, therefore, collection development has been created to deal with exactly the kind of crisis we currently face, so that a test of collection develop- ment is now under way: if methods can be devised and resources chanelled to meet the competing information needs of dif- ferent faculty user groups in the face of the rapidly declining purchasing power of li- brary budgets, then collection develop- ment will have demonstrated its utility to its parent institutions. Different constituencies among current faculty represent only one of the compet- ing needs of library users. There are at least two other essential categories of com- peting user needs which collection devel- opment is expected to address. First, there is the competition between undergradu- ate and faculty needs. Graduate students probably do not form an immediately ap- parent, separate constituency, because their needs are in many cases identical with those of the faculty, but undergradu- ates often require very different material from that pursued and used by faculty. For most subjects in most academic li- braries, the instructional function will be the highest priority. The academic library must therefore acquire material specifi- cally intended for and used by undergrad- uates in fulfillment of its instructional function. When the notification and the instructional functions begin to compete vigorously for strained resources, it be- comes an urgent responsibility of collec- tion development to ensure that the ca- pacity of the collection to support education is not undermined by the li- brary's obligation to foster communica- tion among scholars. ''Responding to the historical func- tion is difficult, because the constitu- ency to be served has not yet arrived, while the other, competing functions (notification, documentation, in- structional) all serve the needs of cur- rent users." In addition to the competing needs of current faculty, and the competition for resources between faculty and undergrad- uates, there is a third category of compet- ing user needs, which is certainly the most difficult to mediate: it is the conflicting re- quirements of present and future users. To serve the needs of future scholars is the library's historical function. Materials no longer necessary for notification (or even documentation, such as superseded edi- tions) must be maintained-not every- where, but somewhere-for future histor- ical research. This consumes space and staff resources which could be applied to the fulfillment of the other library func- tions. Responding to the historical func- tion is difficult, because the constituency to be served has not yet arrived, while the other, competing functions (notification, documentation, instructional) all serve the needs of current users. The larger the research library, moreover, the more criti- cal becomes the historical function, al- though academic libraries of all sizes can and must contribute to the effort. The real- ization of the historical function can in fact only be achieved effectively by the coordi- nation of collection decisions among aca- demic libraries. The Collection Development Community The successful development of aca- demic library collections, especially dur- ing periods of budgetary distress, de- pends upon the exchange of information and the coordination of planning and op- erations among collection development officers at different institutions. One effort to improve coordination has taken the form of standards and guidelines to en- Old Forms, New Forms 511 sure adequate and equitable service to cur- rent and future clientele in all institu- tions.11 Such published standards are essential, but also abstract, so that the value of their application is difficult to as- sess. The other, more practical method to im- prove coordination has been cooperative collection development, which has been a goal of academic libraries for many years. 12 The first decade of College & Research Li- braries contains several calls for improved cooperation in the development of library collections. 13 The arguments and the rec- ommendations presented in those articles are not at all unlike positions still taken to- day, which is evidence of how modestly we have progressed in this area. There are a variety of cooperative programs now in operation, but few of these seem to be having demonstrable effects. The recent survey by Joe Hewitt and John Shipman on cooperation among ARL libraries re- vealed that cooperative programs "must, for the most part, still be described as somewhat poorly delineated or even em- bryonic. The most important finding of the study relates to the level of interest and activity directed toward establishing cooperative collection development rela- tionships, rather than specific program ac- tivities.' ' 14 While there is great enthusiasm for coc:;>peration, there has been consider- able difficulty actually implementing such programs. Why cooperative programs have not worked as well as expected, de- spite the significant quantities of time, money, and intelligence devoted to them, remains a source of continuous specula- tion and frustration for the collection de- velopment community. Joseph Branin has recently compiled a list of the standard reasons for program inadequacies, and he has also provided some sound sugges- tions for solutions. 15 The simple fact may be, however, that the historical function, which cooperative collection develop- ment is primarily intended to promote, is being given a lower priority in most aca- demic libraries despite our efforts to sup- port it. The Publishing Community The contact group that is least under- 512 College & Research Libraries stood, most alien, and increasingly dis- trusted, is the publishing sector. The con- cerns and motives of publishers remain obscure to libraries, because publishers are normally not directly connected with either the institution or the profession. Publishers are critically important to the fulfillment of library functions, but unlike the other three contact groups, they are not proponents of any particular library function. The commercial publishers es- pecially operate on the basis of a value sys- tem which is relatively foreign to those of both the academy and the library profes- sion. It has become clear recently, that the values and aspirations of at least some members of this group are having the most significant impact on the library's abili~ to accomplish its multifold mis- sion. 6 THE DRIVE FOR SYSTEM The professionalization of collection de- velopment derives in part from the real- ization that subject knowledge is a neces- sary but insufficient prerequisite for selection. Another special form of knowl- edge is needed to ensure the equitable use of resources and the creation of balanced collections which were sometimes jeop- ardized when selection was done exclu- sively by faculty. The primary motivation behind the burgeoning literature of collec- tion development has been to create a system-a coherent, self-validating process-which can rationalize collection decision-making. The origination of such a system, we must also acknowledge, has some clear rhetorical benefits, in that it can be used to emphasize the care and profes- sionalism with which the development of collections is now being conducted by li- brary staff. This systematization of collection devel- opment has been achieved, for the most part, by regulating or formalizing features which have always been central to the col- lection development process . Much of the literature on budget allocation, for exam- ple, has been concerned primarily with the use of formulas. But a budget formula is merely the automatic application of pre- determined factors relating to such issues as needs and use, which are routinely September 1989 taken into account in the course of alloca- tion anyway. In a sense, therefore, budget allocation is always based on unwritten formulas, which are merely imperfectly applied. The formula simply ensures that those factors are articulated and invariably considered. The problem is that the de- mands on the acquisitions budget are so various, the competing needs so diverse, that no ''magicformula'' can possibly take all such factors into account. 17 The major attraction of the budgeting formula is rhetorical: it serves "to con- vince faculty members and departments that their allocations are fair.'' 18 Most col- lection development budgeting remains in any case necessarily imprecise, because of the inability of the library to predict publication costs and patterns. (This is perhaps partially a result of the poor rela- tionship and communication between li- braries and the publishing community.) Most budget allocation is therefore based upon past spending rather than upon pro- jections.19 · The creation of a unified collection pol- icy is also intended to articulate and ren- der consistent criteria which are often al- ready being applied by selectors. The purpose of the policy is to raise those crite- ria to consciousness, to compare and to co- ordinate them, occasionally as a prelude to adjusting them so that they meet the varied and competing needs of the institu- tion as consistently as possible. But the collection policy, like the formula budget, while certainly a significant step in the di- rection of systematic decision making, re- mains defective as a coordinating tool. The reason is that most of our policy state- ments (includirig the Conspectus) are pri- marily descriptive: they merely articulate the current condition (''existing collection strength") and direction(" current collect- ing intensity") of the collections. 11Collection policies . .. fail to stipu- late in detail how future collecting should be adjusted in response to changing economic and technical conditions." Some policies provide . an indication of the direction in which the collection should be moving (''desired collecting in- tensity"), but even the inclusion of this feature cannot compensate for the policy's lack of prescriptive authority. Collection policies, in other words, fail to stipulate in detail how future collecting should be ad- justed in response to changing economic and technical conditions. The collection policy also frequently fails to reflect clearly the broader goals of the library. The pursuit of system has also been in- tense in the most fundamental area of col- lection development literature, selection theory. Publications on selection written in the 1940s and 1950s were for the most part elegantly phrased opinions of learned men who seldom doubted their capacity to distinguish between significant and in- ferior publications. Their major criticism of academic library collections was that the stacks were being clogged with materi- als of questionable quality. 20 Once the li- brary assumed responsibility for collec- tion development, however, it quickly found itself beset by precisely the same in- ability (previously presumed to be a fac- ulty malady) to distinguish important from less essential publications. This prompted Margit Kraft to warn in 1967 that American libraries had "forsaken the responsibility for judging quality," and have thus become ''enamored with quan- tity."21 A decade later Daniel Gore was still making essentially the same charge. 22 Little progress has, in fact, been made re- fining and coordinating selection criteria. While academic libraries can no longer af- ford to collect as broadly as they did in the 1960s, the qualitative basis for their reduc- tions have been poorly articulated and for the most part uncoordinated. The desire to systematize selection may have reached a kind of apex in the recent work of Robert Losee, who has devised se- lection formulas, which he urges collec- tion development officers to use in order to render selection ''more scientific and thus more productive." 23 Like formula budgeting, such a quantitative approach to selection has great rhetorical value, but whether it is possible or desirable to apply such a system in the real world of com pet- Old Forms, New Forms 513 ing information needs must remain open to question. All selection methods, and especially those used for cooperative collection de- velopment, must be founded, to be sure, upon some kind of articulated gradation of source qualities. Of the many formula- tions of utility actually used in libraries and in consortia! agreements, the most fa- miliar and most frequently applied is probably the concept of the core. The word "core" certainly has its rhetorical value, too, because most people probably associate the word with the ultimate ob- jective science, nuclear physics. Despite the frequent and confident use the term receives, however, it usually remains not much more than a metaphor for "impor- tant material.'' This is not to say that there is no core, or that there are no core publi- cations. Such a statement would be c~­ cal and counterproductive. Certainly there are core journals, core documents, core editions and texts which anyone fa- miliar with the relevant field could iden- tify. Such items are viewed by consensus as indispensable for research and educa- tion. A real core must have a periphery- some boundary which separates it from the remainder of the universe of publica- tion. Our effort to establish that bound- ary, to distinguish core from non-core ma- terials, has been so far singularly unsuccessful, except through such retro- spective methods as citation analysis or the use of circulation records. 24 For pur- poses of planning, budgeting, or coordi- nation, the concept of the core, for all its use, is practically useless. Something be- tween the algorithms urged upon us by our colleagues in information science and the currently vague metaphor of the core needs to be established, if our effort to de- velop a system for collection decision- making is to move forward. It should be noted, finally, that the drive to create a systematic basis for collection development is also partially a response to and an application of the increasing serial- ization of scholarly information. Periodi- cals have long played an essential role in scholarly communication, but recently we have become especially conscious of the 514 College & Research Libraries extent to which they have come to domi- nate our collections and our budgets. Be- tween 1978 and 1987, the number of jour- nals published in the sciences quadrupled. 25 This prevalence of serials will probably continue, so we had best learn as much as we can about their special bibliographic and epistemological quali- ties. It is clear that the distinguishing charac- teristic of the serial, as opposed to the monograph, is its diachronic context: each article can be perceived as a dependent component of a single text which is the en- tire, ongoing journal. Thus the quality of any article published in a scholarly journal is at least partially anticipated and judged by the reader on the basis of his or her con- clusions about articles read previously in the same journal. Each article is a kind of chapter in an ever expanding treatise. Every selector is well acquainted with the problems this causes when there is a need to undertake cancellations: when a journal is cancelled, its users invariably interpret such action as an amputation. The backfile of the can- celled journal is then perceived as a defec- tive part which is no longer useful, be- cause it has been separated from its whole. To cancel a journal is to interrupt a conversation. This is traumatic not only for the user but also for the library, be- cause it appears as a reduction or even a repudiation of that systematization which collection development strives with such zeal to create and maintain. FUTURE DIRECTIONS Much has been achieved in academic li- brary collection development since the as- sumption of selection responsibilities by the library began almost thirty years ago. Collection development has created at least a rudimentary system for the bal- anced fulfillment of competing functions, and that system has been successfully ap- plied. But it remains equally clear, that more must now be done, more responsi- bility assumed, more control sought, more boundaries spanned, if the success of collection development is to be sus- tained. The remainder of this paper will focus upon three specific recommenda- September 1989 tions to improve further the current effec- tiveness of collection development: cate- gorization by function, core definition, and refinements in policy. Categorization by Function Collection policies, materials budgets, and the distribution of selection responsi- bilities among selectors tend to be divided along subject lines. 26 Collection develop- ment's heavy reliance upon subject divi- sions is at least partially a vestige of the time when selection was primarily a fac- ulty responsibility. Many acquisitions budgets, for example, continue to be di- vided as if they were being allocated to ac- ademic departments. One disadvantage of the subject divi- sion of materials budgets is that it throws into stark relief the competition among subjects for library resources. It is for this reason that the frequent reliance upon subject divisions has been a serious im- pediment to cooperative collection devel- opment. Such cooperative arrangements often entail the agreement by participat- ing institutions that they will collect in cer_. tain subject areas to the exclusion or dimi- nution of others. Faculty concerned with the subjects targeted for deemphasis un- derstandably oppose the imElementation of such cooperative plans. 27 We need, therefore, a more refined method of col- lection categorization-not to conceal the competition among subjects, which unde- niably exists, but rather to clarify its com- plexities and to create a basis for more practical collection goals. Such categories should reflect the use of library materials, so that the effects of collection planning upon research and instruction will be . more apparent to local clientele. While it remains impractical to abandon subject categories entirely, we might im- prove selecting, budgeting and coopera- tion, by subdividing subjects according to function. For purposes of this discussion let us simply apply the five functions iden- tified earlier as a basis for distinguishing sources (or, more precisely, source access regardless of format). For each subject we might identify the following: 1. Notification sources. These are mainly journal articles and monographs written by scholars for other scholars in the same or related fields of research. 2. Documentation sources. In this cate- gory are all primary materials. Examples of these include data sources for the social sciences, original publications such as dia- ries or newspapers used by historians, and the original works and authoritative editions of standard authors for the hu- manities. 3. Instructional sources. This includes summaries of knowledge, such as text- books or manuals, intended to provide in- troductions to and exercises in standard subjects taught at the institution. 4. Historical sources. These are sources which are no longer in demand, but which may be needed one day for historical re- search. 5. Bibliographical sources. These are reference sources which organize and pro- vide access to all other sources. Notification and bibliographical sources are essential for all disciplines, and re- spond directly to the needs of scholars at the institution. The extent of the instruc- tional sources required will depend pri- marily on the size and use of the institu- tion's academic programs. The most divergent needs are met by the documen- tation sources. Some subjects, notably the sciences, have very little requirement for documentation sources (always depend- ing upon how we define them), while sub- jects which view the library as their labora- tory are heavily dependent upon documentation sources. The division of sources by function would improve opportunities for estab- lishing priorities among and between sub- jects. If we are to work with faculty to make the best use of increasingly shrink- ing resources, we must have the ability to divide sources in a manner clearly related to their actual use. We must be willing to decide when the fulfillment of one func- tion can be reduced in order to maintain or enhance another, and that decision should be reflected in our collection build- ing and management. At the same time, the library must also have the capacity to ascertain when weaker constituencies are not receiving adequate collection support, and to shield weaker constituencies from Old Forms, New Forms 515 stronger ones. This can only be achieved systematically and openly by designating and monitoring source functions. "Certain types of sources, such as most instructional sources and many notification sources, cannot be shared effectively among institu- tions, but must be owned.'' Cooperative collection development could also benefit from a method that ba- ses cooperative agreements upon func- tional categories. Certain types of sources, such as most instructional sources and many notification sources, cannot be shared effectively among institutions, but must be owned. Faculty must receive as- surances that cooperative agreements will not affect their access to such sources. Specific functional categories, with a di- rect relationship to use, therefore, should enhance communication among most of collection development's contact groups. Core Definition We can conceive of the functional cate- gories as a kind of horizontal division of a subject. This is only a first step toward the kind of specification which will be needed if the library is to assume an even more re- sponsible and active role in the reconcilia- tion of competing demands on inadequate resources. No matter how carefully or cre- atively we categorize information sources, we are still obliged to devise some vertical or qualitative criteria within each func- tional category as a basis for selection. While such criteria will be necessarily dif- ferent at each institution, they must also have some common characteristics among all institutions in order to maintain stan- dards and to foster cooperation. A more exact and applicable definition of core ma- terials is'' essential to the rationality of col- lection development in the future." 28 It is normally assumed that the core will vary from one institution to another de- pending upon local needs, but core hold- ings should overlap significantly. The core 516 College & Research Libraries should ideally serve as a kind of common vocabulary for all those engaged in re- search on the subject-the accepted refer- ence point, to which all work in the field orients itself. The only fair measure of progress in a field is by the relationship of current work to a consensually estab- lished core of information. Compatible re- search depends upon such common points of reference, as does the coordina- tion of education at all levels, which pre- supposes a well-defined core of informa- tion to which all students are exposed. If the core concept is to become truly useful we must be prepared to work to- ward the definition of a standard core, which would be consensually accepted as such by scholars and libraries. The most important attribute of such a core should be that it is endorsed in detail by the aca- demic community at large. (From the standpoint of bibliographic administra- tion, what constitutes that core is rela- tively unimportant; what is important is rather that everyone involved agrees on what constitutes that core .) This can only be achieved by defining specific titles as core items. Defining core titles, at least for notification sources, should be accepted, therefore, as a fundamental, ongoing re- sponsibility of the academic library com- munity.29 The consensual designation of core sources would have an immediately bene- ficial impact on acquisitions budgeting for academic libraries of all sizes. If it were possible to achieve some general agree- ment among all libraries as to which sources should be included initially in the core, each library could begin its budget- ing process by projecting the funding nec- essary to acquire and maintain such mate- rials. Our ability to compare the purchasing power of acquisitions budgets at different institutions would also be greatly improved by such a unified core definition; comparisons could be based upon what libraries have to spend once the core materials have been budgeted. Moreover, because the items defined as belonging to the core would be scrutinized by all participating libraries, the costs of these materials could be carefully tracked, routinely compared, and widely publi- September 1989 cized. If certain items were found to be sig- nificantly overpriced, the suitability of those items for the core could be reconsid- ered. Finally, a common definition of core ma- terials would improve cooperative collec- tion development. The first step in a coop- erative program should not be to try to divide collection responsibilities for low- use materials, because it is so difficult to agree on which materials fall into that cat- egory. The first objective of cooperation should be to decide upon which items should be duplicated among all participat- ing libraries. Once that has been achieved, at least for notification sources, then po- tential areas for cooperative collection de- velopment can be much more easily nego- tiated. Our ability to rate aspects of our collections in relationship to each other would also be significantly enhanced, but only if we dare to define the dividing line between core and specialized materials. How can a core definition be achieved? How are we to take charge of scholarly in- formation in order to guarantee access to our users, if the utility of that information can only be gauged very imprecisely, and only after the item has already been ac- quired? One answer lies in the materiality of information. Libraries seldom control information directly. Rather they manipu- late the containers the information is moved about in. To define a core is, there- fore, to define the containers in which fu- ture core information will appear. One possibility is to exploit the contextuality afforded by the serialization of scholary information. At least for notification sources, it should be possible for scholars and experienced subject or area bibliogra- phers to arrive at a consensus as to which journals in each subject field should be categorized as core journals on the basis of the nature of the articles already pub- lished in them. The same could be done with databases, on the basis of the quality of information previously retrieved. It may also be possible to designate different types of core lists by level of academic pro- gram; a larger core might be defined for li- braries with graduate programs in the subject than for libraries that support only undergraduate programs. ''The Collection Management and Development Committee of the Re- search Libraries Group (RLG) has al- ready taken a decisive step in this di- rection by working on lists of journals in selected subject areas nec- essary to build a very strong, research-level (4 +) collection.'' The Collection Management and Devel- opment Committee of the Research Li- braries Group (RLG) has already taken a decisive step in this direction by working on lists of journals in selected subject areas necessary to build a very strong, research level ( 4 +) collection. The purpose of the lists will not be to ensure duplication, but to make certain that at least one copy of each journal on the list will be available somewhere in the RLG consortium. These lists are not core lists because they are in- tended to represent works needed in the aggregate for exceptionally powerful re- search collections, but this potentially very effective project now being initiated by RLG has demonstrated that lists of es- sential periodicals can be assembled. The RLG project has also shown that the construction of such lists requires a high degree of cooperative organization and collaboration. If core lists are to be devised at a national level, improved organization and communication among academic li- braries will be required. The links between larger and smaller academic libraries will also need to be strengthened. It is essen- tial that smaller libraries participate in the process, so that their users can be assured access to the same basic core materials in each subject as the users of larger libraries. This should improve the fulfillment of the notification function by smaller libraries. Care would need to be taken to ensure that the cost of the core does not exceed the budgets of smaller libraries. If such a core were defined, of course, all academic libraries, but especially smaller libraries, would be able to communicate their budget needs much more accurately and forcefully to their institutions. Accredita- Old Forms, New Forms 517 tion might also eventually take core hold- ings into account. The identification of core sources pub- lished in monographic formats is much more problematic. The retrospective cir- culation method could be used. An alter- native for monographic notification sources might be to designate certain care- fully selected monographic series in each subject as core series. This would doubt- less be even more controversial than des- ignating core periodicals, but it should not be impossible, especially if we are willing to work with editors and scholars in the field. There would also need to be mecha- nisms to review and update core lists peri- odically. By designating core titles in this man- ner, the competition for publication in such sources should increase substan- tially, so that we could expect the quality of that material to remain consistently high. But would the definition of core ma- terials constitute a form of censorship? Probably. Like all bibliographic decision making, core definition would necessarily involve the rejection of some materials or sources of information in favor of others. This is unavoidable, and it is a key aspect of the bibliographic function. As increas- ingly large volumes of data become avail- able online, moreover, the art of biblio- graphic discrimination will become even more important to scholarly communica- tion than it is now. No matter who makes the bibliographic selection, bibliographer or end user, it remains a collection devel- opment responsibility to ensure that the decisions are made consciously and ac- cording to consistently defined criteria. Refinements in Policy If functional categories could be estab- lished within subjects, and if title-specific cores could be defined, then the next step in improving control of and access to scholarly and instructional information would be to work on the design of pre- scriptive collection and access policies . . Each institution must compare and prioritize the primary functions for the li- brary as a whole, and for each subject area. Once access to the core materials in each functional category has been estab- 518 College & Research Libraries lished, the remaining funds can be allo- cated on the basis of those priorities. The political difficulties should be allayed by the definition of core notification sources, because most faculty users at most institu- tions would thereby be able to depend upon the guaranteed availability of stan- dard sources as nationally defined. Com- munication among scholars would be pro- tected. The extent to which the library wants to develop its collections beyond those standard sources either individually or cooperatively would be based upon in- stitutional directions and resources. If each institution works toward a policy which is truly prescriptive, then the amal- gamation of those policies should provide a clear indication not only of the current condition of the national collection, but also of the transformation the national col- lection would undergo in the event of sub.., stantial economic or technical changes .. Only in this way can we have adequate control of access to scholarly information at the national level, and negotiate policy adjustments among libraries to ensure continued access. Institutional policies afford opportuni- ties for planning and decision making. If collecting were categorized by function, there can be little doubt that each institu- tion would quickly confirm what we al- ready know, namely that materials bud- gets are being spent increasingly upon notification sources, especially in the sci- ences. A prescriptive policy would deter- mine the extent to which such notification sources should be permitted to consume the budget, or the degree to which the col- lection of other materials should be re- duced in order to compensate for the in- creasing costs of scientific notification. Before this situation gets out of control, our policies must finally set functional lim- its to ensure that the needs of all constitu~ encies are consistently met within the con- fines of current economic conditions. September 1989 The prescriptive collection policy must have the capacity to serve as a component of a general library policy regulating the use of all library resources. A clear and dis- tinct link should be set between collection policy and all other library operations so that the effects of other operations on col- lections and access will be clarified for fac- ulty clientele. CONCLUSION There are really only two ways to build a collection: on the basis of publication, or on the basis of use. Selection based on publication seeks to acquire a broad share of what has been published on the subject, while the use-based method imports only materials specifically applicable to current user needs. Most college libraries have al- ways applied some form of the use-based method for most subjects, but larger uni- versity libraries have managed until re- cently to build many segments of their col- lections on the basis of publication. Today, however, even university library collections are becoming increasingly use- driven; they are being tailored to fit the special needs and interests of current us- ers, because the publication-based ap- proach is no longer economically feasible. Financial constraints are forcing a return to a kind of indirect selection by users. But things are now very different from the way they were thirty years ago. The agency of collection development has begun to assume some control over the information needs of the academy. A system to regulate and focus selection and access, imperfect as it still may be, is now in place. The challenge facing collection development is to calibrate its operation more precisely, to define its rationale more persuasively, and to apply its meth- ods more rigorously in preparation for the unprecedented economic and technical changes which we have only begun to ex- perience. REFERENCES AND NOTES 1. Mark Sandler, maintains, in "Organizing Effective Faculty Participation in Collection Develop- ment," Collection Management 6:64 (Fall/Winter 1984), that "the teaching faculty continue to serve as the primary agents of selection in the great majority of academic institutions." Old Forms, New Forms 519 2. The University of Florida Libraries, for example, continued to allocate its firm order budget to aca- demic departments until1987. Letter from Sam Gowan, 15 May 1989. 3. J. Periam Danton, Book Selection and Collection: A Comparison of German and American University Li- braries (New York: Columbia Univ. Press, 1963) p .61-82. See also Jasper G. Schad and Ruth L. Adams, "Book Selection in Academic Libraries: A New Approach," College & Research Libraries 30:437-42 (Sept. 1969); and John D. Haskell, Jr ., "Subject Bibliographers in Academic Libraries : An Historical and Descriptive Review," Advances In Library Administration and Organization 3:73-84 (1984). 4. Charles B. Osburn, Academic Research and Library Resources: Changing Patterns in America (Westport, Conn .: Greenwood, 1979), p.3-34. 5. Douglas Waples and Harold D. Lasswell, National Libraries and Foreign Scholarship (Chicago: Univ. of Chicago Press, 1936), p.71, showed that the New York Public Library, which relied even then upon bibliographers, held collections of European social sciences materials which were superior to faculty-built collections in large academic libraries . 6. Allen B. Veaner, "1985 to 1995: The Next Decade in Academic Librarianship, Part I," College & Research Libraries 46:219-25 (May 1985). 7. Elaine F. Sloan, ''The Organization of Collection Development in Large University Research Li- braries," Ph .D. diss ., Univ . of Maryland, 1973, p.47. 8. Gerard B. McCabe, "Austerity Budget Management," in Austerity Management in Academic Li- braries, ed. John F. Harvey and Peter Spyers-Duran (Metuchen, N .J. : Scarecrow, 1984), p.231 . 9 . See F.W. Lancaster, "Evaluating Collections by Their Use," Collection Management 4:15-43 (Spring /Summer 1982). 10. Robert L. Houbeck, Jr., "If Present Trends Continue: Responding to Journal Price Increases," Th e Journal of Academic Librarianship 13:214-220 (Sept. 1987). 11. See Guidelines for Collection Development, ed . David L. Perkins, Chicago : American Library Assn ., 1979. Revisions of all of the guidelines included in this publication are now being undertaken by the Collection Management and Development Committee of the Resources Section, Resources and Technical Services Division, A.L.A . 12. David C. Weber, "A Century of Cooperative Programs among Academic Libraries," College & Research Libraries 37:205- 21 (May 1976) . 13. See, for example, Louis R. Wilson, "The Challenge of the 1930's and the 1940's," College & Re- search Libraries 1:131 (Mar. 1940); Verner W. Clapp, "Cooperative Acquisitions," College & Re- search Libraries 8:99- 100 (Apr. 1947); Ernest Cadman Colwell, "Cooperation or Suffocation," Col- lege & Research Libraries 10:195-98, 207 (July 1949). 14. Joe A. Hewitt and JohnS. Shipman, "Cooperative Collection Development among Research Li- braries in the Age of Networking: Report of a Survey of ARL Libraries, " Advances in Library Auto- mation and Networking 1:225 (1987). 15. Joseph J. Branin, "Issues in Cooperative Collection Development: The Promise and Frustration of Resource Sharing," in Issues in Cooperative Collection Development, ed . June L. Engle and Sue 0 . Medina, Atlanta: Southeastern Library Network, 1986, p .15-36. 16. See the Report of the ARL Serials Prices Project (Washington, D.C.: Association of Research Li- braries, 1989). 17. Richard Hume Werking, "Allocating the Academic Library's Book Budget: Historical Perspec- tives and Current Reflections, " The Journal of Academic Librarianship 14:143 (July 1988). 18. Jasper G. Schad, "Fairness in Book Fund Allocation," College & Research Libraries 48:480 (Nov . 1987). 19. See the reactions of Hugh F. Cline and Loraine T. Sinnott, Building Library Collections: Policies and Practices in Academic Libraries (Lexington, Mass.: Lexington, 1981), p.79-81. 20 . See, for example, Mortimer Taube, "Libraries and Research, " College & Research Libraries 2:22-26, 32 (Dec. 1940), and Garrett Hardin, "The Doctrine of Sufferance in the Library," College & Research Libraries 8:120-24 (Apr. 1947). 21. Margit Kraft, "An Argument for Selectivity in the Acquisition of Materials for Research Li- braries," Library Quarterly 37:285 (July 1967). 22 . Daniel Gore, "Farewell to Alexandria: The Theory of the No-Growth, High-Performance Li- brary," in Farewell to Alexandria: Solutions to Space, Growth, and Performance Problems of Libraries, ed . Daniel Gore (Westport, Conn .: Greenwood, 1976), p.164-180. 23. Robert M. Losee, Jr ., "Theoretical Adequacy and the Scientific Study of Materials Selection," Col- lection Management 10, no. 3:25; (1988). See also his" A Decision Theoretic Model of Materials Se- lection for Acquisition," Library Quarterly 57:269-83 (July 1987) . 24 . Richard W. Trueswell, "Growing Libraries: Who Needs Them? A Statistical Basis for the No- 520 College & Research Libraries September 1989 Growth Collection,'' in Farewell to Alexandria, p.72-104. In any case, to use circulation as a basis for core definition, one must first buy the document and see how it circulates, before one can deter- mine whether it qualifies as part of the core. This is, of course, the weakness in the argument that a library should make the best use of its resources by acquiring only core materials. 25. Ulrich's News 1, no.3:1 (Apr. 1988). 26. The major exception to this rule is for area studies; area selectors are frequently responsible for many subjects from or relating to a single geographical area. 27. Fremont Rider recognized and criticized this same problem nearly fifty years ago in his The Scholar and the Future of the Research Library (New York: Hadam Press, 1944), p.68-85. 28. Charles B. Osburn, "Toward a Reconceptualization of Collection Development," Advances in Li- brary Administration and Organization 2:180 (1983). 29. The definition of the canon or the "classics" (i.e., materials belonging to the highest class), was a responsibility of the library at Alexandria. See Georg Luck, "Scriptor Classicus," Comparative Lit- erature 10:152 (Spring, 1958).