Libraries on the Web:

An Examination of Library Web Servers

Tracy Seneca

LIS 296, Spring, 1994

  1. Introduction
  2. The Server as a Source of Information for Local Users
  3. The Server as a Navigational tool
  4. The Server as a Source of Information for Library Staff
  5. The Server as a Library Brochure
  6. The Server as a Medium of Communication
  7. The Server as an Online Library/Collection
  8. Comparison of Academic, Public and Special Library Servers
  9. Three Outstanding Library Web Servers
  10. Conclusion


As new resources become available and popular on the internet, Libraries have contributed to the development and proliferation of different internet resources. Libraries play a role both as providers and organizers of information, and would seem to have a natural place in the creation of Internet resources. For the last couple of years, gopher has been the most popular internet tool. Universities and organizations from around the world set about the task of creating gophers, making local information available on line, and also providing access to distant documents. What has resulted is a sometimes useful, sometimes hopelessly frustrating tangle of menus and files. Libraries have participated in building this vast structure, and have been involved in the debate about its role in the world of information. The last year has seen the increasing presence of Library-originated sources on World Wide Web, an arena of the internet which provides the ability to write documents in hypertext, to provide images, sounds and film as part of the body of a document, and still provides access to any of the resources available on gopher or WAIS. Many Library staff and users have barely begun to incorporate gopher into their searching habits and workflow, when it is time again to build a new service, a Library Web Server.

This paper will examine several examples of Web documents created by libraries. The question I will try to answer is: "When a Library creates a Web server, what, exactly, is being created?" To some degree these documents are texts, searching tools, organizing tools, public relations devices, and possibly maps or mirrors of the structure of the library. Furthermore, the audience for a server is entirely unknown. This network of documents must address the needs of different groups of people - library staff, local library patrons, and the general public. The degree to which a server can make these distinctions clear is important. A home page must be comprehensible, and hopefully appealing, to all of these people. I will examine the way in which different servers go about fulfilling these different roles.

I have used the "United States" section of Thomas Dowling's list of Library Web Servers as the source for my choices. This list is a comprehensive one, and the sources on it range dramatically in quality and style. After browsing through these sites, I then composed a list of criteria to be applied to all of the servers I looked at. This list was intended to clarify the different roles that a web server can play, and to judge the strengths and weaknesses of each site. This list was broken into two sections. The first concerned the general use of hypertext in the server, and whether the server made the best possible use of the qualities that the medium has to offer. The criteria for judging the use of http were as follows:

  • Images - are they neccessary? Are they functional? Are they well - placed on the screen?
  • Links - are links back to the previous page and the home page provided as part of each document?
  • Is there more than one route to the same information?
  • Is the overall style of the site different from what would be possible with a gopher or a plain text document?
  • Does the server show evidence of clear planning?
  • Are there warnings for large files, indicating the file size?
  • If the site is still 'under construction', does it say so?

    The second part of this list of criteria concerned the different roles that a library web server can play. I identified 6 distinct roles for a library web server. These were:

    1. Source of information for local users
    2. Gateway/navigating tool for the internet
    3. Source of information for Library staff
    4. Publicity/web server as a Library brochure
    5. Medium of communication
    6. Online library/collection

    In the following discussion, I will focus on the 6 roles listed above and provide examples of the way different libraries approach those roles. As I begin discussing each role, I will list the specific criteria I used to judge it. Some servers did an excellent job of addressing all of the 6 roles, while others focused on one or two. I will examine what happens when a server is strong or weak in certain roles and will also discuss the different approaches of University, public and special libraries. Finally, I will examine 3 of the most outstanding sites - those which address all 6 roles as well as exhibiting good general use of the medium.

    As I examine these sources, the reader should keep in mind that these sites may change between the time I initially looked at them and the time the reader sees them. As a critic, I must also bear in mind that some of these sites are still being created, whether the authors warn me of this or not. What might seem to me to be a weakness may have been fixed by the time a reader finds this paper.


    Specific Criteria for this role:

    If nothing else, this should be the one role that a server is strong in. The extent to which local information is made available and is easy to find is what makes a server a tool as opposed to a document. While users could easily turn to other sources for WWW guides or internet resources, this role is where the library provides unique information and services. Some libraries were very complete here, and had clearly given a great deal of thought to the types of questions that patrons commonly have about the library.

    One server which was strong here was designed by Thomas Dowling of the University of Washington, the person who compiled the list of library web servers used as the basis for this examination. This server had a generally strong assortment of the kinds of information listed above: a list of library hours, a guide to instruction offered through the library, a guide to CD ROMS and databases available in the library and descriptions of library departments and services. What made this particular site unique was that the designers made clear distinctions about types of local users, and addressed their differing needs in different areas of the server. This was the only server I found that offered information about student employment in the library, which indicates that the designers were thinking clearly about the needs of their student audience. It also contains an outstanding guide to the library for faculty members. This guide functions both as publicity for the library and as a practical guide for library services. This allows the writers to take the right tone for addressing members of the University faculty. The overall tone that a server takes or presents is important, and by making clear distinctions about the intended audience, the designer of this server enables it to be flexible - to convey the same information in different ways. It also demonstrates that the library is considering and anticipating the needs of the faculty. The reference services area of this server was also very well done. It includes a general description of the service, hours, maps and a guide to reference resources.

    Boise State University was another site that was good at conveying information for local users. This server was actually strong in other roles as well, but here it provided exceptionally good use of hypertext in order to convey local library information. One of the things that I looked for when judging the strength of local information was some kind of guide to the electronic resources available in the library. While many offered a list of CD ROMS available, and told users where to find them, these were often just plain text lists. Only a few libraries included hypertext links to each database name, leading the user to instructions or descriptions of each database. This particular site offered the most comprensive and creative use of hypertext for this purpose. The ABI Inform selection from the list of CD ROMS is a complete guide to using this database, including sample screens. This guide might perhaps be a little too realistic because at first glance one might think they had actually made it available online. What is important, I think, is to provide more than just a list of databases, but to convey to local users which databases will be helpful to them. This site also contained information about construction which is taking place on its campus. This was also thoughtful, as local users who are inconvenienced by construction might benefit from knowing about the changes taking place in the library building. Another strong site for local information was the Johns Hopkins University Welch Medical Library server. This server was well organized, and information was easy to find. The "Ed ucation Opportunities" section of the server included course descriptions, schedules, forms for registering for classes, instructional guides and tutorials.

    For the sake of comparison, look at a home page from a site that does not, and is perhaps not intended to convey local information or address itself to local users. Central Connecticut State University's server appears to be primarily a gateway to other services and contains technical jargon. It does not serve as a library guide for local users in the sense of acquainting them with local services, aspects of the building, or library policies. To be fair, I cannot tell what the intended purpose of this server really is. It does not display the name of the library prominantly, and therefore may not be supported by the library. It might be out of place on Dowling's list, as there are other documents coming from Central Connecticut State University which pertain to its library, but these appear to still be under construction, or intended for purposes so local that I cannot understand them. But this site is typical enough of the servers that neglect local information to have a basis of comparison for sites that show great effort in addressing their local audience.

    A common aspect of servers that were strong in local information was the use of maps. Because http allows you to link parts of an image to certain kinds of information, these maps could be more or less complicated. Some were straightforward map images, others had special features. If a library took the trouble to provide a map, that generally meant that they had included other types of local information as well. Boise State University's map is both simple and well done. The different color schemes for each room are helpful, and the link up to the next floor is also a good touch. California State University, Fresno appears to be building a map - it does not yet contain all of the floors of the library, and some of its links do not yet work. It also does not allow the user to navigate to other sections of the building once s/he has choosen a particular floor to look at. For now, it is also a simple image map, with no special features but the initial description leads me to think that other features will be added later on. It also seems that they intend to include campus maps. General campus information or links to other campus servers were another indication of good anticipation of the needs of local users. Rice University's maps were also under construction, which they clearly stated, but already include features the other two maps did not have - for each area of the library, you can click to find additional information about the unit or service. Links have been provided from the map to the descriptions of library services. Purdue has by far the best map arrangement I've seen. Users can view the maps either by asking for maps of particular floors, or by choosing a type of service to see where the service is located. One can choose "computers and terminals" for instance, and see where all of the computer terminals in the library are located, and find out whether the terminal is a MAC, a PC or a library catalog terminal. They can also find out where different call number ranges are located - a frequent question in libraries.

    Other concerns for local users are related to the general use of hypertext in the server, or the style in which the information is presented. By 'general use of hypertext' I don't neccessarily mean the inclusion of lots of images. In fact, large image files would be something of a deterent for local users, as they take a long time to appear on the screen. An important distinction for local users is the frequency with which they consult the server. If the server is made widely available throughout the library, or if the campus community has good access to World Wide Web, and if the server is designed to be a frequently used tool, then hypertext can present certain pitfalls. It is helpful, for instance, to stay away from very long texts standing in front of the links needed to use the server. While a simple list of choices at the top of the screen is a little dull - a long introduction to the library, however well done, will get annoying if frequent users have to scroll past it every time they look at the server. Another issue is one of style. It seems best to use a style that is accessible to many different kinds of people and that will not wear thin for frequent users. One example of this is Dakota State University's server mascot, the bookworm. Now this is actually a very well designed server. It is quite complete in a variety of different roles and it does include consistant links back to the home page and to other areas of the server, which is important. (Users will be viewing the server from any number of different client programs, some of which are easier to control than others. If the server provides its own tools for moving from one document to another and back again, all users will have equal ease exploring the server.) However, this link is an anthropomorphized 'bookworm' which is there to help you "navigate, learn, and keep you company". There is something about the 'cuteness' of this concept that might grate on one's nerves after a while, particularly if the user recognizes the icon from WordPerfect clip art. For local users, a server should be designed both practically and aesthetically with frequent use in mind.


    Specific criteria for this role:

    I saw this particular role as being closely connected with the role of a server as a source of information for library staff. If the library provides access to world wide web for its staff, and also supplies its own web server, odds are that the first page that staff members will see when they access world wide web will be the one designed by the library. In that case, they will be using the library web page as the point of departure for their web exploration. For this purpose, they will benefit from plenty of information about world wide web, html, and some guide by subject for the web. Now, this particular role may also be opaque to me as a remote user of a library's web server, because it is possible to provide all of this to local users only. That is, there may be links to services that I can only see if I am using a terminal from within that particular library.

    Most servers provided a telnet connection to the library's catalog, so that by choosing it, you would exit the web page and enter the local catalog. As a remote user, I appreciated those libraries that provided instructions for using their catalog clearly next to the telnet link. I also appreciated libraries that put some thought into which other catalogs they would provide access to. It does little good to provide random general access to online catalogs if the patron has no way of knowing which one is best to search. For local users, it would be most helpful to provide links to nearby libraries, libraries with special lending arrangements, or to libraries with a similar subject focus (if appropriate). North Carolina State University's server evidenced some clear planning along these lines, and was a well designed server overall. If you scroll down the home page to "Information Systems of the Triangle Research Libraries Network", you'll see that the libraries that NCSU provides telnet connections to were specifically chosen for local users.

    One feature that went above and beyond providing telnet access were those servers that had designed a WWW interface for their library catalog. Rather than providing a telnet link that takes the user out of World Wide Web and into a remote system, these libraries designed WWW forms that connect to their catalog and conduct queries without the user having to know how that system works. Butler University was one site that provided this capability. Here, it still took some brief instructions, and one still had to type in search codes such as "au" or "ti" to narrow the search to a particular field. They did, however, provide access to other library catalogs in the state, and allowed the user to search more than one catalog with the same command, thereby creating a union catalog out of separate systems. They also provided access to RLIN, OCLC, and periodical indexes through this searching form. St. Johns University and the College of St. Benedict offered a search form with even more features. My only arguement with this catalog is that it never stated the name of the Universities, and it took some digging to find out whose catalog I was looking at. This searching tool also offers access to other catalogs, and automatically combines the catalogs of the two Universities, but does not seem to allow the user to combine any other catalogs into a single search. Other catalogs must be selected and searched individually. It does however, provide point and click control for every command. The user does not have to remember any search or display commands and can still specify, combine, limit and truncate a search. Furthermore, the final results of the search are also in hypertext, leading the user on to other logical search choices for his/her topic. By providing hypertext links for call numbers and for Library of Congress Subject Headings, the user can click to find more books on the same topic, without having to type lengthy and totally unmemorable terms. This particular system is about as user-friendly as a catalog can get. The only words a user has to type are the terms they are looking for.


    Specific criteria for this role:

    I have discussed this role of library web servers somewhat already, in discussing the sever as a navigational tool. As I mentioned previously, my perception as a remote user might be hindered by the way web servers function. Not only can libraries provide local access to things that distant users cannot see, but they may also provide home pages for their staff that are distinct from the home pages for the public. In this way, they can provide access to such things as handbooks, policies and other information that the staff needs and the public does not. Still, there were some sites that clearly had library staff members in mind for some of the items available on their servers. I appreciated this, because there is another, perhaps more obscure audience consisting of the professional community at large. I appreciated, for instance, clear statements of library policy for Interlibrary Loan and Bibliographic Instruction, both of which I am involved and interested in. This allowed me to see the different approaches that other University libraries take to the same issues I work with. I especially appreciated Northwestern University's inclusion of a task force report "Visions of Library Technology".

    Other items of benefit to staff members and local users alike would be organization charts, staff directories, and easy access to campus directories. One server that provided all of this as well as some totally unique features that would be valuable to staff members was Purdue. In the "Application Toolbox" area of the server, they somehow managed to provide access to wordprocessing programs, database programs such as Paradox, and spreadsheets as well as including documentation for each program. I am not able to tell how well this arrangement works, since as a remote user, I do not have access to programs provided locally. I do not know if these programs are available only to staff or to library patrons as well. Indiana University at Bloomington provided access to the usenet version of PACS-L, which I thought was a good idea, since it would save staff members the trouble of having their e-mail inundated with listserv messages. This server, however, appeared to be entirely geared toward staff members. This made it strange looking in comparison to other servers, because it presented no information about the library whatsoever.


    Specific criteria for this role:

    This particular role was the easiest to get an immediate sense of, but is the hardest to convey, since it is almost entirely the general design or appearence of the server that influences this role. The underlying question here is: were the designers aware that their product would convey some image of the library, and did that awareness influence their work? Some sites conveyed a very clear impression of the physical environment of the library, the nature of its collection and its role on the campus, so that the server not only conveyed practical information, but gave the user an overall impression of the library. The general use of hypertext and the extent to which a server adressed all of the roles I've listed also conveyed an impression of the library. The more thought and planning that went into making the server easy to use and comprehensive, the better the impression it left.

    In this role, too, it is helpful to have some basis for comparison. Washington University's server, while clearly still under construction, might best have been left under wraps until it was done. Again, I do wonder how Mr. Dowling compiles his list of library web servers, and whether the library in question has a chance to say when they are ready to be listed there. The term 'joker' just didn't convey the tone I would expect. Along these lines, it seemed best to stay away from links to the Melrose Place newsgroup as part of the electronic text page, and to also steer clear of technical jargon that might put people off. Some libraries resolved these problems by created a page that was clearly intended just for fun. This let them take different tones for different purposes, in the same way that defining the audience for specific areas of the server allowed for more variety in the style of the server.

    Among the things that conveyed a a good impression of the library were pictures, maps, statements of library philosophy and histories of the library. Among the sites that provided a history of the library were the University of Georgia, and Rice University.. Rice is one of the three outstanding servers to be discussed more fully later, but it was intentional enough in its role as a brochure that there was even a page directed toward potential donors to the library. Most library servers stayed away from large image files, but many included a picture of the library on the first screen or a Library or University logo. The University of Texas at Austin is a typical example of a server where the designers were clearly aware of the visual impression that users would take from the server, and that the look of the server also conveyed an impression of the library. Exhibits of special collections were also impressive. These will be discussed separately, but they did provide good publicity for the library.

    There was sometimes some conflict between the server's role as a brochure and its role as a searching tool. As mentioned previously, any lengthy introduction or description of the library is going to stand in the way of a frequent user's access to the links that the server provides. A large picture of the library or a complicated graphic design will do the same. The space within the first screen of any page is the most valuable section of the server, because here is where the designer can make a good first impression or provide immediate ease of use for the server. Some sites resolved the conflict between these two different goals by creating graphics that were both images and links to other areas of the server. This meant that in the first screen, the user could get some sense of the scope of the server, without having to scroll down or meander through to find out what was there. A creative example of this solution is the first screen to the server at California State University, Fresno. Here the designers managed to combine an image simply for aesthetics with a menu selection of icons.


    Specific criteria for this role:

    Most servers contained at some point the name and address of a contact person for feedback, advice or problems. Some servers, however, used the full potential of World Wide Web by allowing users to communicate and conduct library transactions by creating forms for specific purposes. This perhaps says more about the size and staffing of the library than it does about the designers of the server. Any library that has already been able to incorporate e-mail requests for services into its workflow would make an easy transition to having business take place over the web. In fact, the structure of WWW forms might make the transactions easier, because they can be easily tailored to the needs of the library; the library would be less likely to need to contact the patron for further information. The forms can also restrict transactions to qualified users, which means that the library does not have to concern itself with finding out if the patron is eligable to be requesting services electronically. (This feature also prevented me from being able to examine these forms; I was only able to tell if the service was offered, but not to judge the method the library used.) But many libraries have not been able to been able to provide access by e-mail. Units which have a difficult time keeping pace with the patrons who request services in person will have a hard time opening up to the increased workflow from electronic access. I was very impressed to see libraries which offered a wide variety of services electronically. While a few libraries provided forms for requesting interlibrary loans, both Rice and Purdue offered a wide variety of services such as reference, document delivery, retrieval of items from storage, and requests to place items on reserve for classes. At Rice, it is clear that the library had incorporated electronic transactions before the creation of the web server, since the forms are actually part of a gopher menu.


    Specific criteria for this role

    What I looked for here, more than anything else, was some kind of organization to the material that was provided. Most servers had some links to electronic texts or exhibits, but these were often simply listed under a heading such as 'e-texts' with no further distinctions made. I appreciated libraries that imposed some organization on the documents they provided. This generally meant that there were enough there to organize, and that they were relatively easy to find. North Carolina State University was one site that did a good job of providing organized access to electronic texts. If you look at the section for reference materials you'll see that this is further subdivided by almanacs, handbooks, etc. The only difficulty with these is that they were accessible through an icon labelled "study carels", which didn't indicate anything to me; I chose it only to find out what the name implied. Another server that seemed to have a rich electronic collection was from the University of Virginia. I did find this server to be confusing to navigate, and I still cannot tell whether the items on it are provided by the library, or whether the library merely provided links to them. There is an Electronic Text Center at the University of Virginia, and they have made some of their material available and searchable on the web. It was also difficult to tell whether or not the library was involved in creating the online exhibits here, such as the "African Art; Aesthetics and Meaning" exhibit. There was quite a range in the sorts of exhibits that libraries made available. I imagine that many of these were online correlates to exhibits that the libraries might have put up in their exhibit halls of unusual or interesting material in the collection. Texas A&M University, for instance, presents and exhibit entitled "Towards a Better Living for Rural Texas Blacks, 1939-1960" which is clearly compiled from their archives. The University of Texas at Austin had a remarkable Exhibits section, consisting of both local exhibits from the collections, such as "Texas, Texans and the Alamo", and a large number of links to other World Wide Web exhibits. The University of Georgia had made digital images available of interesting items in their special collections, such as rare maps and old stereoscopic slide images. I appreciated these exhibits whenever I found them; WWW naturally lends itself to this kind of document, and I like the chance to see the oddities in a library's collection. The absolute best source for online exhibits is the Library of Congress. Some libraries simply provided a link to the Library of Congress as their exhibits section, and one finds links to the LC exhibits throughout the web. They are beautifully done, and are also taken from exhibits that took place at the Library of Congress.


    Because most of the servers on Dowling's list were academic libraries, they established the norm, or set the standard for this examination. That is, the six roles I've identified are derived largely from looking at the tendencies of academic libraries. Public and special libraries differed because they tended to be strong in particular areas. Special libraries tended to very geared toward the roles of the server as a navigational tool and an online collection, providing extensive access to the subject covered by the library's own collection. The Rutgers School of Law Library was an example of this, providing extensive access to internet sources in law. They offered a full range of access via links to other web sites, telnet connections and to gophers from other law schools and to firms. The State Library of North Carolina offered special services that made perfect sense for a state library, providing access to statistics for the state and to an online Encyclopedia of North Carolina. The Library of Congress has specialized in putting together a series of extemely popular WWW exhibits on a variety of topics. Public libraries tended to provide more in the way of local or community information. The St. Joseph County Public Library, serving South Bend, Indiana, not only provided good coverage of library information, but had extensive information available about community events. This calendar made good use of hypertext, and certainly evidences the library's involvement in the community.


    You will already have seen some of the things that these servers have to offer throughout the previous discussion. The servers that stood out above the others on Dowlings list were the University of Indiana - Purdue, Rice University, and Johns Hopkins University Welch Medical Library. These were chosen mainly because they address most or all of the roles they could have, because they show very clear and thoughtful organization and because they make good use of the medium of hypertext, offering a service that they could not have offered via a gopher server. I would recommend that anyone involved in designing a library web server take a look at these three sources to see what can be done.


    This site was admirable because it was not neccesarily flashy. The home page is not loaded with images, and seems to be a fairly straighforward list of menu choices. What is noteworthy is that this server has some depth, and these choices are well organized. There is much to be found beyond this orignal list, and it is fairly easy to find. The same information is available in more than one place when appropriate; forms for initiating library transactions are available both in a section on the home page for "forms" and within the description of each service that offers online forms. At the bottom of the home page is group of icons, providing links to other parts of the server. This group of icons appears at the bottom of documents throughout the server, allowing the user to move not just back to the home page, but throughout the server from any point.


    Though this is technically a special library, it provided good coverage of a variety of web server roles. It was also very well designed. It has a very different appearance from the server at Rice, as it starts the user out with a grid of icons, making the best possible use of the full first screen. These icons divide the server up into sections which organizes the information in ways that a straight list of selections would not. Like rice, it also has an icon bar which appears at the bottom of documents throughout the server. These icons differ from the ones at the top of the screen, so the server is more extensive than it appears at first. JHU includes links to campus information and shows good anticipation of local users needs by including directions to the library and information about building details such as the location of copiers and telephones. The descriptions of databases available in the library link to descriptions of each database, and the server shows general good use of hypertext.


    I found this server to be the most impressive one on the list. It too begins with a grid of icons, and from the first screen you can see that it covers all of the roles I've been looking at. These icons repeat throughout the server, allowing the user to navigate easily through it. It also included the fullest use of hypertext capabilities of any of these servers. I have already pointed out the maps, which the user can change depending on what sort of detail they are looking for. I have also mentioned the "applications" section of this server, where the library is providing services well beyond the scope of most others. This site also includes forms for online transactions for most library services.


    Having examined these servers, I do feel that a web server can provide library users with a valuable service. Under the best conditions, a server can make the catalog easier to use, provide users with important visual information about the library, and, if the designers make good use of hypertext, information can be much easier to find than if it were lurking somewhere in a gopher menu. All of this can be presented in a format that is pleasing to the eye and that conveys a sense of place for the library. I did feel that if a server did not take advantage of any of the unique capabilities of hyptertext, or did not organize their information well, that there was little point to having a web server available. In some instances it seemed that the server was there because World Wide Web is the 'next exciting thing' on the internet, but the library was not able to create an effective tool with it. Another concern about these servers is the question of the kind of access patrons have to them. I have been unable to tell much about the availability of WWW to patrons on each campus; I am infering that libraries which have provided very complex and well planned servers have also provided access to them from within the library. Not all libraries are able to do this, and it may be that the complexity of a server reflects the access that patrons have to such resources in the first place. It certainly would not make sense to undertake the time consuming project of designing and writing a server on a campus where practical access to the machinery is a problem. For those campuses that are able to provide the machinery, a web server would be the best option for providing local users with internet access and online library information.