A paper published in Archaeological Computing Newsletter 48
R M Yorston
One of the more exciting developments in computing in recent years has been the invention of the Java programming language by Sun Microsystems. What has caught the attention of many people is Java's ability to generate applets, small programs, which can be downloaded over the Internet and run on Web browsers. This has been used to add dynamic features to Web pages which would be quite impossible to achieve with previous technologies.
Although caught up in the hype associated with the Internet, Java has many features which make it useful in other contexts. In "The Java Language Environment: A White Paper", (Gosling and McGilton, 1995) Java is characterised in terms of a list of buzzwords: it is a simple, object-oriented, distributed, interpreted, robust, secure, architecture neutral, portable, high-performance, multithreaded and dynamic language. Each of these can be a powerful argument for using Java, but this is not the place to expand on this list. Interested readers are referred to the expanding literature on the language (Flanagan, 1996, Arnold and Gosling, 1996).
One of the major uses of Java is in the development of applets to run in Web browsers. An applet is a small program which can be incorporated into a Web page. The executable code of the applet is downloaded along with the text of the page. Once the code arrives from the server it is executed by the browser on the local machine. This allows the programmer to do many things which would be impossible in a normal Web page. Among other things the applet can display dynamic graphics on the Web page, it can allow the user to interact with it and it can connect back to the server to download further information or query a database there. Some of these capabilities are illustrated in the present work.
The infrastructure required to support the applet is built into the Web browser. This has a number of consequences. Firstly, the applet will run on any platform: the environment provided by the browser insulates the applet from the underlying operating system and allows it to function on any platform on which a Java-enabled browser is available. This is obviously a key feature in allowing use over the Internet. Secondly, the applet code is compact. All of the standard libraries for such things as the graphical user interface and network access are provided on the client platform. The only code which needs to be transferred across the Internet is that which is unique to the applet.
There are undoubtedly many ways in which applets can be used to present archaeological information. The approach taken here is essentially an extension of the map-based illustration which is a common feature of the literature. This has the advantage of being familiar and easily interpreted. The dynamic nature of the applets provides the opportunity to add new features to a familiar medium.
The applets described here were developed as part of a larger project to reexamine the relationship between the landscape and monuments of the Stonehenge region. They have been used in a number of capacities during this project. Initially they were developed as a means of exploring the data set. The greater part of the data processing was performed on a GIS at the Birmingham University Field Archaeology Unit. The author did not have access to such powerful facilities and wrote the applets as a means of visualising data generated in Birmingham. Secondly, applets provided a means of communicating results between the geographically separated members of the multi-disciplinary team working on the project. As well as making the applets available on the Internet it is also possible to distribute them on floppy disks to standalone machines. Finally, the applets provide a means of publishing some of the results of the project. They have already been used in two different ways: as part of a Web site describing the work and as an adjunct to more traditional visual aids during a presentation at the Liverpool TAG conference. In future they may also be published on CD-ROM.
The basic intervisibility applet, showing the control panel.
The highlighted monuments are those visible from Stonehenge.
(Full sized version)
The first applet was developed to gain familiarity with the study area and to illustrate the intervisibility of the monuments (Figure 1). The display consists of a map of the region around Stonehenge with all of the monuments plotted. Positioning the mouse pointer close to a monument causes the monument name to be displayed in the browser status bar and all the other monuments visible from that point to be highlighted. A separate control panel allows the user to customise the appearance of the display. There are controls to change the backdrop, to turn the highlighting of visible monuments on or off, to enable different forms of highlighting based on barrow type and richness, and to turn the display of different types of monument on or off.
This way of presenting information has a number of advantages over the use of static diagrams. To associate a descriptive legend with each of several hundred points would add considerable clutter to a map. Putting the identifying text in the browser status bar is an efficient use of space. Furthermore, to display intervisibility information for all of the monuments would require as many maps as monuments, which is clearly impractical. Allowing the user to customise the appearance of the display lets them concentrate on aspects of the data which interest them.
The henge viewshed applet. The grey area represents the viewshed
of Stonehenge. The highlighted monuments are those within 100 metres of
the edge of the viewshed.
(Full sized version)
The second applet (Figure 2) retains many of the features of the first but has backdrops showing the viewsheds of each of the henges in the Stonehenge region. This applet was written to investigate the relationship between the visible monuments and the viewshed edges. Monuments within the viewshed are highlighted differently depending on their distance from the viewshed edge. A text box in the control panel lets the user change the relevant distance.
This illustrates a further advantage of the dynamic nature of applets: the viewer can be provided with the means to interrogate the data set. In selecting values for the viewshed distance parameter they are not restricted to pre-computed values selected by the author: arbitrary values can be used. All the code necessary to calculate statistics based on the downloaded data is present in the applet.
A snapshot of the animation of a walk along the Avenue. The
viewpoint is close to the crest of the King Barrow ridge. The backdrop
represents the topography.
(Full sized version)
The third applet (Figure 3) displays animations of the monuments visible from a number of points along extended features like the Stonehenge Avenue and the Greater Cursus. Again there is a control panel, this time to allow the user to select the backdrop, the path to be animated and the speed of motion.
The use of animation allows the viewer to see how vistas open and close as the viewpoint moves across the landscape. This style of display is well suited to extended monuments which can be considered to delineate processional routes. Similar animation techniques could be used to illustrate temporal development as well a spatial motion.
There are other advantages to the dynamic display of results. The type of display used here is a middle route between static maps and publication of the full data set. Most readers of a paper do not require access to the original data: they will have neither the time nor the inclination to perform a complete reinterpretation. The sort of applets described here allow authors to increase the amount of information they make available without the need for any additional interpretative effort on the part of their readers.
Moreover, there may be sound reasons, such as copyright or commercial confidentiality, why it is not possible for the original data to be published. Often it would be permissible to print a static map of such data because the resolution of the diagram would limit the precision of the information which could be gleaned from it. Where such sensitive data is to be downloaded into an applet for interactive manipulation a similar technique can be applied: the data points can be provide at a precision sufficient for the final display but without revealing the full detail of the original data set.
The technology used here is still relatively immature. There are bugs and inconsistencies in the implementation of the Java Virtual Machine and Abstract Windowing Toolkit on different platforms. These manifest themselves in inexplicable failures: sometimes the animation fails to start in Microsoft's Internet Explorer; some of the components of the control panel fail to appear in Sun's HotJava; and attempting to run the applets in early versions of Netscape Navigator cause a fatal crash.
Another problem is the speed of the Internet. The applets consist of a number of class files, graphical images and data files. It can take some time for all of these to be downloaded. There are ways of mitigating this limitation of current technology.
The class files can all be combined into one archive file, so avoiding the need to make a number of separate connections to the server. Connection setup is often the slowest part of downloading a file from a Web server. A drawback of archive files is that there are currently three different technologies in use: uncompressed zip files, Microsoft's cabinet files and Sun's JAR files.
Another workaround for the slow speed of the Internet is to use the multithreading capabilities of Java to permit some limited operation of the applet while the required data is being downloaded. Thus data which is vital to the operation of the applet can be fetched first with less significant files, such as the different backdrops, being fetched later. As data arrives the different features of the applet are progressively enabled.
A third technique is to minimise the amount of information which needs to be transmitted. For example, the positions of the monuments are encoded, not as full grid references, but in terms of their pixel positions within the final display. The sets of monuments visible at different points along the animation have been encoded into a GIF file. Such files are normally used for graphical images, but are used here to transmit raw data because the GIF standard includes compression which considerably reduces the file size. (Version 1.0 of Java includes support for GIFs. Version 1.1 has support for compressed data streams, which provide a more natural and more general solution.)
The above considerations suggest that the generation of efficient and effective applets currently requires careful design and cunning programming. In future, as these techniques are developed, it may become possible to package the technology so that non-technical users can publish information with applets as easily as they can now produce tables and charts from a spreadsheet.
The work described here forms part of a project sponsored by The Nuffield Trust. The other members of the project team are Sally Exon, Vince Gaffney and Ann Woodward, all of Birmingham University Field Archaeology Unit.
Flanagan, D. 1996: Java in a Nutshell. Sebastopol, O'Reilly & Associates.
Gosling, J. and McGlinton, H. 1995: The Java Language Environment: A White Paper. Mountain View, Sun Microsystems.
Arnold, K. and Gosling, J. 1996: The Java Programming Language. Reading, Addison-Wesley.