In the application of computer technology to archaeology, the nature of technological change and the purpose of archaeology needs to be considered. Technologies go through three phases: when first developed they are used to replace some previous technology; subsequently they are used to meet needs which did not exist previously; and finally; they are themselves replaced, thereby being reduced to a hobby or niche technology. Thus, steam trains were first used to replace horses on railways that linked coal mines to the ports. Innovation came from Thomas Cook who pioneered the use of railway transport to develop package holidays. Now steam trains are kept running by enthusiasts in the developed world, whilst still being used in some third world countries. Much of what is called gardening is really the continued use of agricultural technologies which have ceased to be used for economic food production. Computers appear to be in all three phases simultaneously. They are still being used for straight replacement tasks (such as accounting); they are being used for tasks which did not exist when they were invented (such as spacecraft navigation); and they have become toys. In considering computer; archaeologists must avoid using them as replacements for paper-based data recording systems. To avoid this trap the real objective of archaeology needs to be re-addressed.
Data is sitting in the ground. It has been put there by a variety of processes. The ground itself is a database in which an imperfect memory of those processes is retained; the data in this database is deteriorating with time. Counterbalancing this is the increasing sensitivity of scientific instruments and technique. Since the data stored is analogue, more data can be extracted by more precise recording instruments. The ground database has two drawbacks. Firstly, the data is not easily read; it can only be accessed by someone actually on site. Secondly the processes of reading this database destroys much of the data. The purpose of archaeological recording is to transfer the ground-based record system into a form accessible not just to the site archaeologist, but to all potential users, including other archaeologists, naval architects, environmentalists, historians and geologists.
The technology used to record the data is therefore not a secondary concern, a mere bureaucratic irritation, but is central to the activity of site archaeology properly considered. The use of the new technology should not be approached from 'this is what we do now, computerise it', but should start from 'this is what we are really trying to do'. Field archaeology is a system for planning and managing the recovery of the soil database, recording deposited materials, modelling their dynamics, and deducing the processes by which the deposits were laid down from the data collected. This activity requires a Dynamic Context Recording and Modelling System (DCRAMS).
In considering what sort of system archaeology needs, it is the nature of the output which should be considered first. To start from the data input tends to limit the system too closely to current practice. Once the output requirement is determined then the input activities can be defined.
The archaeological site data-user dons the virtual-reality helmet. The user selects the visual input wanted. To study ship structure, for instance, the wood can be displayed along with metal objects directly related to the wood (copper sheeting, iron nails). Displayed all around in three-dimensional space is the shipwreck. In one hand are the controls which enable the viewer to fly around the wreck, look at it from above or below, and change its scale. The other hand wears an instrumented glove which allows the wearer to 'touch' and turn over objects, examining them in detail whilst calling up stored data relating to them, and looking to see where they have come from.
Similarly, another user may set the wood to be semi-transparent, blue-and-white glazed pottery to be fully present in its after cleaned state, white glazed sherds to be displayed as orange semi-transparent, and everything else invisible. This allows the outline of the ship to be seen for the purpose of orientation, but at the same time any relevant sherds can be seen even when hidden by wooden structures. The user wears data-gloves on both hands, selecting sherd images and fitting matching sherds together, in virtual reality, without having to touch the actual artefacts.
However, the artefacts are not in the positions they lay in when the ship sank; various natural, biological and human processes have moved them about. The archaeologist provides a model of the burial process. This can include how strata are deposited over time, how artefacts move within the strata, how the strata is disturbed by marine organisms, and other biological processes. Once the algorithm has been set up, the user can run time backwards and watch the movement of the artefacts back to their projected starting zones. When an artefact is shown as moving to the surface of the sea-bed it shows the time when the wreck sank, or when a salvage operation exposed or disturbed the artefact, or when natural erosion exposed the site (a big storm for example). Where the historical dates for these are known they can be used to calibrate the algorithm used. This information may also help marine biologists and climatologists to understand the relationship between the wreck and the local ecosystem.
Archaeologists trying to reconstruct complex broken artefacts or structures can use the projected original positions of objects to select items to match together. The reconstruction of broken objects can be used in turn to calibrate the dynamic model.
The marine biologist using the virtual-reality system may make the wreck and its artefacts semi-transparent, but wish to see selected species of coral or shells. The marine biologist may then refine the model for the growth and movement of the key species in the area.
There will, of necessity, be limits to backwards projection; the calculated original positions will always be subject to approximation. This is especially true where a large heavy object has sucked smaller adjacent objects into its slipstream. For example, a large brass artefact may disturb potsherds during its movements. Concretions also pose problems. Looked at backwards in time they will shrink, then the different objects embedded in them will break free and thereafter move independently. Of course, where the user elects to see artefacts in their cleaned state, clusters of objects, and not concretions, are seen lying together.
The user will be able to call images of standard objects from the library, such as types of pottery and coins. These images can be used to overlay or merge the possibly corroded coins retrieved. Where the standard coin extends beyond the surface of the recovered coin this may show in blue, where the recovered coin extends beyond the standard this shows in red, and where they match, in white. In this way coins can be rapidly identified. Such a matching system can, in principle, also be used to determine how many moulds or stamps were used to produced moulded or stamped artefacts.
The images can also be subject to enhancement, both by computerised image enhancement and by illustrators enhancing the image by means of electronic paint brushes. Pottery vessels can be reconstructed and the missing parts 'painted' in. where sherds are suspected of coming from a standard type of vessel, fitting the three dimensional image of the sherd onto the image of the standard vessel allows the approximate original position to be determined. This in turn helps with the reconstruction process.
Given the outputs possible with computer-based data systems much more detailed data has to be recorded than occurs at present. Time and money constraints which underwater working impose will have to be overcome m order to obtain the detailed and accurate data the virtual-reality output can handle. Clearly an automated data capture system is needed.
For this an accurate positioning system is set up, and the bottom scanned with remote-sensing equipment to outline the site before archaeological intervention. Then an automatic collector starts work. The automatic collectors are equipped with high definition TV cameras and other instruments, to detect and record the exact location and type of environment of all objects in the search area.
Once an object is picked up it is filmed by TV cameras in the packaging and lifting system. These cameras are linked to intelligent image-recognition computers to give an initial classification to the object. if the object has an associated magnetic field then this is recorded by the magnetometers inside the collector. The image of the object plus supporting data is displayed to the on-site archaeologist along with the computer's initial classification. The site archaeologist can accept or reject that classification. The computer's learning algorithm incorporates any such corrections into its image-processing system-so it learns by its mistakes.
The classification of the object determines the object's destination. If it is potsherd it may be transferred to the individual identification and recording system where it will he automatically numbered and wrapped in plastic awaiting further work on it. If it is a lump of coral it may be identified by type and bagged with other lumps of the same type, with random sampling for analysis of age and chemical composition.
The automatic collector can operate at varying speeds depending on the conditions found. Thus, it may operate quite quickly at first when removing easily recognised biological layers, sand and shells. When it reaches the first artefacts it will proceed slowly until its image-recognition system is working efficiently. Each wreck contains different types of artefacts, and the image-processing system has to be taught how to recognise each type and their fragments. Whilst some objects, such as potsherds, may be similar for most wrecks, other objects, such as clay pipes may only occur on a few. The computer-based image processing system will have to learn how to distinguish these from worm tubes, for example.
Once a level has been removed, the remote-sensing systems can be used again to analyse the next layer and the process is repeated. The images collected by the collector are used to create the virtual-reality space described above. Additional recording units are used to document vertical structures as they are exposed.
As objects are processed in the laboratory their records are updated. Thus the potsherd is recorded in its as-found state and its after-cleaning state. Objects may be reclassified at this stage. Concretions are tracked as they are radiographed and chemically processed.
At the outset of the dig the archaeologist will have to supply the computer with an outline of what the team expects to find. How big the ship is, over what area the remains are believed to be scattered, how deep in the sediment the remains are, and what speeds the collector should work at in specific areas. This initial model will be revised throughout the dig as more data becomes available. This data is linked into planning tools which are used to schedule the activities of the dig, and to estimate the number of containers and other materials which will be needed. This kind of management information is then available to the sponsors of the excavation, and to those who have made grants towards it.
The technology on which this system is visualised is the computer controlled compact disc technology now coming onto the market. The most profitable use for this technology will he interactive pornography; users will see on their screens TV quality images and will be able to interact with the computer to determine how the 'story' proceeds. The images will be full TV standard, and so considerably superior to current virtual-reality images. They will be held on laser disc similar in concept to those used in compact disc players. Ultimately this technology is going to be priced for the consumer products market.
High-definition digital TV will probably he required to provide the image quality necessary for archaeology and other uses. However, its release is anticipated and should be integrated into future plans. Specifying specialist hard-ware to support this system is a trap to avoid, since that will be many times more expensive than the consumer product. Also, the finished product, the wreck as it was when it sank, may well be saleable as a home entertairunent product. Publication of the results of a dig will, therefore, include two forms: the data on video disc, and the analysis of the results in learned articles.
The DCRAMS system will be multi-user and networked. The dig, laboratory work and the classification of objects all proceed simultaneously. This requires a networked virtual-reality system. However, such Systems will be, or are being, developed, with the object of providing competitive virtual-reality games.
The archaeological system outlined will be expensive to develop. This is not, however, a drawback, but an asset. Compare and contrast the sums of money sought and obtained by astronomers and nuclear physicists with those for archaeologist. Archaeologists do not benefit from the direct and indirect sponsorship of industry that is enjoyed by researchers working in fields which are major consumers of industrial goods and software. Astronomers spend money with the electronics and aerospace industries, and nuclear physicists spend on complex construction projects. Archaeologists must seek industrial sponsors and promote an expensive high-tech approach to obtain comparable funding.
The basic system required for archaeology has wide non-archaeological uses as well. These range from cleaning up and decommissioning nuclear sites, investigation of highly disturbed geological formations on the Moon or Mars, to police investigations. The system could even be included as a requirement for treasure hunters to use to ensure that at least some archaeological data is obtained, while also acting to protect the interests of share-holders and the licensing authority by ensuring all finds are recorded.
The development strategy should not be for a specifically archaeological system since the market is too small and too poor, but for a series of tools to record deposits/contexts and model the process of accumulation to be used in a variety of application areas. The key developments needed are the data interchange between the different modules. These are:
The resulting system is unlikely to be developed as a single unified project. It will evolve as the different modules are developed and then linked together. The sooner the standards of data storage and exchange can be agreed then the easier it will be to grow the different parts into a unified whole. Many of the modules will be produced as a result of developments outside the field of archaeology and may be available soon-the evolution of a unified system will take much longer.
Since archaeological records must be available to archaeologists other than the originators, excavation data should be stored in an industry compatible format. As data storage methods change, data from past excavations should be moved to the new media, or they will be lost forever. For example, data currently stored on magnetic media may need to be transferred to compact disc to avoid becoming unreadable and lost a decade later.
Use of the system will need careful planning, and intensive effort to set up and maintain once in operation. The amount of preparatory work will increase, but the excavation work on the sea floor should speed up. This will partly be due to the ability to run the system, whenever conditions permit, 24 hours a day. Priority should therefore be given to fully automating the underwater collector so that the physiological limitations of divers cease to constrain routine excavation. The diving tasks required to install and maintain the system underwater will still be constrained by permitted bottom times.
The proposed system does not eliminate archaeologists. it will reduce the number of people physically excavating, leaving them to do the set-up tasks and the one-off tasks, such as recovering large complex structures. The on-site archaeologists will have much greater control over routine excavation underwater. They will see each object at the time it is retrieved from the sea bed, and can stipulate how it is to be handled from then on.
Governments fail to take archaeology seriously because archaeologists fail to ask for big money which will ultimately be channelled back into industry. The expenditure archaeologists make on trowels, plastic bags, paper and people is simply too small a proportion of the markets in those commodities for archaeologists to be taken seriously. This paper recommends the development of innovative products needed to promote state-of-the-art technology. This will appeal to industry and to governments. It may also result in better archaeology.
Published in International Journal of Nautical Archaeology Vol 23/2 May 1994.