| PREVIOUS | CONTENTS | FORWARD |
Recently the Archaeological Data Service in York provides a service for digital data making entire site archives from around the world accessible on the internet ads.ahds.ac.uk/. In long terms this may help to apply common data standards.
Computerise as many as possible records because then they can easily be sorted for a report or made accessible in your local or worldwide archives. Some excavation projects such as in Heslerton, North Yorkshire, have recorded straight into hand-held computer devices on site (Powelsland 1986). At the Museum of London we have tried inputting data on site from context sheets into a database in 1982, but it proved unpractical in a crowded and dusty environment of urban rescue excavations. However, since we know little about the long-term legibility of diskettes, paper printouts and/or microfiche security copies of all records still seems to be a safe form of archive and digital backup copies should be updated to current standards or eventually put by ADS on the Internet (Hodder, Holtorf 1995).
The parts of the field record to be computerized are:
The Bonn Harris Program (downloaded from AL001@MIL1.RRZ.UNI_KOELN.DE or from http://www.uni-koeln.de/al001/basp.html) helps to check the correctness of matrices and presents them in a neat form. The program also deletes redundant relationships if physical relationships had been entered. For complicated stratigraphy, instead of entering the context numbers from the context sheets, it is recommended to enter first the plan matrices per grid-square each separately, before linking one by one the other squares to the first one. This is done by producing ".lst" files, which are exported into a word-processing file, merged and saved under a new name file. When merging, it is important to make sure that the two files link without a heading or spare paragraph line. The sorting is done automatically when re-imported to WinBasp. Merging in stages has the advantage that inconsistencies or stratigraphic circles will become obvious step by step. The missing fills above the cut numbers can then be added to complete the matrix.
Digitised single context CAD plans can quickly be assembled into any wanted groups. If for economic reasons not all plans can be digitised, provide at least all structural elements in situ including floors with a straight edge (to show where a wall once was that left no traces), roads and yards. Such context plans can be printed in any wanted order and combination, for instance when using a database in conjunction with ArchView. For analysis it is best if all printouts are using the same orientation (always north at the top) and scale (usually 1:100 or 1:200), because this makes it easier to keep track of structural alignments and occupation clusters.
Context sheets, particularly if there were many, hand-written and described by various recorders take time to read. Although it is not essential to have the context descriptions translated into coded computerised form, I found it a very helpful tool for research to obtain quickly comparable context details. In listed form columns of coded components and inclusion can easily be sorted and provide an overall view. The MoL Site Manual shows how to estimate and express percentages of particle size and composition (MoL 3.2, fig 11, Hodgson 1974).
The list only translates percentages in broadest terms but as the initial record was made by judgment of the recorder it does not matter. If a layer is described on the context sheet as 'silty clay' this could be coded by an estimated percentage of 'cla60, sil40', while clay with a little silt would read 'cla80, sil20'. The inclusions are added in separate columns in their quantities 'frequent' (3 of a kind, 'moderate '2' or occasional '1'). Special finds and context dates can join the list as early as their information becomes available. You find an example of such a context list in coded form below in 'Fictive Excavation' (Appendix 1.3).
All these elements of single context records and their codes form the basis for structural analysis. Their coding helps to sort and compare records easily. Although the context number will remain the 'pivot' reference in the database to link structural evidence with finds and samples, a more meaningful higher interpretation for events and dating is established through grouping.
| RETURN TO TOP |