My apologies to readers who come to the blog for discussions of heritage. This is a post about a technical matter in archaeological recording. I’ve written very quickly to get some very old ideas out there to be useful to others.
I’ve been having a Twitter discussion this morning about how the Harris Matrix is used in archaeological analysis and how that can be supported by digital systems. As often happens it was chaotic and high energy discussion. In the course of the discussion I realised that some ideas I presented at WAC5 in 2003 but have not published separately from the English Heritage Research Reports Series are still worth thinking about and still hold opportunities.
The core of the issue is that in single context recording there are two different systems representing the relationships between archaeological contexts, plan and matrix. The plan represents the spatial relationships and the matrix represents stratigraphic relationships (which are the basis of the temporal relationships).
GIS allows us to interrogate data based on spatial relationships. So we can draw a plan of all the contexts in which a certain pottery type was found. But there is no similar software to allow us to show the location of that pottery within the matrix.
So in 2003 I exported a matrix that had been drawn in CAD into ArcGIS as if it were a plan. This allowed me to display the data ‘in matrix’ in the same way that you would ‘in plan’ The data is from the Stanwick Roman Villa in Northamptonshire and I’m grateful to Vicky Crosby, the project manager, for making the data available and to Liz Muldowney who created the matrix in CAD.
The slides here show some data presented that way. Because the data was not collected with this kind of tool in mind, they aren’t well organised for displaying on. So a bonding agent in a wall may be described in 15 different ways and its hard to display them all at once. The key to making a system like this work is, as ever, data standards. Both standards for describing the attributes of contexts and also for recording their relationships. If we implemented these kinds of standards we could both interrogate our data more easily in Post Ex and also make our data more useful to future researchers.