This post will outline one of the most important tasks in post-excavation analysis – working out and showing how features relate to one another. I’ll discuss how records of data retrieved from the Crickley excavations might be used to establish stratigraphic relationships, and illustrate one common way of showing relationships – a type of diagram known as the Harris Matrix. I’m currently undertaking this task in I’m preparation to digitise plans of a building in a GIS programme (see 7. ‘ Digitising Crickley Plans and Using GIS‘). I’ll begin with an example of how a matrix might be used, in conjunction with context records.
Here’s an example of how matrices might be used (this matrix is the diagram that I’m working on today: it’s therefore not quite complete. So I need to flip between the context records and my earlier report more than I might need to ordinarily, in order to double check that all of the details are correct, and to see if I’m missing any info). This is something I had to do earlier, to verify details for the summary site report that I put online today (5. ‘Interim publication: Roman and early Medieval Crickley Hill‘). My task was to clarify exactly where a Roman coin (probably dating to the 3rd century) fits in the sequence of activity. I went to my dissertation to consider the information on the coin:
Then I looked in the finds records (knowing that this building was excavated in 1983), to double check where it was found (in the database, it can be seen in X94, with the coordinates):
This coin was recorded as being found amongst F (= feature no.) 5452: an area of collapse of the modern stone field wall (F5455). I took a note of the coordinates of this find (COORD_X & Y – the X & Y COORDS are those used in the previous plans created within AutoCad, which I’m now redoing in GIS: see post 7. ‘Digitising Crickley plans and using GIS’), from which I could see by looking on the plans was not clearly related to any definite period 4 features. So I used the matrices to consider the relationships of associated features in more detail:
The context F5452 can be seen to underlay topsoil (F5451), overlay F5526, F5509, F5460, and F5449, and be contemporary with F5461=5471 and F5455 (the modern field wall). I went back to the context records to read the descriptions of, and locations of, these contexts, check if any of them were associated with the period 4 settlement, and to see (considering the location coordinates of the coin) whether any of these might coincide geographically with the coin. F5526 was a possible post-hole in the area of the period 4 settlement, but was not near the location of the coin. F5509 represents the house platform of the first phase of the building in the centre of the period 4 settlement, but again the coin was not associated with this feature. I could see that the associated feature F5461=5471 was clearly a continuation of the wall tumble into adjacent cuttings (with F5471 representing this feature within the same cutting as the coin):
Under the collapse F5471 on the east side of the wall, part of the old ground surface (F5472) was recognised, but this could not be directly associated with the location of the coin.
I could say nothing more about the date of the contexts from this find. I might conjecture that the coin had lain on the old ground surface behind the house (close to the most direct route from the entrance through the settlement enclosure and across the settlement zone), which the collapse of the house walls may have disturbed this surface at a later date (the finds records also recorded 17th and late 18th-early 19th century ceramics from this context). However, I couldn’t demonstrate this possibility, and neither was I able to use the coin to provide anything but a ‘TPQ’ (terminus post quem) – the period or time during or after which a context might be dated: these features (the modern wall and its collapse) were anyway clearly later. This is often the way it goes in archaeology!
I examined the relationships when undertaking research for my MA in 1997-99, but as I noted in a previous post, one of the problems I’ve encountered has been the failure of a number of data files since this time. I therefore have to re-do a many of the Harris Matrix diagrams (schematic plans showing stratigraphic relationships features / strata). The context databases are used to examine relationships between the features. Many of the relationships were also recorded within my MA dissertation – which I am now re-writing to provide greater clarity for the report on Roman and early Medieval activity on the site, and to incorporate discoveries I have made over the last decade:
The re-creation of matrices has involved using a computer programme that automatically arranges the relationships when the archaeologist inputs the data (though this information usually has to be arranged further manually). This programme is very good (and I certainly recommend it), but can be frustrating when it repeatedly crashes, after having spent time arranging the diagrams!
Having spent several hours recreating a Harris Matrix diagram for a two-phase 5th century building in the centre of the enclosed settlement at Crickley Hill (during which the programme crashed 11 times – so two steps forward and one step back!), I now have to end this task for the day, as the programme is refusing to work. This is where I’ve had to leave it for the moment – this diagram includes earlier, and a few later, features (which is why it’s so large) that I’m in the process of disentangling from the period 4 activity.
In my next post I’ll outline another job I’m doing at the moment – 7. ‘Digitising Crickley Plans and Using GIS’.