Context Relationships: Stratigraphic v. Physical

Properly understanding the relationships between excavated contexts is of vital importance in archaeology. Most often we think about these relationships in terms of stratigraphy but, given the nature of KAP’s recording system, another type of relationship is important to us as well.

Though we employ a stratigraphic excavation method at Kaymakçı, it is often the case that stratigraphic relationships don’t accurately reflect the physical relationships between contexts; to put it another way, which contexts “touch” which other contexts. These physical relationships are very important in the production of the context volumes that are one of the many outputs of KAP’s digital spatial recording system.

Below are some guiding principles for the recording of stratigraphic and physical relationships.

Stratigraphic Relationships #

Our main purpose in recording stratigraphic relationships is to understand the formation of the archaeological record; most project members will be familiar with this way of thinking about archaeological deposits. However, it is worth going over some guidelines that are particular to KAP’s recording system, and which have been articulated here to ensure consistency in interpretation between excavation areas and over time.

Creating relationships. Stratigraphic relationships can be recorded and viewed in the Excavation form in the database. All relationships are expressed “upwards” (e.g., context 2 is “earlier than, covered by” context 1), but reciprocal relationships are visible via an automatically generated table (e.g., context 1 is “later than, covers” context 2). In the case of “Equal to, same as” relationships that could be made in either direction, be careful to ensure that relationships are not duplicated. Relationships currently available in the database are the following:

1. Earlier than, covered by
2. Earlier than, abutted by
3. Earlier than, cut by
4. Earlier than, filled by
5. Equal to, same as (only used for contexts that are the same deposit)
6. Contemporary with, contained by (used only for objects that are contained by a larger fill)

Stratigraphic relationships can also be recorded and viewed in the Harris matrices produced for each excavation area. It should be noted, however, that not every context in the database will be represented in the Harris matrix (e.g., objects are not included in the matrix). See here for a free copy of Harris’s Principles of Archaeological Stratigraphy and here for an overview of how we produce Harris matrices at KAP.

Cleaning contexts. Due to the nature of KAP’s volumetric recording system, some actions that might be undertaken more casually at other excavations (e.g., scraping the scarps before taking pictures) must be formally recorded, including a photobatch of the bottom of the context. The solution we have found to this is the “cleaning context.” Cleaning contexts should be rare; they should remove the minimal amount of material necessary; and they should only be considered a “cleaning context” if they have no stratigraphic value. Examples include the initial cleaning contexts at the beginning of the season, scraping of the scarps, cleaning of walls/features prior to removal, etc. (Note that sediment pedestals are generally not considered cleaning contexts.) Because they have no stratigraphic value, cleaning contexts should have no stratigraphic relationships (they will, however, have physical relationships, as described below.)

Group contexts. Group contexts are used to group together related contexts and can be sorted into two conceptual types: features and interpretive groupings. Group contexts for features allow us to give a consistent name to a feature and to associate all related contexts as “children” of the group. Because the feature is a single stratigraphic unit, all stratigraphic relationships to other deposits should be made to the group; the “children” of the group might have stratigraphic relationships to each other, but they should not have stratigraphic relationships with other deposits or features externally.

Interpretive groupings, on the other hand, are used to associate contexts into interpretive units, some of which may be stratigraphic units. Examples include grouping fills that are considered to be the same deposit or grouping multiple walls that make up an architectural unit. Because these associations are interpretive—and subject to reanalysis—stratigraphic relationships should be made to the children of the group rather than the group itself. (See more on group contexts and their stratigraphic relationships here.)

Negative Features/Interfaces. There are some instances in the formation of the archaeological record where material is removed rather than deposited; this might include the digging of a pit, the robbing of wall stones, or the erosion of a deposit over time that is later buried again. At KAP, the only negative interfaces we record currently are anthropogenic negative features: pits, post-holes, identifiable cuts in the bedrock, etc. Recording these interfaces requires the creation of a new context that has appropriately descriptive relationships with the excavated units surrounding it; for example, fill 4 is cut by pit 3 (the negative feature), and pit 3 is filled by pit fill 2.

NOTE: Some contexts in the KAP archive do not exist volumetrically because they were not closed properly (see below). Assuming there is sufficient data to do so, however, you should still record the stratigraphic relationships that these contexts have with surrounding contexts.

Physical Relationships #

Our main purpose in recording physical relationships is to aid the production of context volumes. The process of producing volumes can be reviewed here and begins with the identification of point clouds that together create a “watertight” envelope around the spatial context. These point clouds are the bottom photobatch of the context in question and the context’s top, which is actually the bottom(s) of the context(s) above/around it.

Recording the bottom of a context is an essential part of closing the context during excavation; if a context is not properly closed and a bottom is not recorded, the context cannot be represented volumetrically. It is possible to close multiple contexts with the same bottom photobatch, but only if the contexts in question are clearly separated from each other physically. Examples of this might include removing two small features in the same area that are non-contiguous (not touching), or using the final photobatch of the excavation area to close contexts in opposite corners of the area at the end of the season. Closing multiple contexts with the same photobatch can save time but might lead to lower quality data, so these factors should be weighed against each other when making decisions. The bottom photobatch number for a context can be found in the Excavation form under the entry for the context in question; if multiple contexts have been closed with the same photobatch, this should be noted in the context description so that it is clear to later users.

Determining the top(s) of a given context is more variable and can be complicated. While some contexts have a single top—perhaps because they were located completely beneath a larger context, or because they were pause contexts (see here for more information)—it is often the case that the top of a context must be created from multiple point clouds that represent the bottoms of contexts above and around the context in question. Some basic principles for identifying tops include the following:

1. The goal is not to determine the fewest number of bottom point clouds that cover the top of the context in question; the goal is to identify the last bottom point cloud to intersect with each part of the context in question. This sometimes means that there are many tops for a single context.
2. Point clouds will generally contain material that is considered outside the boundaries of the context, with the boundary being indicated in Metashape and the orthomosaics by the locations of coded targets. (The boundaries of the context are also indicated by the spatial context polygons in the GIS data, drawn by excavation area supervisors.) Areas of a point cloud outside the coded targets will necessarily be less accurate the the areas inside the targets, so do not rely on these parts of the point clouds to provide coverage of the top of a spatial context.
3. Whereas cleaning contexts are excluded when considering stratigraphic relationships (see above), these are often very important when establishing physical relationships. For example, the cleaning contexts from the beginning of each season provide the best top point clouds for the first spatial contexts excavated that season.

Given how complicated it can be to identify the top(s) of a given context, it is often useful for excavators to record the specific top of a context in a new photobatch, with the eventual goal of making the construction of that context volume easier and more accurate. This most often occurs when recording features that are about to be removed and pause contexts (see examples below), but may be useful in other circumstances also.

Features. In the case of features that have been left in situ for an extended period during or across excavation seasons, it is often difficult to find or construct a clear top of the feature. For example, consider a wall: while parts of the wall will be captured in the bottoms of many other contexts, it is unlikely that the whole wall would be captured except in large photobatches from the beginning and end of the season. Before removing a wall like this that has been left in situ for an extended period, the recommended course of action is to open a cleaning context for the feature (see above); in recording the bottom of the cleaning context, you can ensure that you produce a high quality photobatch of the whole feature that will serve as the top point cloud during volume construction. Such a cleaning context will also allow you to remove any excess material (e.g., soil that was never properly scraped off) so that it is not in the final volume and also not in the types of high-quality photographs that need to be taken before features are removed (i.e., physically destroyed). (See here for more discussion on how to approach removal of a feature.)

Pause Contexts. These are most often used for objects found during the excavation of a larger context, such as a complete or nearly complete ceramic vessel. In a pause context, both a top and a bottom are recorded for the context with the same photobatch number, with the top photobatch allowing the volume of the object to be isolated from the volume of the surrounding open context, thereby facilitating subsequent volume construction.