3D Imagery of Caddo Vessels from the Washington Square Mound Site (Selden)

I recently began collecting 3D scans of Caddo ceramics in an effort to preserve something of a digital proxy, since the collections that I’m working with articulate with human remains (these are NAGPRA collections, which will be discussed later). Since it was never my objective to simply catalog the vessels, I began searching for new methods of analyzing and documenting them. Pulling from plenty of guidance from the generous folks at Geomagic (whom fielded more than their fair share of questions), I was able to create 3D sketches of the motifs that could be viewed with or without the vessel.Image

Vessel image reproduced courtesy of the Anthropology and Archaeology Laboratory.

This image illustrates the design that occurs around the periphery of the vessel–which is broken, and missing a rather large chunk on the opposite side–both with and without the image of the vessel itself. While this was a productive exercise, and one that yielded some visually exciting results, I was curious what the utility of this method would be within the framework of a more traditional ceramic analysis. After scanning around 100 vessels, and revisiting some of them later, I found that it was often difficult to make out some aspects of the designs in the scans. Due to the nature of this preservation effort (the vessels were being reburied, and would no longer be available for study), I decided to revisit those ceramics with motifs that were not easily discernible to document the design using a 3D sketch. This–like most work with 3D objects–is a time-consuming process, so in an effort to capitalize on the time invested, only those elements that were difficult to see were included in the 3D sketch (which can be revisited and completed subsequent to reburial, if needed).

Image

Vessel image reproduced courtesy of the Anthropology and Archaeology Laboratory. 

Of the more intriguing approaches, to me, for analyzing 3D data is called geometric morphometrics, which is–simply–an analysis of shape. Pulling from numerous conversations with an old roomate of mine (a biological anthropologist named Tim Campbell) while I was at Texas A&M, I had learned quite a bit about the process, and often marveled at the results of those studies, wondering what might come of a similar undertaking aimed at Caddo ceramics. As a test of the method, I opted to first explore the variation in compound bowls, since that was the largest category of vessels within my collection at that time. To do this, I knew that I would need to collect a few strategically-placed homologous points (landmarks) from each of the vessels to capitalize on variation. In the end, I decided on 41 landmarks that would work well within and across currently defined categories, which I hope to compare in future analyses as more scans are completed.

Image

Vessel image reproduced courtesy of the Anthropology and Archaeology Laboratory.

 I should probably point out here that the above image is a carinated–and not compound–bowl. At the time of this posting, this was the best illustration that I had, and I think that it illustrates the point well enough.

So what happens after the points are collected? That’s a great question, and one that I spent plenty of time mulling over. Within traditional analyses of morphometrics, there are several options when it comes to the software used to analyze these kinds of datasets. In the end, I opted for version 2.5 of Morphologika. Now the only thing in my way was finding out how to export the points from Geomagic to a file type that Morhpologika would read. Of all of the steps in this process, this–for me–was the most tedious. In the end, perseverance paid off, and the dataset was imported.

Image

Some of you are–no doubt–thinking, wow…that looks like more data than would come from three compound bowls. Well, you’re right. This was raw data from a subsequent analysis, but it’ll work for explaining the process. After importing the points into Morphologika, they appear like they’re all mixed up with vessels of varying size pointed in almost every direction. What we do next is called a generalized procrustes analysis (or GPA) to standardize the size of the vessels (so that vessels of varying size can be used in the comparative sample). After the GPA is complete (below), the vessels are ready for an analysis of principal components (PCA).

Image

This next step is the most exciting, since it illustrates (statistically) the amount of variation in vessel shape within–what has until now been–a single category of vessel form. This is particularly interesting to me, since it adds yet another dimension to the toolkit of ceramic analysts globally (this method of analysis is certainly not limited to this dataset). The results of this preliminary analysis (below) point to a fairly wide range in vessel form, which is something that should be–and is–further clarified in future analyses. Additionally, this represents a fairly substantial increase in the utility of these 3D data, and is an avenue of research that is currently ill-represented in the academic literature. With any luck, we can solve that in the not-so-distant future.

Image

Vessel image reproduced courtesy of the Anthropology and Archaeology Laboratory.

 

***Next week I tackle the 11 carinated bowls from the Vanderpool site. Are there discrete groups of vessel forms within the carinated bowls? Could that imply potential variation in vessel function? Visit us again next week to find out.***

Written by zselden

Selden (PhD, Texas A&M University, 2013) is a husband, father, US Marine Corps veteran, cyclist, kayaker, backpacker, hiker, climber, fisherman and general all-around outdoor enthusiast. His research is focused at the confluence of archaeological methods and digital technology, and he is particularly interested in the application of 3D technologies to archaeological problems, geometric morphometrics, network analyses, predictive modeling, archaeological theory, and archaeological science.