Digital Fingerprints

The Origin and Developement of the Cuneiform System of Writing. A table showing the forms of eighteen representative signs from about 3000 B.C. to about 600 B.C.

Our first project is being funded by the National Endowment for the Humanities Digital Humanities Initiative Program to explore the feasibility and power of using laser scanning and three-dimensional quantification and analysis of a collection of cuneiform tablets. Our aim is to see if our approach is helpful in telling tablets apart from eon another, in grouping those that are similar and distinguishing those that are different in an objective and repeatable way. This is a first step toward tracing the origins of tablets without interpreting the meaning of the writing, which would provide researchers with an independent way of testing studies based on translation of cuneiform text. Our research will help the students of cuneiform writing, epigraphers and Assyrologists, as well as archaeologist, with new tools for studying how tablets are produced, where they came from and who may have written them.


While most Assyriologists are concerned with what is written on a tablet, we are concerned with how it was written and what we can learn from the way it was written. To test our methodology, we selected a study group of tablets on which we can reliably find the same signs, with the same meaning. The questions we asked: Were these signs made in a consistent uniform way, as if they were made by one person? Or, are they so variable that we must assume the signs were made by different people?

The Sample

Thirty one small triangular tags from an unknown site in Mesopotamia, CUNES 50-15-003 through CUNES 50-15-033, are a series that was made sometime during the Old Babylonian period. Judging by their appearance, and the style of the seal impressed on them, we think they were made about 3,700 years ago, during the late 18th century B.C. The tags are remarkably well preserved. Only a few have small portions broken and missing. (ILLUSTRATION) Most of the original surface of the tablets where the writing appears is pristine. Each tag has a date written on it, along with a message. The dates cover a period of 45 days. In most cases, we have only one tag per date. However, for days 1, 2, 11, and 22 of the sequence, we have two tags.

Each tag has a narrow channel cut into it. When they were made, the clay was formed around a knotted piece of string running through the edge of one side. (ILLUSTRATION). While none of the string has been preserved, a set of two holes remain in opposite corners of the tags, connected by the channel. The exact purpose of the threaded string is not known. It may have been used to hang a tag on a storage hook or peg, or from a belt or loop for carrying. Whatever the explanation, it seems that the tags were not meant to last long. We think they were written quickly and rather carelessly. The messages are very short and simple, without any extraneous information. And, they were written in a pattern, like a list. In fact, the messages are so condensed that they are difficult to translate.

All the tags deal with the same contents. They record a single occurrence of a transaction that was repeated frequently. Except for a few variants, they read as follows:

  1. 2(ban2) duh-a
  2. AN.HA.NE
  3. giri3 i-bi-{d}nin-szubur
  4. (month)
  5. (day)

The inscription may be translated as: "Two liters of bran, . . . , (disbursed) by (orders of) Ibbi-Ninshubur."

All are sealed with the same seal, which is inscribed as follows:

  1. {d}na-bi-um
  2. dub-sar sag-ila
  3. ki-ag2 {d}AMAR.UTU

This seal inscription means: "Nabu, the scribe of the Sagila (Temple), beloved of Marduk."

This is a common type of seal inscription, giving the name and nickname or label of a major deity. Nabu held a prominent place in the pantheon of Babylon. As scribe of the gods, he was the patron god of all scribes. He was also the father of Marduk, the chief god of Babylon. The fact that the seal inscription honors Nabu suggests the seal was carved in the territory and time of the First Dynasty of Babylon, and it seems likely that the tags were written in that state as well.

Since the same seal appears on each tag, we hypothesize that the tags were written all by the same scribe, who was the seal owner. We cannot test whether the seal owner was identical to the scribe, but we can test for evidence that the tags were written by more than one scribe.


So far we have chosen three symbols to landmark, Ha, An and Pa. To date, 18 Ha, 8 Pa and 9 An symbols have been digitized. For each symbol, all measured "individuals" were then assembled into a single file and processed using the morphometric software morphologika. This software uses a technique called Generalized Procrustes Analysis (or GPA) to superimpose all the shapes on top of each other, scale them, and remove any rotational or translational differences that might be present. These rotational or translation differences refer to potential positional differences in orientation and position of each laser scanned tablet or tag within the visualization software. The result is a set of adjusted coordinates where geometry is preserved. Any differences between landmarks are real shape differences between symbols that can then be explored in further detail. Thus the spatial relationships between the different wedges that make up a sign can be explored in a true three dimensional sense (as can the differences in relative size of each wedge).

We then subject these superimposed shapes to a multivariate statistical procedure called Principal Components Analysis (PCA). This essentially compares all the shapes to each other, taking the geometric position of each landmark into account. A new series of summary variables (which are divided into separate "axes") are produced, which can be displayed on conventional scatter-plot graphs. Theoretically, each axis explains a different aspect of observed shape differences in the sample, and most importantly, these shape changes can actually be visualized by warping wireframe renditions of the wedges and signs. The elegance of PCA is that the majority of variation in the sample is expressed in the first few axes.

As an example of work in progress using these techniques, the graph displays the results of a PCA of a series of individual An signs taken from a series of tags (blue circles), and a single large tablet (red squares). Even based on this preliminary analysis, it can be seen that there is separation between the An signs depending on whether they come from a tag or the large tablet. There is also some significant separation between signs on the different tags (as seen on the x-axis), which is interesting and requires further investigation.