An Introduction to Seriation - Scientific Dating Before Radiocarbon

01
of 06
What is Seriation?

Clay Pots from Egypt from Various Times and Places
"Egyptian Pots": An illustration of clay pots from Egypt from various times and places, published in 1800. Description De L'Egypte, 1800

Seriation, also called artifact sequencing, is an early scientific method of relative dating, invented (most likely) by the Egyptologist Sir William Flinders Petrie in the late 19th century. Petrie's problem was that he had discovered several predynastic cemeteries along the Nile River in Egypt that seemed to be from the same period, but he needed a way to put them in chronological order. Absolute dating techniques were not available to him (radiocarbon dating wasn't invented until the 1940s); and since they were separately excavated graves, stratigraphy was no use either.

Petrie knew that styles of pottery seemed to come and go over time--in his case, he noted that some ceramic urns from the graves had handles and others had just stylized ridges in the same location on similarly shaped urns. He assumed that the change in styles was an evolutionary one, and, if you could quantify that change, he surmised it might be used to indicate which cemeteries were older than others.

Petrie's notions about Egyptology, and archaeology in general, were revolutionary. His worrying about where a pot came from and what period it dated to and what that meant to the other objects buried with it were light-years away from the ideas represented in this photo dated to 1800, in which "Egyptian pots" was considered enough information for the thinking man. Petrie was a scientific archaeologist, probably close to our first example.

Sources and Further Information

See the bibliography for a list of sources and further reading.

02
of 06
Why Seriation Works: Styles Change Over Time

A 78 rpm Gramophone Player from 1936
A 78 rpm Gramophone Player from 1936. Zecas

The seriation method works because object styles change over time; they always have and always will. A good example of a change in artifact type is the development of hand-held PDAs from those first enormous cell phones. Beam me up, Scotty! As an example of how change through time works, consider the different music recording methods that were used in the 20th century. One early recording method consisted of large plastic disks which could only be played on a huge device called a gramophone. The gramophone dragged a needle in a spiral groove at a rate of 78 revolutions per minute (rpm). The gramophone sat in your parlor and certainly couldn't be carried along with you and your earbobs. Thank goodness for mp3s.

When 78 rpm records first appeared on the market, they were very rare. When they became popularly available, you could find them everywhere; but then the technology changed and they became rare again. That's change over time.

Archaeologists investigate trash, not shop window displays, so we measure things when they are discarded; in this example, we're going to use junkyards. Archaeologically, you would expect no 78s to be found in a junkyard that was closed before 78s were invented. There might be a small number of them (or fragments of them) in the junkyard which stopped taking junk during the first years 78s were invented. You would expect a large number in one closed when 78s were popular and a small number again after 78s were replaced by a different technology. You might find a small number of 78s for a long period after they were pretty much done. Archaeologists call this kind of behavior "curation" -- people then, just like today, like to hang on to old things. But you would never have any 78s in junkyards closed before they were invented. The same is true for 45s, and 8-tracks, and cassette tapes, and LPs, and CDs, and DVDs, and mp3 players (and really, any kind of artifact).

Sources and Further Information

See the bibliography for a list of sources and further reading.

03
of 06
Seriation Step 1: Collect the Data

Percentage of Six Musical Media Types in Six Junkyards
Percentage of Six Musical Media Types in Six Junkyards. K. Kris Hirst

For this seriation demonstration, we're going to assume that we know of six junkyards (Junkyards A-F), scattered in the rural areas around our community, all dated to the 20th century. We don't have historical information about the junkyards--they were illegal dumping areas and no county records have been kept on them. For a study we're doing on, say, the availability of music in rural locations during the 20th century, we'd like to know more about the deposits in these illicit junkyards.

Using seriation at our hypothetical junkyard sites, we will try to establish the chronology--the order in which the junkyards were used and closed. To start out, we will take a sample of the deposits in each of junkyards. It isn't possible to investigate all of a junkyard, so we'll pick a representative sample of the deposit.

We take our samples back to the laboratory, and count the kinds of artifacts in them, and discover that each of the junkyards have broken pieces of musical recording methods in them--old broken records, pieces of stereo equipment, 8-track cassette tapes. We count the types of musical recording methods found in each of our junkyard samples, and then work out the percentages. Of all the music recording artifacts in our sample from Junkyard E, 10% are related to 45 rpm technology; 20% to 8-tracks; 60% are related to cassette tapes and 10% are CD-Rom parts.

The figure on this page is a Microsoft Excel (TM) table showing the results of our frequency count.

Sources and Further Information

See the bibliography for a list of sources and further reading.

04
of 06
Seriation Step 2: Graph the Data

Percentage of Musical Media Expressed as Stacked Bar Chart
Percentage of Musical Media Expressed as Stacked Bar Chart. K. Kris Hirst

Our next step is to create a bar graph of the percentages of the objects in our junkyard samples. Microsoft Excel (TM) has created for us a lovely stacked bar graph for us. Each of the bars in this graph represents a different junkyard; the different colored blocks represent percentages of artifact types within those junkyards. Larger percentages of artifact types are illustrated with longer bar snippets and smaller percentages with shorter bar snippets.

A good source of information about how to make charts in Excel is Ted French's Excel Chart Tutorial (for several different versions of Excel).

Sources and Further Information

See the bibliography for a list of sources and further reading.

05
of 06
Seriation Step 3: Assemble Your Battleship Curves

Seriation of Musical Media - Exploded Bars
Seriation of Musical Media - Exploded Bars. K. Kris Hirst

Next, we break apart the bars and align them so that all of the same colored bars are positioned vertically next to the others. Horizontally, the bars still represent the percentages of musical recording types in each of the junkyards. What this step does is create a visual representation of the qualities of the artifacts, and their co-occurrence at different junkyards.

Notice that this figure does not mention what kind of artifacts we're looking at, it just groups similarities. The beauty of the seriation system is that you don't necessarily have to know the dates of the artifacts at all, although it helps to know which is earliest. You derive the relative dates of the artifacts--and the junkyards -- based on the relative frequencies of artifacts within and between sites.

What the early practitioners of seriation did was use colored strips of paper to represent the percentages of artifact types; this figure is an approximation of the descriptive analytical technique called seriation.

Note: Ashleigh S. points out that Excel can't do the "exploded bars" step for you, you'll need to copy each of the colored bars with the Snipping Tool and arrange them in another part of Excel to make this graph.

Sources and Further Information

See the bibliography for a list of sources and further reading.

06
of 06
Seriation Step 4 - Arranging the Data

Seriated Junkyards
Seriated Junkyards. K. Kris Hirst

Finally, you move the bars vertically until each artifact percentage bar group lines up together in what is known as a "battleship curve", narrow at both ends, when the media shows up less frequently in the deposits, and fatter in the middle, when it occupies the largest percentage of the junkyards.

Notice that there is overlap--the change isn't an abrupt one so that the previous technology isn't instantly replaced by the next. Because of the stepped replacement, the bars can only be lined up in one of two ways: with C at the top and F at the bottom, or vertically flipped, with F at the top and C at the bottom.

Since we know the oldest format, we can say which end of the battleship curves is the starting point. Here's a reminder of what the colored bars represent, from left to right.

  • 78 rpm
  • 33 1/3 rpm
  • 45 rpm
  • 8 Track
  • Cassette
  • CD Rom
  • DVD

In this example, then, Junkyard C was likely the first opened, because it has the largest quantity of the oldest artifact, and lesser amounts of the others; and Junkyard F is likely the most recent, because it has none of the oldest type of artifact, and a preponderance of the more modern types. What the data doesn't provide is absolute dates, or length of use, or any temporal data other than the relative age of use: but it does allow you to make inferences about the relative chronologies of the junkyards.

Why is Seriation Important?

Seriation, with some modifications, is still in use today. The technique is now run by computers using an incidence matrix and then running repeated permutations on the matrix until it falls out in the patterns shown above. However, absolute dating techniques have made seriation a minor analytical tool today. But seriation is more than a footnote in the history of archaeology.

By inventing the seriation technique, Petrie's contribution to chronology was an important step forward in archaeological science. Completed long before computers and absolute dating techniques such as radiocarbon dating were invented, seriation was one of the earliest applications of statistics to questions about archaeological data. Petrie's analyses showed that it is possible to recover otherwise "unobservable hominid behavior patterns from indirect traces in bad samples", as David Clarke would observe some 75 years later.

Sources and Further Information

Timing is Everything: A Short Course in Dating Techniques

Sampling

McCafferty G. 2008. Seriation. In: Deborah MP, editor. Encyclopedia of Archaeology. New York: Academic Press. p 1976-1978.

Graham I, Galloway P, and Scollar I. 1976. Model studies in computer seriation. Journal of Archaeological Science 3(1):1-30.

Liiv I. 2010. Seriation and matrix reordering methods: An historical overview. Statistical Analysis and Data Mining 3(2):70-91.

O’Brien MJ and Lyman LR 1999. Seriation, Stratigraphy, and Index Fossils: The Backbone of Archaeological Dating. New York: Kluwer Academic/Plenum Publishers.

Rowe JH. 1961. Stratigraphy and seriation. American Antiquity 26(3):324-330.