Mean Ceramic Dating

adopted from Barber (1994) Doing Historical Archaeology #15

Dating is one of the basic tasks of historical archaeology and without it research will provide few insights into the past (it is the basic building block that we need to get over in order to talk about cultural stuff). 

Stanley South (AKA "Mr. Science") argued that it is possible to generate formulas that will allow archaeologists to simply plug in values from an archaeological deposit and calculate the date when it was formed. 

Artifacts, South argues, change in frequency in a predictable pattern: They initially are rare, rise to numerical prominence, then decline to rarity and finally disappearance (Like the battleship curves in Deetz's gravestones).  Consequently, at any given moment, the assemblage in use is a mix of the individual trajectories of the several artifact types that make it up.  This means that it should be possible to calculate a quite accurate date for a deposit by averaging the dates of manufacture for the artifact types within it, adjusting so that more frequently occurring artifacts have greater impact on the calculation.  This is the basis of his mean ceramic dating.

Mean ceramic dating is a method of calculating the date of a deposit on the basis of the ceramic types found within it.  A wide variety of types have been assigned median manufacture date, dates that approximate the average dates of manufacture for those types.  These dates have been approximated on the basis of archaeological experience, based on occurrences of ceramic types in dated contexts.  The most useful series of median date so far are for English ceramics in the Mid-Atlantic states and for Spanish majolica in the northern part of Spanish America.

To calculate a mean ceramic date, one uses the following formula:

 

O.K. calm down. . . . I know, its a formula, but its not that hard. . . really.

d₁ = median manufacture date of the type i and f₁ = the frequency of type i.  Stated in other terms, one calculates the mean ceramic date for a deposit by:

1. multiplying the frequency of each type of median manufacture date for that type,
2. adding these products together and
3. dividing this sum by the sum of the frequencies of the individual types.

In essence, this produces some sort of average date for the specimens in the assemblage.  Ideally, frequencies will be expressed in terms of numbers of vessels, but counts of sherds (broken pieces of pottery) are often used.

For example, let’s examine a fictitious late twentieth-century ceramic assemblage from a short-term deposit in Bella Vista, a well-to-do town in Northwest Arkansas.  The assemblage and its relevant characteristics are:

Type	Number of Sherds	Median Manufacture Date
Terminal American Grotesque ware	11	1991
Beverly Hills artware	22	1983
crystalware	33	1967
Black vitreous ware	44	1986
	---------
Total=	110

To calculate the mean ceramic date, multiply the number of sherds for each type times the median manufacture date for that type:

11X1991=21,901
22X1983=43,626
33X1967=64,911
44X1986=87, 384

add these products together:

21,901+43,626+64,911+87,384=217,822

and divide that sum by the total number of sherds:

217,822/110=1980.2 

The mean ceramic date for the Bella Vista deposit , therefore, is 1980.2.

A mean ceramic date supplies a single date to summarize a deposit’s age.  It is a precise date that can be calculated easily and can be reproduced independently by different analysts.  By using a mechanical procedure, every analyst should derive about the same date for the deposit, unlike the results obtained with more impressionistic dating methods.  These latter approaches to dating call on an analyst to examine the artifacts in a depost and produce a date that is reasonable in light of an overall assemblage.  In the hands of a knowledgeable and experienced analyst (like Henry Miller or James Davidson), impressionistic dating can provide accurate and useful dates, but choosing between conflicting results produced by different analyst can be difficult, often devolving to the reputation of the analyst.  Mean ceramic dating, for better or worse, reduces the impact of the analyst’s knowledge and experience.

 

Limitations of Mean Ceramic Dating

 Every method has limitations that must be weighed against its advantages.  Mean ceramic dating’s primary limitations are:

1. Obviously, mean ceramic dating is usable only where median manufacture dates have been approximated on enough common ceramic types so that most of the specimens in any assemblage can enter into the calculation.  Consequently, the method has been most successful along the eighteenth-century southern coastal areas of British North America (Maryland to South Carolina) and in sixteenth and seventeenth-century Spanish America.
2. Difference between the median dates of manufacture and the median dates of discard can distort mean ceramic dating.  If, for example, a particular ceramic was typically produced in 1750 but typically was broken and discarded in 1790 that 40-year lag will cause mean ceramic dates to be too early.  Some archaeologists have criticized the method on this account.
3. Related to the last issue is the problem of an assemblage with a few items that were very old at the time of discard.  A single antique, for example, coud have a major effect on a mean ceramic date, especially in a small sample.  For example, a deposit formed in 1800 with a single vessel from 1700 and four from 1800 will have a mean ceramic date of 1780, fully 20 years too early.  This process where some items are preserved by their owners and make their way into the archaeological record considerably after their manufacture is called the curation effect.
4. If a deposit forms over a long period of time, no single date will be adequate summary of its age.  If a deposit formed continuously from 1750 to 1840, a mean ceramic date of 1795 is hardly the only critical chronological information about that deposit.  Under these circumstances, the analyst must be aware of the range of artifacts occurring on the site—including whether some of the earlier types had ceased being manufactured before some of the later ones had been developed.  Using this line of thinking, an alert analyst usually will be able to get a good idea whether a deposit was formed over a long or short span.
5. The function of a site may affect how well mean ceramic dating works.  South is convinced that mainstream and frontier sites along the southern Atlantic coast of North America received goods from Britain essentially at the same time. (He refers to this phenomenon as horizon, where items of a particular type appear over a broad area rapidly.)  While this may be true there, it may be true everywhere else that mean ceramic dating could be applied.  Also, industrial sites or domestic sites of the poor may show consistent patterns of misdating, perhaps as result of using secondhand or handed down items.
6. Calculating mean ceramic dates with sherd counts may not produce as accurate results as calculating them with vessel counts.  Given the assumptions behind mean ceramic dating, vessel counts are the logical way to quantify ceramics.  Unfortunately, ceramics usually occur archaeologically as sherds.  There are ways to calculate minimum numbers of vessels for a ceramic assemblage, but they can be time-consuming and the results are not always satisfactory.  Consequently, sherd counts usually are used for men ceramic dating.  This means that sherds from a vessel that was smashed into tiny but identifiable fragments will have a greater impact on the mean ceramic date that one that was split neatly into three pieces.
7. Small sample sizes can produce distorted mean ceramic dates.  If you have a large sample of ceramics, the oddball sherd has only a minor effect.  The effects of a single curated specimen (or even a piece that was carried into the deposit by a rodent after the rest of the deposit was formed) will be overwhelmed by the combined effects of the large numbers of other, more typical specimens.  With a small sample, however, the likelihood of a few oddballs affecting the calculated date in a significant way becomes greater.  There are no fixed rules about how large a sample must be to avoid this problem, although there is a piece of archaeological folklore that suggests that a sample of 30 will usually be adequate.  South and others sometimes have been used mean ceramic dating on samples under ten and with apparent success, but such small samples should inspire concern in the heart of the analyst.

Your Assignment

In the early 1970s, Dr. Cecilia Bump excavated and analyzed the remains from a series of deposits in Virginia, North Carolina and South Carolina.  All the deposits were habitations in the British-American tradition.  At the time of Bump’s work, mean ceramic analysis was new and was not so widely used as it would become later, and she used more impressionistic methods for dating the deposits.  Today, you are an undergrad at Miskatonic University, studying under Leslie Squarkmuffin.  You are completing a senior honors thesis in historical archaeology and part of it involves seeing whether you can refine the Bump dates using mean ceramic dating.  Bump’s published description of the materials, using the same typology that South used in his method, is in the library and Squarkmuffin has agreed to contact Bump to get additional information, if you need it.  The ceramic assemblages from five of Bump’s deposits are summarized in Tables 1-5, using standard type numbers.  The type numbers, type names, median manufacture dates and manufactured date ranges from South (1977) are given in table 6.

Your assignment is to calculate mean ceramic dates for the five archaeological deposits excavated and described by Bump.  Your report should include a general introduction and, for each deposit, the mean ceramic date you have calculated and any appropriate discussion about its probable accuracy.

 

Table 1

TYPE	Number of Sherds
16	21
26	4
29	49
37	6
39	7
43	23
48	24
54	6
56	7
61	1
Total=	148

 

 Table 2

TYPE	NUMBER OF SHERDS
16	12
26	14
29	14
37	19
39	12
43	19
48	12
54	17
56	21
61	2
Total=	142

 

 Table 3

TYPE	NUMBER OF SHERDS
2	22
3	9
4	14
5	6
7	2
31	1
36	14
56	52
58	7
59	3
Total=	130

 

Table 4

TYPE	NUMBER OF SHERDS
2	4
69	1
Total=	5

 

 Table 5

TYPE	NUMBER OF SHERDS
2	44
5	1
22	23
69	1
Total=	69

 

Table 6

Type Number	Type	Median Manufacture Date	Range of Known Manufacture
2	Whiteware	1860	1820-1900+
3	Ironstone	1857	1813-1900
4	Polychrome pearlware	1830	1820-1840
5	Canton porcelain	1815	1800-1830
7	Enameled Chinese Porcelain	1808	1790-1825
16	Molded white slat-glazed stoneware	1753	1740-1765
22	Creamware	1791	1762-1820
26	Chinese export porcelain	1730	1660-1800
29	Jackfield ware	1760	1740-1780
31	English porcelain	1770	1745-1795
36	“clouded” wares	1755	1740-1770
37	Refined red stoneware	1733	1690-1775
39	Underglaze blue Chinese porcelain	1730	1660-1800
43	White slat-glazed stoneware plates	1758	1740-1775
48	Slip-dipped white salt-glazed stoneware	1745	1715-1775
54	British brown stoneware	1733	1690-1775
56	Lead-glazed slipware	1733	1670-1795
57	Plain delft	1775	1750-1800
58	Blue Rhenish stoneware	1668	1650-1725
59	Gray Renish stoneware	1700	1690-1710
61	North Devon ware	1713	1650-1775
69	Late Ming Chinese porcelain	1609	1574-1644