Thoughts Towards A User's Guide to Ceramic Assemblages, Part I:

Lumping Sites into Mega-assemblages by Those That Cannot Tell Time.

George L. Miller URS Corporation
Council for Northeast Historical Archaeology Newsletter, No. 18, April 1991 pages 2-5

What constitutes meaningful data from excavated ceramic assemblages and the
scrutiny to which it is subjected is a gray area in the archaeological literature. There
seems to be an unstated assumption that once an assemblage has been described and
quantified, it is as good as the next one as a building block in reconstructing and
understanding the past. Quantified assemblages are presented as representing the whole
period of a site's occupation with few questions asked in terms of what the assemblage
does or does not represent. This paper is meant to raise issues for further discussion that
will hopefully lead towards the establishment of some guidelines for the evaluation of
ceramic assemblages. This essay will start with what seem to be the most common
problems and the unstated assumptions that they represent.

The most common format encountered in dealing with site reports is the
phenomenon of lumped assemblages in which all of the ceramics are combined into a
single laundry list that is presented as representing the whole period of the site's
occupation. This technique, or one should say lack of a technique, no doubt goes back to
our origins as prehistorians where the ability to "tell time" is limited compared to what is
possible with the chronologies available to historical archaeologists.

Archaeologists have established basic standards for dealing with excavated

collections, which, at a minimum level, involve recording the grid and level locations for
excavated artifacts. This time-consuming process is one of the things that makes
archaeology so expensive. Why bother to expend all the time and energy used to
establish and maintain locational and associational data if the artifacts are just going to be
lumped together in one assemblage whose only context to be considered is the site? Is
there a difference between a lumped collection from a carefully excavated site and a pot
hunter's collection where the only provenance is the site? Clearly, the lumping of
excavated collections, particularly in sites occupied for long periods of time, is a cop out
on one of the main objectives of archaeological research which is the study of changes
through time. Instead of dealing with change within sites, much broader lumped periods
are being used to study change in which sites are used as the basic building blocks.

There are sites for which lumping is a very appropriate approach, for example,
those which were occupied for periods of less than 10 years and have not produced
enough artifacts for meaningful breakdowns. Even for sites occupied for short periods of
time it is sometimes possible to break the collections into sub-assemblages. For example,
Meredith Moodey was able to use locational data in combination with information on
cross-mends and ceramic chronology to segregate a plowzone collection into three sub-
assemblages representing the sequence of deposition at the Franklin Glass Works, which

1
was only occupied for eight years. This breakdown establishes a sequence of acquisition
for the site's eight short years of occupation which enabled Moodey to examine the
impact of economic stress of the failing glass works on ceramics purchases (Moodey
1988). Moodey clearly demonstrated that it is possible to break down a plowzone
assemblage from a site only occupied for eight years. That level of control of our data is
necessary to deal with questions of acquisition and deposition of ceramics, and of course
the site formation process will be better understood by establishing better control of the
element of time.

Lumping of archaeological collections carries with it some implied assumptions

which need to be examined. Perhaps first and foremost is the assumption that they are
representative of what was on the site. Time and again, the ceramics from an excavation
are presented without any discussion as to how representative they are of what was on the
site. Rarely is there any indication of what percentage of the site was excavated, or an
estimate by the person presenting the collection as to what percentage that they think the
collection represents of what was deposited on the site. There seems to be an assumption
that as long as the number of vessels recovered is fairly large then they are representative
of what was on the site. Looking at the distribution of vessels from 19th century sites
that have been occupied for a long period of time, one is frequently struck by how
skewed the assemblage is in representing only a segment of the time that the site was
occupied.

John Otto's Cannon's Point Plantation, 1794-1860: Living Conditions and Status
Patterns in the Old South illustrates some of the problems in the lumping of ceramic
assemblages, the implicit assumptions involved and how they do not hold up to close
scrutiny. Otto set out to excavate assemblages from the households of the plantation
owner, the overseer, and a typical slave cabin. This he accomplished; however, his
control of the data beyond that has several major problems which were compounded by
the lumping of his ceramic assemblages and assumptions as to what they represent in
time periods.

Consider the main plantation house. A fair amount of historical research

established this basic outline of the plantation's history. Cannon's Point Plantation was
built by John Couper in 1794 to take advantage of a strong market and high prices for
long-staple cotton. Previously he had been a merchant. John was born in 1759, married
in 1792, and moved his family to the plantation in 1796. The initial phase of the
plantation was very successful, and by 1804 Couper had built two cotton gin houses and
owned several hundred slaves. However, there were setbacks. In 1804, a hurricane
devastated the cotton crop. Shortly after that the Embargo of 1807 followed by the War
of 1812 put further economic strain on the operation which caused Couper to mortgage
290 of his slaves to borrow $100,000. During the War, 60 of his slaves were carried off
by the British. The market for long-staple cotton recovered after the War briefly but then
fell during the panic of 1819 and remained low through the mid-1820s. In addition,
another hurricane devastated the cotton crop in 1824, and in 1825 a crop was lost to
caterpillars.

2
 In 1827 John Couper declared bankruptcy as the price of cotton fell and took with
it the value of his slaves and land. While Couper lost another plantation and other
investments, he came out of his bankruptcy still owning his Cannon's Point plantation
and 100 slaves. By 1828, Couper was on the way to recovery, and prices for long-staple
cotton rose to a peak in 1837 of between 45 to 50 cents a pound. After the panic of 1837
set in, the prices fell to a low of 18 cents a pound in 1842. John Couper retired to
Hopeton Plantation in 1845 at age 86. Five years later John Couper died, and the
ownership of Cannon's Point passed to his son James. "From 1845 to 1861 the only year-
round white residents of Cannon's Point were the hired overseers who supervised the
slave force . . ." (Otto 1984:124). The management of the Cannon's Point Plantation was
taken over by John's son, James Couper, who managed it from his residence on another
plantation. After 1845, according to Otto's research, the Couper family used Cannon's
Point as a summer home, residing there during the malarial season. In 1862, Union Army
troops occupied the area. The plantation house was described in 1864 by a Union naval
surgeon, who included the following statement:

In the basement, large quantities of [fossil] bones and minerals of all sizes and
kinds are scattered around the floor. Broken furniture, dilapidated paintings, and
broken crockery by the boat load are strewn around the rooms (Otto 1984:30).

John Otto pulled together all of the above data, presented it, and then ignored its
potential for providing insights on his excavated assemblages. From the historical
synopsis, it is clear that John Couper set up his household after marriage in 1792 and
moved his family to Cannon's Point in 1794. He appears to have done very well until the
hurricane of 1804, the embargo of 1807, and the War of 1812. The fall of cotton prices
following the Panic of 1819 and losses of the 1824 crop to another hurricane and the
1825 crop to caterpillars led to a bankruptcy in 1827. In the 1830s the plantation seems
to have flourished. One would guess that the ceramics from initial setting up of the
household would be quite different from those in use three decades later after the
plantation had recovered.

Instead of using this information, Otto presented a lumped assemblage from the
excavation of the midden associated with the Couper's plantation kitchen as being
representative of the whole 66 years of the plantation's occupation. Several aspects of the
ceramic assemblage and its associations indicate that this assemblage probably represents
a period after the hurricane of 1824 until the 1860 abandonment of the plantation. One is
that almost 80 percent of the 935 nails recovered from the kitchen were machine cut and
headed which places them no earlier than the 1820s. John Couper lists several
outbuildings and 12 slave cabins being lost in the 1824 hurricane, suggesting the kitchen
midden with which the ceramics are associated probably began accumulating in the
1820s after the kitchen was built.

Supporting this supposition is the makeup of the ceramic assemblage. Otto

presented the 1242 sherds recovered from the planter's kitchen as representing the whole

3
period of the Couper family's occupation of the site and came up with a ceramic mean
date of 1818 for the assemblage. Otto states that 1818 is a good fit for the site because he
feels that represents the main period of the site's occupation which was from 1796, when
John Couper moved his family to Cannon's Point, to 1845, after which it was just
occupied during the summer season. The median of this period would be 1820.

While the date generated from Stanley South's formula is within a couple of years
of the median of peak occupation period, it appears to be far too early for the ceramic
assemblage. Just over 2 percent of the sherds from the Planter's Kitchen were
creamware, which seems far too low for an assemblage that began accumulating in the
1790s. Pearlware, on the other hand, made up almost 86 percent of the ceramics, while
whiteware only accounts for 4.5 percent of the sherds. Those proportions suggest a site
that was occupied for a short period in the early 1820s. A more plausible explanation is
that Otto's definition of pearlware was too broad and that any trace of blue in the foot ring
would have cast the vessels into the pearlware category. When dealing with blue printed
wares this can be a problem as the definition of what constitutes pearlware is an arbitrary
one. Given that whiteware began showing up in American assemblages shortly after the
War of 1812 and had become very common by the 1830s, particularly with the growing
popularity of red, green, brown, and purple printed wares during that decade, one would
expect a much higher proportion of whiteware in the Planter's Kitchen assemblage.
Unfortunately, Otto does not mention the colors of the printed wares he is dealing with.

The Planter's Kitchen assemblage is presented as one lumped context. However,

in Otto's earlier article in Stanley South's Research Strategies in Historical Archaeology,
Otto presented the kitchen assemblages broken down into "zones" as shown in Table 1.

These zones appear to have integrity as a meaningful time sequence. In this table,
the whiteware category includes the decorated as well as undecorated whitewares,
whereas the printed category are those listed as printed pearlwares by Otto. Assuming
that these "zones" represent layers, the TPQ artifact in layer III would be the white
granite wares, suggesting that the level was accumulating material until at least the mid-
1840s. Level II therefore must postdate ca. 1845, which would have been a period of just
summertime occupation of the plantation. Given that sequence of events, I doubt that the
printed pearlware made up 64 percent of the sherds. There must be more printed
whitewares than Otto has identified.
Again, looking at the above levels, they seem to suggest that shell edge may have
been the site's earliest tableware, which was probably replaced by printed wares after the
War of 1812. However, to confirm that assumption one would need to see the shell edge
and printed wares. Why Otto chose to lump these three zones in his book is not clear, nor
is there any information presented on what was found in Zone I. On page 66 of his book,
Otto discusses changes in the styles of transfer prints which he summarizes as follows.
The earlier decades of production were dominated by oriental patterns, which were
replaced by English, American, and Near Eastern scenes that remained popular to the
1840s, which were then replaced by floral patterns. Unfortunately, Otto did not use this
information to provide his readers insight to the dating of his assemblages that would

4
have gone beyond the ceramic mean dates that he generated. None of the transfer printed
patterns are broken down into pattern styles.

In summary, Otto segmented his historical research from his archaeological

analysis and limited the dating of his assemblages to what could be derived from Stanley
South's mean ceramic date formula, which for the 19th century does not work very well.
Despite the evidence of the predominance of machine cut and headed nails from the
kitchen which suggests a building date sometime after the early 1820s, Otto concludes
that the ceramics associated with the kitchen's midden represent the whole period of the
site's occupation by the Couper family, i.e., from 1794 to 1860. From the minimal
information presented in the book, I would estimate that the assemblage represents a
collection generated from the mid-1820s to the early 1840s. One of the variables that
Otto claims to have under control is the element of time. Lumping is not a way to control
the element of time as it masks changes that took place, which is one of the major
subjects of archaeological inquiry.

Even if Otto's assemblage was representative of the whole period of the

occupation, how meaningful would it be? Let's assume that the ceramic assemblages
from slave cabin and the overseer's house were also representative of the same 66 year
period of occupation. What would the samples tell us? The period from the 1790s to the
Civil War saw a major decline in English ceramic prices (Miller 1991:1-4). Ceramics
have a flexible demand curve, which means as they got cheaper people adjusted their
consumption patterns accordingly. This is seen in two ways: one was that more ceramics
were consumed, and the second was that more decorated ceramics were used as they
became cheaper. The 1790s was a period still dominated by plain creamware with some
shell edged tableware and painted teas.

After the War of 1812, plain creamware began to be replaced by decorated wares,
including shell edge, dipt, painted, and printed wares. In addition to decoration becoming
more common, the variety of forms and quantity of ceramics being purchased and used in
households increased. When one lumps ceramics from such a long period as the first half
of the 19th century, it would be very difficult to come to meaningful conclusions about
the differences in consumption patterns. The one place where Otto's argument holds up
is in the area of vessel forms, i.e., bowls versus flatware. These differences can be seen
through time. In short, lumping obscures the process of change that we are trying to
observe as archaeologists.

References

Miller, George
1991 A Revised Set of CC Index Values for Classification and Economic
Scaling of English Ceramics from 1787 to 1880. Historical
Archaeology 25(1):1-25.
Moodey, Meredith C.

5
 1988 Ceramics from the Franklin Glassworks: Acquisition Patterns and
Economic Stress. Unpublished M.A. thesis, Department of
Anthropology, College of William and Mary, Williamsburg, Virginia.
Otto, John S.
1977 Artifacts and Status Differences: A Comparison of Ceramics from
Planter, Overseer, and Slave Sites on an Antebellum Plantation. In
Research Strategies in Historical Archaeology, edited by Stanley South,
pp. 91-118. Academic Press, New York.
1984 Cannon's Point Plantation, 1794-1860: Living Conditions and Status
Patterns in the Old South. Academic Press, New York.

***************************************************

Thoughts Towards a User's Guide to Ceramic Assemblages, Part II:

What Does This Assemblage Represent?
George L. Miller URS Corporation

Council for Northeast Historical Archaeology Newsletter, No. 20, November 1991 pages 4- 6

In Part I of this series on evaluating archaeological collections, the problem of

lumping site collections into mega-assemblages was discussed in light of what they may
or may not represent. The issue of what an assemblage represents should be one of the
starting points in the analysis process: i.e., what am I looking at and what does it mean?

If someone presented you with a list of ceramics which they said are a sample
pulled from a 19th-century account book, that would immediately generate a series of
questions. Among those questions would be how was the sample selected, what
percentage does it represent of the whole account book, and what time periods are
represented? Beyond that, is the sample representative of what is in the account book?
However, when presented an assemblage of vessels from a site, rarely does anyone ask
these very basic questions. Through a leap of faith, most archaeologists assume that an
excavated sample, if it is "large" enough, is representative of the archaeological
population. Archaeologists have varying definitions for what constitutes a "large"
sample, which can range from 100 sherds to a minimal vessel count of at least 50 vessels.

As was demonstrated in the discussion of Otto's Cannon's Point site, excavated

assemblages may or may not be representative of the whole period of a site's occupation.
This is particularly true with sites which have long occupation periods (Miller 1991).
How does one begin to go about assessing an assemblage, and what are the questions to
be addressed? One way to think about this question is to consider the information one
would like to have available in an ideal world. Clearly, it would be great to know what
percentage of the archaeological population had been recovered. Second, is the
collection representative of the whole occupation period of the site? Third, what were the
major changes during the site's occupation that impacted what was used, broken, lost, and

6
discarded there? This list could be expanded considerably; however, these questions will
be enough to occupy our time for the present. While the above questions cannot be
answered, consideration of them will bring to mind some of the types of data that can be
assembled in a consistent manner that would be useful for comparison and examination
of assemblages.

The question of what percentage of the archaeological population has been

recovered is a difficult one to answer. A simple model for estimating the size of the
vessel population on isolated rural sites has been published in Historical Archaeology
(Miller and Moodey 1986). This model is based on one developed by biologists for
estimating the number of fish in a pond through capture and recapture of a tagged
population. For urban sites with dumped-in fill and organized garbage removal,
establishing an estimate of the size of the vessel population becomes much more complex
and may be impossible to answer. Given the problems associated with urban sites, this
discussion will be limited to rural sites where there are limited amounts of intrusive
materials.

What types of information can help in gaining a handle on how representative an

excavated sample is from a site? Often site reports contain information that is ignored
when assemblages from sites are brought together for comparison. For example, a simple
estimate of what percentage of the site was excavated would be a starting point. While it
is true that different parts of a site have different intensities of use, archaeologists have a
tendency to excavate the more intensely used areas such as around structures and where
there are concentrations of artifacts. In other words, if a report provides an estimate that
roughly a fourth of the site was excavated, then there is a good chance that probably more
than 25 percent of the ceramics were recovered.

This rough estimate, of course, can be influenced by the way in which the site was
excavated. In plowed shallow sites such as 17th century post-in-the-ground structures of
the Chesapeake or log cabins without minimal foundations, most of the artifacts will be
found in the plowzone. If the archaeologists have bulldozed the plowzone to get down to
features and postholes, most of the collection will be lost from ever being recovered. In
sites where the whole plowzone has been bladed off into oblivion to expose all of the
features, the resulting sample will clearly be a fairly low percentage of the archaeological
population.

The house area of the Franklin Glass Works site in Portage County, Ohio, can
shed some light on what can be lost if the site has been bulldozed. A total of 1330 square
feet of domestic area of the site was hand excavated and screened. This area represented
between 15 and 25 percent of the house area of the site (Miller and Moodey 1986:61).
Six small trash pits were exposed below the plow zone. Sherds to a minimum of 141
vessels were recovered from the plowzone and trash pits. All of these vessels had sherds
from the plowzone, whereas only 28 percent of them had sherds from the features. In
other words, bulldozing would have blown away 72 percent of the vessels and greatly
limited what could be done with the recovered sample. Using the biologists' model for

7
estimating population size, it was possible to generate an estimate of between 144 and
152 vessels in the archaeological population for the house area. This suggests that the
excavated sample represents between 93 and 98 percent of the archaeological population
(Miller and Moodey 1986:62). While there can be some doubt as to how well the
formula borrowed from biology works on archaeological populations, it still provides a
starting point towards understanding what one is looking at and a better handle on
quantification of data. Quantification in archaeology and history seems to have different
meanings. In history they are generally dealing with known quantities, whereas in
archaeology we are more often dealing with ratios and samples from populations where
the size is not known.

Unfortunately, the information presented with minimal vessel counts rarely

includes data on how much of the site was excavated, whether or not the site was hand
excavated and screened or bulldozed down to the features. All of these factors clearly
affect what was recovered, and how well that sample represents the archaeological
population of the site. When extracting data on minimal vessel counts, one should make
an effort to gather the information on how the site was excavated, and what percentage of
the occupation area was excavated. Beyond these simple considerations, one can begin
to look at the minimal vessel counts in relationship to the number of years that a site was
occupied as a rough gauge of the significance of the sample. For example, the Franklin
Glass Works was occupied from 1824 to 1832, after which the site reverted to
agricultural land. The data from this site can be summarized as shown in Table 1.

This simple summary provides the reader some usable information about the
quality of the data from the site and can be used to gauge this site against others in terms
of how completed they appear to be. Let's now look at Table 2, which contains similar
data recently published in Anne Yentsch's excellent article on "Minimal Vessel Lists as
Evidence of Change in Folk and Courtly Traditions of Food Use" (1990).

Arranging the information in this format begins to suggest that some assemblages,
i.e., those with a higher ratio of vessels to years of occupation, are more likely to be
representative of what was discarded, abandoned, or lost on a given site. Again, one
would like to have an estimate of what percentage of the site was excavated or, failing
that, the types of deposits or at least the amount of square footage excavated. Some of
the above sites, such as Pettus and Utopia, were partially destroyed by bulldozing away
their plowzone layers to expose the features while others, such as the Van Sweringen
sites, had all levels excavated and screened. Clearly the proportion of the population
recovered in the latter sites would be much greater than that of the two bulldozed sites.

It is well known that the level of ceramic usage increased as ceramics became
cheaper and replaced treen and pewter wares (Martin 1989). Therefore, one would
expect more ceramics from 19th century sites than from 17th century sites. It is not that
simple, however, because there will be a difference in ceramic holdings related to wealth
of the sites' occupants. Pettus and Utopia would be a case in point. Both sites were from
Kingsmill Plantation, occupied for roughly the same period of time, and excavated by

8
Bill Kelso (1984). However, Pettus was a house of a fairly well-off planter family, while
Utopia appears to have been a tenant house which was only a fifth the size of Pettus. The
difference in the quantities of ceramics could be reflecting either the socioeconomic
differences or the differential proportion of the archaeological populations recovered
from each site.

If it was known what percentage of the archaeological population had been

recovered from each site, then one would know if the differences in ratio of vessels to
years of occupation reflect different quantities in each household or a difference in the
recovery of the artifacts. Fortunately, Henry Miller has analyzed the food bone from
these sites to provide an estimate of the amount of meat they represent (Miller 1979). His
meat estimates suggest that the sites may be very comparable in terms of the amount of
population that was recovered. Consider the comparisons shown in Table 3.

Given that the house at Utopia was about a fifth the size of the house at Pettus and
would have been occupied by a smaller household, that family probably consumed less
meat. If that is the case, then the sample from the tenant site, Utopia, probably represents
a greater proportion of the archaeological population than was recovered from Pettus,
which has a much larger vessel population. While food bone or, more accurately, the
meat they represent seems to work for this situation, it clearly is not a good solution
because bone preservation varies considerably from area to area.

While the meat estimates suggest that the Utopia sample probably is better than
the Pettus sample, there could still be a large time distortion when one is dealing with
sites occupied for 60 years. Consider the data from Fraser Neiman's excavations of Clifts
Plantation. The site was occupied from 1670 to ca. 1730, which is 60 years. From that
site there were a minimum of 321 vessels, which works out to 5.4 vessels per year of
occupation. However, Fraser was able to separate four distinct components of the site
which had occupation periods ranging from 10 to 20 years. If these units had been
lumped into a single mega-assemblage, the distribution would have been as in Table 4.

From the Table 4 data, it can be clearly seen that the Clifts' site assemblages are
skewed towards the last 10 years of occupation, which accounts for 57 percent of the
vessels recovered. Given the level of skewing that can take place, one would be leery of
the data from Pettus and Utopia. The Pettus site has some documentation related to a
change in generational occupation of the site. Perhaps with further work on the
collections, the vessels could be separated into generational components.

The objective of this discussion has been to suggest some simple questions that
can be asked of archaeological assemblages which will help researchers sort out
assemblages according to their potential for comparative research. Further, this
discussion is a call for data to be included on what an archaeological assemblage
represents in terms of an estimate of how much of the site was excavated, how it was
excavated, and other factors which can lead towards a better understanding of our data. I

9
would appreciate comments on these thoughts and suggestions for other ways in which
the archaeological data base can be improved.

Table 1 & 2 combined

Site Dates Years Vessels Vessels/Year

Franklin Glass Works House 1824-1832 8 141 17.6
The Maine 1618-1626 8 88 11.0
Pasbebay Tenement 1625-1650 25 18 0.7
Kingsmill Tenement 1625-1650 25 78 3.2
Pettus 1640-1700 60 335 5.6
Utopia 1640-1700 60 55 0.9
Clifts I 1670-1685 15 34 2.3
Van Sweringen I 1672-1700 18 64 3.6
Gov. Drummond II 1680-1710 30 102 3.4
Clifts II (tenant) 1685-1705 20 32 1.6
Van Sweringen II 1700-1720 20 58 2.9
Clifts III 1705-1720 15 70 4.7
Clifts IV 1720-1730 10 185 18.5
Van Sweringen III (tenant arm) 1720-1745 25 84 3.4
John Hicks 1721-1740 19 263 13.8
Gov. Calvert 1728-1735 7 148 21.1
Wellfleet (C-9) 1690-1740 50 236 4.8
John Howland 1710-1730 20 136 6.8

Table 3

Site Vessels Vessels/Year Bone Animals Meat

Pettus 335 5.6 707 50 7,121
Utopia 55 0.9 994 71 7,973

Table 4

Site Dates Years Vessels Vessels / Years

Clifts I-IV 1670-1730 60 321 5.4
Clifts I 1670-1685 15 34 2.3
Clifts II (tenant) 1685-1705 20 32 1.6
Clifts III 1705-1720 15 70 4.7
Clifts IV 1720-1730 10 185 18.5

References
Kelso, William
1984 Kingsmill Plantation, 1619-1800: Archaeology of Country Life in
Colonial Virginia. Academic Press, New York.

10
Martin, Ann Smart
1989 The Role of Pewter as Missing Artifact: Consumer Attitudes Toward
Tablewares in Late 18th Century Virginia. Historical Archaeology
23(2):1-27.
Miller, George L.
1991 Thoughts Towards a User's Guide to Ceramic Assemblages. Council for
Northeast Historical Archaeology Newsletter 18:2-5.
Miller, George L., and Meredith Moodey
1986 Of Fish and Sherds: A Model for Estimating Vessel Populations from
Minimal Vessel Counts. Historical Archaeology 20(2):59-85.
Miller, Henry
1979 Pettus and Utopia: A Comparison of the Faunal Remains from Two Late
Seventeenth Century Virginia Households. The Conference on Historic
Site Archaeology Papers 1978 13:158-175.
Yentsch, Anne
1990 Minimal Vessel Lists as Evidence of Change in Folk and Courtly
Traditions of Food Use. Historical Archaeology 24(3):26-53.

*************************

Thoughts Towards a User’s Guide to Ceramic Assemblages, Part III:

Breaking Archaeological Assemblages into Functional Groups.

George L. Miller URS Corporation

Council for Northeast Historical Archaeology Newsletter, No. 22, July 1992 pages 2-4

Response to Part II

Part II of this series on “Thoughts Towards a User’s Guide to Ceramic

Assemblages” drew a long letter of response and just criticism from Bill Kelso. I had
hoped that Bill would send his letter stating his views on the subject to the [Council for
Northeast Historical Archaeology] Newsletter to add to the discussion of ways in which
we present data and excavate sites. I shall take the liberty of presenting this summary of
some of Bill’s comments.

Kelso’s comments on the way in which I presented my case for the importance of
the plow zone were right on target. Clearly, in the real world, the excavation of a site
represents a contest between limited funding and available time versus the compromises
made in getting the most information from a site given those constraints. My bemoaning
what is lost when the plow zone is “bladed away” is not a solution to the problem of the
information loss and the compromises that have to be made, particularly in environmental
archaeology. Bill also objected to my use of the terms “bulldozed” and “blown away.”
These terms are value laden and were not necessary to the discussion of the need to have

11
a description of what an excavated sample represents. I thank Bill for pointing out my
use of these value-laden terms.

Another point that Kelso raised was that plow zone from the Pettis site was
screened, but off site. The site was to be graded down 20 feet to make a marina and time
was limited, so the plow zone, which was “loaded” with artifacts, was trucked back to the
lab yard (16 truckloads), where it was screened over the next four years. Therefore, the
recovered sample from that site was a very high proportion of the archaeological
population.

My use of Pettis and Utopia collections for a comparison of two assemblages was
not aimed at making a comment on the quality of the archaeology. I merely wanted to
suggest how other artifact categories such as food bone can sometimes shed some light
on the ceramic samples in terms of how well they represent the population from which
they were recovered. I hope that Bill Kelso and others will respond in letters to the
CNEHA [and this] Newsletter because we need open discussions of the issues involving
archaeological research. Before leaving this topic, I would like to restate that the main
thrust of Part II of the “Thoughts Towards a User’s Guide to Ceramic Assemblages” was
to encourage archaeologists and others using their data to provide much more information
on just what their excavated samples represent and to provide their readers with some
suggestive estimate of what proportion of the archaeological population was recovered.

Before going on with the discussion of use of ceramic assemblages, I would like
to comment on Edward F. Heite’s letter in the April issue of the [CNEHA] Newsletter.
Mr. Heite’s letter was in response to Julia King’s review of the COVA Symposium on
Historical Archaeology in which she stated that “Chesapeake historical archaeologists are
adopting more mainstream methodologies, including the standard use of screens and the
recovery of data from plow zone contexts.”

Mr. Heite took a strong stand against screening which he supported with a quote
from Ivor Noel Hume published in 1969 on the loss of artifacts and information by
relying on the screen. It should be pointed out that Colonial Williamsburg’s Department
of Archaeological Research has made the screen a standard tool of their excavations since
Marley Brown came to Williamsburg in 1982. I have had eight years of looking at
assemblages in the Colonial Williamsburg collections from both screened and unscreened
contexts, and it is my impression that the screened contexts have higher counts of small
finds such as straight pins, small bone fragments, and mini-sherds. In the field, I have
seen the finely troweled soil taken to the screen, which resulted in the recovery of
additional artifacts. I would add that this impression has not been put to any tests
because few of us working in Williamsburg questioned the value of screening. Given
that Julia King’s above statement also mentions the recovery of data from the plow zone,
one wonders if Mr. Heite trowels the plow zone of the sites he excavates. The screen is a
tool, just as is the gradall, and all have their place in the excavation of sites. We may
have different opinions on when and how to use them, but to suggest that screens are not
an appropriate tool for historical archaeology is going a bit far.

12
Breaking Archaeological Assemblages into Functional Groups

The analysis of ceramic assemblages has come a long way in the last couple of
decades. Early site reports often included doll parts, marbles, and floor tiles in their
listing of ceramic counts, which were duly calculated into the percentage tables of
recovered ceramics. We are still looking for meaningful functional breakdowns for the
analysis of archaeological assemblages. Anne Yentsch, in an article (Historical
Archaeology 1990) titled “Minimal Vessel Lists as Evidence of Change in Folk and
Courtly Traditions of Food Use,” provided a useful breakdown for food-related vessels
into the following classes: 1) food preparation and storage, 2) kitchen, and food
consumption, 3) food distribution, 4) beverage distribution, and 5) beverage
consumption. Organizing assemblages along these lines will begin to bring order to the
data that will greatly facilitate comparison among collections. Yentsch’s breakdown
covers most of the food-related types of vessels typically found in archaeological
assemblages and appears to work very well for assemblages from the 17th and 18th
centuries.

I have been using a simpler functional breakdown for ceramic assemblages in an

attempt to establish the “market basket” of ceramics typically available in country stores
for the period 1780 to 1880. The market basket research grew out of a project titled
“English Ceramics in America, 1760 to 1860: Marketing, Prices and Availability” that
was funded by the National Endowment for the Humanities (RO-21158-86) [published in
1994 as “Changing Consumption Patterns: English Ceramics and the American Market
from 1770 to 1840" by George L. Miller, Ann Smart Martin, and Nancy S. Dickinson, pp.
219-248, in Everyday Life in the Early Republic, edited by Catherine E. Hutchins, Henry
Francis du Pont Winterthur Museum, Winterthur, Delaware]. Additional funding for this
research was provided by two fellowships to Winterthur Museum, which has an
incredible collection of invoices, account books, and price lists. The “market basket” is
being generated from invoices for ceramics imported to the American market and
invoices for ceramics purchased by country merchants. Economists use the “market
basket” concept as a tool in establishing the consumer price index. The “market basket”
represents the best estimate of what the American consumers were purchasing at a given
point in time. For the ceramic “market basket,” I have used invoices to establish the
makeup of wares. These wares have been broken down into four basic functional groups,
which are 1) tableware, 2) tea ware, 3) kitchen ware, and 4) toilet ware. Most of the
available invoices were limited to refined English or American wares and did not include
the coarse wares such as would have been used in storage and dairying. The
classification system being used is the one used by the Staffordshire potters in their price-
fixing lists, and the categories hold together very well when compared over a long period
of time. There are some choices which some would question. For example, the potters
had a grouping titled “Mugs and Jugs ware,” which included bowls. In this classification,
jugs are pitchers. The English still call pitchers “jugs.” These wares all came in dipt
decorations such as mocha, common cable, annular bands, and other types of decoration
which are rarely found on table or tea wares. Therefore, bowls in this classification,

13
unless they are clearly identified as tea wares such as slop bowls, have been classified as
kitchen ware. Bakers and nappies, on the other hand, have been classified as tableware
because they have been classified as such by the potters. Clearly, some bakers would
have been used to prepare food in the kitchen, just as some bowls would have been used
to serve food at the table. No classification system is perfect, and one is reminded of
recorded incidents where chamber pots have been used as soup tureens. Table 1 presents
the percentage of these functional types for invoices dating from 1783 to 1858.

Table 1
Percentages of Vessels by Major Functional Groups

Year Vessels Tea Table Kitchen Toilet

1783 5,058 68.0 22.1 8.5 1.4
1806 39,018 52.8 27.0 18.6 1.6
1824 1,475 60.2 30.0 9.8 --
1825 893 57.4 35.5 5.7 1.3
1826 2,244 59.1 25.7 13.5 1.7
1827 2,715 53.9 34.0 10.7 1.3
1828 1,917 55.8 26.7 16.9 0.6
1829 1,371 58.6 31.1 9.4 0.9
1830 564 62.2 34.0 1.6 2.1
1856 2,463 63.5 26.7 6.0 3.8
1857 4,344 54.8 33.5 7.7 4.0
1858 6,725 53.1 38.3 6.1 2.4

From the distributions shown in Table 1, it can be seen that there was a high level
of consistency among these groups from 1783 to 1858. Over half of the vessels in all of
the above assemblages was tea ware. Tableware seems to gradually increase through this
period, which probably is a reflection of the declining use of pewter. The surprising
category is toilet ware, which seems rather low. This may be because a large number of
chamber pots would have been made in red wares and salt-glazed stonewares. Toilet
wares were on the increase by the 1850s, which seems to be related to the introduction of
white granite wares.

Most Common Types of Vessels

Some archaeologists feel intimidated in their identification of vessels because of

the great variety that were produced. For example, the 1789 Whitehead pattern book lists
57 different types of vessels, while the 1814 Leeds pattern book lists 86 different vessel
forms. The number of available forms was expanding as ceramic prices dropped during
the first half of the 19th century. For example, the 1796 Staffordshire potters’ price-
fixing agreement only lists 35 vessel types. The potters had expanded the types of
vessels available to 93 types by the 1846 price-fixing list. This seems like a lot of forms

14
to attempt to identify; however, most of these vessel types are rarely found listed in
invoices or in archaeological assemblages.

Fortunately, over 90% of the vessels in our assemblages can be accounted for by
just 16 types of vessels. Table 2 lists these forms and the percentages of the vessels that
they account for in five assemblages from 1783 to 1858. These are simple forms that
most archaeologists can identify from small sherds. If you are willing to break down
your assemblages into these basic simple vessels, then you will have greater interplay
between documentary and archaeological assemblages.

Table 2
Forms and Percentages of Vessels for Six Assemblages, 1783-1858

1783 1806 1826 1828 1856 1858

TEA WARE Cups & Saucers 42.7 41.7 52.9 52.6 57.5 48.2
Coffees 17.1 -- -- -- 2.9 0.7
Bowls & Saucers -- 6.2 -- -- -- --
Teapots 4.3 3.2 3.5 0.8 3.2 0.6
TABLE WARE Platters 1.4 0.8 2.7 1.0 1.0 2.4
Plates & Twifflers 14.2 16.9 13.9 7.5 14.1 14.6
Muffins -- 7.4 7.1 16.3 10.2 17.8
Bakers & Nappies -- 0.5 1.6 0.6 1.0 1.8
KITCHEN WARE Bowls 5.7 14.2 11.1 13.8 3.9 3.4
Mugs 2.8 1.8 -- 1.3 0.7 0.4
Pitchers -- 2.0 2.3 1.9 1.3 2.2
TOILET WARE Chambers 1.4 1.0 1.1 0.6 0.6 0.7
Ewers & Basins -- 0.7 0.6 -- 3.1 1.3
TOTALS 89.6 96.4 96.8 96.4 99.5 94.1
Other types of vessels 10.4 3.65 3.2 3.6 0.5 5.9

The information presented here is from the project on establishing the “market
basket” of ceramics available in country stores from the 1780s on to 1900. If you have
any invoices for ceramics that you would care to share, I would appreciate having a xerox
copy. I will be more than willing to pay the cost of reproduction. Once again, responses
to this column are encouraged and welcomed.

*****************************************

Thoughts Towards a User’s Guide to Ceramic Assemblages, Part IV:

Some Thoughts on Classification of White Earthenwares
George L. Miller URS Corporation

15
 Council for Northeast Historical Archaeology Newsletter, No. 26, November 1993 pages 4-7

Classification of English white-bodied refined wares has presented a problem for

archaeologists. Most classification systems for these wares have been based on visible
traits, such as cream color or blue tint to the glaze, or vitrification of the body. Classic
definitions for cream, pearl, white, and white granite wares are static and oversimplified.
All of these wares evolved during their periods of production, and there was a great deal
of variety within the types due to the fact that there were well over a hundred potters
producing them. William Evans (1970) published a collection of formulas for bodies and
glazes from a number of potters in 1846 which illustrates the great diversity of formulas
used by the various potters for these wares and their glazes. When attempting to define
the refined white earthenwares by one or two simple traits, one limits our understanding
of the evolution of those wares. In the case of the distinction between whiteware and
white granite, the resulting identification will be inconsistent with the way the potters
classified and priced their products. Unless the changes in these wares are taken into
consideration, the resulting classification will be inconsistent with the classification
system used by the potters who made them and the merchants who sold them. In short, if
our classification does not match that of the potters and merchants selling the wares, it
will not be possible to scale them for the study of purchase patterns.

Static definitions of cream, pearl, white, and white granite wares are useful in
establishing chronological control because the types can be associated with limited time
periods. Unfortunately, the wares themselves were not static. They evolved as tastes
changed, and as a result of competition between the potters to produce cheaper wares.
For example, creamware became lighter through time. These changes mean the
creamware of the 1760s is different from the CC ware of the 1780s and different still
from the CC ware of the 1820s and 1860s. White granite, which evolved from the stone
chinas and ironstone in the early 1840s, was almost always vitrified in the early period of
production. By the 1870s, however, much of it was fired below the temperature
necessary to produce a vitrified ware. If one defines white granite on the basis of
vitrification, then much of what the potters and merchants would have called white
granite in the post-1870 period would be classified as a white ware. This becomes a
problem when one is trying to establish the expenditure patterns represented by an
archaeological assemblage. For the study of the wares in terms of their cost, it is
essential that the classification of the vessels be as consistent as possible with that used
by the potters and merchants dealing in the wares. This paper is an attempt to provide
some insights on the classification of these wares as they evolved at different points in
time.

Thoughts on the Relationship Among the White Earthenwares, Chinese Porcelain,

and Bone China

Creamware represented a major change in direction for the Staffordshire pottery

industry away from white salt-glazed stoneware and the quest for the secret of porcelain.
When Wedgwood perfected his version of creamware, he and his partner Thomas

16
Bentley went on to market it by selling sets to Queen Charlotte of England and Catherine
the Great of Russia. Wedgwood and Bentley were able to promote creamware to the
point where it could compete with porcelain in status. This was a major breakthrough for
the Staffordshire ceramic industry, and the production of creamware expanded
tremendously in the 1760s and 1770s. Popularity, however, does not last forever. As the
public grew tired of creamware, the other potters began to experiment to find a product to
catch the consumers’ attention. The setback that the English porcelain industry had
suffered due to the demand created for creamware began to ease, and the porcelain
industry again began to advance.

In 1768 William Cookworthy took out a patent for producing a Chinese-style

hard-paste porcelain using kaolin and china stone from Cornwall, England. Richard
Champion was able to renew Cookworthy’s patent in 1774; however, the use of the
kaolin and china stone was available to others so long as they did not produce porcelain
(Hughes 1960:110). Having the materials to produce porcelain, but not being able to
produce it because of Champion’s patent, some of the Staffordshire potters began to
produce a ware that they called “China glaze” by at least 1775. This is the origin of
pearlware (Miller 1987). China glaze can be distinguished from pearlware and provide
archaeologists with a meaningful chronological indicator. As a ware type, it would have
the following characteristics: 1) A blue-tinted glaze that gave the whole vessel a bluish
tint in imitation of Chinese porcelain; 2) Blue painted and printed patterns in a Chinese-
style pattern; and 3) Some of the vessel forms would be in a Chinese style, such as the
handleless Chinese tea bowl shape for cups and undercut footrings on plates.

As a group, these traits cluster between ca. 1775 and ca. 1812. They seem to be
pretty much out of style by the end of the War of 1812. The important thing to keep in
mind here is that the bluing was added to copy Chinese porcelain in a product that was
called China glaze.

Josiah Wedgwood was under pressure from his partner Thomas Bentley to
produce something along the lines of China glaze in the late 1770s (Miller 1987). He
referred to his new product as “Pearl White” and referred to it as a whiteware.
Wedgwood used the cobalt to make his ware white in appearance, not to make it look like
Chinese porcelain. His term never really caught on, and later scholars changed it to
pearlware. If we consider the floral painted patterns on vessels in which the bluing can
only be seen in the footring as pearlware, then we have a ware that would date from ca.
1780 to ca. 1830. Pearlware starts showing up in underglaze colors such as mustard
yellow, olive green, brown, and blue around 1795, according to Ivor Noel Hume. These
polychrome painted wares are rarely in Chinese-style patterns (Reimer 1991).

For the most part, these wares will show up on American sites following the
Revolutionary War. The exception would be in areas that were occupied by the British.
Pearl white went through its own evolution at the Wedgwood factory, where there were
at least six formulas for the pearlware body from 1815 to 1846 (Delhome 1977).
Pearlware production at the Wedgwood plant continued up into the 20th century. The

17
later pearlware, however, does not have any blue tint so would be classified as a
whiteware by historical archaeologists. Llewellynn Jewitt described Wedgwood’s
pearlware of 1865 as being “not a pearl of great price, but one for ordinary use and of
moderate cost” (Godden 1969:396). Some post-1860 Wedgwood pieces are impressed
with the “Pearl” as part of the maker’s mark. Those that I have seen would be classified
as whiteware by archaeologists, which points out the problems in our classification
system. Archaeologists have a definition for ware types that is related to a period of time,
which is fine. However, there is another classification used by the potters and merchants,
which evolves. That needs to be kept in mind when working with prices and consumer
behavior.

Whiteware has been a major problem for historical archaeologists. Its origins are
poorly understood, and until recently very little documentation had been published on its
development. We have been using 1820 as the introduction date for whiteware, which is
the date that Ivor Noel Hume (1970:130) estimates that pearlware was being replaced by
whiteware. In a recent article John des Fontaines (1990:4) documented the production of
whiteware by the Wedgwood factory by early 1805. Like China glaze, whiteware
appears to have been developed as a copy of porcelain. Josiah Spode’s success with his
bone china appears to have been the impetus for change. Spode’s bone china fires very
white, and its popularity led the earthenware potters to move towards a whiter-looking
ware (des Fontaines 1990:7; Miller 1980:17).

There were different ways of producing a whiteware. Given that there were over
100 potters in Staffordshire, it is not surprising that there were different approaches taken
to the problem. The simplest solution was to cut back on the amount of the cobalt used
so that it just countered any yellow tint in the glaze, but did not create a blue tint to the
ware. We have all seen wares that are white except for a very light blue tint in the glaze
gathered around the footring. Many people have classified these as pearlware. We need
to keep the intent of the potter in mind. If it was to produce a whiteware, then the vessel
should not be classified as pearlware because of a small amount of cobalt used to achieve
a white appearance. White wares with a small amount of blue in the footring area
probably show up on American sites as early as the end of the War of 1812 and seem to
last into the 1840s.

White wares without any indication of the presence of cobalt were also probably
showing up on American sites after the War of 1812, and they are still in production.
Definition of these wares becomes very complicated because pearlware and CC wares
appear to merge together. For example, Wedgwood’s wares of the 1840 to 1860 period
bearing the impressed mark “PEARL” do not have any indication of the use of cobalt
(des Fontaines 1990:6).

Use of the terms China glaze, pearlware, and whiteware are very rare in
Staffordshire potters’ price-fixing lists and invoices from 1780 through the 19th century.
The only undecorated wares that I have seen listed in invoices for wares sent to the
American market prior to the early 1840s are CC ware. Clearly, all of the China glaze,

18
pearlware, and whitewares for this period were decorated and would have been classified
by their type of decoration, e.g., edged, dipt, painted, or printed. In invoices for the
period between 1824 and 1858, CC ware ranged from between 5% and 13% of wares
sold to country stores (Miller 1990). CC ware is the potters’ shorthand for cream color or
creamware. Again, our definition of what is creamware falls short of what creamware
became following the 1820s. It is very white and clearly is being classified as a
whiteware by most historical archaeologists. Given that undecorated China glaze,
pearlware, and whiteware are not in the potters’ price lists or invoices, the plain
undecorated vessels that we recover from contexts dating before the early 1840s are most
certainly CC ware. Here again, we have two typologies. These vessels are whiteware by
our chronological typology, but CC ware in the potters’ terms, and that is how one needs
to classify them to work with economic scaling of assemblages.

Early in the 1840s another plain undecorated ware begins to be imported in

quantity to the American market. That ware is what most of the potters called white
granite ware. It has been called ironstone by most archaeologists. Ironstone is the name
that Charles Mason gave to his stone china in his 1813 patent. In a sense, ironstone is a
brand name that became generic. The use of the term ironstone and its dates of
production have led to some confusion in the dating of late 19th century assemblages. In
his article on mean ceramic dating Stanley South lists “Ironstone and Granite China” and
gives the dates 1813 to 1900 with a mean date of 1857. Mason’s ironstone was but one
of several stone chinas that began production around 1800. These wares, like China
glaze, were most commonly copies of Chinese porcelain and also had blue-tinted glazes.
The stone chinas were rarely undecorated, and they are rather rare on American sites
(Miller 1991:9-10). The shift comes in the early 1840s when the potters began producing
what they called white granite.

White granite, like the other ware types, went through its own evolution. In the
beginning period, the term pearl came back into use in names such as pearl stone china,
pearl white ironstone, and pearl white granite. Sometimes the blue is a tint in the glaze,
and sometimes it is a tint added to the body (Miller 1980:18-19, 1991:9-10). The wares
from the 1840s through the 1860s are generally vitrified. Molded marlys such as the
Ceres pattern are common as are geometric shapes with eight, ten, and twelve sides.
These give way to plain round shapes without molding. By the late 1870s it is not
uncommon to find white granite wares that are not vitrified. Its price had been dropping,
and the potters were cutting their production costs. When we define white granite wares
as vitrified, we again are locking into a definition that might be helpful for chronological
purposes, but it presents problems in scaling collections for the study of consumption
patterns.

These wares began to change after the Civil War. During the war, the American
tariff on imported ceramics was raised to over 50%. A large greenback currency was
issued to finance the war, resulting in an inflated currency. Under these conditions, the
cost of English ceramics almost doubled, which encouraged a number of English potters
to move to Trenton, New Jersey, and begin production of American-made white granite.

19
It took these potters a period of adjustment to the new clays and other materials they had
to work with, so many of the early American white granite wares were heavily crazed.
Thus, white granite with heavy crazing probably is American made and dates from ca.
1865 to ca. 1890. This problem was worked on, and the wares improved as the American
industry mastered their raw materials.

As white granite began losing its appeal, the potters cut its price and found ways
to make it cheaper. One way was to fire it at lower temperature, and thus, the later white
granite ware is often not vitrified. The later wares are usually unmolded, whereas the
earlier ones commonly had embossed molding on the marly. In summary, the following
traits would be helpful to keep in mind when separating white wares from white granite
wares.

1) For pre-1845 assemblages, there will not be any white granite ware.

2) After ca. 1820, plain undecorated vessels are CC ware.

3) After the War of 1812, CC ware was mostly confined to toilet and kitchen wares
such as bowls and chamber pots. Some plates are still being sold in CC ware, but
teaware is very rare.

4) Importation of white granite ware began in the early 1840s. In post-1840s

contexts the undecorated teaware is most likely white granite. The same is true to
a lesser extent for tableware.

5) If it is vitrified, it most likely is white granite.

6) If it has embossed molding around the marly, it most likely is white granite.

7) If the body, rather than the glaze, has been tinted with cobalt to make it look light
blue or gray, it is most likely white granite. Some white granite also has blue-
tinted glaze. The term pearl came back into use in the post-1840 period in names
such as “Pearl China,” “Pearl Stone Ware,” “Pearl White Ironstone,” “Pearl
White,” and “Opaque Pearl.” Despite incorporating the name pearl, these marks
occur on white granite wares.

There will be body sherds where one cannot distinguish white granite from
whiteware. For those cases, one could use a category of whiteware/white granite. One
should keep the use of this category to a minimum.

It should be kept in mind that while undecorated or molded white granite wares
were the dominant type for the period ca. 1850 to ca. 1890, these wares also came with
standard types of decoration such as shell edging, painted, and printed patterns. When
you find white granite wares with color decoration, the decoration will be more important

20
for classification than the ware type. Therefore, one does not need to be quite as
concerned with identification of the type of ware as in the case for undecorated vessels.

Acknowledgments

My research on the chronology and identification of white-firing earthenwares has

been funded by an NEH Fellowship to Winterthur Museum in 1991 and by an NEH grant
(RK-20004) titled “English and American Ceramics, 1846-1917: Prices, Index Values,
and Chronology.” This paper is meant to generate a dialogue on the problems in
separating and dating of white-firing earthenwares and is a call for information on the
subject. A more formal paper on the issues raised here will be one of the products of my
NEH grant on English and American Ceramics, 1846-1917. I would like to thank Wade
Catts, Rebecca Tinsman, and Justine Withers for reading and commenting on this paper.

References
Delhome, M. Mellanay
1977 Pearlware. Wedgwood: Its Competitors and Imitators 1800-1830.
Wedgwood International Seminar 22:61-95.
Evans, William
1970 Art and History of the Potting Business, compiled from the most practical
sources for the especial use of working potters. Journal of
Ceramic History 3:21-43.
des Fontaines, John K.
1990 Wedgwood Whiteware. Proceedings of the Wedgwood Society (London)
13:1-8.
Godden, Geoffrey A.
1971 The Illustrated Guide to Mason’s Patent Ironstone China and Related
Wares: Stone China, New Stone, Granite China -- and Their
Manufacturers. Praeger Publishers, New York.
Hughes, G. Bernard
1960 English and Scottish Earthenware 1660-1860. Abbey Fine Arts, London.
Miller, George L.
1980 Classification and Economic Scaling of 19th Century Ceramics.
Historical Archaeology 14:1-40.
1987 Origins of Josiah Wedgwood’s Pearlware. Northeast Historical
Archaeology 16:80-92.
1990 The “Market Basket” of Ceramics Available in Country Stores from 1787
to 1880. Paper presented at the Society for Historical Archaeology
Annual Meeting, Tucson, Arizona.
1991 A Revised Set of CC Index Values for Classification and Economic
Scaling of English Ceramics from 1787 to 1880. Historical Archaeology
25(1):1-25.
Noël Hume, Ivor
1970 A Guide to Artifacts of Colonial America. Alfred A. Knopf, New York.
Reimer, Gretchen M.

21
1991 Painted Pearlware: An Analysis of Painted Pearlware in the Study
Collection of Colonial Williamsburg Foundation. Ms. on file at Colonial
Williamsburg.

22