4

Compositional Data Analysis

in Archaeology
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

Ronald L. Bishop and Hector Neff

Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

Conservation Analytical Laboratory, Smithsonian Institution,

Washington, DC 20560

As compositional analysis has become more routine in archaeological

investigations, deficiencies in the numerical techniques used for data
reduction and summary have become more apparent. A brief over-
view of techniques commonly used in the analysis of compositional
data is presented as well as an example illustrating how data modeling
(as opposed to data summary) can facilitate both the recognition of
relevant data structure and inferences from data structure to un-
derlying natural and cultural processes.

THE APPLICATION OF CHEMICAL ANALYTICAL TECHNIQUES to archaeolog-

ical questions has a l o n g history that extends back to the late 1700s (I). M a n y
of the earlier investigations dealt w i t h the compositional characterization of
objects to elucidate aspects of t h e i r properties, such as color. Yet, as H a r -
bottle (2) has n o t e d , b y the e n d of the 19th c e n t u r y , chemists l i k e D a m o u r
a n d H e l m v i e w e d the c h e m i c a l analysis of artifacts as a means of d o c u m e n t i n g
long-distance traffic i n p a r t i c u l a r materials. T h e basic approach of d e t e r m i n -
i n g a c h e m i c a l c o m p o s i t i o n for an object a n d t h e n c o m p a r i n g that profile to
others s i m i l a r l y d e r i v e d has b e e n elaborated since that t i m e . Today, the
c h e m i c a l characterization of artifacts constitutes a basic archaeological ap-
proach that can be u s e d to address not only p r o b l e m s p e r t a i n i n g to l o n g -
distance exchange b u t to intraregional p r o d u c t i o n a n d d i s t r i b u t i o n (3), d e -
v e l o p m e n t of craft specialization (4), and typological r e f i n e m e n t (5, 6), a m o n g
other issues.
D e s p i t e the v o l u m e of data generated a n d the variety of applications,
d e v e l o p m e n t a n d testing of data-handling techniques have lagged. T h e r e is

0065-2393/89/0220-0057$08.50/0
© 1989 A m e r i c a n C h e m i c a l S o c i e t y

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 58 ARCHAEOLOGICAL CHEMISTRY

not a " c o o k b o o k " approach to data analysis any more than there is some
i d e a l group or n u m b e r of e l e m e n t a l concentrations to d e t e r m i n e for a l l
applications. C o m p l e x natural a n d c u l t u r a l interactions can account for m u c h
of the o b s e r v e d c o m p o s i t i o n a l v a r i a t i o n , a n d one m u s t b e aware of these
interactions to achieve a greater u n d e r s t a n d i n g of the data. I n the discussion
to follow, w e w i l l b e c o n c e r n e d w i t h aspects of m u l t i v a r i a t e data analysis
that l e a d us t o w a r d the position that many of the questions b e i n g addressed
i n a c o m p o s i t i o n a l investigation r e q u i r e m o d e l i n g rather than m e r e l y s u m -
m a r i z i n g the data.

Background
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

T h e d e v e l o p m e n t of increasingly sophisticated analytical i n s t r u m e n t a t i o n

Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

that allows n u m e r o u s e l e m e n t a l concentrations to be d e t e r m i n e d i n a r e l -

atively short t i m e a n d increased t h r o u g h p u t of specimens has h a d a d e c i d e d
i m p a c t o n archaeology. It has e v e n b e e n c l a i m e d (7) that the availability of
analytical capability is partially responsible for concentrated archaeological
attention to m a t e r i a l exchange d u r i n g the 1970s. W i t h the interest a m o n g
archaeologists a n d the technological advances has c o m e a staggering a m o u n t
of analytical data. N u m e r i c a l s u m m a r i z a t i o n of these data, assisted b y the
increasing speed of the c o m p u t e r a n d availability of general-purpose statis-
tical packages, is a v i t a l l i n k b e t w e e n the generation of data a n d its i n t e r -
pretation w i t h i n the archaeological context.
D a t a analysis has not b e e n neglected b y archaeologists a n d "archaeo-
m e t r i c i a n s . " N u m e r o u s papers have d e s c r i b e d various techniques a p p l i e d
to specific sets of data. O t h e r s have d e s c r i b e d h o w particular options of
readily available c o m m e r c i a l programs are u s e d (8-11). A t times, r o u t i n e
n u m e r i c a l procedures a p p l i e d to w e l l - d e t e r m i n e d data have b e e n i n t e r -
p r e t e d i n a m a n n e r that fails to c o n t r i b u t e to increased archaeological u n -
derstanding. I n m a n y of these efforts there is an inappropriate use of
statistics, failure to u n d e r s t a n d the c o m p o n e n t nature of the m a t e r i a l b e i n g
analyzed, or a failure to b r i d g e f r o m the analytical data to the archaeological
context. I n m o r e general terms, the speed of data p r o d u c t i o n and c o m p u -
tation has outpaced the logic of the investigation.
Because w e recognize that the m e r g e r of archaeological investigation
w i t h p h y s i c o c h e m i c a l analysis is still e v o l v i n g , w e w i l l t r y to a v o i d reference
to specific applications w h e r e w e b e l i e v e basic mistakes w e r e made. Instead,
w e w i l l discuss p r o b l e m s a r i s i n g i n the compositional analysis of archaeo-
logical materials i n an abstract or generic m a n n e r . T h i s approach m a y s u b d u e
the i n c l i n a t i o n some investigators may feel to engage i n v i t r i o l i c r e b u t t a l
l i k e that w h i c h f o l l o w e d Thomas's (12) general c r i t i q u e of statistical practice
i n archaeology (13).
A l t h o u g h m a n y of the c o m m e n t s i n this chapter are applicable to sit-
uations e n c o u n t e r e d d u r i n g the analysis of data from diverse types of ar-

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 59

chaeological materials, w e w i l l illustrate o u r points i n a later section w i t h

examples d r a w n f r o m c e r a m i c compositional systems. O u r examples w i l l
incorporate v e r y w e l l - u n d e r s t o o d a n d artificial (or " d u m m y " ) data, as a p -
p r o p r i a t e i n a discussion of methodology.

Goals of Data Analysis

C o m p o s i t i o n a l analysis of archaeological materials entails a series of n o n d i s -
crete steps of research design:

• problem formulation,
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

• sample selection,
• analytical approach,
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

• data analysis, a n d
• data integration.

T h e nature of the specified p r o b l e m w i l l suggest w h i c h samples a n d

h o w m a n y w i l l b e c o n s i d e r e d , w h e t h e r r a w source materials w i l l b e i n c l u d e d
i n the investigation, the spatial a n d t e m p o r a l extent of s a m p l i n g , etc. C e r -
tainly, s a m p l i n g of an i n t e r r e g i o n a l investigation w i l l differ considerably f r o m
the m o r e d e m a n d i n g r e q u i r e m e n t s for an intraregional focus (14). O n c e the
p r o b l e m is f o r m u l a t e d a n d samples are specified, selection of an appropriate
analytical t e c h n i q u e i d e a l l y d e p e n d s u p o n the sensitivity a n d p r e c i s i o n r e -
q u i r e d to address the p r o b l e m at h a n d . O n a m o r e practical l e v e l , one cannot
dismiss considerations of i n s t r u m e n t a l availability a n d cost.
A data m a t r i x p r o d u c e d b y compositional analysis c o m m o n l y contains
10 or m o r e m e t r i c variables (elemental concentrations) d e t e r m i n e d for an
e v e n greater n u m b e r of observations. T h e b r i d g e b e t w e e n this m u l t i d i m e n -
sional data matrix a n d the d e s i r e d archaeological i n t e r p r e t a t i o n is m u l t i v a r -
iate analysis. T h e purposes of m u l t i v a r i a t e analysis are data exploration,
hypothesis generation, hypothesis testing, a n d data r e d u c t i o n . A p p l i c a t i o n
of multivariate techniques to data for these purposes entails an a s s u m p t i o n
that some f o r m of structure exists w i t h i n the data matrix. T h e n o t i o n of
structure is therefore f u n d a m e n t a l to compositional investigations.
S t r u c t u r e w i t h i n a compositional data set is the differential o c c u r r e n c e
of data points i n the n-space d e f i n e d b y e l e m e n t a l concentrations. O n e s i m p l e
k i n d of structure consists of points g r o u p e d a r o u n d two centroids, or centers
of mass, i n the e l e m e n t a l concentration space. S t r u c t u r e w i t h i n a c o m p o -
sitional data set is assumed, i m p l i c i t l y or e x p l i c i t l y , to reflect the u n d e r l y i n g
process responsible for the data. T h u s , i n the case of the t w o - c e n t r o i d struc-
ture j u s t m e n t i o n e d , an u n d e r l y i n g process, such as p r o c u r e m e n t of clay
from two sources, is assumed.
Different operational levels m a y exist for the inference of process from
structure. F o r e x a m p l e , p r i n c i p a l - c o m p o n e n t s analysis (a m e t h o d d e s c r i b e d

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 60 ARCHAEOLOGICAL CHEMISTRY

later i n this chapter) p e r m i t s compression of m u l t i v a r i a t e data into a few

d i m e n s i o n s a n d y i e l d s a scatterplot of w h a t appear to be two groups ( F i g u r e
1). T h e two groups are r e a d i l y recognizable b y u s i n g several different k i n d s
of cluster analysis; the group separation is r e a d i l y c o n f i r m e d w i t h d i s c r i m -
inant analysis. I n fact, one group was f o r m e d f r o m the other b y m u l t i p l y i n g
all e l e m e n t a l concentrations b y 0.66. T h i s example approximates the effect
of a r e l a t i v e l y p u r e t e m p e r (e.g., quartz sand) o n the clay composition of
ceramics (16). P a r t i c u l a r l y i n c e r a m i c p r o d u c t i o n , an o b s e r v e d c o m p o s i t i o n a l
profile m i g h t relate not o n l y to the natural r e a l m (source rocks, w e a t h e r i n g ,
erosion, transportation, etc.), b u t also to the c u l t u r a l r e a l m (social a n d i n -
d i v i d u a l patterns o f materials p r o c u r e m e n t a n d preparation).
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

T h e search for structure proceeds a c c o r d i n g to some m a t h e m a t i c a l

m o d e l that can organize a n d represent the i n f o r m a t i o n i n a data matrix.
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

P a r t i c u l a r k i n d s of associations b e t w e e n data entities or variables may be

e x a m i n e d — b u t always relative to the p a r t i c u l a r m o d e l u s e d (17). T h e s e
models are at the same t i m e s t r u c t u r e - r e v e a l i n g a n d s t r u c t u r i n g . T h i s c o n -
cept is illustrated b y three natural groups s h o w n relative to concentrations
of F e a n d Sc i n the scatterplot i n F i g u r e 2. Because of i n t e r e l e m e n t a l cor-
relations, the groups f o r m elongated ellipses, yet are fully separable at the
9 5 % confidence i n t e r v a l . I f a h i e r a r c h i c a l cluster analysis based on E u c l i d e a n
distances calculated f r o m logged F e a n d Sc values w e r e c a r r i e d out, the

1 -
0.9 -
0.8 -
0.7 - CP
0.6 -
a
=5: 0.5 -
C
0.4 -
z 0.3 - •
Ixl
Z 0.2 - cP •
oa. 0.1 -
o 0 -
o
+ a
-0.1 -
<! -0.2 - •
Q. -0.3 - +
O
z -0.4 - +
££
CL -0.5 -
-0.6 -
-0.7 -
-0.8 -
-0.9 - +
-1 -
-0.6 -0.4 -0.2 0.2 0.4 0.6
PRINCIPAL COMPONENT #2
a Non-diluted + Diluted

Figure 1. Example showing how proportional dilution may create compositional

subgroups. "Diluted" specimens were created by multiplying 17 elemental
concentrations in the nondiluted specimens by 0.66.

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

2?
o

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
Figure 2. Scatterplot of Fe and Sc values for three distinct groups.
 62 ARCHAEOLOGICAL CHEMISTRY

r e s u l t i n g s p h e r i c a l clusters w o u l d confound the group m e m b e r s h i p . If, h o w -

ever, a h i e r a r c h i c a l cluster analysis w e r e c a r r i e d out b y u s i n g E u c l i d e a n
distances calculated from standardized p r i n c i p a l - c o m p o n e n t scores, the r e -
s u l t i n g clusters w o u l d c o r r e s p o n d to the groups e v i d e n t i n the scatterplot.
W i t h o u t any p r i o r k n o w l e d g e of the structure i n the data set i n F i g u r e
2, one m i g h t b e g i n a search for structure w i t h an analysis of the straight-
l i n e distances b e t w e e n points i n the data set. H i e r a r c h i c a l cluster analysis
of E u c l i d e a n distances i m p l e m e n t s a systematic approach to the analysis of
straight-line distances. (Euclidean distance is the n - d i m e n s i o n a l g e n e r a l i -
zation of straight-line distance, a n d is discussed i n greater d e t a i l later. Cluster
analysis is a m e t h o d of r e p r e s e n t i n g the E u c l i d e a n distances i n two d i m e n -
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

sions, a n d is also discussed later). H o w e v e r , the cluster analysis approach

i n this case fails to represent the k n o w n relationships a m o n g the k n o w n
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

groups, a l t h o u g h groups of closely similar samples are f o r m e d .

T h e p r o b l e m lies i n the m o d e l . T h e E u c l i d e a n distance calculation is
i n a p p r o p r i a t e for use w i t h correlated variables because it is based o n l y o n
pairwise comparisons, w i t h o u t r e g a r d to the elongation of data p o i n t swarms
along p a r t i c u l a r axes. I n effect, E u c l i d e a n distance imposes a s p h e r i c a l c o n -
straint o n the data set (18). W h e n correlation has b e e n r e m o v e d f r o m the
data, (by d e r i v a t i o n of standardized characteristic vectors) E u c l i d e a n distance
a n d average-linkage cluster analysis r e t u r n the three groups.
M o s t of the t i m e , w e do not have absolute a p r i o r i k n o w l e d g e r e g a r d i n g
the n u m b e r of groups i n a data set, or the relationships a m o n g the variates.
S e v e r a l rather distinct populations w i t h differing patterns of i n t e r e l e m e n t a l
correlations m a y b e r e p r e s e n t e d . I n such cases, i n s p e c t i o n of correlations
whose p a t t e r n is p o o l e d o v e r a l l samples m a y not be informative (although,
as i n the example j u s t p r e s e n t e d , i n s p e c t i o n o f scatterplots alone may p r o v i d e
information o n the n u m b e r of groups l i k e l y to be found).
A s s u m i n g that a data set has some natural or o p t i m u m structure a n d
that a g i v e n m u l t i v a r i a t e approach w i l l be able to reveal it is a b l i n d approach
to data analysis. Because m a t h e m a t i c a l p a t t e r n - r e c o g n i t i o n techniques not
only reveal structure b u t may impose structure as w e l l , more i n f o r m e d
application of multivariate techniques is n e e d e d . T h e choices a m o n g data
analytical approaches m u s t b e m a d e w i t h reference to the stated research
p r o b l e m a n d an awareness of the r e q u i r e m e n t s a n d assumptions of the v a r -
ious m u l t i v a r i a t e techniques. Different groups w i l l be f o r m e d or different
aspects of the data investigated d e p e n d i n g u p o n specific p r o b l e m f o r m u l a -
t i o n . F r o m this p e r s p e c t i v e , one may reject outright naive notions o f u n i f o r m
methodology i n v o l v i n g m u l t i v a r i a t e data analysis (11).

Multivariate Techniques and the Search for Structure

T h e literature d e a l i n g w i t h the m u l t i v a r i a t e techniques of p a t t e r n r e c o g n i -
t i o n , n u m e r i c a l taxonomy, g r o u p evaluation, etc. is extensive (e.g., reference

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 63

19). T h i s discussion provides o n l y a n o u t l i n e o f the techniques that have

b e e n u s e d to search for structure i n compositional data matrices generated
b y t h e analysis o f archaeological materials. Before m a n y of t h e t e c h n i q u e s
are u s e d , h o w e v e r , some p r e t r e a t m e n t of t h e data m a y b e necessary.

Data Representation. Transformations can b e a p p l i e d to t h e data

so that they w i l l m o r e closely follow the n o r m a l d i s t r i b u t i o n that is r e q u i r e d
for c e r t a i n procedures or for r e m o v i n g (or lessening) u n w a n t e d influences.
C e r t a i n l y for data analysis i n w h i c h major, m i n o r , a n d trace e l e m e n t a l c o n -
centrations are u s e d , some f o r m o f scaling is necessary to k e e p t h e variables
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

w i t h larger concentrations f r o m h a v i n g excessive w e i g h t i n t h e calculation

of m a n y coefficients o f similarity.
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

A n o t h e r f o r m of scaling involves e q u a l i z i n g the extent of variation a m o n g

the variables. I n some instances, t w o transformations o f t h e data are of
interest (e.g., transforming t h e data to l o g - n o r m a l d i s t r i b u t i o n s a n d t h e n
e n s u r i n g e q u a l w e i g h t t h r o u g h additional scaling). U s i n g transformations
that equalize t h e m a g n i t u d e of the measurements a n d the a m o u n t of variation
is i n k e e p i n g w i t h one of t h e premises of n u m e r i c a l taxonomy: that i s , no
i n d i v i d u a l variable s h o u l d assume m o r e w e i g h t t h a n another i n a n analysis
i n v o l v i n g the calculation of resemblance (19, 20). H o w e v e r , w h e n o n e is
m o d e l i n g , rather than m e r e l y s u m m a r i z i n g the data, variable contributions
m a y b e adjusted according to different c r i t e r i a . O n e example n o w u n d e r
investigation is w e i g h t i n g c h e m i c a l determinations as a f u n c t i o n of t h e i r
analytical errors.
T y p i c a l transformations i n c l u d e calculation o f logarithms; standardiza-
t i o n (mean of 0, standard d e v i a t i o n of 1); p e r c e n t range; a n d p e r c e n t of t h e
m a x i m u m value. A different t y p e o f transformation has b e e n u s e d for s u m -
m a r i z i n g c h e m i c a l data from t h e analysis of steatite or soapstone. A r g u i n g
from p r i n c i p l e s o f geochemistry, A l l e n a n d coworkers (21-23) n o r m a l i z e d
the rare earth concentrations relative to abundances i n c h o n d r i t i c meteorites.
F o l l o w i n g Sayre (24) a n d H a r b o t t l e (20), w e use base-10 logarithms i n most
of t h e examples discussed later i n this chapter. A p e r c e n t range transfor-
m a t i o n is also e m p l o y e d for one operation.

Ordination. O r d i n a t i o n procedures place a sample data p o i n t i n a

variable space to represent some t r e n d o r variation. N o assumptions n e e d
to b e made r e g a r d i n g the n u m b e r of groups. A s i m p l e type of o r d i n a t i o n
w o u l d b e to p l o t t h e coordinates o f a sample relative to t w o variables as i n
F i g u r e 2. F o r p-variables, h i g h e r d i m e n s i o n a l i t y p r o h i b i t s easy i n s p e c t i o n ,
so most o r d i n a t i o n techniques attempt to s u m m a r i z e t h e i n f o r m a t i o n w i t h i n
a data set a n d r e d u c e the d i m e n s i o n a l i t y (e.g., F i g u r e 1).
D i m e n s i o n a l i t y r e d u c t i o n a n d ordination have h a d three m a i n uses i n
compositional investigations. T h e y have b e e n u s e d

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 64 ARCHAEOLOGICAL CHEMISTRY

1. to inspect the data to see i f a general size c o m p o n e n t , one

s t e m m i n g from a p r o p o r t i o n a l rather an absolute relationship,
is present i n the data (16, 25)

2. to project the data into a standardized space that offers a dif-

ferent, possibly m o r e appropriate, perspective o n i n t e r p o i n t
distances (26)

3. to f o r m a set of reference axes of r e d u c e d d i m e n s i o n a l i t y for

graphical display of g r o u p e d sample distributions d e t e r m i n e d
b y some other t e c h n i q u e (27)
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

T h e most w i d e l y u s e d o r d i n a t i o n methods are based o n extracting e i g e n -

values a n d eigenvectors (also c a l l e d characteristic roots a n d characteristic
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

vectors) from a m i n o r p r o d u c t matrix, X ' X , or major p r o d u c t matrix, X X ' ,

of a data matrix, X (28). I f the data matrix is centered b y columns before
calculating X ' X , the m i n o r p r o d u c t matrix is a variance-covariance matrix.
I f the data matrix is first not o n l y centered b u t standardized, the m i n o r
p r o d u c t matrix is a correlation matrix. W h e t h e r the starting point of the
analysis is a v a r i a n c e - c o v a r i a n c e matrix or a correlation matrix, the e i g e n -
vectors of X ' X are usually called principal components. T h e eigenvectors of
this matrix may be m u l t i p l i e d b y t h e i r c o r r e s p o n d i n g eigenvalues to p r o d u c e
factors that m a y be rotated to enhance t h e i r i n t e r p r e t a b i l i t y (in this case,
the analysis is c a l l e d a factor analysis).
I n p r i n c i p a l , n e i t h e r c e n t e r i n g nor standardization is necessary i n e i -
genvector analyses, a n d each m a y be u n d e r t a k e n w i t h o u t the other. O r l o c i
(29) a n d N o y - M e i r (30) discuss the effects of c e n t e r i n g i n ecological a p p l i -
cations. To o u r k n o w l e d g e , there has b e e n no careful consideration of the
effects of c e n t e r i n g a n d standardization o n compositional data matrices. T h e
p o p u l a r i t y of standardized, centered eigenvector analyses has m o r e to do
w i t h the availability of software than w i t h the appropriateness of the as-
sumptions.
T h e first p r i n c i p a l c o m p o n e n t accounts for the d i r e c t i o n of m a x i m u m
variance t h r o u g h the data, w i t h each successive c o m p o n e n t accounting for
the m a x i m u m of the r e m a i n i n g variation. T h e l e n g t h of each vector is d e -
t e r m i n e d b y the square root of the associated eigenvalue. T h e d e r i v e d c o m -
ponents constitute a n e w set of reference axes that are l i n e a r combinations
of the o r i g i n a l m e a s u r e m e n t , b u t that n o w are orthogonal; that is, the v a r i -
ance of the o r i g i n a l data is p r e s e r v e d b u t the covariance has b e e n e l i m i n a t e d .
D e p e n d i n g o n h o w m u c h of the variance one wishes to preserve i n the
analysis, the n u m b e r of c o m p o n e n t s may be truncated, b u t o n l y at the loss
of some information. F a c t o r analysis includes an explicit statistical assump-
tion that a l l m e a n i n g f u l variation i n the data is accounted for b y m u n d e r l y i n g
factors, w i t h m<p, w h e r e p is the n u m b e r of o r i g i n a l variables (28). V a r i a t i o n

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 65

r e m a i n i n g after extraction of the m factors is assumed to be p u r e l y stochastic.

T h e d e r i v e d factors may also b e rotated to some s i m p l e structure. A n example
is varimax rotation, i n w h i c h orthogonality is p r e s e r v e d b u t the factors are
rotated so that the variance of the loadings o n each factor is m a x i m i z e d .
Rotation aids i n t e r p r e t a t i o n of the factors.
Q - m o d e factor analysis is based o n a major p r o d u c t matrix, X X ' .
W h e r e a s the R - m o d e analyses focus o n interrelationships a m o n g variables,
Q - m o d e analyses focus o n interrelationships among objects. A c c o r d i n g l y ,
the major p r o d u c t matrix is usually a distance or similarity matrix. F o r m a l l y ,
Q - m o d e a n d R - m o d e factor analyses are closely related because the n o n z e r o
eigenvalues of the major p r o d u c t matrix are i d e n t i c a l to the eigenvalues of
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

the m i n o r p r o d u c t matrix, a n d the eigenvectors are easily d e r i v e d f r o m one

another (28).
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

W h e n a p p l i e d to geological m i x i n g p r o b l e m s , Q - m o d e factors are

thought of as i d e a l i z e d e n d - m e m b e r s w h e n the rotated factor matrix is n o r -
m a l i z e d (31-33). W h e n t h r e e such e n d m e m b e r s are extracted, the data
points can be p l o t t e d o n ternary diagrams, the apices of w h i c h can be thought
of as r e p r e s e n t i n g the c o m p o s i t i o n of components c o n t r i b u t i n g to the m i x -
tures i n the data set. M i x t u r e s d e r i v e d f r o m m o r e than three components
are m o r e difficult to represent graphically.
I f groups have b e e n d e f i n e d , t h e n factor analysis can b e u s e d to d e t e r -
m i n e to what extent elements are r e p r e s e n t e d b y the c o m m o n p o r t i o n of
the factor. N o t s u r p r i s i n g l y , elements that are k n o w n to be m o b i l e i n several
different e n v i r o n m e n t s ( N a , K , B a , C s ) , or that are d e t e r m i n e d w i t h less
analytical p r e c i s i o n , often have less of t h e i r variance r e p r e s e n t e d i n the factor
solution.
O r d i n a t i o n has b e e n useful for i d e n t i f y i n g regional p a t t e r n i n g of c o m -
positional data o n natural or c u l t u r a l variables. F o r instance, i n an effort to
m o d e l n e u t r o n activation data f r o m the analysis of pottery r e c o v e r e d along
an escarpment of the C h i a p a s highlands, B i s h o p (16) u t i l i z e d R - m o d e c o m -
mon-factor analysis. N o n c h e m i c a l data w e r e projected onto c h e m i c a l l y based
factor plots. A f t e r 13 years, a n d a c e r t a i n d i m i n i s h i n g of m u l t i v a r i a t e zeal,
it can be q u e s t i o n e d w h e t h e r the same patterns c o u l d not have b e e n d e r i v e d
f r o m m o r e straightforward p r i n c i p a l components plots. T h e use of a c o m m o n -
factor m o d e l s e e m e d w a r r a n t e d at the t i m e as a means to p a r t i t i o n analytical
data into noise a n d signal components. I n a similar effort, B i s h o p (25) p r o j -
ected n o n c h e m i c a l data onto p r i n c i p a l - c o m p o n e n t s plots a n d t h r e e - c o m p o -
n e n t Q - m o d e factor plots i n an effort to identify p a t t e r n i n g of natural a n d
c u l t u r a l variables o n compositional data.
A s i d e f r o m p r i n c i p a l components, R - m o d e factor analysis, a n d Q - m o d e
factor analysis, other techniques that have b e e n u s e d to reduce d i m e n -
sionality i n u n g r o u p e d compositional data i n c l u d e m u l t i d i m e n s i o n a l scaling
(34) a n d correspondence analysis (35).

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 66 ARCHAEOLOGICAL CHEMISTRY

Grouping. T h e most c o m m o n l y e m p l o y e d techniques of data a n a l -

ysis i n compositional investigations are those that seek to p a r t i t i o n a data
set into smaller groups that contain samples that are m o r e similar to others
i n the g r o u p t h a n to other samples i n the data set. C l u s t e r analysis, i n c l u d i n g
b o t h h i e r a r c h i c a l a n d n o n h i e r a r c h i c a l variants, encompasses v i r t u a l l y the f u l l
range of g r o u p i n g p r o c e d u r e s u s e d i n compositional data analysis.
T h e m a n n e r i n w h i c h sample-to-sample resemblance is defined is a k e y
difference b e t w e e n the various h i e r a r c h i c a l c l u s t e r i n g techniques. S a m p l e
analyses may be s i m i l a r to one another i n a variety of ways a n d reflect interest
i n d r a w i n g attention to different u n d e r l y i n g processes or properties. T h e
selection of an appropriate measure of similarity is d e p e n d e n t , therefore,
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

o n the objectives of the research as set forth i n the p r o b l e m d e f i n i t i o n .

E x a m p l e s of different s i m i l a r i t y measures or coefficients that have b e e n used
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

i n compositional studies are average E u c l i d e a n distance, correlation, a n d

cosine. M a n y others that c o u l d be a p p l i e d are discussed i n the literature
d e a l i n g w i t h cluster analysis (15, 18, 19, 36, 37).
O n c e a matrix of resemblance coefficients has b e e n created, i t can be
s u m m a r i z e d b y partitions created to b r i n g together the samples that share
strong measures of resemblance. M o s t frequently, this p a r t i t i o n i n g is c a r r i e d
out b y c l u s t e r i n g the samples, l i n k i n g together the pairs of most s i m i l a r
samples, a n d t h e n stepping t h r o u g h the r e m a i n i n g samples u n t i l a l l are
l i n k e d i n a large t r e e l i k e structure or d e n d r o g r a m .
H i e r a r c h i c a l cluster analysis is a m e t h o d for o b t a i n i n g a q u i c k approx-
i m a t i o n to g r o u p i n g t e n d e n c y i n a set of data, b u t it is an insufficient r e p -
resentation of the data o n w h i c h to base data analytic or archaeological
inferences. I n a d d i t i o n to the n e e d to make an appropriate choice of r e s e m -
blance measure a n d c l u s t e r i n g a l g o r i t h m , the r e d u c t i o n of the m u l t i v a r i a t e
s i m i l a r i t y matrix to a t w o - d i m e n s i o n a l d e n d r o g r a m can introduce c o n s i d -
erable distortion. O n e measure of such distortion is the c o p h e n t i c correlation
(38). A l t h o u g h the resemblance relationships of the l o w e r linkages of the
d e n d r o g r a m are usually w e l l - r e p r e s e n t e d , the relationship a m o n g the h i g h e r
linkages is not. I n u s i n g a h i e r a r c h i c a l cluster m o d e l , no assumptions n e e d
to b e made r e g a r d i n g the n u m b e r of groups present i n the data set. I n -
spection of the d e n d r o g r a m does not i n f o r m d i r e c t l y o n the best-levels l i n k -
age for g r o u p arrangement, although a m e t h o d for testing the distinctness
of clusters has b e e n p r o p o s e d b y Sneath (39). A large change i n cluster l e v e l
of a d e n d r o g r a m m a y be a necessary c o n d i t i o n for cluster break, b u t i t is
not itself a sufficient c o n d i t i o n (40). A d d i t i o n a l problems arise i n that once
two samples are l i n k e d , that l i n k can not be b r o k e n . F i n a l l y , h i e r a r c h i c a l
cluster p r o c e d u r e s do not seek o p t i m u m partitions, a n d are usually f o u n d
to p e r f o r m p o o r l y unless natural groups are well-separated.
A different approach to cluster analysis involves the use of some e x p l i c i t
c r i t e r i o n ; the objective of that is to indicate the o p t i m u m n u m b e r of groups

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 67

i n a data set. S e v e r a l measures have b e e n p r o p o s e d that are based o n the

f u n d a m e n t a l p a r t i t i o n equation (41):

T = B + W (1)

w h e r e T is the m a t r i x of the total variation i n the data set, consisting of B ,

the v a r i a t i o n b e t w e e n the groups, a n d W , the variation w i t h i n the groups.
A f o r m a l discussion of these m a t r i x components is g i v e n i n reference 42. B y
u s i n g these relationships, one can use m i n i m u m t r a c e - W , d e t e r m i n a n t - W ,
or m a x i m i z a t i o n of trace W B , a m o n g others (43, 44). L i k e the resemblance
_ 1

measures m e n t i o n e d for h i e r a r c h i c a l c l u s t e r i n g , these c r i t e r i a have strengths

Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

a n d weaknesses. F o r example, the use of t r a c e - W w i l l result i n s p h e r i c a l

clusters. I f one uses the d e t e r m i n a n t - W it is not necessary to assume that
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

the natural groups exist i n s p h e r i c a l d i s t r i b u t i o n s , b u t o n l y that t h e i r shapes

are r o u g h l y the same (45, 46).
T h e c r i t e r i a j u s t d e s c r i b e d are found i n m a n y of the k-means cluster
programs that f o r m disjoint partitions i n to k-clusters. I n use, the programs
step f r o m a m i n i m u m n u m b e r of groups to some specified m a x i m u m n u m b e r .
Initial clusters are f o r m e d a n d samples are t h e n m o v e d i n an interactive
fashion u n t i l the c r i t e r i o n value is c o n s i d e r e d o p t i m u m (43, 47). I n f o r m a l
measures can b e u s e d to indicate w h i c h p a r t i t i o n represents the n u m b e r of
natural groups i n the data (43). k - M e a n s procedures approach a global so-
l u t i o n b y d e r i v i n g the best p a r t i t i o n i n g relative to a stated c r i t e r i o n value.
U n f o r t u n a t e l y , t h e y have b e e n f o u n d to be q u i t e sensitive to o u t l y i n g s a m -
ples, the p a r t i o n i n g of w h i c h can result i n s u b o p t i m u m solutions.
A basic q u e s t i o n of w h e t h e r h i e r a r c h i c a l or n o n h i e r a r c h i c a l cluster a n a l -
ysis is u s e d deals w i t h the correct or best n u m b e r of groups i n a data set.
T h e n o t i o n of " b e s t " relates not o n l y to a c r i t e r i o n value o r large break i n a
d e n d r o g r a m , b u t to the research objectives as w e l l . W e can not resist q u o t i n g
from E v e r i t t (48) w h a t is p r o b a b l y the u l t i m a t e w o r d r e g a r d i n g the n u m b e r
of groups:

In theory, of course, the problem is simple, to quote Dr. Idnozo

Hcahscror-Tenib, that super gaiactian hypermetrician [sic] who
appeared in Thorndike's 1953 Presidential address to the Psy-
chometric society, "Is easy. Finite number of combinations. Only
563 billion billion billion. Try all. Keep best."

Group Evaluative Procedures

G r o u p evaluative procedures p r o v i d e an assessment of the compactness of
the groups r e s u l t i n g f r o m the p r i o r application of the o r d i n a t i o n a n d g r o u p i n g
procedures j u s t discussed. A l t h o u g h group evaluation logically s h o u l d follow

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 68 ARCHAEOLOGICAL CHEMISTRY

group formation, groups are sometimes f o r m e d b y reference to the n o n c o m -

positional data (e.g., t y p e , f o r m , provenance), w i t h group evaluation p r o -
cedures t h e n u s e d (as ad hoc group formation techniques) to i n c l u d e or
exclude specimens o n the basis of composition. F o r example, d i s c r i m i n a n t
analysis may b e a p p l i e d to a n a l y z e d specimens i n groups c o r r e s p o n d i n g to
several different sites to c o n f i r m a hypothesis of m u l t i l o c u s p r o d u c t i o n . T h e
p r o b l e m is that d i s c r i m i n a n t analysis is v e r y good at m a k i n g groups out of
w h a t e v e r groups are assumed i n the first place, p a r t i c u l a r l y w h e n the cor-
relation or v a r i a n c e - c o v a r i a n c e matrix is p o o l e d over a l l groups.
T h e techniques used for group evaluation are based o n m u l t i v a r i a t e
generalizations o f univariate statistics such as the c e n t r o i d (a m u l t i v a r i a t e
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

m e a n , or m e a n vector) a n d v a r i a n c e - c o v a r i a n c e matrix (a m u l t i v a r i a t e g e n -
eralization of the variance). G r o u p evaluative procedures can be classified
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

as single group p r o c e d u r e s , i n w h i c h specimens are evaluated as to t h e i r

l i k e l i h o o d of m e m b e r s h i p i n a single g r o u p , a n d m u l t i g r o u p p r o c e d u r e s ,
such as d i s c r i m i n a n t analysis a n d canonical variates, i n w h i c h several groups,
or m u l t i v a r i a t e centers of mass are assumed to be r e p r e s e n t e d i n a data set.
S i n g l e - g r o u p evaluative p r o c e d u r e s , because they are not i m p l e m e n t e d
i n most w i d e l y available statistical packages ( S A S , S P S S , B M D P , etc.), are
u s e d less often i n c o m p o s i t i o n a l data analysis t h a n m u l t i p l e - g r o u p evaluative
procedures. A D C O R R , a p r o g r a m d e v e l o p e d at B r o o k h a v e n N a t i o n a l L a b -
oratory (24), performs single-group multivariate evaluation based o n
M a h a l a n o b i s (generalized) distance and the related H o t e l l i n g T statistic 2

(the latter is a m u l t i v a r i a t e generalization of the w e l l - k n o w n Student's t).

M a h a l a n o b i s distance, w h i c h is central to m u l t i p l e - g r o u p as w e l l as single-
group e v a l u a t i o n , is d e f i n e d for a group c e n t r o i d a n d each i n d i v i d u a l m e m b e r
of the group as follows:

D 2
= (X — X ^ J ' S - K X - X m e a n ) (2)

w h e r e (X - X m e a n ) is a vector of differences b e t w e e n the values for an ob-

servation a n d the m e a n values a n d S " is t h e inverse of the group v a r i -
1

a n c e - c o v a r i a n c e m a t r i x . H o t e l l i n g T is e s s e n t i a l l y e q u i v a l e n t to
2

M a h a l a n o b i s distance for i n d i v i d u a l points. T h e probabilities of m e m b e r s h i p

i n the core g r o u p for each s p e c i m e n are t h e n readily obtainable f o l l o w i n g
transformation of the T statistic to a r e l a t e d , F - d i s t r i b u t e d statistic (28).
2

D i s c r i m i n a n t analysis evaluates the distance b e t w e e n i n d i v i d u a l points

a n d several centroids h y p o t h e s i z e d to exist i n the hyperspace d e f i n e d b y
e l e m e n t a l concentrations. D a v i s (28) provides a clear a n d concise d e s c r i p t i o n
of the algebra i n v o l v e d i n two-group a n d m u l t i p l e - g r o u p d i s c r i m i n a n t a n a l -
ysis, s h o w i n g that d i s c r i m i n a n t functions are e q u i v a l e n t to the eigenvectors
of W B , w h e r e W " is the inverse of the w i t h i n - g r o u p sums of products
_ 1 1

matrix, a n d B is the b e t w e e n - g r o u p sums of products matrix. T h e M a h a l -

anobis distances f r o m an u n k n o w n p o i n t to each of the alternative centroids

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 69

p r o v i d e c r i t e r i a for evaluating the relative probabilities of m e m b e r s h i p of

the s p e c i m e n i n each of the groups.

Modeling vs. Summarizing Compositional Data: A Ceramic

Example. M o d e l i n g is i m p o r t a n t ; it involves i n f o r m e d i n t e r a c t i o n b e -
t w e e n the researcher, his or h e r objectives, analytical data, a n d m u l t i v a r i a t e
data presentation. T h e p o t e n t i a l for revealed data structures to vary d e -
p e n d i n g u p o n the choice of n u m e r i c a l p r o c e d u r e is a c k n o w l e d g e d e x p l i c i t l y .
I n a d d i t i o n , this approach requires the objectives of a particular investigation
and the complexities of the compositional data base to b e k e p t i n m i n d
d u r i n g the stage of data analysis. T h e m o d e l provides the e x p l i c i t rationale
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

for reasoning from r e v e a l e d data structure to inferences about h u m a n b e -

havior. T h e r e m a i n d e r of this chapter illustrates some of the implications of
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

this approach for the analysis of ceramic compositional data.

A n y g i v e n a n a l y z e d sample of pottery is a s m a l l subset of a larger c e r a m i c
system. P o t t e r y is f o r m e d f r o m clays a n d nonplastic constituents a c c o r d i n g
to shared customs of the local p o t t e r y - m a k i n g group as w e l l as idiosyncratic
or stochastic effects. T h e compositional profile that is d e r i v e d from the c h e m -
ical analysis of a c e r a m i c sample, therefore, is a w e i g h t e d expression of b o t h
natural a n d c u l t u r a l constraints.
O n e of the m o r e obvious examples of this interaction involves the a d -
d i t i o n of t e m p e r to a clay matrix (temper may be another clay, b u t is m o r e
often a nonplastic material). T h e effect of t e m p e r i n g varies; a r e l a t i v e l y p u r e
material, such as q u a r t z , m a y r e d u c e e l e m e n t a l concentrations i n a c e r a m i c
paste b y a constant p r o p o r t i o n (49). A d d i t i o n of other k i n d s of t e m p e r or
clay w i l l result i n a c o m p l e x relationship of d i l u t i o n a n d e n r i c h m e n t (14, 2 5 ,
50). Because e l e m e n t a l concentrations i n sediments vary d e p e n d i n g u p o n
grain size (e.g., references 5 1 - 5 3 ) , the size distributions of the a d d e d n o n -
plastics also c o n t r i b u t e to compositional c o m p l e x i t y . If behavioral inferences
are to be d r a w n , the c u l t u r a l l y i n d u c e d e l e m e n t a l variation arising from
texture a n d t e m p e r differences a m o n g pottery p r o d u c e d f r o m a single clay
resource requires m o r e than s i m p l e g r o u p i n g a n d s u m m a r y statistics.
F a i l u r e to consider the effect of t e m p e r d u r i n g the analysis of c o m p o -
sitional data m i g h t l e a d to spurious inferences i n the f o l l o w i n g way: Suppose
the object of an investigation is to identify local a n d nonlocal pottery at some
site. I f b i m o d a l amounts of t e m p e r w e r e a d d e d b y local potters to c e r a m i c
paste (for example, d e p e n d i n g o n w h e t h e r the potter p l a n n e d to make a
s e r v i n g d i s h or a w a t e r storage jar), a n a l y z e d pottery from the site m i g h t
fall into two compositional groups e v e n though a single set of ceramic r e -
sources was u s e d . I f t e m p e r is left out of the m o d e l , the compositional data
m i g h t suggest that one of the compositional groups is local a n d the other is
nonlocal.
I n the last example, the investigator's p r o b l e m orientation m i g h t p r e -
dispose h i m or h e r to accept the plausible b u t incorrect inference that local

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 70 ARCHAEOLOGICAL CHEMISTRY

a n d nonlocal specimens are represented i n the analyzed collection. D a t a

analysis techniques that do not p e r m i t recognition of the source of g r o u p -
separating variation can p r o m o t e such erroneous interpretations of c o m p o -
sitional p a t t e r n i n g .
N o data analysis t e c h n i q u e is foolproof. O b t a i n i n g v a l i d inferences from
compositional analysis d e p e n d s , above a l l , o n c o n s i d e r i n g alternative m o d e l s
of the processes responsible for data structure. T o d e v e l o p useful a n d p l a u -
sible alternative models, n o n c h e m i c a l as w e l l as c h e m i c a l information must
be considered. I n the last h y p o t h e t i c a l case, observation of a textural dif-
ference b e t w e e n s e r v i n g vessels a n d water storage jars m i g h t suggest b i m o d a l
t e m p e r i n g of the same local clay, so that local p r o d u c t i o n of b o t h c o m p o -
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

sitional groups c o u l d be suggested as an alternative to multilocus p r o d u c t i o n .

O u r goal is to s h o w h o w data analysis techniques, although they may not
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

always lead to v a l i d inferences, m a y at least facilitate the search for relevant

data structure (see reference 54). T h e illustrations focus o n the p r o b l e m of
t e m p e r i n g because it is an i m p o r t a n t consideration i n most ceramic c o m -
positional investigations.

Data Used for Illustration. O u r approach to the p r o b l e m of h o w

to fit data analysis techniques to expectations about structure i n the data
involves h e u r i s t i c use of artificially s t r u c t u r e d data. I n particular, w e have
r e l i e d o n analyses of ceramics a n d ceramic raw materials from contemporary
h i g h l a n d G u a t e m a l a n p o t t e r y - m a k i n g towns. A n a l y z e d clays, tempers, a n d
ceramic sherds from three towns i n the n o r t h e r n V a l l e y of G u a t e m a l a ( D u r -
azno, Sacojito, a n d Chinautla) constitute a basic data set of 113 observations.
W i t h these data, a h y p o t h e t i c a l data set is constructed, the structure of
w h i c h w e w i l l examine b y u s i n g some of the data analysis techniques just
d e s c r i b e d . T h e analyzed specimens w e r e f u r n i s h e d b y D e a n A r n o l d . A r n o l d s
ethnographic observations (55-57) p r o v i d e baseline expectations about pat-
t e r n i n g i n the compositional data. N e u t r o n activation analyses w e r e c a r r i e d
out b y using the standard procedures d e v e l o p e d at B r o o k h a v e n N a t i o n a l
L a b o r a t o r y (47).
W i t h i n the n o r t h e r n V a l l e y of G u a t e m a l a w h i t e w a r e t r a d i t i o n , there is
a basic d i c h o t o m y b e t w e e n S a c o j i t o - C h i n a u t l a w h i t e w a r e a n d D u r a z n o w h i -
teware. D u r a z n o potters exploit sources of w h i t e w a r e clay different from the
source used b y S a c o j i t o - C h i n a u t l a potters (55). D u r a z n o w h i t e w a r e pottery
and raw materials therefore should be distinguishable from C h i n a u -
tla-Sacojito whiteware.
T h e t e m p e r i n g material u s e d i n D u r a z n o is volcanic ash from the ash
blanket that covers the n o r t h e r n V a l l e y of G u a t e m a l a . C h i n a u t l a - S a c o j i t o
potters exploit the same volcanic ash, albeit i n different locations. A l t h o u g h
potters i n a l l three towns t e m p e r their clay, it is instructive to consider the
potential effect on compositional p a t t e r n i n g of variable t e m p e r i n g practices
(e. g., a fine p a s t e - m e d i u m paste dichotomy). S u c h a d i c h o t o m y characterizes

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 71

m a n y archaeological collections. I f n o r t h e r n V a l l e y of G u a t e m a l a potters

made fine-paste p o t t e r y , its c o m p o s i t i o n w o u l d resemble the c o m p o s i t i o n
of r a w clay specimens i n c l u d e d i n the present data set; the a n a l y z e d sherds
i n the data set are taken to represent m e d i u m - p a s t e pottery.
B y artificially m i x i n g clays a n d t e m p e r s , any n u m b e r of h y p o t h e t i c a l
sherds can be a d d e d to the basic data set. I n p r i n c i p a l , the h y p o t h e t i c a l
mixtures w e i g h h e a v i l y i n favor o f finding suitable n u m e r i c a l techniques for
d i s c o v e r i n g relevant structure i n the data set. I n the f o l l o w i n g example
h y p o t h e t i c a l fine-paste a n d m e d i u m - p a s t e mixtures are a d d e d to the basic
data set of clays a n d t e m p e r s . W e feel that " s t a c k i n g the d e c k " i n this m a n n e r
is not o n l y appropriate b u t necessary i n a study a t t e m p t i n g to evaluate
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

methodology.
H y p o t h e t i c a l c e r a m i c paste mixtures are generated f r o m clays a n d t e m -
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

pers o f k n o w n concentrations b y calculating e l e m e n t a l concentrations ac-

c o r d i n g to

S, = VT(Td + PC(C«) (3)

w h e r e S represents the e l e m e n t a l concentrations i n the c e r a m i c , T r e p -

t {

resents the e l e m e n t a l concentrations i n the t e m p e r , a n d C represents the

t

e l e m e n t a l concentrations i n the clay. B o t h T a n d C have b e e n d e t e r m i n e d

{ {

analytically. F T a n d P C are the proportions of t e m p e r a n d clay, r e s p e c t i v e l y ,

m i x e d to make the ceramic paste.
T h e s u m of P T a n d P C m u s t e q u a l one. Because any recipe for m i x i n g
components, no matter h o w strictly followed, w i l l show some variation, the
PTs a n d P C s are chosen r a n d o m l y w i t h m e a n , jx , as the p r o p o r t i o n of clay
P

or t e m p e r specified b y the r e c i p e a n d standard d e v i a t i o n , a n d r j i n d i c a t i n g

P

the relative standardization (i.e., h o w strictly the recipe was followed). I n

this example, the m e a n clay p r o p o r t i o n i n fine-paste mixtures was set at
0.85, a n d the m e a n p r o p o r t i o n i n m e d i u m - p a s t e mixtures was set at 0.65.
Relative h o m o g e n e i t y was a s s u m e d , a n d the standard d e v i a t i o n was set at
0.05 i n b o t h cases. T h e c o m p o n e n t s (T a n d C-) to be m i x e d i n the r a n d o m l y
t

chosen proportions are chosen r a n d o m l y from the analyzed data sets of clays
a n d tempers. U n f o r t u n a t e l y , o n l y four samples of the clay u s e d to m a k e
D u r a z n o w h i t e w a r e w e r e a n a l y z e d , so the h y p o t h e t i c a l fine-paste a n d m e -
d i u m - p a s t e m i x t u r e s c o u l d e x h i b i t some bias d u e to the s m a l l n u m b e r o f
clays to choose from.

Analysis. T w o separate patterns s h o u l d be d i s c e r n i b l e i n the artificial

data set: sherds, clays, a n d h y p o t h e t i c a l m i x t u r e s from C h i n a u t l a - S a c o j i t o
s h o u l d b e separable f r o m r a w materials a n d products from D u r a z n o a n d fine-
a n d m e d i u m - p a s t e subgroups s h o u l d b e recognizable w i t h i n the major
groups. T h e first p a t t e r n involves shape variation, that is, differing m e a n
vectors a n d differing v a r i a n c e - c o v a r i a n c e structures that arise t h r o u g h nat-

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 72 ARCHAEOLOGICAL CHEMISTRY

u r a l processes of clay formation. T h e second pattern can be thought of as

size variation, i n w h i c h t e m p e r i n g introduces l i n e a r changes i n the e l e m e n t a l
concentrations. D i l u t i o n b y a constant (e.g., b y quartz sand) is an extreme
example of size variation.
T h e e l e m e n t a l concentration data w e r e first transformed to base-10
logarithms to counteract the i m p l i c i t w e i g h t i n g due solely to variation i n the
abundance of elements i n nature (cf. reference 20). P r i n c i p a l c o m p o n e n t s
w e r e t h e n calculated f r o m the v a r i a n c e - c o v a r i a n c e m a t r i x of the l o g c o n -
centration data. A p l o t of the first two p r i n c i p a l components ( F i g u r e 3) shows
two major groups; one that is h i g h o n c o m p o n e n t N o . 1 a n d one that is l o w
o n c o m p o n e n t N o . 1. T h e groups represent the two clay sources; C h i n a u -
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

t l a - S a c o j i t o - d e r i v e d materials scored h i g h a n d D u r a z n o - d e r i v e d materials

scored low. T h e group separation is due largely to C r , R b , a n d C s c o n c e n -
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

trations, that t e n d to be h i g h e r i n the C h i n a u t l a - S a c o j i t o materials. Table I

contains coefficients of the o r i g i n a l variables o n t h e p r i n c i p a l c o m p o n e n t s
along w i t h the percentage of variance e x p l a i n e d b y each c o m p o n e n t .
H i e r a r c h i c a l a n d k-means cluster analysis distinguishes the two major
source-specific groups e v i d e n t o n the scatter plot. T h e highest l e v e l d i v i s i o n
is b e t w e e n C h i n a u t l a - S a c o j i t o m a t e r i a l a n d D u r a z n o m a t e r i a l . L o w e r - l e v e l
branches s u b d i v i d e the D u r a z n o m a t e r i a l first, a n d the C h i n a u t l a - S a c o j i t o
texture subgroups r e m a i n i n a single s u b d i v i s i o n above a E u c l i d e a n distance
of 0.11. k - M e a n s cluster analysis allows the recovery of the major structure
i n the data, although the m o r e subtle structure remains o b s c u r e d , e v e n
w h e n eight groups are f o r m e d . D i s c r i m i n a n t analysis confirms the true struc-
ture i f the true group m e m b e r s h i p is u s e d as the classification c r i t e r i o n , or
it confirms a six-group structure d e r i v e d from k-means a n d h i e r a r c h i c a l
cluster analysis (if that structure is used as a basis for classification). T h e
application of g r o u p i n g procedures followed b y d i s c r i m i n a n t analysis t h e r e -
fore w o u l d result i n an erroneous i n t e r p r e t a t i o n of this data set.
I f this data set w e r e archaeological, a further p r o b l e m of i n t e r p r e t a t i o n
w o u l d i n v o l v e the o r i g i n of separation b e t w e e n the two major groups. I n
other w o r d s , w h a t w o u l d constitute the e v i d e n c e that group separation o n
p r i n c i p a l c o m p o n e n t N o . 1 i n F i g u r e 3 represents a clay-source d i s t i n c t i o n
rather than t e m p e r - r e l a t e d variation? T h e p r o b l e m is analogous to the p r o b -
l e m of separating size f r o m shape i n the field of multivariate m o r p h o m e t r i e s
(see references 5 8 - 6 2 ) . T h e o r e t i c a l l y , general size manifests itself as c o n -
sistently positive coefficients of the o r i g i n a l variables o n the first p r i n c i p a l
c o m p o n e n t (59). H o w e v e r , because a d d i t i o n of a volcanic ash t e m p e r almost
certainly w i l l not p r o d u c e u n i f o r m e n r i c h m e n t or d i l u t i o n , the variation
i n t r o d u c e d b y t e m p e r i n g cannot be c o n s i d e r e d analogous to general size.
Because some elements m a y be d i l u t e d w h i l e others are e n r i c h e d , the signs
of the coefficients of the o r i g i n a l variables o n the first p r i n c i p a l c o m p o n e n t
w i l l differ. Because the amount of e n r i c h m e n t or d i l u t i o n is l i k e l y to v a r y ,
the m a g n i t u d e of the coefficients also may differ w i d e l y .

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

0.6

+ * • « <,

+ +
0.1 -
+ +
0 4

A
A

A
V A
I
A
X A
-0.3- A AA A A
A A

A A A
X X x
A\ A AA A
-0.5-

In Archaeological Chemistry IV; Allen, R.;

x
XXX x x y

-0.6-
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
Principal Component #2
A
+ Chinautla/Sacojito O Chinautia/SacoJIto Durazno X Durazno

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.

Medium Paste Fine Paste Medium Paste Fine Paste

3. Principal components plot (components Nos. 1 and 2) based on the entire data set used for illus-
tration.
 74 ARCHAEOLOGICAL CHEMISTRY

Table I. Principal Component Coefficients for the

Example D a t a Set
Element No. 1 No. 2 No. 3
K 0.634 -2.421 2.242
Sc 0.030 1.492 -0.073
Cr 1.868 1.135 -1.038
Fe 0.144 1.263 0.240
Zn 0.618 1.040 0.022
Rb 1.223 -0.906 0.859
Cs 1.039 -0.175 -0.017
Ba -0.310 -1.136 2.148
La 0.015 0.540 1.677
Ce 0.106 0.628 1.456
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

Sm 0.087 1.276 1.316

Eu -0.109 1.561 1.347
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

Yb -0.108 1.286 1.435

Lu -0.159 1.307 1.347
Hf -0.238 0.608 1.318
Ta 0.349 0.337 0.797
Th 0.293 -0.550 1.225
Percent of variance explained 54.817 17.327 14.420

M o r e serious than the p r o b l e m s w i t h the analogy b e t w e e n size a n d

t e m p e r i n g , the fact that t e m p e r may interact differently w i t h clays from the
two distinct sources makes it u n l i k e l y that a single d i m e n s i o n analogous to
a l l o m e t r i c size w i l l b e recognizable i n the data set. N o n e t h e l e s s , i f one
possesses some i n f o r m a t i o n about the c o m p o s i t i o n of l i k e l y t e m p e r i n g m a -
terials, that i n f o r m a t i o n can b e u s e d to h e l p identify p o t e n t i a l t e m p e r i n g
d i m e n s i o n s . T h e volcanic ash t e m p e r u s e d i n the example data set is k n o w n
to b e e n r i c h e d , relative to local clays, i n K a n d B a a n d d i l u t e d , relative to
local clays, i n most other elements i n the analysis (Th tends to be e n r i c h e d
i n t e m p e r , b u t the p a t t e r n is inconsistent). A l t h o u g h this i n f o r m a t i o n was
o b t a i n e d b y n e u t r o n activation analysis of t e m p e r o b t a i n e d f r o m potters, it
can be i n f e r r e d f r o m other sources, such as p u b l i s h e d studies of the c o m -
p o s i t i o n of h i g h l a n d G u a t e m a l a n volcanic ash (63). I n an archaeological sit-
uation, the i n f o r m a t i o n c o u l d have b e e n o b t a i n e d t h r o u g h compositional
analysis of l i k e l y t e m p e r i n g materials c o l l e c t e d d u r i n g a carefully p l a n n e d
raw-materials s a m p l i n g p r o g r a m or b y separation a n d analysis of the nonclay
c o m p o n e n t . U s i n g this i n f o r m a t i o n , w e expect the t e m p e r i n g components
to b e recognizable because K , B a , a n d possibly T h s h o u l d have h i g h - m a g -
n i t u d e coefficients of same sign.
T h e coefficients of the o r i g i n a l variables o n the p r i n c i p a l components
(Table I) do not contain a single d i m e n s i o n that is clearly related to t e m p e r .
K a n d B a , w h i c h are thought to b e e n r i c h e d i n t e m p e r relative to clay, have
opposite signs o n the first c o m p o n e n t , so the first c o m p o n e n t is not a t e m -
p e r i n g d i m e n s i o n . T h e coefficients of K a n d B a are fairly large a n d have the

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 75

same sign o n c o m p o n e n t N o . 2, so c o m p o n e n t N o . 2 m a y carry some t e m p e r -

related variation. S i m i l a r reasoning suggests that p r i n c i p a l c o m p o n e n t N o .
3 also m a y reflect t e m p e r i n g .
A d d i t i o n a l insight into the significance of group-separating variation can
be o b t a i n e d from further use of o r d i n a t i o n . I f the first three factors from a
Q - m o d e factor analysis of data (expressed as a percentage of range) are
n o r m a l i z e d , data points are expressed as i f they w e r e mixtures of three
components. T h e data points can t h e n be p l o t t e d o n ternary diagrams i n
w h i c h positions are d e t e r m i n e d b y the percentage c o n t r i b u t i o n from each
factor. F i g u r e 4 shows the data for the present example p l o t t e d against the
first three varimax-rotated factors d e r i v e d f r o m Q-analysis. T w o apices cor-
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

r e s p o n d to the two clay sources r e p r e s e n t e d i n the data. T h e t h i r d apex,

w h i c h expresses p r i m a r i l y the a m o u n t of K , B a , a n d T h i n the m i x t u r e ,
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

p r o b a b l y corresponds w i t h t e m p e r . F i g u r e 5, w h i c h is d e r i v e d f r o m a Q -
m o d e factor analysis of the o r i g i n a l data set plus the tempers u s e d to f o r m
the h y p o t h e t i c a l m i x t u r e s , confirms the identification of the t h i r d apex as
t e m p e r - r e l a t e d . T h e s e t h r e e - c o m p o n e n t plots identify the o v e r a l l structure
i n the data set m o r e clearly than any of the previous o r d i n a t i o n a n d g r o u p -
f o r m i n g techniques.
F i g u r e 5 illustrates w h y it m a y be useful to i n c l u d e p o t e n t i a l t e m p e r i n g
materials i n a c o m p o s i t i o n a l study, e v e n i f exact sources p r o b a b l y have not
b e e n s a m p l e d . I f t e m p e r i n g materials u s e d i n ceramics from a p a r t i c u l a r
r e g i o n can b e assumed to b e r e l a t i v e l y u n i f o r m (like volcanic ash, w h i c h is
a c o m m o n t e m p e r i n p r e h i s t o r i c ceramics from Mesoamerica), t h e n i n c l u s i o n
of compositional data o n the m a t e r i a l i d e n t i f i e d as t e m p e r may, as i n the
present case, elucidate the interactions b e t w e e n t e m p e r a n d several different
clays.
I n s u m m a r y , although d i v i s i o n of the data set into two major groups
attributable to clay sources can be a c c o m p l i s h e d b y several methods ( d i v i s i o n
is especially clear w h e n o r d i n a t i o n is c o m b i n e d w i t h group formation p r o -
cedures), it is m o r e difficult to detect a n d i n t e r p r e t p a t t e r n i n g w i t h i n the
major groups. T h i s difficulty arises because processes operating o n two h i -
erarchical levels are b e i n g confused: the major d i v i s i o n is p r o d u c e d b y natural
variation a r i s i n g i n the process of clay formation; whereas l o w e r - l e v e l pat-
t e r n i n g is a result of two distinct c u l t u r a l processes, D u r a z n o potters m i x i n g
t e m p e r w i t h t h e i r o w n u n i q u e clay a n d C h i n a u t l a - S a c o j i t o potters m i x i n g
t e m p e r w i t h t h e i r o w n u n i q u e clay. I n the present example, a t h r e e - c o m -
p o n e n t o r d i n a t i o n clarified b o t h levels of structure, b u t this h a p p y result i n
part reflects the artificial nature of the data a n d the fact that a l l r a w materials
h a d b e e n analyzed. M o r e c o m p l e x data structures w o u l d have r e m a i n e d
o b s c u r e d i n the Q - m o d e factor analysis.
A m o r e generally useful approach is to incorporate into the a s s u m p t i o n
that m u l t i p l e processes may create structure into the m o d e l , a n d to recognize
that coarse-grained processes m a y obscure fine-grained processes. F o r i n -

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 76 ARCHAEOLOGICAL CHEMISTRY
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
Figure 5. Plot of specimens in the example data set plus northern Valley of Guatemala volcanic ash tempers
relative to three components defined by Q-mode factor analysis.
 78 ARCHAEOLOGICAL CHEMISTRY

stance, h a v i n g d i s c o v e r e d two source-specific groups i n the example data,

a logical next step i n the analysis w o u l d b e to examine the subgroups i n -
d i v i d u a l l y . T h e f o l l o w i n g paragraphs describe this approach a p p l i e d to the
C h i n a u t l a - S a c o j i t o subgroup.
D e s p i t e the complexities i n t r o d u c e d b y t e m p e r i n g w i t h a n o n i n e r t m a -
terial, the analogy w i t h a l l o m e t r i c g r o w t h s h o u l d h o l d w h e n , as i n the present
case, the data set was generated b y a group of potters w o r k i n g w i t h a single
set of c e r a m i c resources. T e m p e r i n g can b e thought of as g r o w t h that involves
change i n shape or p r o p o r t i o n a l i t y . A s s u m i n g that the data are a l l d e r i v e d
from a single, r e l a t i v e l y homogeneous clay source (as i n the present example)
a n d , f u r t h e r , that the t e m p e r differs substantially f r o m the c o m p o s i t i o n a l
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

profile of the clay (as i n the present example), a l l e l e m e n t a l concentrations

s h o u l d show l i n e a r relationships (positive or negative) w i t h the p r o p o r t i o n
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

of t e m p e r i n the m i x t u r e . T h e major d i m e n s i o n of variation (the first p r i n c i p a l

component) s h o u l d carry most or a l l of the variation related to g r o w t h (tem-
pering), p a r t i c u l a r l y i f t e m p e r - r e l a t e d variation is sufficient to create b i p o l a r
extremes recognizable as compositional groups.
Coefficients of the e l e m e n t a l concentrations o n the first p r i n c i p a l c o m -
p o n e n t (Table II) show the p a t t e r n expected for a t e m p e r - r e l a t e d d i m e n s i o n :
the signs are positive for a l l variables except K a n d B a , w h i c h are negative.
T h u s , the group separation o n p r i n c i p a l c o m p o n e n t N o . 1 s h o w n i n F i g u r e
6 can b e i n t e r p r e t e d as t e m p e r - r e l a t e d . A s expected, the two major groups
e v i d e n t o n the p l o t are (I) h y p o t h e t i c a l fine-paste mixtures along w i t h r a w
clay a n d (2) h y p o t h e t i c a l m e d i u m - p a s t e mixtures along w i t h r e a l sherds.

Table II. Principal Component Coefficients for the

Chinautla-Sacojito Data Set
Element No. 1 No. 2 No. 3
K -0.669 2.231 4.865
Sc 1.374 -0.879 1.534
Cr 1.617 -1.271 5.321
Fe 1.127 -0.302 2.857
Zn 1.031 -1.253 -1.078
Rb 0.410 -0.059 2.541
Cs 0.665 -0.586 3.745
Ba -0.528 3.314 1.254
La 0.999 1.451 -0.911
Ce 1.199 1.233 -0.223
Sm 1.585 0.742 -1.947
Eu 1.647 0.551 -2.672
Yb 1.439 0.892 -2.337
Lu 1.368 0.678 -2.158
Hf 0.602 0.882 0.768
Ta 0.721 0.378 2.338
Th 0.379 1.384 1.054
Percent of variance explained 44.419 31.947 8.047

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

0.6 - i
w
0.5 - as
o o
o o
0.4 -
+
o * w
0.3 - o A •n

0.2 - *• o
+
UJ o O
z
o 0.1 - 0
A
A 1
0 A OA rA
8 o
A £ s
< -0.1
D_ D

O . /A
+ a
Z -0.2
an ° +
n f f
CL o
-0.3 H +
-0.4 +

In Archaeological Chemistry IV; Allen, R.;

•

-0.6
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
PRINCIPAL COMPONENT #2

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.

+
'Medium P a s t e ' + 'Pine
'Fine Paste'
Paste* oo Raw
Raw Clav
Clay A Sherds

Figure 6. Principal components plot (components Nos. 1 and 2) based on Chinautla-Sacojito data only.
 80 ARCHAEOLOGICAL CHEMISTRY

B o t h k-means a n d h i e r a r c h i c a l cluster analysis y i e l d two major groups at-

t r i b u t a b l e to t e m p e r - r e l a t e d variation (the m e d i u m - p a s t e group i n c l u d e d six
fine-paste mixtures). B o t h techniques also extracted a small t h i r d group o f
outliers (the seven data points i n the l o w e r left c o r n e r o f F i g u r e 6). D i s -
c r i m i n a n t analysis confirms the existence e i t h e r of groups r e c o v e r e d b y
cluster analysis or groups d e f i n e d p r e v i o u s l y as m e d i u m - p a s t e or fine-paste.
Single group evaluation w i t h the p r o g r a m A D C O R R also confirms the m e m -
b e r s h i p of the m e d i u m - a n d fine-paste core groups, a l t h o u g h a few n o n -
m e m b e r s show greater than 5 % p r o b a b i l i t y of m e m b e r s h i p i n each g r o u p .
F i g u r e 6 shows an u n e x p e c t e d result. T h e two major groups that are
i n f e r r e d to be t e m p e r - r e l a t e d are separated along p r i n c i p a l c o m p o n e n t N o s .
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

2 a n d 1. T h i s separation suggests that, contrary to expectations d e r i v e d f r o m

analogy w i t h the b i o l o g i c a l m o d e l , g r o w t h d u e to t e m p e r i n g is not c o n f i n e d
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

to the major d i m e n s i o n i n the data. T h e coefficients of the e l e m e n t a l c o n -

centrations of c o m p o n e n t N o . 2 (Table II) f u r t h e r support this i n t e r p r e t a t i o n ,
a fact i n d i c a t i n g that K a n d B a , the two elements e n r i c h e d b y t e m p e r , are
major d e t e r m i n a n t s of variation o n this second d i m e n s i o n . T e m p e r - r e l a t e d
variation (size), therefore, is expressed o n b o t h the first a n d second d i m e n -
sions of variation i n the second example data set.
F a i l u r e to isolate the effect of t e m p e r i n g o n the first p r i n c i p a l c o m p o n e n t
appears to result from u n e x p e c t e d i n h o m o g e n e i t y i n the data. T h e seven
outliers i d e n t i f i e d b y h i e r a r c h i c a l a n d k-means cluster analysis cause t i 1

major axis of variation to v e e r away from the d i r e c t i o n of t e m p e r - r e l a t e d

size. W h e n the outliers are r e m o v e d a n d the p r i n c i p a l components r e c a l -
c u l a t e d , c o m p o n e n t N o . 1 (Table I I I a n d F i g u r e 7) m o r e closely approximates

Table III. Principal Component Coefficients C h i n a u t l a -

Sacojito D a t a with Outliners Removed
Element No. 1 No. 2 No. 3
K -1.401 2.778 6.074
Sc 1.639 0.205 1.256
Cr 2.102 0.792 5.323
Fe 1.237 1.326 1.021
Zn 1.373 -1.457 2.029
Rb 0.436 0.572 3.564
Cs 0.917 0.563 3.685
Ba -1.669 3.675 -0.930
La 0.375 2.001 -2.424
Ce 0.636 2.014 -1.398
Sm 1.145 1.215 -2.902
Eu 1.238 0.740 -2.928
Yb 0.934 1.168 -3.183
Lu 1.004 1.163 -4.114
Hf 0.245 1.537 -0.478
Ta 0.591 1.546 1.199
Th -0.138 1.956 0.155
Percent of variance explained 50.529 23.255 7.281

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

0.6
o
0.5
0
0.4
o o o
_ 0.3 o +
>
^ 0.2
Ld +
+
O 0.1 + +
a. + +
:§ A
A
8 0 + +
< -0.1
Q_
o

CL
-0.3 D
a • rj i
•

In Archaeological Chemistry IV; Allen, R.;

•

-0.6
-0.6 -0.4 -0.2 0.2 0.4 0.6
PRINCIPAL COMPONENT #2
D

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.

'Medium Paste' + 'Fine Paste' o Raw Clay A Sherds

Figure 7. Principal components plot (components Nos. 1 and 2) based on Chinautla-Sacojito data after removal
of seven outliers.
 82 ARCHAEOLOGICAL CHEMISTRY

the e x p e c t e d a l l o m e t r i c size d i m e n s i o n . T h e fine- a n d m e d i u m - p a s t e groups

that w e r e separated o n c o m p o n e n t s N o s . 1 a n d 2 i n the o r i g i n a l data set
w e r e separated p r i m a r i l y o n c o m p o n e n t N o . 1 i n the data set w i t h outliers
r e m o v e d . (Judging from the c o m p o n e n t coefficients i n Table I I I , some t e m -
p e r - r e l a t e d variation m a y still b e on c o m p o n e n t N o . 2.)
O n c e the effect of size has b e e n r e c o g n i z e d , it m i g h t b e useful i n some
cases to r e m o v e its effects a n d reexamine the g r o u p i n g tendencies i n the
data. G r o u p i n g tendencies m a y be e x a m i n e d i n d e p e n d e n t l y of size-related
variation b y c l u s t e r i n g the data b y u s i n g p r i n c i p a l c o m p o n e n t scores, e x c l u -
sive of scores o n the size-related c o m p o n e n t (principal c o m p o n e n t N o . 1 i n
this case). A n o t h e r means of a c c o m p l i s h i n g the same e n d is to regenerate
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

the data from the p r i n c i p a l c o m p o n e n t s a n d t h e i r associated characteristic

roots (eigenvalues), e x c l u d i n g the characteristic root of the t e m p e r - r e l a t e d
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

d i m e n s i o n . G r o u p i n g procedures based o n these two techniques y i e l d i d e n -

tical results.
A f t e r r e m o v a l of size from the data set (outliers r e m o v e d previously),
h i e r a r c h i c a l a n d k-means cluster analyses fail to d i s t i n g u i s h the two texture
groups i n the C h i n a u t l a - S a c o j i t o data. D i s c r i m i n a n t analysis w i t h the size-
c o r r e c t e d data yields nonsignificant test statistics for the effect of group
(tempered vs. n o n t e m p e r e d ) a n d misclassifies 22 of 76 observations.
C o m p l e x e n r i c h m e n t - d i l u t i o n , l i k e s i m p l e p r o p o r t i o n a l d i l u t i o n , is ex-
pressed o n the major axis of variation i n the data, unless heterogeneity i n
the source clay o v e r w h e l m s the effect of t e m p e r i n g . H o w e v e r , i f h e t e r o -
geneity i n the clay o v e r w h e l m s t e m p e r - r e l a t e d v a r i a t i o n , one p r o b a b l y w i l l
not have i d e n t i f i e d subgroups anyway, a n d the p r o b l e m is moot. If, as i n
the present example, two or m o r e major groups are i d e n t i f i e d along w i t h a
smaller set of outliers, the outliers can be r e m o v e d from the data set a n d
the p r i n c i p a l c o m p o n e n t s recalculated to isolate the effect of t e m p e r . A l -
t h o u g h b i m o d a l t e m p e r i n g m a y create two compositional groups i n a data
set of ceramics made f r o m a single clay source, the effect of t e m p e r i n g can
be i d e n t i f i e d analytically a n d r e m o v e d i n o r d e r to reveal the u n d e r l y i n g
homogeneity.
T h i s example shows that, w h e n a single process (potters i n t e r a c t i n g w i t h
a single clay a n d t e m p e r source) is i n v o l v e d , it is possible to isolate t e m p e r -
related effects i n compositional data. S u c h effects are m o r e difficult or e v e n
i m p o s s i b l e to identify w h e n m u l t i p l e c u l t u r a l a n d natural processes are c o n -
f o u n d e d w i t h i n a compositional data set. A s R e a d (64) has p o i n t e d out, the
p r o b l e m of m u l t i p l e u n d e r l y i n g processes also renders statistical theory i n -
applicable.
I n general, one can expect o n l y the most coarse-grained patterns i n a
compositional data set to be r e a d i l y apparent a n d i n t e r p r e t a b l e . I n the o r i g -
i n a l analysis, t e m p e r - r e l a t e d compositional p a t t e r n i n g was largely o b s c u r e d
b y major c h e m i c a l differences b e t w e e n the clay sources a n d the fact that
t e m p e r interacted differently w i t h the two clays r e p r e s e n t e d i n the data,

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. BISHOP & N E F F Compositional Data Analysis 83

w h i l e second stage analysis of the C h i n a u t l a - S a e o j i t o data r e a d i l y r e v e a l e d

such p a t t e r n i n g . T h e p o t e n t i a l for a h i e r a r c h y of processes to create p a t t e r n
i n c o m p o s i t i o n a l data suggests that pattern-recognition techniques s h o u l d
be a p p l i e d h i e r a r c h i c a l l y . S u c h an approach to the present data set involves
a second-stage search for p a t t e r n w i t h i n each of the source-specific subgroups
i d e n t i f i e d i n the first stage of the analysis.

Conclusion
T h i s chapter stresses the n o t i o n of m o d e l i n g as i t pertains to a structure or
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

structures c o n t a i n e d w i t h i n a compositional data matrix a n d as r e v e a l e d or

i m p o s e d b y choice of a l g o r i t h m i c approach. B y u s i n g a generated example,
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

the influence of such factors as outliers, transformations, i n t e r e l e m e n t a l

correlation, choice of resemblance coefficients, g r o u p i n g procedures, a n d
group s u m m a r y evaluation have b e e n discussed. A l l of these factors are
variable w i t h i n the context of specific p r o b l e m f o r m u l a t i o n .
A l t h o u g h the discussion has b e e n r e s t r i c t e d to c e r a m i c systems, the
basic theory holds for the s u m m a r y of analytical data from stone (jade, t u r -
quoise, limestone); metals (native c o p p e r , iron); a n d glass. T h e s e materials
are frequently c o m p l e x m i n e r a l o g i c a l assemblages r e q u i r i n g that the m u l -
t i p l e sources of variation b e u n d e r s t o o d a n d m o d e l e d i n r e l a t i o n to the
p r o b l e m b e i n g investigated. E v e n s i m p l e p r o p o r t i o n a l i t y can be a pervasive
influence i n o b s i d i a n data, as the trace e l e m e n t a l concentrations may vary
i n r e l a t i o n to the silica content i n a single flow, d e p e n d e n t u p o n the stage
of discharge.
T h e concept of m o d e l i n g tends to l i m i t discussions of a best n u m b e r or
group of elements to use i n compositional analysis (cf. 65, 66). W h a t may
be useful for one p r o b l e m m a y not be useful for another. F o r example, at
the global l e v e l of analysis, c h r o m i u m concentration is an i m p o r t a n t r e g i o n -
ally sensitive d i s c r i m i n a t o r of ancient M a y a pottery i n the southwestern
lowlands of C h i a p a s , M e x i c o . Yet, at the i n t r a r e g i o n a l l e v e l of investigation,
the range of variation f r e q u e n t l y confuses rather than contributes to the local
l e v e l of refinement.
M o d e l i n g b y u s i n g the structure c o n t a i n e d w i t h i n a data set is a p r o b l e m -
o r i e n t e d process, a n d as such it is f u n d a m e n t a l l y o p p o s e d to notions of a
static m o d e of group formation. I n the recent archaeometric l i t e r a t u r e , it
was p r o p o s e d that, as a c o n v e n i e n t m e t h o d of c o m m u n i c a t i o n b e t w e e n r e -
searchers, the classification functions d e r i v e d from d i s c r i m i n a n t analysis
c o u l d be t r a n s m i t t e d rather t h a n the actual data (II)! S u c h a proposal seems
to b e insensitive to p r o b l e m orientation, c e r a m i c processes, or statistical
influence.
Statistical m o d e l i n g of compositional data, u s i n g b o t h p a r a m e t r i c a n d
n o n p a r a m e t r i c aspects, m u s t b e c o n s i d e r e d along w i t h p r o b l e m d e f i n i t i o n ,

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 84 ARCHAEOLOGICAL CHEMISTRY

research design, s a m p l i n g , etc., as these factors are part of the investigative

p r o g r a m . A c c o r d i n g l y , what o n first glance may be v i e w e d as a misuse of
some statistical approach m a y have less to do w i t h some specific a l g o r i t h m
than it does w i t h its use i n isolation from the program's objectives. T h i s
isolation too frequently arises f r o m the functional divisions of labor b e t w e e n
the archaeologist a n d scientist. Part of a m e a n i n g f u l collaboration m u s t e n t a i l
consideration of h o w compositional data is to be m o d e l e d w i t h i n the focus
of a specific p r o b l e m . W i t h i n such a context, archaeologically significant
inferences w i l l gradually replace object d e s c r i p t i o n as the m a i n p r o d u c t of
compositional investigations.
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

Acknowledgments
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

I n this p a p e r w e have d r a w n o n analyses c a r r i e d out as part of the M a y a

Jade a n d C e r a m i c s Project, a collaborative p r o g r a m of the M u s e u m of F i n e
A r t s , B o s t o n a n d B r o o k h a v e n N a t i o n a l L a b o r a t o r y d u r i n g 1977-1983. W o r k
at B r o o k h a v e n was c o n d u c t e d u n d e r the auspices of the U . S . D e p a r t m e n t
of E n e r g y . E x p l o r a t i o n into the interface b e t w e e n archaeological objectives,
compositional variation a n d statistical m o d e l i n g is an endeavor of the S m i t h -
sonian A r c h a e o m e t r i c Research C o l l e c t i o n s a n d Records ( S A R C A R ) facility
located at the Smithsonian's C o n s e r v a t i o n A n a l y t i c a l L a b o r a t o r y . N e f f s p a r -
ticipation i n this research is made possible b y a S m i t h s o n i a n I n s t i t u t i o n
M a t e r i a l s A n a l y s i s Postdoctoral F e l l o w s h i p .

References
1. Caley, E . R. J. Chem. Ed. 1949, 26, 242-247.
2. Harbottle, G . In Contexts for Prehistoric Exchange; Earle, T. K . ; Ericson, J.
E . , Eds.; Academic: New York, 1982; pp 13-51.
3. Rands, R. L.; Bishop, R. L . In Models and Methods in Regional Exchange; F r y ,
R. E., Ed.; S A A Papers No. 1; Society for American Archaeology: Washington,
D C , 1980; pp 19-46.
4. Rice, P. In The Ceramics of Kaminaljuyu, Guatemala; R. K. Wetherington, Ed.;
The Pennsylvania State University Monograph Series on Kaminaljuyu: Univer-
sity Park, 1978; pp 401-510
5. Neff, H . P h . D . Thesis, University of California, Santa Barbara, 1984.
6. Bishop, R. L.; Lange, F. W.; Lange, P. C. In The Art and Archaeology of Costa
Rica; Lange, F. W., Ed.; Phillips Exeter Academy through P D Q Printing: U r -
bana, 1987.
7. Earle, T.; Ericson, J. E. In Exchange Systems in Prehistory; Earle, T.; Ericson,
J. E., Eds.; Academic: New York, 1977; pp 3-12.
8. Attas, M.; Fossey, J . M.; Yaffee, L . Paper presented at the 1984 Symposium on
Archaeometry, Washington, D C .
9. Attas, M.; Fossey, J . M.; Yaffee, L . Advances in Computer Archaeology 1985,
2, 1-30.
10. Vitali, V.; Franklin, U . M. Paper presented at the Symposium on Archaeometry,
Washington, D C , 1984.
11. Vitali, V.; Franklin, U . M. J. Arch. Sci. 1986, 13, 161-170.

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 4. B I S H O P & NEFF Compositional Data Analysis 85

12. Thomas, D . H . American Antiquity 1978, 43, 231-244.

13. Hole, B. Annual Review of Anthropology 1980, 9, 217-234.
14. Bishop, R. L.; Rands, R. L.; Holley, G . In Advances in Archaeological Method
and Theory; Schiffer, Μ. Β., Ed.; Academic: New York, 1982; Vol. 5; pp 275-330.
15. Bezdek, J . C. Pattern Recognition with Fuzzy Objective Function Algorithms;
Plenum: New York, 1981.
16. Bishop, R. L . P h . D . Thesis, Southern Illinois University, Carbondale, 1975.
17. Kemeny, J . G.; Snell, J . L. Mathematical Models in the Social Sciences; G i n n -
Blaisdell: Waltham, 1962.
18. Everitt, B . , In Exploring Data Structures; O'Muircheartaigh, C. Α.; Payne, C.;
Wiley and Sons: New York, 1977; Vol 1; pp 100-130.
19. Sneath, P. Η. Α.; Sokal, R. R. Numerical Taxonomy; W. H . Freeman: San
Francisco, 1973.
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

20. Harbottle, G . In Radiochemistry: A Specialist Periodical Report; Newton, G .

W. Α., E d . ; The Chemical Society, Burlington House: London, 1976; pp 33-72.
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

21. Allen, R. O.; Luckenback, A . H.; Holland, C. G . Archaeometry 1975, 17, 69-83.
22. Allen, R. O.; Pennell, S. E . In Archaeological Chemistry-II; Carter, G . F., Ed.;
Advances in Chemistry Series No. 171; American Chemical Society: Washington,
D C , 1978; pp 230-257.
23. Moffat, D . ; Buttler, S. J . Archaeometry 1986, 28, 101-15.
24. Sayre, Ε. V. Brookhaven Procedures for Statistical Analysis of Multivariate
Archaeometric Data; Brookhaven National Laboratory: New York, 1975, u n ­
published.
25. Bishop, R. L . In Models and Methods in Regional Exchange; F r y , R. E., Ed.;
SAA Papers No. 1; Society for American Archaeology: Washington, D C , 1980;
pp 47-66.
26. Mertz, R.; Melson, W.; Levenbach, G . Proceedings of the 18th International
Symposium on Archaeometry and Archaeological Prospection, Bonn, 14-17
March 1978, 1979, pp 580-596.
27. Bishop, R. L . Mesoamerica 1984, 7, 103-111.
28. Davis, J. C. Statistics and Data Analysis in Geology, second edition; Wiley and
Sons: New York, 1986.
29. Orloci, L. J. Ecology 1966, 54, 193-215.
30. N o y - M e i r , I. J . Ecology 1973, 61, 329-341.
31. Klovan, J . E. In Concepts in Geostatistics; McCammon, R. B . , Ed.; Springer-
Verlag: New York, 1975; pp 21-69.
32. Miesch, A . T. Computers and Geosciences 1976, 1, 147-159.
33. F u l l , W. E.; Ehrlich, R.; Klovan, J . E . Computers and Geosciences 1981, 7,
331-342.
34. Hammond, N.; Harbottle, G.; Gazard, T. Archaeometry 1976, 18, 147-168.
35. Underhill, L . G . ; Jacobson, L.; Peisach, M. Paper presented at the 1984 Sym­
posium on Archaeometry, Washington, D C , 1984
36. Anderberg, M. R. Cluster Analysis for Applications; Academic: New York, 1973.
37. Romesburg, H . C. Cluster Analysis for Researchers; Lifetime Learning: B e l ­
mont, California, 1984.
38. Sokal, R. R . Rohlf, F. J. Taxon 1962, 11, 33-40.
;

39. Sneath, P. H. A. Mathematical Geology 1977, 9, 123-144.

40. Everitt, B. S. Biometrics 1979, 35, 169-181.
41. Wilks, S. S. Mathematical Statistics; Wiley and Sons: New York, 1962.
42. Cooley, W. W.; Lohnes, T. R. Multivariate Data Analysis; Wiley and Sons: New
York, 1971.
43. Rubin, J.; Friedman, H. P. A Cluster Analysis and Taxonomy System for Group­
ing and Classifying Data; I B M , Scientific Center: New York, 1967.

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.
 86 ARCHAEOLOGICAL CHEMISTRY

44. MacRae, F. K . Behavioural Science 1971, 16, 423-424.

45. Marriot, F. H. C. Biometrics 1971, 27, 501-514.
46. Bishop, R. L.; Rands, R. L . In Analysis of Fine Paste Ceramics; Sabloff, J . Α.,
E d . ; Memoirs of the Peabody Museum of Archaeology and Ethnology; Peabody
Museum: Cambridge, 1982; Vol. 15, No. 2; pp 283-314.
47. Bishop, R. L.; Harbottle, G.; Sayre, Ε. V. In Analysis of Fine Paste Ceramics;
Sabloff, J. Α., Ed.; Memoirs of the Peabody Museum of Archaeology and E t h ­
nology; Peabody Museum: Cambridge, 1982; Vol. 15, No. 2; pp 272-282.
48. Everitt, B. S. Cluster Analysis; Heinmann Educational: London, 1974.
49. Olin, J . S.; Sayre, Ε. V. In Science and Archaeology; Brill, R. H, Ed.; The M I T
Press: Cambridge, 1971; pp 196-209.
50. Rice, P. In The Ceramics of Kaminaljuyu, Guatemala; R. K. Wetherington, Ed.;
The Pennsylvania State University Monograph Series on Kaminaljuyu: Univer­
Downloaded by UNIV OF PITTSBURGH on June 1, 2014 | http://pubs.acs.org

sity Park, 1978; pp 511-542.

51. Hirst, D . M. Geochimica Cosmochimica Acta 1962, 26, 1147-1187.
Publication Date: July 1, 1989 | doi: 10.1021/ba-1988-0220.ch004

52. Slatt, R. M. Can. J. Earth Sci. 1975, 12, 1346-1361.

53. Slatt, R. M.; Sasseville, D . R. Canadian Mineralogist 1976, 14, 3-15.
54. Carr, C . In For Concordance in Archaeological Analysis: Bridging Data Struc-
ture, Quantitative Technique, and Theory; Carr, C., Ed.; Westport: Kansas City,
1986; pp 18-44.
55. Arnold, D . E . I n The Ceramics of Kaminaljuyu, Guatemala; Wetherington, R.
K., E d . ; The Pennsylvania State University Monograph series on Kaminaljuyu:
University Park, 1978; pp 327-400.
56. Arnold, D . E . In Spatial Organisation of Culture; Hodder, I., Ed.; University
of Pittsburgh: Pittsburgh, 1978; pp 39-60.
57. Arnold, D. E.; Rice, P. M.; Jester, W. Α.; Deutsch, W. H.; Lee, Β. K.; Kirch,
P. V. In The Ceramics of Kaminaljuyu, Guatemala; Wetherington, R. K., Ed.;
The Pennsylvania State University Monograph series on Kaminaljuyu: University
Park, 1978; pp 543-546.
58. Reyment, R. A . In Handbook of Statistics; Krishnaiah, P. R.; Kanal, L. N., E d s . ;
North Holland: N e w York, 1982; Vol. 2; pp 721-745.
59. Reyment, R. A . Mathematical Geology 1985, 17, 591-609.
60. Reyment, R. Α.; Blackith, R. E.; Campbell, N. A . Multivariate Morphometries;
Academic: New York, 1984; 2nd ed.
61. Humphries, J. M.; Bookstein, F. L.; Chernoff, B . ; Smith, G . R.; Elder, R. L.;
Poss, S. G . Systematic Zoology 1981, 30, 293-308.
62. Somers, Κ. M. Systematic Zoology 1986, 35, 359-368.
63. Drexler, J. W.; Rose, W. I., J r . ; Sparks, R. S. J.; Ledbetter, M. T. Quaternary
Research 1980, 13, 327-345.
64. Read, D . In For Concordance in Archaeological Analysis: Bridging Data Struc­
ture, Quantitative Technique, and Theory; Carr, C., E d . ; Westport: Kansas City,
1986; pp 45-86.
65. Widemann, F.; Picon, M.; Asaro, F.; Michel, H. V.; Perlman, I. Archaeometry
1975, 17, 45-59.
66. Winther-Nielsen, M.; Conradsen, K . ; Heydorn, K . ; Mejdahl, V. In Scientific
Studies in Ancient Ceramics; Hughes, M. J., Ed.; Occasional Paper No. 19,
British Museum: London, 1981; pp 85-92.

RECEIVED for review June 11, 1987. ACCEPTED revised manuscript March 1, 1988.

In Archaeological Chemistry IV; Allen, R.;

Advances in Chemistry; American Chemical Society: Washington, DC, 1989.