university alabama press w c mckern and the midwestern taxonomic method nov 2002

The Midwestern Taxonomic Method as an Aid to Archaeological Culture Study... McKern’s system of taxonomic units and the Linnaean biological taxonomy for the Class Mammalia, Order Artioda

M i dw e s t e r n T a x o n o m i c M e t h o d

Stephen Williams, Series Editor

W C McKern and the

Midwestern Taxonomic Method

R L e e L y m a n a n d M i c h a e l J O ’ B r i e n

The University of A l a b a m a Press

Tuscaloosa and London

The University of Alabama Press

Tuscaloosa, Alabama 35487-0380

All rights reserved

Manufactured in the United States of America

Typeface: Janson Text

∞

The paper on which this book is printed meets the minimum requirements of American National Standard for Information Science–Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984.

Library of Congress Cataloging-in-Publication Data

Lyman, R Lee.

W C McKern and the Midwestern Taxonomic Method / R Lee Lyman and Michael J O’Brien.

p cm — (Classics in southeastern archaeology)

Includes historical documents by W C McKern and others.

Includes bibliographical references and index.

ISBN 0-8173-1221-8 (alk paper) — ISBN 0-8173-1222-6 (pbk : alk paper)

1 Indians of North America—Middle West—Antiquities—Classi¤cation.

2 Archaeology—Middle West—Methodology 3 McKern, W C ( Will Carleton), 1892– 4 Middle West—Antiquities 5 Archaeology—United States—History.

I O’Brien, Michael J (Michael John), 1950– II McKern, W C ( Will Carleton), 1892– III Title I V Series.

E78.M67 L96 2003

977′.01—dc21

2002007554 British Library Cataloguing-in-Publication Data available

List of Illustrations vii

Preface and Acknowledgments xi

PART I: W C McKern and the Midwestern Taxonomic Method

2 Taxonomic Classi¤cation and Biological Taxonomy 13

3 Developing the Midwestern Taxonomic Method, 1930–1935 51

4 Subsequent Developments, 1935–1940 101

5 Applications, Comments, and Later Proposals 135

6 The Midwestern Taxonomic Method in Light of Biological Systematics 179

PART I I: Historical Documents

7 Culture Type Classi¤cation for Midwestern North

10 The Problem of Culture Classi¤cation

11 Certain Culture Classi¤cation Problems in Middle

Western Archaeology

Contents

12 Some Assumptions and Implications of the McKern

15 The Midwestern Taxonomic Method as an Aid to

Archaeological Culture Study

1.1 Will Carleton McKern, 1939 6 1.2 Will Carleton McKern, 1925 7 2.1 Symmetrical hierarchical classi¤cation producing eight classes 23 2.2 Asymmetrical hierarchical classi¤cation producing seven classes 23 2.3 Hypothetical ¤ve-level taxonomic classi¤cation of projectile points 24 2.4 Hypothetical taxonomic classi¤cation with unweighted

characters and character states 26 2.5 Twelve entities differing in shape, size, and shading 26 2.6 Unweighted, symmetrical taxonomic classi¤cation 27 2.7 Model of differentia and other characters of two genera and six species 30 2.8 Nested (aggregative) structure of the Linnaean biological taxonomy 35 2.9 Positional-structure hierarchy of descent for four generations

2.10 Hierarchical structures showing phylogeny and the

relationships of higher and lower taxa 382.11 Differences and similarities between a model of phylogenetic

relationships and a model of formal similarity 402.12 Hypothetical phenogram showing the arrangement of nine taxa 452.13 Kinds of characters and their relationship to determination

of phylogenetic history 47 3.1 W C McKern’s system of taxonomic units and the Linnaean

biological taxonomy for the Class Mammalia, Order Artiodactyla 79 3.2 Alignment of a taxonomic classi¤cation of organisms

and a taxonomic classi¤cation of artifacts 91 3.3 Status of a culture trait as determinant, diagnostic, or linked 93 4.1 Thorne Deuel’s midwestern taxonomic method 111

Illustrations

4.2 Units compared on the categorization of traits as diagnostic,

linked, or determinant 114 4.3 Analytical relations within the midwestern taxonomic method 126 4.4 System of units comprising the midwestern taxonomic method 128 4.5 Determinant traits of a phase change 129 5.1 Models of cultural change 151 5.2 System of roots, stems, branches, and phases 161 5.3 Model of pottery evolution 167 5.4 Interrelationships of components, phases, traditions,

2.1 De¤nitions of Classi¤cation, Systematics, and Taxonomy from the

Biological Literature 153.1 Evolution in Names of Units in the Midwestern Taxonomic Method 535.1 Alignment of Archaeological Units Used in the Southwest with

Midwestern Taxonomic Method Units 1615.2 Taxonomy of Southwestern Pottery 1638.1 The Relationship of Taxonomic Terms and Cultural Taxa 202

Tables

Beginning late in the nineteenth century, archaeologists in North Americafound increasing evidence of a diversity of prehistoric cultures Given theiranthropological training and exposure to ethnological theory and ethno-graphic data, it is not surprising that many of the archaeological collec-tions they generated and studied after the turn of the century were called

“culture X” or characterized as representing the “W culture.” There were

no standards or algorithms for specifying when one had a culture distinctfrom or identical to someone else’s culture in an adjoining or distant re-gion As a result, by the late 1920s there was a plethora of prehistoric cul-tures that displayed disparate geographic and chronological distributions.Given the rapid growth in terminology that accompanied the discovery ofthese cultures, it is not surprising that someone would eventually raise ared ®ag because he was getting lost in the classi¤catory confusion.That someone was Will Carleton McKern of the Milwaukee PublicMuseum, who in 1929 began to work on what he thought might comprise

a solution to the problem It became known as the midwestern taxonomicmethod (MTM) At the suggestion of one of his assistants McKern chose

a particular model for his solution: Linnaean biological taxonomy He chosethat model because he wanted the ability to monitor relationships, particu-larly historical and cultural relationships, among the various identi¤ed ar-chaeological cultures His reasoning was simple: Because the Linnaeantaxonomy implied historic and genetic relationships among biological taxa,

a cultural taxonomy no doubt would reveal similar sorts of relationshipsamong cultures As simple as this reasoning was, it belied the rough watersthat lay ahead as McKern and his colleagues tried to put the method intopractice After several ¤ts and starts they abandoned it in favor of a methodbased more on intuition than on methodological rigor

McKern was the chief architect of the MTM, but he did not workPreface and Acknowledgments

alone—a point overlooked for the most part in previous essays on theMTM In particular, Carl E Guthe of the University of Michigan was notonly a major source of good ideas; in his role as chairman of the Commit-tee on State Archaeological Surveys of the National Research Council, heensured that various of the early statements on the method were distrib-uted to a wide audience for comment Thorne Deuel of the University ofChicago helped McKern and Guthe write one of the early statements, and

he applied a version of the method in his dissertation Deuel also wrotetwo signi¤cant but previously unpublished papers on the method James B.Grif¤n and William A Ritchie also made notable contributions to the de-velopment of the method—the former attempting an early application of

it, the latter proposing some critical terminology that became a signi¤cantpart of the ¤nal published version These and other aspects of the history

of the development and use of the MTM have not received the attentionthey deserve One result has been that when modern archaeologists com-ment on the method, they often discuss how people other than McKernapplied it rather than on how the method itself was intended to be used Inaddition, previous commentators on the MTM, ourselves included, havevariously suggested that the method is analogous to one of three ratherparticular methods of biological classi¤cation Such claims, however, havebeen made with little attendant analysis of either the pertinent literature

on biological systematics or the relevant archaeological literature

It is our goal in this volume to ¤ll these voids, but more than that wewant to show that many of the thought processes that McKern and hiscolleagues went through were parallel to, and in some ways anticipated,the thought processes that biologists and paleontologists went through inthe decades following the evolutionary synthesis of the early 1940s Theontological and epistemological issues with which McKern, Guthe, andDeuel wrestled are not unique to archaeology but underlie any attempt toclassify segments of the natural world McKern and his colleagues wereapparently not aware of the debates in biology and paleontology, but themethod they eventually proposed was remarkably similar to one that wouldcause a stir in biology when it was introduced in the late 1950s Thatmethod was phenetics, sometimes referred to as numerical taxonomy Phe-neticists might have been interested in knowing that a handful of archae-ologists several decades earlier had wrestled with such issues as identifyingcharacters and character states and attempting to understand how variouscharacters were linked The method used by that small group of archae-ologists was nowhere near as sophisticated as that developed by the phe-neticists, but the basic approach was the same: Use any and all available

characters to determine relationships among taxa, irrespective of the gin of those characters.

ori-Perhaps not coincidentally, both methods ultimately fell into disuse forthe same reasons: They left time, and by implication genealogy, out of theequation And if anything unites archaeology and biology it is an emphasis

on chronological and genealogical ordering Any classi¤cation that doesnot place primary emphasis on those two facets of the organic world isbound to fail That, perhaps, is the take-home message of our look at theMTM Its architects had the best of intentions when they started out inthe early 1930s, but within a few years even they jettisoned it in favor of aculture classi¤cation built around time In biology, phenetics had its mo-ment in the sun, but it, too, fell prey to other classi¤catory methods thatwere based at least in part on genealogical ordering Time simply plays tooimportant a role in our attempts to understand both culture and organ-isms for it to be left out of the game

Our archival research was aided by a number of people and tions Jay Satter¤eld of the University of Chicago’s Regenstein Library,Special Collections, provided access to the papers and correspondence ofFay-Cooper Cole Pat Burg of the Illinois State Museum provided access

institu-to the papers and correspondence of Thorne Deuel Daniel Barbiero of theNational Research Council arranged for us to use the archives of theCommittee on State Archaeological Surveys Dan Glover sorted throughthe papers of W C McKern archived at the Milwaukee Public Museum.Interlibrary Loan personnel of the University of Missouri-Columbia’sEllis Library were most helpful in obtaining several critical documents.Lyman thanks his relatives ( Jay, Karen, Gracie, Thomas, and Claire) andfriends ( Jim and Sue Scott) for their hospitality when the book was near-ing completion but still lacked certain critical bits of information

We thank Dan Glover for producing the line drawings and Robert D.Leonard, Charles McNutt, and E J O’Brien for reading the manuscript

in its entirety and providing numerous comments on how to improve it.Permission to publish the archived materials was provided by BruceMcMillan ( Illinois State Museum), Patricia Zimmerman ( Illinois StateAcademy of Science), Alex Barber and Susan Otto (Milwaukee Public Mu-seum), Daniel Barbiero ( National Research Council), the Delaw are Ar-chaeological Society, the American Anthropological Association, and theSociety for American Archaeology

PART I

W C McKern and the Midwestern Taxonomic

Method

[T]here is no better method for scientists of one period to bring tolight their own unconscious, or at least undiscussed, presuppositions(which may insidiously undermine all their work) than to study theirown subject in a different period (A J Cain 1958:144)

Early last century Franz Boas (1902:1) indicated that “in the study ofAmerican archaeology we are compelled to apply methods somewhat dif-ferent from those used in the archaeology of the Old World.” Although he

was not clear about why this was so, part of the reason appears to have been

that the then generally accepted time depth of the American cal record was much shallower than that of the Old World By the middle

archaeologi-of the nineteenth century, European paleontologists had found logical evidence that humans had walked the earth alongside large mam-mals that even then were known to have become extinct at the end of thePleistocene (Grayson 1983; Van Riper 1993) No such evidence was thenavailable for the Americas, and the only well-understood and generally ac-cepted evidence of prehistoric people in the New World pointed to a hu-man occu pancy that was only a few thou sand years old (Meltzer 1983,1985; Stewart 1949) That would change in the 1920s and 1930s with thediscovery of human tools in association with extinct bison and mammoth

archaeo-at several localities in eastern New Mexico (Meltzer 1983, 1985, 1991).The problem that arose after the New Mexico discoveries was identi¤ed

by A V Kidder (1936a) as one of ¤lling in the cultural gap between latePleistocene archaeological materials and late prehistoric materials Theproblem was in how best to classify the myriad archaeological cu ltu resthat lay in that large tract of time Kidder addressed this problem in theSouthwest by founding the annual Pecos Conference (Kidder 1927), at

1 Introduction

which chronological issues could be debated The early conferences sulted in a proposed chronological sequence of cultural manifestations forthat region, but the sequence was limited in geographic extent No suchframework existed for the Southeast or Midwest, two regions that werebeginning to witness more and more attention from professional and avo-cational archaeologists (O’Brien and Lyman 1999a, 2001).

re-Will Carleton McKern of the Milwaukee Public Museum addressedthis issue in the early 1930s, producing what eventually became known asthe midwestern taxonomic method (MTM) As we document in later chap-ters, the method attracted considerable attention throughout the 1930s asarchaeologists, primarily those working in the Midwest, worked to re¤nethe method to suit their needs No one was ever completely satis¤ed with

it, and few other than McKern really understood how it operated Unlikethe southwestern classi¤cation, the MTM explicitly excluded time fromconsideration, and it was this feature that contributed to its demise Per-haps because of its short life span, roughly from 1932 to 1940, the MTMhas been treated as little more than a historical curiosity in recent text-books (Fagan 1997) Or, as is more common, it is not mentioned at all (forexample, Sharer and Ashmore 1993; Thomas 1998) We have a differenttake on the matter In our opinion, what McKern attempted to do with theMTM has important lessons for all archaeologists McKern and his col-leagues wrestled with many of the same issues facing archaeologists today.One of these is how to classify archaeological remains in order to analyzethem Archaeologists are not the only natural scientists who classify thephenomena they study, and McKern borrowed a biological model of clas-si¤cation as the template for the MTM

This volume is divided into two parts Part 1 contains six chapters thatexplore select aspects of the MTM The remainder of chapter 1 presents abrief biography of McKern Chapter 2 examines the epistemology and on-tology of taxonomic classi¤cation and describes biological taxonomy Chap-ters 3 and 4 examine the historical development of the MTM based onpublished information and extensive correspondence between McKernand his collaborators During the 1930s and 1940s, several archaeologistsattempted to apply the method in their geographic areas of concern, andthey often published comments on the method, prompting responses fromMcKern Various applications and statements are discussed in chapter 5

In chapter 6 we place the MTM within the context of broader tion systems that attempt to order the natural world

classi¤ca-In part 2 we reprint the several versions of the MTM penned in the1930s and also several published addenda and comments and unpublisheddiscu ssions of the method There is much to be gleaned from reading

these statements, and throughout part 1 we make extensive reference tothem We use several conventions in the following text Papers that arereprinted in part 2 are signi¤ed by a chapter number in brackets within astandard citation (for example, McKern 1932[7]) Throughout part 1 weuse “MPM” to denote the archives of the Milwaukee Public Museum;

“NRC” to denote the archives of the National Research Council, ington, D.C.; “ISM” to denote the archives of the Illinois State Museum,Spring¤eld; and “UC” to denote the Special Collections, Regenstein Li-brary, University of Chicago

Wash-W C MCKERN

There is no obituary for McKern in American Antiquity, the leading

ar-chaeological journal in the United States Instead, a short notice appeared

in the March 1989 issue of the Bulletin of the Society for American

Archae-ology It read simply:

Society Founder Dies

At the age of 96, Will Carleton McKern, one of the founders of theSociety for American Archaeology, died in Waukesha, Wisconsin on

November 20, 1988 He was the ¤rst editor of American Antiquity

from 1935 to 1939, and President of the Society in 1940 Until hisdeath he was Emeritus Director of the Milwaukee Public Museum

A bit longer obituary was published in the American Anthropological

As-sociation’s Anthropology Newsletter in January 1989 (Anonymous 1989),

and McKern’s colleague Alton Fisher (1988) published an even longer one

in The Wisconsin Archeologist, a journal in which McKern regularly

pub-lished in the 1930s and 1940s and for which he served as editor in 1931–32(Overstreet 1999)

McKern (¤gures 1.1 and 1.2) was born in Medicine Lake, Washington,

on July 6, 1892 He seldom used his full name, preferring to be identi¤ed

as W C McKern professionally and to be addressed as “Mac” by hisfriends He earned a B.A in anthropology under A L Kroeber at theUniversity of California, Berkeley, in 1917 He received a fellowship tosupport his studies of California Indians but was inducted into the mili-tary and sent to France in 1918 After serving in World War I, McKernmarried Clara Florence on October 22, 1919 He taught anthropology

at the University of Washington in 1919 and conducted ethnographic(McKern 1922, 1923, 1924) and archaeological (McKern 1929) research

on the island of Tonga in Polynesia as part of the Dominick Expedition of

the Bernice P Bishop Museum in Honolulu from 1920 to 1922 He thenserved as an assistant archaeologist for the Bureau of American Ethnologyfrom 1922 to 1924, working with J Walter Fewkes at Mesa Verde insouthwestern Colorado (Basile 2000; Fisher 1988).

McKern became head of the anthropology department of the kee Public Museum on January 1, 1925, and held that position until No-vember 27, 1943, at which time he became director of the museum Hewas awarded the Wisconsin Archeological Society’s Lapham ResearchMedal in 1930 (Overstreet 1999) and served as president of the AmericanAnthropological Association in 1933 and as president of the Society forAmerican Archaeology in 1940–41 McKern received an honorary doc-torate from Marqu ette University in 1956 “in recognition of his manycontributions to anthropology, Wisconsin, and the City of Milwaukee”

Milwau-Figure 1.1 Will Carleton McKern, 1939 Courtesy of

Milwaukee Public Museum (negative number 423187).

( Fisher 1988:208) He retired from the Milwaukee Public Museum in

1958, having ensured that a new museum building would be constructed

McKern was the founding editor of American Antiquity, serving from

1935 to mid-1939 On assuming the editorship from McKern, Douglas S.Byers (1939:1) noted that “To Mr McKern who has prescribed the for-mula that has brought the [journal] through the trying years of infancy isdue not only credit, but also the heart-felt thanks of the Society As a newEditor attempts timorously to take over his duties he realizes only too wellthe debt that the Society owes its ¤rst Editor who fou nded its editorial

policies Mr McKern has made of American Antiquity a jou rnal that is

above all readable; while keeping the pages open to all who wish to

con-Figure 1.2 Will Carleton McKern, 1925, in

archae-ological ¤eld camp at Green Lake Mounds,

Wiscon-sin Courtesy of Milwaukee Public Museum (negative

number 118255).

tribute he has maintained a standard for contributions that holds the nal high in the rank of scienti¤c publications.”

jour-McKern penned numerous editorials for issues of American Antiquity,

often encouraging readers to send in letters “pertinent to the subject ofarchaeology, factual or controversial, as may re®ect your interests and ex-perience” (McKern 1935:2) If the topic was controversial, McKern warnedthat the correspondent should be prepared to see a response published be-cause “science thrives on criticism, and languishes on an exclusive diet ofindulgence” (McKern 1935:2) He perceived the value of contributionsmade by avocational archaeologists as well as professionals, but he alsorecognized that the former would bene¤t from training by the latter Inhis editorials McKern wrote about the importance of classi¤cation andwhether or not collections generated by nonprofessionals should be pur-chased by museums, noting that a collection was valuable only if “the exactorigin of [the specimens] is positively known to the purchaser of materials,and [further that] the placement of such specimens in a well cataloguedcollection, renders a valuable direct service to all students who may haveoccasion to examine these materials in the course of research” (McKern1936:179) These topics were still being discussed by members of the So-ciety for American Archaeology in the 1990s

TH E MIDWESTERN TAXONOMIC METHOD

Of all his accomplishments, McKern is best remembered as the chief chitect of the MTM, designed to classify archaeological remains from the

ar-upper Midwest This is the reason for the name “midwestern taxonomic method” rather than, say, “archaeological taxonomic method.” Fisher (1997)

presented a revealing paper about the origins of the MTM at the ond Indianapolis Archaeological Conference in 1986, noting that by 1929McKern’s archaeological ¤eldwork in Wisconsin was suf¤cient for him tosuspect that there were relationships among the materials he had exca-vated There was, however, “no comparative system in general use in theMidwest at that time to facilitate analysis of subtle as well as overt culturetraits so as to suggest possible relationships among them” (Fisher 1997:118) Fisher had joined the Milwaukee Public Museum staff early in 1927

Sec-at McKern’s invitSec-ation, and he remained there until September 1933 ing that time, he and McKern often ate lunch together and discussed vari-ous archaeological matters Fisher (1997:119) reported that late in 1929these lunchtime discussions “began to concentrate on how a cultural clas-si¤cation system could be designed to serve the archaeological needs of theWisconsin area.” Because no means was available for ascertaining the age

Dur-of much Dur-of the archaeological material they had collected, they agreed that

“temporal considerations would have to be ignored,” at least until logical information became available They believed that hypothetical evo-lutionary sequences would not suf¤ce as either a classi¤catory or temporalframework and ¤nally decided that “the system that was needed so u r-gently would have to be based on morphological or typological considera-tions alone” (Fisher 1997:119) Fisher by that time had recently completedcourses in biology, and he later stated that he

chrono-had become quite familiar with the taxonomic system of Linnaeus

It was based primarily on relationships of form If that tion system could show morphological relationships between animalforms as diverse as mastodons and earthworms, might it not be pos-sible to show some relationship between the creations of man as dem-onstrated by form or stru ctu re alone? When this su ggestion wasproposed to McKern he was interested but he was not convinced that

classi¤ca-it would work After considerable discussion and thought on the ter, however, he began to test the idea with data he had collected, and

mat-he was pleased to ¤nd that it often was successful This encouragedfurther testing and more successes followed When it became evi-dent that there might be a reasonable prospect of success at designingdifferent levels or degrees of relationship between lithic and bone ar-tifacts, pottery, earthworks, and burials and between complexes ofsuch cultural manifestations, the need to become speci¤c in de¤ningthe various proposed categories of relationship claimed Mac’s atten-tion He worked almost alone on this phase of the development of themethod, bringing the results of his thinking and writing to meevery few days for criticism There was no need to be fussy aboutterminology as long as there was no duplication of the Linnaean no-menclature and each term that was chosen had a reasonable conno-tation of the scope of inclusiveness of the class of relations that wasimplied (Fisher 1997:119–120)

This statement indicates that it was Fisher who suggested to McKern thatthe Linnaean biological taxonomy might serve as a good model for build-ing a system of archaeological classi¤cation McKern may have been ame-nable to such a suggestion because he minored in paleontology during hisundergraduate education (McKern to Julian H Steward, December 15,

1939 [MPM]) Whether it was Fisher’s suggestion or McKern’s idea itially, perhaps later reinforced by Fisher’s suggestion, will likely forever beunknown

in-The proposed levels of relationship mentioned by Fisher would becomethe focus, aspect, phase, pattern, and base units of the ¤nal version of theMTM (McKern 1939[15]) McKern obviously did not duplicate the Lin-naean nomenclature of species, genera, families, and so on Whether hedid in fact “duplicate the scope of inclusiveness of the class of relations”among the units composing the various levels is dif¤cult to assess As wepoint out in chapter 2, the scope of inclusiveness is still debated amongbiological taxonomists Further, the morphological relationships denoted

by the Linnaean nomenclature implied several kinds of relationships tween mastodons and earthworms to biologists after 1859 These multi-ple and not necessarily complementary relationships were recognized byMcKern, but he seldom spoke of them in explicit terms When he men-tioned relationships between his cultural units, it was generally with am-biguous phrasing such as “focus A is related to focus B and both are sub-sumed under aspect 1, but A and B are not related to focus C which issubsumed under aspect 2” and the like

be-DISCUSSIONMcKern was working at a time when explicitly worded explanatory theorywas not part of archaeological discussion (Lyman et al 1997) Rather, as henoted in some of his correspondence, the archaeology of the 1930s had fewstandardized ¤eld and laboratory methods Writing the history of cultureswas, however, an analytical goal that many Americanist archaeologists ac-tively sought throughout the ¤rst six decades of the twentieth century(Lyman et al 1997) They developed and perfected stratigraphic excava-tion, seriation, and interdigitation as ways to measure time’s passage (Ly-man and O’Brien 1999; Lyman et al 1998; O’Brien and Lyman 1999b),and they operated under the assumption that typological similarity re-

®ected the “historical relatedness” ( Willey 1953) of archaeological

phe-nomena (for example, Rouse 1939) They were not explicit about what

histori-cal relatedness meant, but by implication it meant some linkage resulting

from contact and information exchange, often glossed as diffusion or culturation They did not acknowledge the fact that mere historical succes-sion, such as could be demonstrated with a strati¤ed sequence of artifacts,was not necessarily the same as succession along a lineage of informationtransmission Nor were Americanist archaeologists often explicit aboutcultural transmission as the mechanism of heritable continuity within acultural lineage (Lyman 2001)

en-Transmission of information, whether genetic or conceptual, as in

teaching an individual how to decorate pottery, is what results in heritable

continuity—a particular kind of historical continuity—because characters

or character states are replicated Heritable continuity and the netic histories of his cultural units were what McKern in part sought viaapplication of the MTM to archaeological materials, but these were minorgoals His contemporaries did not disagree; they themselves were seekingsuch nonexplicit relationships among cultural units (see chapter 4) With-out the attendant explicit development of an explanatory theory that couldaccount for various sorts of relationships, discussion of the MTM reduced

phyloge-to endless squabbles over the method itself rather than what its resultsmight signify Those discussions re®ected the “typical methodologizing

of the period” (Dunnell 2000:371)

McKern was explicit and vocal about the purpose of the method andwhat its intended results would accomplish First, the MTM was intended

to produce a terminology for a set of classi¤catory units that would hance communication between archaeologists working in different areas.Second, it was supposed to facilitate comparative analyses and thus helparchaeologists decipher culture history, particularly “cultural af¤nities”(McKern 1939[15]:303) Ignoring the ambiguity of cultural af¤nities,McKern’s contemporaries protested that the method itself was ®awed be-cause it ignored critical variables, particularly time and geographic space,and was qualitative rather than quantitative What McKern’s contempo-raries displayed by such arguments was the lack of agreement on the pur-poses and analytical goals of the MTM In turn, that lack of agreementre®ected the ambiguity of cultural af¤nities If they could not agree on

en-what those were, then how could they possibly agree on a method

suppos-edly aimed at ¤nding them?

Gordon Willey, a graduate student in the 1930s when the MTM wasunder discussion, recently described what he and James A Ford took to besome of the central problems with the MTM:

Jim was uneasy with the way many eastern archaeologists used thevarious categories of cultural af¤liation, and I was too As an example,should the Louisiana Troyville period and complex be grouped un-der the same classi¤catory category as the obviously Hopewellian-related Marksville period and complex? Troyville pottery had clearantecedents in Marksville; in other ways, however, as in the presence

of what appeared to be temple mounds, Troyville’s relationships peared to go in other taxonomic directions In our opinion the Mid-western taxonomy ran too fast toward conclusions, building up,which [James B.] Grif¤n had done [in 1935], the big picture of cul-tural relationships for the whole of the East before we had local situ-

ap-ations, especially local chronologies and sequences of developmentworked out ( Willey 1999:x)

McKern, of course, thought he had the algorithm for ¤nding cultural

af-¤nities: Build a cultural classi¤cation modeled on the Linnaean taxonomy,and those af¤nities would be self-evident The af¤nities supposedly re-

®ected by the Linnaean taxonomy in the 1920s and 1930s were debatable,

so it is little wonder that McKern’s effort failed Because it was the naean taxonomy that served as the model for the MTM, we begin our dis-cussion with a consideration of taxonomic classi¤cation in general and bio-logical classi¤cation in particular

Lin-To know is one of the fundamental human appetites and classi¤cation

is the method whereby apparent ®ux and chaos of the external worldare reduced to quiescence and unity and become the object of intellec-tual contemplation ( W R Thompson 1952:2)

The conceptual structure of what became known as the midwestern nomic method originated with Swedish physician and botanist Carl VonLinné, better known as Carolus Linnaeus Understanding the Linnaeanbiological taxonomy—how it was built and what it might signify—is criti-cal to grasping why the MTM took the form it did As implied in chap-ter 1, understanding why the MTM failed to accomplish the tasks it wasintended to necessitates understanding the ontology and epistemology ofthe Linnaean taxonomy and how Darwin’s (1859) ideas in®uenced inter-pretations of that taxonomy Because the Linnaean biological taxonomy ishierarchical in structure, to make sense of the MTM requires a basic un-derstanding of classi¤cation in general and taxonomic classi¤cation in par-ticular

taxo-CLASSIFICATION

Although we suspect most people would agree that one purpose of si¤cation is to place similar specimens together and then separate themfrom dissimilar specimens, the terminology of classi¤cation is diverse We

clas-initially use category and its derivatives as a generic term for the results and

processes of sorting phenomena into piles such that like goes with like

2 Taxonomic Classi¤cation and

Biological Taxonomy

TerminologyCategorization has several purposes, not the least of which is to simplifyvariation by reducing it to a small, manageable number of kinds more eas-ily discussed than is each individual specimen Categorization occurs invirtually all endeavors A librarian must decide if a newly published book

is a work of ¤ction, a work of history, or a work of historical ¤ction tronomers must decide if a newly discovered celestial body is a star, aplanet, a moon, an asteroid, or something else Pedologists must decidewhich category of soil occurs in a particular area The systems, processes,and results of categorization are referred to by a plethora of terms thatoften have no common meaning, even within a discipline For example,classi¤cation, systematics, and taxonomy are some of the most discussedterms in the biological literature Sometimes any two of these might bede¤ned as synonyms Borgmeier (1957:53), for example, equated systemat-ics and taxonomy as “that branch of biological science which explores theorder existing in the Plant and Animal Kingdoms and represents it bymeans of a system of concepts (the categories) [ It is] the science of or-der a pure science of relations [based only on] a simple juxtaposition ofdifferent conditions of form.” Not all biologists equate systematics andtaxonomy, but a sample of de¤nitions for the terms given by biologists(Table 2.1) shows that they also do not agree on the de¤nitions Standarddictionaries do not clarify things because their included de¤nitions pre-sent ideals rather than the meaning of the terms as they are used in par-

As-ticular real-world situations Webster’s Third New International Dictionary (1993) de¤nes classi¤cation not only as a verb—“the act or a method of

classifying,” where classifying comprises “distributing into groups” and

“grouping or aggregating in classes that have systematic relations, usuallyfounded on common properties or characters”—but also as a noun—“theresults of classifying: a system of classes or groups or a systematic division

of a series of related phenomena.” Note that the process of the ¤rst tion demands the categories of the second de¤nition This same dictionary

de¤ni-de¤nes taxonomy as “the study of the general principles of scienti¤c

clas-si¤cation: systematics,” which itself is de¤ned as “the science of tion, usually the classi¤cation and study of organisms with regard to theirnatural relationships: taxonomy.” De¤nitions of terms we present in thenext several paragraphs are those we use in the remainder of the book.Classi¤cation involves the creation of new units and the modi¤cationand revision of old units by stipulating the necessary and suf¤cient condi-tions for membership within a unit (Dunnell 1971) The necessary and

classi¤ca-from the Biological Literature

proper-— “the grouping of species and higher categoriesproper-—the actual building ofclassi¤ed systems” (Blackwelder and Boyden 1952:26)

— “zoological classi¤cation is the ordering of animals into groups (or sets)

on the basis of their relationships, that is, of associations by contiguity, larity, or both” (Simpson 1961:9)

simi-— “the product of the activity of the taxonomist” (Mayr 1969:4); “the limitation, ordering, and ranking of taxa” (Mayr 1969:400)

de-— “the name of a logical activity, that of ordering concepts into classes(categories in the sense of biologists)” (Grif¤ths 1974:90)

— “the ordering or arrangement of objects into groups or sets on the basis

of their relationships” (Sokal 1974:1116)

— “arrangement of organisms into hierarchic groups; modern biological

classi¤cations are Linnaean and classify organisms into species, genus, family,

order, class, phylum, kingdom, and certain intermediate categoric levels”(Ridley 1993:632)

— “the tabular or hierarchical end result of [taxonomy]” (Padian 1999:357)

systematics (from the Greek word systema)

— “the entire ¤eld dealing with kinds of animals, their distinction,

classi¤cation, and evolution” (Blackwelder and Boyden 1952:26)

— “the scienti¤c study of the kinds and diversity of organisms and of anyand all relationships among them” (Simpson 1961:7)

— “the scienti¤c study of the kinds and diversity of organisms and of anyand all relationships among them” (Colless 1967:6, after Simpson 1961)

— “the science dealing with the diversity of organisms” (Mayr 1969:413)

— “that branch of science which deals with the systematic structure of theworld” (Grif¤ths 1974:89-90)

— “a near synonym of taxonomy” (Ridley 1993:641)

— “the philosophy of organizing nature” (Padian 1999:357)

taxonomy (from the Greek words taxis, meaning arrangement, and nomos,

meaning law)

— “the science of arranging the myriad forms of life” (Simpson 1945:1)

suf¤cient criteria for membership are the de¤nitive criteria of a unit They

de¤ne the members of the unit and at least partially describe those bers; thus it is important to emphasize that other characters and character

mem-states are simply descriptive and not de¤nitive of a unit A unit is a

concep-tual entity that serves as a standard of measurement (Lyman et al 1997;O’Brien and Lyman 2000) A centimeter is a unit constructed explicitly tomeasure linear distance; the degrees on a compass are units constructedexplicitly to measure direction or orientation

As conceptual entities, units must be explicitly de¤ned to be useful formeasuring (characterizing, describing, categorizing, de¤ning) phenomena.Units can be speci¤ed at any scale Phenomena to be classi¤ed can com-prise discrete objects such as projectile points or organisms; they may

comprise characters (attributes) of discrete objects, such as the bits of

tem-per in pottery or the genes in organisms; or they may comprise sets ofdiscrete objects, such as aggregates of tools variously termed assemblages

or tool kits, or populations of organisms variously termed faunas, ®oras, orcommunities Phenomena are classi¤ed on the basis of the characters theydisplay, such as size, shape, color, frequency, and material Characters used

to classify phenomena occur at a ¤ner, less-inclusive scale than the nomena themselves They are mutually exclusive properties of phenomena,and each character can take any of several values, which are termed char-acter states The character weight, for example, can have the states 1–5

phe-(continued )

— “the descriptive, discriminatory, and nomenclatural phase dealing mostlywith species and lower categories, the data accumulated about them, andtheir names” (Blackwelder and Boyden 1952:26)

— “the general process of sorting things out into groups containing things

of the same kind and of establishing a hierarchy between such groups”(Thompson 1952:3)

— “the theoretical study of classi¤cation, including its bases, principles, cedures, and rules” (Simpson 1961:11)

pro-— “the design and construction of classi¤catory systems” (Colless 1967:6)

— “the theory and practice of classifying organisms” (Mayr 1969:2, 413)

— synonymous with systematics (Grif¤ths 1974)

— “the theoretical study of classi¤cation, including its bases, principles, cedures, and rules” (Sokal 1974:1116)

pro-— “theory and practice of biological classi¤cation” (Ridley 1993:641)

— “the use of sets of organic data guided by systematic principles, whichsort of sets will differ among taxa” (Padian 1999:357)

grams, 6–10 grams, 11–15 grams, and so on; the character length can havethe states 1–3 cm, 4–6 cm, 7–9 cm, and so on.

Taxonomy concerns theories of classi¤cation, their bases, principles,

procedures, and rules (Simpson 1961) Systematics is the study of diversity

of the phenomena of interest, irrespective of the scale or kind of ena, and involves sorting that diversity into sets such that like goes withlike according to some principle of sorting The goal of systematics is toclassify phenomena into sets of individuals that are in some sense similar.Each set should be internally homogeneous such that within-group varia-tion is analytically meaningless and between-group variation meaningful,

phenom-where a group is an empirical unit comprising one or more specimens.

Similar phenomena are often, but not always, conceived of as being not

only formally similar but similar in other ways as well Af¤nity refers to

the particular relation of analytical interest between formally similar mens within a group or between groups of formally dissimilar specimens.Multiple kinds of things may be af¤nes because they are close in form, intime, in function, in ancestry, or in terms of something else Measuring aparticular kind of af¤nity is the ultimate goal of classi¤cation, irrespective

speci-of discipline, and thus speci¤cation speci-of that af¤nity comes from some ciple of taxonomy Classi¤cation is generally problem oriented; it is gearedtoward some larger goal than simply measuring how similar particular

prin-phenomena are to one another The problem takes the general form: Why

are phenomena A and B similar to one another (what is their af¤nity)whereas they are dissimilar to phenomenon C? Answering such why ques-tions demands an explanatory theory, or taxonomic principles

Ontological FrameworkArtifacts are classi¤ed so that we can conduct analytical work There arethree important implications of that goal First, a set of phenomena can beclassi¤ed in a virtually in¤nite number of ways There is no assumptionthat natural groups exist, and in this sense all classi¤cation is arbitrary

This does not mean that classi¤cation is capricious or willy-nilly; rather, it

means that specimens are sorted into groups for some analytical purpose,and as the purpose varies, so, too, will the groups and the units used tocategorize them Second, the analytical validity of the units produced byclassi¤cation must be testable Do the units measure the kind of af¤nitysought? Third, how do we know which characters we should measure and

how we should measure them? There must be some theory—explanatory

statements—that guides analysis, because it is theory (taxonomic ples) and its derivative propositions that suggest which characters are rele-vant and at what scale they should be measured The theory one chooses

princi-to follow depends in part on the particular analytical problem one seeks princi-tosolve Problems concerning change in culture over time demand someform of evolutionary theory; problems concerning how artifacts work orfunction demand theories of mechanics, engineering, and physics.Different theories rest on one of two different ontologies In the fourthcentury b.c., Aristotle sought the underlying essence—the essential charac-teristics—of each kind of organism (Hull 1965) In the twentieth cen-

tury his ontology came to be known as essentialism, or typological thinking (Mayr 1959) Typological thinking holds that types are real and ¤xed—a

statistical average comprises an essence—and variability between viduals within a kind has no analytical importance Essentialism in®u-enced all classi¤cations until Charles Darwin (1859) proposed an alterna-

indi-tive, today termed materialism, or population thinking Darwin focused on

the uniqueness of phenomena Although the basic form of individualswithin a set of similar phenomena can be captured by, say, a statistical av-erage, Darwin realized that such measures of central tendency are abstrac-tions and in no sense real (Padian 1999)

Either the ontology of essentialism or that of materialism underpins allclassi¤cations (Dunnell 1980; Mayr 1959; O’Brien and Lyman 2000, 2002).This can lead to a misunderstanding of the meaning of a particular clas-si¤cation, but it does not mean that one ontology is always preferred overthe other Essentialism is advantageous when we want to predict how kinds

of things interact regardless of their positions in time and space The riodic table is founded in essentialism, and it serves its intended purposewell However, when history is the focus of study, materialism is preferredbecause each particular event is unique in terms of its potentially criti-cal characters, despite the fact that we can construct a classi¤cation thatplaces all international con®icts in one group, plagues in another group,and droughts in yet another group Materialism holds that things are con-stantly in the process of becoming something else In biological evolution,

pe-a populpe-ation of interbreeding orgpe-anisms representing pe-a species cpe-an (butmight not) constantly evolve via mutation and differential reproductioninto another species Each species is a spatiotemporally unique entity and,unlike a chemical element, does not exist forever

Under the ontology of materialism, there is a parallel between the logical evolution of a lineage of multiple, temporally successive species andthe cultural evolution of a lineage of multiple, temporally successive cul-tures, tool traditions, and the like As we will see, McKern was hesitant tosay it explicitly, but it is precisely the evolutionary (phylogenetic and ge-nealogical) af¤nities of the cultural units constructed using the MTMthat he thought might be revealed

bio-Epistemological Issues of SystematicsBecause theory is the source of our ideas on causes of af¤nity, it has to bethe ¤nal arbiter of which units are applicable for which kinds of analyticaljobs Theory and an analytical problem dictate which characters out ofthe almost in¤nite number that could be selected are actually chosen bythe analyst for measurement, and theory in particular speci¤es the values(character states) of those characters The characters and values chosen are

the units used to construct types, and as such they are conceptual, or

idea-tional, units (Dunnell 1971, 1986) The specimens we classify are

pheno-menological, or empirical, units Ideational units, or classes (Lyman et al.

1997; O’Brien and Lyman 1999b, 2000), can be descriptional units, usedmerely to characterize or describe a property or a thing, or they can betheoretical units, which are created for speci¤c analytical purposes Forexample, in light of a proposed causal relation between function and edgeangles of stone tools, edge-angle units such as 1–30°, 31–60°, and 61–90°

could be constructed as theoretical units A theoretical unit is an ideational

unit that has explanatory signi¤cance speci¤cally because of its theoreticalrelevance for the solution of an analytical problem

Ideational units can be constructed or de¤ned in either of two ways An

intensional de¤nition comprises the necessary and suf¤cient conditions for

membership in a unit; it explicitly lists the de¤nitive characters that a nomenon must display in order to be identi¤ed as a member of the unit(Buck and Hull 1966; Dunnell 1971) Importantly, there is no necessaryreference to a particular set of real, empirical specimens when the unit isconstructed other than to specify that, say, projectile points rather thanpottery comprise the phenomena to be classi¤ed The basic procedure is

phe-to write a set of intensional de¤nitions and then phe-to identify each specimen

as a member of a particular class

An extensional de¤nition also comprises the necessary and suf¤cient

con-ditions for membership in a unit and is derived by enumerating selectedcharacters shared by the unit’s members That is, the de¤nition is based onobserved characters of the existing members of a unit As Buck and Hull(1966:105) put it with respect to biological taxa, “a taxon is thought of assimply the collection of its members,” and a taxon’s name “is de¤ned ex-tensionally rather than intensionally.” The de¤nitive characters of exten-sionally de¤ned units are not theoretically informed in any explicit man-ner because the group of specimens was formed by some murky processprior to the speci¤cation of the (extensional) de¤nition One common re-sult of extensionally derived de¤nitions is the con®ation of de¤nitive char-acters and descriptive characters Most types traditionally employed in ar-

chaeology are extensionally de¤ned units formed when an analyst vides a collection of artifacts into smaller piles based on perceived simi-larities and differences A summary of the central tendencies of the mem-bers of each pile, or a statement on the normal appearance of specimens ineach pile, comprises the de¤nitive criteria of a type Because the de¤ni-tions depend entirely on the specimens examined, we cannot know if suchextensionally de¤ned units are comparable in terms of the kind of af¤nity

subdi-we hope they measure

Taxonomic Classi¤cationPaleontologist George Gaylord Simpson (1945:14) argued that “any prac-tical system of classifying a large number of things involves a hierar-chy by which minor units are progressively gathered into groups of in-creasingly greater scope.” Although this begs the question of practical forwhat purpose, it may re®ect a tendency of the human mind to generalize

at successively more inclusive levels as an ef¤cient information and-retrieval system (Mayr 1981, 1995), particularly with respect to com-plex phenomena such as organisms (Simon 1962) All German shepherdsare dogs, but not all dogs are German shepherds; all dogs are mammals, butnot all mammals are dogs These are commonsensical observations, andthe resulting taxonomic classi¤cation provides a useful communicationsystem as well as a readily understood framework for organizing such ob-servations More important, a taxonomy generally not only means a hier-archical arrangement of units—a particular architecture—but also carries

storage-an evolutionary implication This implication originates in the modernconnotation of the Linnaean biological taxonomy, as we discuss below We

use the term hierarchical classi¤cation or taxonomic classi¤cation to denote a

particular structure of classi¤cation without any such connotation

A taxonomic classi¤cation sometimes is referred to simply as a taxonomy.Such a classi¤cation comprises a hierarchical arrangement of nested setssuch that less-inclusive, lower-rank units are included in more-inclusive,higher-rank units ( Valentine and May 1996) The hierarchy is a system of

units of various ranks, each rank comprising what we will call a taxon.

Taxa of a particular rank are symbolized as Tj; each particular taxon prises one or more taxa of rank Tj−1 and is itself a member of rank Tj +1.Multiple taxa of rank Tj are parallel taxa, taxa of rank Tj +1 are superiortaxa, and taxa of rank Tj−1 are subordinate taxa relative to taxa of rank Tj

com-In nested taxonomies no two taxa of rank Tj have any Tj−1 member incommon The specimens comprising the lowest, least-inclusive rank are

of rank zero and are themselves not a taxon but a population(s), or gate, of individuals Populations are the “operational units” of classi¤ca-

aggre-tion (Buck and Hull 1966:98) The lowest-ranking taxa are called nal taxa Taxa of rank Tj are said to be nested within taxa of ranks Tj +n

termi-because the latter comprise all members of lower ranks.

Several important properties attend the nestedness of biological taxa in

a hierarchy First, although it sometimes is said that empty taxa are notpossible (Kay 1971)—each taxon must have at least one empirical mem-ber—we ¤nd this untenable because taxa are classes and thus their mem-bership can be zero Knowing which taxa/classes have members and which

do not could be important in some analyses (see below) Second, as noted

in the preceding paragraph, no parallel taxa Tj can include the same ber taxon Tj−1 There is no partial overlap between parallel taxa In ex-treme form, this attends the fact that a taxon of rank Tj is a member of ataxon of rank Tj +1, but the reverse cannot be true (Buck and Hull 1966).Third, not all terminal taxa will be at the same level This means that ter-minal taxa will not always be distinguished from each other in like ways

mem-or to similar degrees Fourth, nestedness is a result of the fact that thede¤nitive criteria of a Tj taxon are also present among its member Tj−1

taxa In particular, superior classes have fewer and/or more-general tive criteria, whereas subordinate classes have more and/or more-speci¤cde¤nitive criteria

de¤ni-There are two methods of building a taxonomic classi¤cation One is towork from the top down The classi¤er starts with some speci¤ed ¤eld ofphenomena and then divides those phenomena into successively more in-ternally homogeneous groups Often this involves dividing a taxon intotwo or more subordinate groups on the basis of two or more characters.Terminal taxa are attained when no further divisions are made The order

in which characters and character states are chosen in®uences the nature

of the terminal taxa Many identi¤cation keys follow a top-down protocol,

which results in the identi¤cation of individual specimens as members of

particular taxa The alternative means of building a taxonomic tion is to work from the bottom up Individuals are grouped on the ba-sis of numerous similarities to form the lowest-rank taxa; those taxa aregrouped to form the next higher-rank taxa on the basis of fewer and/ormore general similarities; and so on until all taxa are grouped into onetaxon

classi¤ca-The results of top-down and bottom-up classi¤cation procedures can

be quite similar, but their epistemologies are obviously different ences in their ontologies are more subtle Both procedures presume an un-derlying hierarchical structure to the phenomena being classi¤ed Thesubtle difference is found in the argument that a top-down procedure as-sumes the classi¤cation will produce the taxa; a bottom-up procedure as-

Differ-sumes the taxa will produce the classi¤cation (Mayr 1969) Here is wheretheory must be called on to account for which procedure is more valid.Some biologists argue that the bottom-up procedure is preferable becausetaxa—in particular species—are real biological units and thus are thebuilding blocks that give the classi¤cation This is a slippery issue, which

we return to below

Architecture of Taxonomic Classi¤cations

A taxonomic classi¤cation can take one of several forms It may be metrical, such as in Figure 2.1, where each member of a taxon is equivalent

sym-to every other member of the taxon in terms of the number of de¤nitivecharacters Thus the least-inclusive, or terminal, taxa—F1aA, F1aB,F1bC, and so on—are de¤ned by three characters Let’s say that the ¤eld

of phenomena to be classi¤ed comprises pottery; thus the ¤rst-place acter (F) is not de¤nitive of the classes within the ¤eld In level II, thesecond place (Arabic) number denotes temper (1 = shell, 2 = grit) In levelIII, the third-place ( lowercase) letter designates vessel-wall thickness (a =0.4–0.5 cm, b = 0.5–0.6 cm, and so on) Finally, in level I V, the fourth-place (uppercase) letter designates surface treatment (A = cord marked, B

char-= stamped, C char-= plain, and so on) All taxa making up each level have thesame number of de¤nitive characters But note as well that there is no pos-sibility of having a grit-tempered vessel with a wall thickness of 0.4–0.5

cm; that is, there is no class F2a Thus the kinds of characters de¤ning a

particular class re®ect both the order of character inclusion constructed

by the analyst and a suspicion regarding the signi¤cance of a particularcharacter or character state for classi¤cation purposes In Figure 2.1, char-acter states a and b of level III are deemed important to class F1 of level

II, whereas character states c and d are deemed important to class F2 oflevel II Here the de¤nitive characters and character states are said to beweighted

Because characters and character states can be weighted, a taxonomicclassi¤cation may be nonsymmetrical, as in Figure 2.2 A nonsymmetricalhierarchy is created by selective omission of certain characters from vari-ous taxa Thus, in Figure 2.2, among the terminal taxa comprising level

I V, one is de¤ned by two characters (F1), another by three characters(F2a), and ¤ve by four characters The level I V taxa not only have variousnumbers of de¤nitive characters, two taxa (F1, F2a) have the same de¤ni-tive characters in level I V that they do in the more-inclusive level III, andone taxon (F1) has the same de¤nitive characters in level II that it does inlevels III and I V Thus, taxon F1 is de¤ned by the same set of characters

at three levels in the hierarchy, and taxon F2A is de¤ned by the same set of

characters at two levels of the hierarchy Taxa within each of the two inclusive levels—III and I V—are not all equivalently different from oneanother in terms of the number of their de¤nitive characters This canresult in a lack of unit comparability There are other, potentially moreserious, problems.

least-Figure 2.3 shows an example of a ¤ve-level hierarchical classi¤cation of

a ¤ctitious collection of projectile points As in Figure 2.2, the three taxathat occur in level V are each de¤ned by four characters and are thus

Figure 2.2 An asymmetrical hierarchical classi¤cation producing seven classes Note that ¤ve of the classes (taxa) in level I V are distinguished by four characters, one class by three characters, and one class by two characters Therefore, each class in either level III or level I V is not equally distinct from every other class in that level.

Figure 2.1 A symmetrical hierarchical classi¤cation producing eight classes Note that (1) each class in a particular level /taxon is equally distinct from every other class in that level in terms of the number of de¤nitive characters and (2) the classes (taxa) in level I V are nested within classes in level III, those in level III are nested within classes in level II, and those in level II are nested within classes in level I.

equally different from each other in terms of the number of de¤nitivecharacters The six other taxa distinguished in level I V are equally differ-ent from each other in these terms but are not as different from each other

as those in level V in terms of the number of de¤nitive criteria Thus allterminal taxa are not equivalently different from each other Note that in

Figure 2.3 A hypothetical ¤ve-level taxonomic classi¤cation of projectile points The leftmost six units are each de¤ned by three characters; the rightmost three units are each de¤ned by four characters The exclusion of a fourth de¤nitive charac- ter for the six units on the left indicates either that base shape does not vary or is analytically insigni¤cant, but it is unclear which applies or if both apply Identifying specimens as members of a particular unit must consider character levels II–V in that order, as reversing the order of levels II and V would signi¤cantly alter unit de¤nitions.

Figure 2.3 the particular weighting of characters re®ected by the order inwhich they are considered in®uences the nature of the resulting units.The character states signi¤ed by c–f concern the character base shape.The problem is that they are distributed across both levels I V and V be-cause stem length is deemed more important for de¤ning straight-stempoints than it is for de¤ning constricted-stem points That’s why stemlength is listed in level I V prior to base shape in level V among members

of taxon B∆’ (level III) but is not considered in taxon B∆ Differences tween units may comprise both differences in the number of de¤nitivecharacters and the kinds of characters or character states making up theclass de¤nition How might such a dif¤culty be avoided?

be-Note that in Figure 2.1 there is no possibility of a taxon F2bC, though such a taxon could be accommodated by adding a third branch,F2b, to the F2 taxon There is yet a simpler way to contend with this, away that in fact reveals potentially important information regarding for-mal variation within a sample A hierarchical classi¤cation need not beasymmetrical nor weight characters An example is given in Figure 2.4.Here, unlike in ¤gures 2.1–2.3, all possible character combinations are in-cluded because each character is added at only one level and each characterstate possible is included under the next-higher taxon The result is thatthe set of terminal taxa listed in level I V includes all possible combinations

al-of character states—all possible terminal taxa are rendered as classes haps only some of those will have empirical specimens present in a givencollection In such a case, knowing which classes comprise empty mor-phospace—classes without empirical members—might represent impor-tant information for some analyses (for example, Gould 1991)

Per-An example may help clarify some of the preceding points Figure 2.5shows twelve phenomena to be classi¤ed Some of them are similar on thebasis of shape and might be grouped on that basis; another combination ofgroups could be derived based on size; yet a third combination would re-sult from the presence or absence of shading Borrowing this examplefrom philosopher Karl Popper (1968:421), entomologist G C D Grif¤ths

(1974:89) stated that “things are similar in different respects; any two things

which are similar from one point of view may be dissimilar in another;similarity in general presupposes the adoption of a point of view or inter-est; and some similarities will strike us if we are interested in one problem,others if we are interested in another problem.” Thus, “many different hi-erarchical arrangements can be made even of these simple geometrical ¤g-ures, and no one such arrangement can adequately represent the overlap-ping (non-hierarchical) distribution of similarities” (Grif¤ths 1974:89) In