You are on page 1of 20

Oceania Publications, University of Sydney

The Use and Abuse of Numerical Taxonomy in Archaeology


Author(s): David H. Thomas
Source: Archaeology & Physical Anthropology in Oceania, Vol. 7, No. 1 (Apr., 1972), pp. 31-49
Published by: Wiley on behalf of Oceania Publications, University of Sydney
Stable URL: http://www.jstor.org/stable/40386153 .
Accessed: 27/06/2014 18:53

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley and Oceania Publications, University of Sydney are collaborating with JSTOR to digitize, preserve and
extend access to Archaeology &Physical Anthropology in Oceania.

http://www.jstor.org

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
THE USE AND ABUSE OF NUMERICAL TAXONOMY IN
ARCHAEOLOGY1

By David H. Thomas*

Abstract
TAXONOMY is examinedin relationto moreorthodoxarchaeo-
logical theory. Althoughtaximetricsfail to providea panacea forall ills of
archaeologicaltaxonomy,phenetictheoryand computerscan equip the archaeologist
with an operational tool-kit. When outfittedwith the weaponry of modern
systematics, the archaeologistis in a betterpositionto attackthe lingering problems
of artifactclassificationand typology. There is little conflictbetweennumerical
taxonomy and long-standing archaeological theory. (Typology, Numerical
Taxonomy,ArchaeologicalTheory,QuantitativeMethodsin Archaeology.)
" Once
upon a time therewas an ugly little caterpillar. All the othersmall
animalsstruttedaround,preeningtheircolorfulfeathersor showingofftheirglittering
coats,whilethelittlecaterpillarhid and feltashamed. Then one day he made up his
mindhe wouldnot restuntilhe changedhimselfinto the mostbeautifulcaterpillar
in the world. He struggled,he puffed,he almostbursthimselftrying,but he did
'
succeed. ' Look at me ', he shouted, I am trulya lovely caterpillar.1But the
otheranimalssnickeredand laughedat himbehindhis back. Finallya wiseold owl,
'
who had been watchingfromabove, said to the deflatedlittle caterpillar: The
othersare notlaughingat you becauseyou are notbeautiful. Don't you knowthere
is no such thing as a beautifulcaterpillar? You have turned yourselfinto a
"
butterfly/ (JohnG. Kemeny,1959 : ix.)
Like Kemeny's caterpillars,archaeologistswho opt for taxonomicelegance
" "
oftenexposethemselvesto seriousabuse frommore practical-minded colleagues.
To paraphraseKemeny,no matterhow hard the archaeologisttriesto discoverthe
scientificprinciplesof classification, he can never do so. If he were to succeed,
"
peoplewillcall hima taxonomist", orworse,accusehimofinterest in " classification
forits own sake ". Thereis an unconsciousarroganceamongacademicswho decry
cross-disciplinaryfertilizationas only productiveof sterilehybrids. One runs no

1The bulk of thispaper was writtenwhileI was a graduatestudentat the University of


California,Davis. The primeideas of the paperare thoseof theorists frombotharchaeological
and biologicalrealmsof systematics.The title,a partialbow to J. O. Brew,shouldindicate
that the tone of the workis mediatingratherthan polarizing. ProfessorD. L. True's many
discussionsofarchaeological
theoryillustratedthenecessityofprobingamongthefoundations of
archaeologicalthought. ProfessorM. A. Baumhoff's seminarson numericaltaxonomyproved
the validityof moderncomputational and statisticalapproaches. W. G. Davis, EdwinA. Cook
and PeterWhiteread and extensively criticizedearlierdraftsof the text. A NationalScience
FoundationTraineeship permittedmethetimeto conducttheabovestudy. I also thankmywife,
Trudy,whosepercipient aid proved,as always,unfailing.I am to be held responsible forthe
shortcomings in the presentpaper.
♦ CityCollegeof the City Universityof New York.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
32 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY
smallriskin proposinga reflective pause in the lemming-like rushesof sometoward
the frontiersof quantificationand objectivity. Recent warnings(e.g. Hodson,
Sneath and Doran, 1966 ; Clarke,1968) expressthe dangersof erectingcomplex,
higher-order structuresupon faultyfoundations. It seemsincongruousto propose
fine-graineddistinctions such as intra-siteactivity areas, functional" tool-kit"
analysesand micro-settlement patternstudieswhenartifacttypesremainsubjective
entitiesdefinedby whimand fancy,and judged largelyupon the reputationof their
Creator. The followingdiscussionof archaeologicaltypology,and its relationship
to biology,is guided by two principles,the firstbeingthat " thereare no ideas in
scienceso well proventhat they cannotbe challenged. There are none so certain
"
that they need not be re-examinedoccasionally (Blackwelder,1964 : 18). This
somewhaticonoclasticpositionis mitigatedby the further propositionthat archaeo-
logistshave not for the past four decades been wasting theirtimein exploringthe
basic natureof archaeologicalsystematics. New methodologicalsuggestionsought
not be proposedin vacuo,but shouldinsteadbe presentedin perspectiveto the work
previouslyattemptedon related problems. Concepts must neitherbe rejected
becausetheyare old norblindlyacceptedsimplybecausetheyare new.
The presentdiscussionexploresthe relevanceof recentnumericaltaxonomic
aids to typologyas it is currentlypractisedamong Americanarchaeologists.To
it is necessaryto examinea coupleoflingering
avoid superficiality, bogeysin archaeo-
logicaltheory,namely the statusof the normative typeconcept and the relevanceof
biologicalanalogiesto cultural phenomena. One disclaimermay be enteredat the
outset: I share Krieger's" bias against theoristswho are unable to demonstrate
theirrecommendations withan actual run of material" (i960 : 145, also quoted in
Chang, 1967a : 84). Such examples are lacking here due to spatial limitations.
To minimizewordgames,rigorousdefinitions are oftenavoidedin favourofcommon-
senseusage. The terms" typology", " taxonomy" and " systematics ", forexample,
are interchangedfreely.

I. Emics and Etics


Ratherthan repeatan in extensoargumentagainstthe use of ernie,normative
types in archaeology(cf. Binford,1965 ; Bayard, 1969 ; Cowgill,1970 ; Harris,
19686), I shall only examine why such an approach is poorlysuited forbedrock
taxonomy. If this appears more a statementof faiththan a valid proof,so be it,
but an explicitpositionis necessaryforproperexpositionin the sectionsto follow.
In the context of basic archaeologicalclassifications, an erniestance lacks both
(1) operationally and (2) the informationalquanta to justifyits being.
Thereare two approachesto establishing the cognitivesaliencyof archaeological
types. The most recent advocate of the firstemic approach to archaeological
typologyis K. C. Chang,who cites the" work of the ethnosemanticists Conklinand
Frake to supporthis positionthat the ' right' categoriesare thosethat reflector
"
approximatethe natives' own thinkingabout theirphysicalworld (1967a : 78)
"
and that archaeological classificationstend to be cognitivelysignificant "

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 33

(19676: 228). Whileconsidering the operationalbasis of such efforts, Changasks :


" How can we our these
put fingers upon (cognitively significant) keysamongso many
others? " (1967α: ηη). This is an excellentquestion, but one which remains
unanswered,at least by Chang. His substitutionof the settlementforthe artifact
typeas archaeology's basicunitand hisassertionthat" archaeological typesthathave
useful must have a structural basis "
provedhistorically (1967a 87) begsthequestion.
:
:
Cowgill(1967 236-37) takes to
exception Chang's criteriaof stabilityand general
measures of Binford : "
workability as cognition. (1967 234) notes that direct
linkagescan be made betweenthe relevantvariableswithouttranslatingtheminto
idealtionalterms". This firstapproach to cognitivelysignificant types is a non-
operationalprocedure at the basic level of archaeological typology.
Deetz arguesfora different structuralrole forcognitivesignificance than does
For " is an rather than an inferential
Chang. Deetz, typology integrative procedure
"
one " (1967 : 63) and it maywellbe thatsometypesare almostperfectdescriptions
ofthetemplatesresponsible forthem; whetherthisis trueor notin no way interferes
with the main aim of typology,that of classification which permitscomparison"
(1967 : 51). Deetz thusplacestheestablishment ofnormativecriteriain sequentially
secondary and operationally inferior posture vis-à-visdefinition of basic classificatory
taxa. In his studyof Arikaraceramics,forexample,Deetz (1965) performs a typo-
logical analysisusing concrete, on the ground correlations of stylistic attributes;
onlya posteriori did he advocate that thesechangesrepresented the natives'model.
His separationofdata frominterpretation cannotbe faulted; scholarsare quite free
to rejectinterpretations and returnto the basic data. Taylor,althoughsuggesting
"
the archaeologicaltaxa oughtto reflectnative categories,admitsthat thereis no
automaticassurancethat the forms,types and classes establishedtoday by the
archaeologist are coextensivewithany separableentitiesthatexistedin the mindsor
lifeways of a bygonepeople. . . any such correspondence is a matterforexplicit
"
hypothesis and testing (1948 121). Similarly, discussingthe relevanceof
: in
componential analysisto archaeologicaldata, Gardin(1965 : 12) concurswithDeetz :
" neitherthe nativetermsnorthe rationaleare of immediate use to the archaeologist
as distinctfromthe linguistor the anthropologist.While the latter is directly
concernedwithsememicmodelsand cognitivesystems. . . the archaeologisthas first
to bringordersof anotherkind into his data : e.g., chronologicalarrangements,
geographicaldistributions, stylisticor more generallycultural groupings,etc."
Harris exhortsarchaeologiststo
Shriveyourselvesof the notionsthat the units whichyou seek to reconstruct
mustmatchthe unitsin social organizationwhichcontemporary ethnographers
have attemptedto tell you exist. . . you are capable of definingentitieswhose
reality,I assureyou,is everybit as well-grounded as the entitieswhichare now
being discussed at great lengthby ethnographers dealing with contemporary
sociopoliticalsystems(1968a 360). :
To do Harrisone better,I assertthatthisis preciselywhat archaeologists have been
doingall along. The errorhas been in the beliefby some thattypes have an empirical

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
34 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

validitygroundedin cognition. How does one a priorideterminecognitionlacking


lexical data and even informants ? This is difficult,if not impossible,given the
natureofarchaeologicaldata and the currentstatusofethnemicanalysis.
Let us rhetorically grant cognitivesaliencyto artifacttypes in orderto ask
" So what? ". Can these
typesreallytell us how the nativesperceivedand inter-
acted withtheirenvironments ? In thismanner,could one actuallyfindFlannery's
" "
(1967) systembehindthe Indian behindthe artifact ? Conklinand Frake seem
to answeraffirmatively althoughtheywouldsurelyregisterseriousobjectionsto such
practicesusing archaeologicaldata. In a similarsituation,Vayda and Rappaport
comparecategoriesderivedfromethnoscience to observedworkingsof the ecosystem
and they concludethat
Thereare no universallyapplicableproceduresforinferring eitherethno-ecology
or the actual relationswithinan ecologicalsystemfromethno-semantics.One
musttherefore demurwhenFrake makesstatementssuch as : " By discovering
whatone mustknowin orderto classifyplantsand otherecologicalcomponents
in Hanunoo fashion,one learnswhat the Hanunoo considerworthattendingto
whenmakingdecisionson how to behave withintheirecosystem" (1962 : 58).
Whetheror not this is so forthe Hanunoo case is an empiricalquestion,and
it is an empiricalquestionin everycase (Vayda and Rappaport, 1968 : 491,
emphasisadded).
My concern is forthosecommittedto provingcognitivecorrelatesto archaeological
"
types, theyseemto imply thatthereare adequate a priorigrounds
for for
>y inferring
ethno-ecology and behavior from ethno-systematics. There are not (Vayda and
:
Rappaport,1968 491). Ethnology has demonstrated that the native modeloften
fallsquite shortof describingexistentialbehaviour,e.g., the distributive functionof
the NorthwestCoast potlatch(Piddocke,1965 ; Suttles,i960), the latentnutritive
functions ofpig festivalsin New Guinea(Vayda,Leeds and Smith,1961 ; Rappaport,
1967), the sacred cows of India (Harris,1966) and the East Africancattle complex
(Schneider,1957). Heider (1967 : 55-57) demonstrateshow a standardarchaeo-
logicaltypologywould failin imitatingthe nativeDugum Dani functionaltypology
(also see Harris, 1964 : 92 ; Gould, 1968 ; Deetz, 1967 : 92 ; White, 1967).
Archaeologists must approach cognitionwith trepidation,forits true characteris
unestablishedeven in workwithlivingsubjects. Sanday, forexample,assessesthe
relationshipof cognitivestructureto American-English kinshipterms. Her results
" the need to be carefulin that thereis onlyone cognitivemodel
suggest assuming
shared by membersof a given society" (1968 : 522). We are not arguingthat
ethnosemantic categoriesare unrelatedto behaviour,but if modernecologistsand
ethnologists cannotassume a prioriexistenceforsuch a correlation, what gives an
ethno-archaeologist that prerogative?
My positioncan be summarizedin threepoints:
(1) Primaryarchaeologicalclassification mustrelyupon integrative ratherthan
inferentialprinciples.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 35

(2) Only rarelycan archaeologistsoperationallydeterminecognitivecorrelates


to archaeologicaltypes.
(3) Even grantedthat suchsignificance could be establishedin everycase, there
is no assurancethat native semanticcategoriesaccuratelyreflect" on the
"
groundbehaviour of ecosystems.
The correlationof archaeologicalentitiesand native categories" constitutesone of
the archaeologist'sgreatestproblems,not one of his self-evident truths" (Taylor,
1948 : 121).
II. The Biological Analogy
Some archaeologists,weary fromwrestlingwith typologicaldifficulties, have
a
explored biologicalanalogy fortaxonomic insights. On the surface,what could be
more obvious than the equation types biologicalspecies (see Krieger,i960) or
of to
the comparison archaeologicalcultureswithbiology'shigherlevel groupingssuch
of
as Class and Order (e.g. Colton,1939 ; Mulvaney,1969). This procedure,termed
"
the " insidiousholdover by Shepard (1956 : 315), receivesharshtreatmentfrom
some archaeologicalquarters(e.g. Steward,1941 ; Brew, 1946 ; Sears, i960). It
is a curiousturnaboutto findmodernbiologists, oftenin a taxonomicmirethemselves,
alluding to inorganicanalyses (Gilmour and Turrill,1941; Sokal and Sneath,
:
1963 90 ; Carmichael,George and Julius,1968) or even archaeologicalcorrelations
(Goodall, 1966) for reinforcement and reciprocalillumination. Before advocating
biological methods for archaeology - and that is the purpose of this paper- the
structure of the biologicalanalogy must be exploredforpossiblepitfalls.
Brew's(1946)enmitytowardbiologytranscendscrassassertionsthat (a) artifacts
"
are not bisexual,(b) theydo not breed,(c) or adapt, and (d) biologicaltaxonomy
would be verydifferent if an oak tree could throughassociationof gardenplants,
"
acquire geraniumleaves and bear strawberries (Steward, 1941 : 366). Brew
stressestwo points: (1) unlike biologicaldefinitionof species,typologicalcriteria
are imposedby the student,and (2) biologicalconceptssuch as evolution,genetics,
and phylogenyare dangerousin archaeologyand ill-definedbiologically. This
firstconcernsubsequentlyeruptedin the Ford-Spauldingdebates,whichis discussed
later. Brew'ssecondpointis extremely welltaken,for" in biology. . . theproblems
" Brewnotes
oftaxonomicmethodare by no meanssettled (1946 : 44). Specifically,
the lack of biologicalobjectivity,failureto establish(or even define)naturalclasses,
and theinadequacyofbiologicalschemesto effectively deal withphylogeny. Brew's
relianceupon the work of the biologistJ. S. L. Gilmour(1940) is criticalto this
discussion,forGilmourfoundeda schoolof biologicalthoughtwhichrevolutionized
modernsystematics - numericaltaxonomy. Gilmour'sretinuerevampedprecisely
those aspects of biologywhichtroubledBrew, whose reservationswere shared by
many biologists. Numericaltaxonomists suggest collocationof objectivityand
" " taxa. In Part III I shall
repeatability withthe definition of natural pursuethe
relevantprinciplesof numericaltaxonomy to demonstrate theirjust relationto the
fabricof archaeologicalclassification.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
36 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

By temporarily deferring discussionon both realityof typesand the profitsof


numericaltaxonomy,only one of Brew's major suspicionsremains: " it has to do
with a seriousmisconception . . . that relationshipsexist betweenculturalobjects
and betweenculturecomplexeswhich are the same as or similarto the genetic
"
relationshipsbetweenliving organisms (Brew, 1946 : 46). Taylor rejoins that,
" if the two
spheres(biologyand archaeology)are so distinctas to make the classi-
ficatorygeneralization ofoneimpertinent to thoseoftheother,thenby thesametoken
the classificatorydeficienciesof one are not pertinentto those of the other"
(1948 : 140). Tayloravers that geneticaffinities can in factexist betweenarchaeo-
logical entities,as in the case of American automobiles. Ideas in the minds of
individuals,argues Taylor, provide a cultural continuumanalogous to biological
"
genetics. Giffordadds that potterytypes are manifestations of ceramicideas,
ceramicimages,held originallyin the minds ' of human beings' . . . ideas, mental
' '"
imagesare the genesof culture (i960 : 346, emphasisadded). But, as under-
scoredby Binford(1965), normativecriteriaare ill-suitedas a substratumforany
archaeologicalclassification.
We need not introducesuch vague notionsas " ideas " or " norms" to retain
a geneticanalogy. Numericaltaxonomistshave recentlyinquiredinto the logical
structureof phylogeneticclassification ; spin-offsfrom those discussionshave
archaeologicalrelevance. Since the Darwinianparadigm,that formalsimilarityis
due to commonancestry,biologicalclassificationhas sortedtaxa on the basis of
presumedcommondescentwith modification - phylogenetics.Withinthe past two
decades therehas arisen a minorityopinionwhich distinguishedanothervariety
of filiation,termedphenetic. These are static relations,distinctfromthose of
phyletics,to be " evaluatedpurelyon the basis of resemblancesexistingnowin the
"
materialat hand (Sokal and Sneath, 1963 : 55). Pheneticists, modernfollowers
of Gilmour,hold that the directevidenceof commonancestryis most elusiveand
therebyprovidesimpotentfootingfor primaryclassification.They proposethat
taxa be organizedon the basis of naturalgroupings,containingthe highestpossible
contentofimpliedinformation and providing themostpowerful predictors. Affinities
are assessedfroma largenumberof unweighted, uncorrelated characters. Phenetic
resemblances are mostoftenexpressedon a phenogram, a treediagramin whichthe
branchesrepresentpheneticdistance,but not necessarilycommonancestry. Subse-
quent to formingpheneticgroupings,one infersphyleticsusing the more direct
evidenceof geologicalsuperposition, relativedatingtechniques,and recently,sero-
logical and protein studies. Lacking such geological information,numerical
techniques have been developed which helpelucidatepossibleevolutionary sequences
(see Camin and Sokal, 1965; Edwards and Cavalli-Sforza,1964). There are two
componentsofphyleticaffinity : cladisticsis the actual branching(genetic)affiliation
betweentaxa and chronistics deals exclusivelywith the taxonomicjuxtaposition
in time (Sokal and Camin,1965). This briefexpositionof the pheneticpositionis
highlyselective and intendedonly to introducerequisiteconcepts. I urge the
reader to consultsuch primarysourcesas Sokal and Sneath (1963), Gilmourand

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 37
Turrill(1941) and Sokal and Camin (1965). Clarke (1968) proffers these principles
and archaeologicalanalogies,but since he is establishinga completelynew order
ratherthanintegrating traditionalconcepts,thefollowing does notnecessarilyadhere
to hisusage. We shouldmentionthatthepheneticpositionis by no meansundisputed
in biology; forthe most cogentcriticisms,see Mayr (1965), Simpson (1965) and
Meeuse (1964).
" "
These conceptsare introducedto prefacethe assertionthat so-called genetic
relationscan, and oftendo, exist among inanimateobjects; thefallacyis in over-
"
simplification. There appears to be no difference between empirical,phenetic
of
classifications livingorganisms and those of inanimate objects except that the
hierarchiesarrivedat in classification of inanimate objects do notnecessarily reflect
"
the courseof descent withmodification (Sokal and Sneath, 1963 : 265, emphasis
added). To denythat a cladisticrelationship, however,existsbetweenthe Spanish
riding saddle and those of the Plains Indians is absurd (Taylor,1948 : 141). To
ignore chronistic superiority of Casa Grande Red-on-buff potteryto Sacaton Red-
on-buffsherdsat Snaketown,forexample,disregardsstratigraphically established
relations; a cladisticaffinity in thiscase may or may not exist. Cladisticcognation
seemswell establishedbetweenthe Na-Dene speakersand the Hypsithermal hunters
ofwesternCanada (Dumond,1969) ; cladisticcontinuity is muchless certainbetween
the historicShoshonean-speakers of the Great Basin and the prehistoricLovelock
Culture. In the following I contrastthesethreerelationships in archaeologicaldata
and discussmethodological considerations foreach. Numericaltaxonomy is suggested
the i.e.
onlyforassessing firstaffinity, phenetics. Brew was in
quite right cautioning
archaeologists againstadaptationofbiologicalmethodsas theystoodin themid-1940* $.
His evaluationofGilmour'sobjectionswas prophetic. But archaeologists shouldnot
ignorerecentadvances in biologicaltaxonomy,forthesenew methodsand theories
vitiateBrew's criticisms. Krieger(i960) quotes Huxley to the effectthat archaeo-
logistslag 40 yearsbehindbiologicaltaxonomists; pheneticmethodsofclassification
will help close that gap.

III. Phenetics and Archaeological Typology


If geneticrelationsexistbetweenarchaeologicalentities,we can utilizebiological
methodsfor explicatingthose relationships.The presentdiscussionis limitedto
archaeologicaltypes. The followingwill not necessarilyhold for hierarchically
superiorlevels of archaeologicalinvestigation. Clarke (1968), it may be noted,
concludesthat numericaltaxonomycan be of assistancewithhigher-order categories
of archaeology; suchdiscussionis beyondthe scope ofthe presentpaper.

(1) Whatis the Functionof an Archaeological TypeΡ


" "
Earlierwritersnote the multiplicity meaningsascribedto the term type
of
(e.g. Krieger,1944; Taylor, 1948; Steward,1954; Chang, 1968; Rouse, 1968).
Willeyand Phillips(1958) commentthat despiteambiguitiesin explicitdefinitions,
archaeologistsmanage to utilize the type conceptand to communicatewith other

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
38 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

archaeologistswhat their types mean, if only by example. Avoiding rigorous


definitionsfor the moment,let me assess what archaeologistshave traditionally
expectedoftheirtypes. First,typesareabstractions fromrealityand nottheartifacts
themselves(Rouse, 1939 ; Chang, 1967a : 4). In addition,types are oftencon-
sideredas reflectionsof the " mentaltemplate" of the maker (Deetz, 1967, 1968),
"
as mentalpatterns (Krieger,1944 : 272), as " fossilizedideas (Deetz, 1967 : 45),
" "
" " "
as intellectualideas (Rouse, 1939 : 20), and comprisedof a seriesof cognitively
"
modes (Chang,19670,1967ο). As discussedin Part I, thisview is not
significant
so much incorrectas it is misplaced. Inferencesof this type cannot exist at the
foundationof a typologicalsystem. Finally, types also possess temporaland/or
spatialsignificance(Krieger,1944 ; Steward,1954; Sears,i960 ; Smith,Willeyand
Gifford,i960 ; Rouse, i960). This most salientcharacteristic of typesprovidesa
refreshingarea of generalagreementin an otherwisequite controversial topic. The
to thatlabelled" historical-index "
latterusage oftypecorresponds type by Steward
" " " class " in the
(r954 : 55)· Steward's morphological type is termed
" " " "following
discussion. No cognitiveelementis impliedin either class or type as used
here.

(2) Are Types Inherentor Imposed?


Even if thereis a workingunityamong archaeologistsas to purposesof the
type,thereappears to be littleagreementas to the epistemiological status of this
concept. The problem of the realityof typeshas resulted in such a formidable
bibliography thatthe subject must be approached withtrepidation. The acrimonious
Ford (1954a, 19546) v. Spaulding(1953, 1954) debate is the classic example of this
knottyproblem,althoughseveralothersalignedthemselveseitherfor(Taylor,1948 ;
Smith,1954 ; Krieger,i960) or against (Rouse, 1939, i960 ; Brew, 1948; Sears,
i960 ; Deetz, 1967) the realityof archaeologicaltypes. Such disputeis not limited
to archaeologyor eventhesocialsciences,as anyonefamiliarwithbiologicalliterature
is aware. If archaeologistshave failedto establishthe philosophicalbasis of their
types,they share this fate with those biologistsworkingwith the species concept
(Ehrlich, 1961 ; Blackwelder and Blackwelder,1962). Numerical taxonomists
"
have been accused (Constance,1964 ; Meeuse, 1964) of implyingan objective
" fortheir
reality entities,the OTU's (OperationalTaxonomicUnits). Sokal (1962)
" The
and Sokal and Sneath (1963) do notmake such claims of realityfortaxa.
fact that a numberof slightlydiffering taxonomiesmay be obtainedby dînèrent
"
statisticalmethodsis clear evidencethat they are not (real) (Sokal and Sneath,
1963: 268). But theygo on to suggestthat once appropriatestatisticalmethods
are chosen,numericaltaxonomistscan objectivelyarriveat a replicabletaxonomy
whichwilloperationally serveits specificpurpose. The crux of the issue,therefore
"
is not reality,but ratherutility. Like Gardin (1965 : 12), we must ask, What
will such a modelteach us ? " Sokal and Sneath (1963 : 269) suggestthe analogy
"
ofthearithmetic mean,X. Let us recruita groupofindividuals,each an intelligent
"
ignoramus (to use thecommonbiologicalsurrogate), and instructthemto " describe

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 39

thelength" ofa seriesofartifacts. We wouldnotreceiveconsistent or usefulreplies.


We could expect more satisfactoryresultshad we given the same subjects the
command" computethe mean length". By providingan operationalalgorithm
(Χ=Σ1ι/η), we may facilitatethe repeatableobjectiveassessmentof an originalset
of itemswithinthe range of samplingerror,assumingan acceptable definitionof
mean is besidethe point;
length. Debate as to realityor capriceof the arithmetic
themeasureis useful; it satisfiestheimmediatepurpose hand,and thisis all that
at
needbe askedofit. In termsof the typeconcept,the pseudo-problem of thereality
of types,like Ford's Easter Egg becomesnon-relevant. We can only ask (1) are
typesoperational,and (2) do the
theysatisfy purpose forwhich theyweredesigned?

(3) How do Archaeologists Form theirTypes?


I shall not make the naive attemptto findthetypologicalmethod,forno such
animalexists. Table 1 liststhepublishedprocedures ofsomeofarchaeology'sleading
taxonomists. Thereare clearlytwostepsinvolvedin thetaxonomyofarchaeological
entities,the firstof whichconsistsof sortingobjects into homogeneousgroupings,
based upon morphological distinctions. Nomenclature forthispreliminary grouping
" "
varies; in addition to those in Table 1, it is termed observedtype (Smith
"
1954: 23), homogeneousgroup of artifacts" (Heizer, 1950 : 56), " provisional
" " "
type (Wheat,1953 : 197), and morphological type (Steward,1954 : 45). Let
us label this firststep alpha taxonomy, afterSokal and Sneath (1963 : 10), and
Edwards and Cavalli-Sforza(1964). We may operationallydefinethis procedure
as thecreationofprimaryclassesofunitsbasedstrictly uponmorphological considerations
" "
withinculturallymeaningfulcontexts.These culturallymeaningfulcontexts
consist of subjectivelydetermined,yet publicly defined,primaryarchaeological
associations. To inject furtherrigidityhere would invalidate the utilityof this
firststep. The primeexampleofculturallymeaningful contextsis an archaeological
component, but otherlimitscould be physiographic province,ecologicalboundaries,
and in somecases,entiresitesorgroupsofsites. Alphaclassesare neithertemporally
significant nor do they necessarilypossess cognitivesaliencyin the mindsof their
makers.
The secondproceduralstep is the definition of temporallysignificant groupings,
i.e. formationof historical-index types. There can be operationallyintermediate
processesinvolved of" tentativetypes" (Krieger,1944 : 280),
here,suchas definition
"
or " varieties (Smith,Willeyand Gifford, i960), but the overallprocedureinvolves
" "
theestablishment of demonstrable meaningin termsofbehavioral
historical patterns
(Krieger,1944 : 292). Thisis wherecontextualanalysiscomesintofullplay. Alpha
classesare appraised,usingcontextualcriteriafortemporaland/orspatialsignificance.
Archaeologists discussedin Table 1 would perhapsarguethat this typeis not only
temporally but also cognitivelysignificant.Such typesmay or may not be fossilized
ideas, but at this juncturethe distinctionis non-relevant.The resultof Krieger's
" we shall termbetataxonomy, is a specially-
(1944) truetypologicalmethod", what " "
constructed whichforms
classification, historical-indextypes (Steward,1954 : 54).

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
40 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

Chang(1968)and Rouse (1968)arequiterightwhentheyarguethatSteward'smorpho-


logical type is of a different orderthan his functionaland historical-indextypes.
Alpha classes are necessary since groupsof items,not individualartifacts,
possess
temporaland spatial significance.Pheneticclasses can providethe buildingblocks
of typology,althoughit would be too much to presentthemas the " basic unit of
archaeology". Classes establishedin alpha taxonomyare not subdividedexcept
alonglinesindicatedon the phenogram(see Sokal and Sneath,1963,Chapter7).
Combiningthis excursioninto biologicalsystematicswiththe above discussion
of alpha and beta taxonomy,certainlimitedparallelscan be drawnsincephenetics,
cladistics and chronisticshave archaeologicalcounterparts.Clarke (1968 : 546)
discussestheseterms,but since Clarke'sexpressionsare not derivedfromAmerican
archaeologicaltheory, I do not feel obliged to follow his usage. I redefine
the operationsin Table 1 as follows:
Biology Archaeology
Phenetics=initial sort into morphologicallyhomogeneousgroupings,i.e. alpha
taxonomy.
Chronistics =determiningmeaningfultemporaland/orspatial relationshipsof
alpha classes, i.e. formationof historical-index
types.
Cladistics=determininggenetic,i.e. cultural,relationships
amongalpha classes,
in the senseof Krieger(1944 : 283) and Rouse (1939 : 14, 22-23).
Such distinctions
echo Krieger'swarningconcerning thedangersofhypothesizing
frommorphological
genetic(cladistic)affinity (phenetic)similarity. Such fallacious
procedure has beenemployed, forexample,bythoseanxiousto proposevast migrations
in space on the evidenceof a few artifactual similarities(cf. Adams, 1968). Pure
phenetics,withoutauxiliarybeta analysis,can give most spuriousresultswhen
attributedchronisticor cladisticsignificance.The leaf-shapedCascade projectile
point is notoriousfor variable temporalposition(chronistics), althoughthe form
is
(phenetics) quite constant(Butler,1961 ; Davis, 1966).
I thereforeadvocate numericaltaxonomicproceduresdevelopedin biologyas
immediateaids to archaeologicalclassification.But I neither(1) suggestirrational
scrappingof traditionalarchaeologicalconceptsnor (2) presentnumericaltaxonomy
as a panacea forall maladiesof typology. In fact,whenpoorlyapplied,numerical
taxonomycarriesthe added spectreof errorswhichare moreflagrantthan thoseof
traditionalmethods,because of a diaphanouscomputermystiqueand the fallacyof
misplacedconcreteness.At this pointnumericaltaxonomicaids are proposedonly
forpheneticanalysis,sincecontextualanalysisrequiredin bothcladisticand chronistic
proceduresis not yet sufficiently
operationalized.

(4) Whatis theRelationofPheneticsto TraditionalArchaeological ?


Systematics
In thepast,suchprocedureshave involvedintuitivenotions,previousexperience
and personalbias- in a word,feel (Phillips,1958). Such a strategycannotserveas
the foundationof a disciplineaspiringtowardobjectivityand repeatability. In the

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 41

•^iijlirffi ¡41
IP 1|%|Ιί!°3ΐ1Ι'ξΚ
isâι ifijjiiïiρ ¡m
'm' I 'õT

I
li ^
1 jitiuiüjiiiiij¡Lulljifm

iit
ΗΊ A
$$ *liil !ÍI
I! Ξ IS
ÎI
.«s
fi«îîïiî III !U
f*8
S.'«I
ΪΙ"8·8|Μ1|« 1*81
I á|
en
J*3,2 CÖ.2 si ^| jrf rt
ΡηΛ Τηο3,5-μ2' I Ho,
Η ü <ü S o<4J υ (β ©W «β ö rt I SC¿
§
Ξ 2 S I<

sìli !í!¡ iSHIÜi:


! J,

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
42 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

longrun,however,theuse oifeelhas beenmoderately in classifying


successful cultural
itemsinto homogenousgroups. Why? It appears to me thatfeel is the sensory
experienceof the totalityof an item, that is, overall morphology.But workers
simplylack the expressiveveracityto communicate such feelingsto theircolleagues.
How oftenhave you seen an otherwisearticulatearchaeologiststammerabout in
searchforwordsto tell us what it is he means by his types? Neologismsabound,
" " "
manyof themendingin -ness and -icity". We are neversurewe comprehend
his entiremeaning; too oftenthe onlyway to see if a new artifactmatcheshis old
typeis to send it forpersonalexamination. It seemsthat thisfeelis the subjective
correlateofpheneticclassification,
the disparitybeingthatpheneticsare operational.
Computationalproceduresof pheneticsare treated in Sokal and Sneath (1963).
Maiseland Hill (1969), Minkoff (1965), Cole (1969), Sokal (1969) and Sneath (1969).
A phenogramis nothingmorethan the objectiveexpressionof thoserelationswhich
archaeologistshave experiencedfordecades as feel.
TraditionalMethods NumericalTaxonomy
(1) Define the taxonomic universe
(2) Examine each artifactand mentally (2) Publiclyselectcharacters(generally
" "
get the feel 100 to 200)
(3) Compare each artifacton the lab (3) Compute matrix of similarityco-
table with the others efficients
(4) Sort artifacts into taxonomic piles (4) Cluster analyse the similarityco-
efficientsand display results on
phenogram
Alpha taxonomy, whether performed on a laboratorytable or a computer,simply
arrays similar items into taxonomicallyhomogeneouslots.
Numericaltaxonomywas developed to isolate " natural" groups in biology
so one may inquireif archaeologicalapplicationsalso discoversuch naturalunits.
Brew (1948) closelyexaminesthe status of naturalnessin archaeologyand rejected
what he called " final-ideal-classifications ". I have already shown how modern
biologistsanswered Brew's earlier criticisms regardingobjectivityand repeatability.
Brew's source on naturalnessis Gilmour,who formulatedthe modernconceptof
naturalgroupsbut failed to operationalizetheirisolation. Gilmourwrites: " In
practice. . . the aim (of findingnaturaltaxa) would neverbe attained,owingboth
to the limitationsof our knowledgeand to the differences of opinionsbetween
taxonomists" (Gilmour,1940 : 461, also quotedin Brew,1946 : 50). But numerical
taxonomyis little more than an operationalizationof Gilmour'sconcept. The
" limitationsof our "
knowledge consistedof man's inabilityto focus sufficient
quantitiesof data on a particularproblem. Computersseem admirablysuited to
fillthis gap. " Differences in opinionbetweentaxonomists" can be minimized
althoughnot eliminated,throughthe explicitand objectiveproceduresof calculating
overallpheneticaffinity.Brewregistered twofurther objections,thefirstbeingthat
"if we attain a classificationwhich ' all the attributesof the individualsunder

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 43
consideration ' are utilized
(as Gilmoursuggests),shall we not end up ... with a
situationwhereineach individualconstitutesa class by itself? " (1946 : 50). The
concept of polytheticunits (Beckner, 1959 : 22 ; see also Solcai and Sneath,
1963: 13-15 and Clarke,1968 : 35-38) eliminatesthis bleak prospectof taxonomic
anomie painted by Brew. Polytheticunits transcendboth the inferenceof an
Aristotelean essence(Taylor's" archetype") and the applicationof simplisticmono-
theticcriteria. Polytheticunitsin archaeologylargelycorrespondto one's intuitive
sense of " natural".
Brew'ssecondmisgivingregardingnaturalclassesis " What have we, and what
do we do with it ? " (1948 : 50). Sokal and Sneath reply:
The reason for the great usefulnessof natural classification is that when the
membersof a group share many correlatedattributes,the " impliedinforma-
tion" or " contentof information " is
high; thisamountsto Gilmour 's dictum
that a systemof classification is morenaturalthe morepropositionsthereare
that can be made regardingits constituentclasses (1963: 19).
Steward'sfunctionaland historical-index typesare not natural,since theywere
definedto reflectonlya singlepurpose. But the entitieswe have denotedas alpha
classes, e.g. Steward's morphologicaltypes, are pheneticand thereforenatural.
" intendedforgeneraluse by all scientists" (Sokal and
Theyare generalarrangements
Sneath,1963 : 12). Being general,theymay not be the mostideallysuitedforall
" "
purposes; thisis the functionof an artificial or specialclassification.

(5) Should Types be Holisticor Partitive?


Binford(1965) suggeststhat classification" should with
proceedindependently
regard to technicaland design attributesand that cross-cutting categoriesshouldbe
used to expressmorphological and decorativevariation". On the faceofit, the call
forpheneticclassesseemsanathemato Binford'spartitivetypes. On closerexamina-
tion,we findthisis not the case ; in fact,carefulattentionto the phenetic-phyletic
distinctioncan operationalizeBinford'ssuggestions. DivisivecriteriawhichBinford
proposed have theircounterparts in biology. Zoologists,forexample,oftenfindit
to
necessary split herbivoresfrom carnivores,especiallywithrespectto the trophic
pyramid. Olson and Miller(1958) discussstatisticalmethodsof such partitions,and
theseproceduresfallquite withinthe generalsweep of numericaltaxonomy.
But wheredoes divisivecategorization stand in relationto the quest fornatural
units? The upperportionofFigure1, adaptedfromGilmourand Turrill(1941 : 219),
illustratesthe proposedrelationshipbetweenpheneticand divisive classification.
Specially-constructedtaxonomies,A, B, C, D . . . n, are coterminouswith those
Binford has termed primary and secondary functional,technomorphological,
decorativedesign,and so on. They " exist in theirown rightforspecial purposes,
and,byutilizingall theseparatetypesofattributeon whichtheyarebased,thegeneral
or (alpha) taxonomicclassificationis constructed ... a class based on a limited
numberofattributes mayeitherbe a specialclassificationor it maybe a stagetowards

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
44 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY
"
necessarilyrestricted
a generalclassification, owingto lack ofinformation (Gilmour
and Turrill,1941 : 219). Operationally,one must firstdesignatethose characters
whichpresumablyreflectthe divisivevariable,such as incisedpotterymotifsfor
decorativedesigns. One thenperforms a numericaltaxonomicanalysison the basis
of thisrestrictedcharacterlist. The resultantphenogramwouldgraphicallydisplay
the relationsof the variable understudy. An alternativemethodis multiváriate
analysis,such as using Q-mode principalcomponentanalysis. One could analyse
these special purposeclasses on a beta level, searchingfor cladisticor chronistic
relations. Natural taxa of the alpha taxonomy,however,should providemore
informational valuebecauseoftheirgenerality. The partitiverelationswhichBinford
seeksmaybe obscuredby suchgenerality, in whichcase divisiveanalysiswouldclearly
be in order. Let me emphasize,however,that naturalgroupscannotbe based upon
suchselectedattributes. Partitiveclassesare,therefore, and unnatural,
interpretative
but completely withinthe scope of numericaltaxonomy.
SPECIALPURPOSE PARTITIVE
^^^ ^^^
CLASSIFICATION (ΊΓ) TYPES
^-s. (Έ^) ^-^

^natural ^ ¿eussesS ^ I Illlliii


"
SCLASSFJCATION ' ■ ■,-.../ Ì ·
™~
ii||p||p|i

CHRONISTICGROUPS. ·.·.«..
M r*w^
. Ρ«β*«ι '
J
CLADISTIC Λ
( 6ene«c 'f BETA
GROUPS . . . ^eqUenCey /CognitivelA TAXONOMY
I
I Significant
NORMATIVEGROUPS V Uni)>J
Figure i.*
If the readeraccepts that (1) archaeologicalclassificationcontainsa phenetic
componentand (2) numerical aids can some
perform of the tasks moretraditionally
performed or
mentallybyfeel archaeologicalprowess, mayhe stillask whywe bother
withcomputers and numbers if an worker
experienced can do the same job, oftenin
less time. The answer to this is, in a word, operationalism.Operationalismis
advocated here only in the more liberalsense of Hull (1968). Operationalclassi-
ficationsexhibitgreaterstability,higherpredictability and objectivity. Stability
can onlybe assessedafterconsiderableexperimentation on archaeologicalmaterials.
Preliminary studiesby Hodson,Sneathand Doran (1966),Johnson(1968) and in the
laboratoriesof the Universityof California,Davis, seem to indicatethat givenboth
the statisticaltechniquesand attributelists,numericaltaxonomiesare quite stable.
Highpredictability is a logicalconsequenceofa highinformation content. In theory,
naturalclassificationsshouldbe mostusefulforthe mostpurposes. For thisreason,
* The cost of this illustration
was met by the author.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 45
it is recommended that the pheneticclasses be establishedas firststeps in defining
phyleticcategories,especiallyhistorical-index types. In some cases special classi-
fications(suchas thoseofBinford)mayprovidebetterchronistic orcladisticindicators.
Experimentation on real data in meaningful
archaeologicalsituations willeventually
provide a more suitable answer.

(6) Does NumericalTaxonomyOperationalizeTypology?


Objectivityis a relativeconcept which is seldom fullyrealized. Numerical
taxonomists do notadvocatetheabolitionofspeculationand inference, butsegregation
of data frominterpretation is central in such rigorousprocedures. There is a
temptationto draw the objective-subjective line between alpha and beta taxonomy,
but such a distinction will not strictlyhold. There are two inferentialsteps before
of
the establishment phenetic classes: statistical considerations and character
selection. The firstof these,numericaltechnique,has receivedenormousattention
in the past decade and the biases of most currenttechniquesare well known.
Methodological stabilizationmay not be faroff. At least the investigator can make
an informed selection of coefficient, clusteringmethod, ordination technique,etc.
Selectionof characters,however,is a moredifficult situation. In the formative
stagesof numericaltaxonomy,pheneticists spoke of the hypothesis ofnon-specificity.
It statedon geneticgroundsthat any randomsampleof charactersfromthe infinite
arrayof such possiblecharactersshould produceroughlythe same set of affinities
(Sokal and Sneath,1963 : 85-91). Recentstudiesby Rohlf(1965) and Sneath(1969)
show this hypothesisto hold only in part; selectionof charactersdoes effectthe
overall outcome of classification.Carryingthis to an extreme,Steyskal (1968)
demonstrates the possibilityof scoringover 1,000characterson the antennaof a fly.
Such a tour de forceis also quite possible using archaeologicalspecimens. The
probleminvolves which charactersto choose. Biologists attempt to distribute
charactersover the entireanimal's body to avoid overemphasisof any particular
regionor organ. Stratifiedrandomsamplingis suggestedaftersufficiently large
attributelists have been generated. Standardizedcharacterlists provideanother
solution,but the prospectof scienceby fiatis particularlyunappetizing. Detailed
attributelists are attemptedby Gardin (1958, 1967) and Binford (1963), among
others. Furtherdifficulties involvethe cullingof unsuitablecharacters; biologists
traditionallyrejectsex-linked,age-gradedand ecologicallyvariablecharactersfrom
theirattributelists. Rouse (1939 : 12, 26) discussesthe same processwithrespect
to invariantor non-cultural modes. Mostarchaeologists probablyagreewithRouse
that artifactswould be classifiedwithan eye to theiroriginalcondition,neglecting
the vagariesof weatheringand decomposition.Taylor (1948 : 127) mentionsthe
possibilityof classifyingcars by the lengthof scratchesin theirpaint and potsherds
by the numberof containedsand grains,but he suggeststhat archaeologists should
reject such criteria. I disagreewith Taylor. Would not a comparison fender
of
scratchesand dents betweena dump truck and a Cadillac limosine provideuseful
criteriain formation of naturaltaxa ? The work of Semenov (1964), forexample,

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
46 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY

presentsa host of new charactersrelatingto the precisefunctionof artifacts. Such


charactersare welcomein a classificationwhichseeksthehighestinformation content.
Specific functionalcharacters also facilitate
formation of partitive,special purpose
classifications.I do not wish,however,to obscurethe factthat attributelistshave
an elementof subjectivityand this remainsno small problemin any applicationof
numericaltaxonomicmethodsof pheneticclassification.
Once numericaltechniquesand characterlists have been decided upon, the
proceduresbecomeobjectiveand repeatableat the alpha level. No such claimsare
made forbeta taxonomywhereobjectivityis less thepointthancompletepublication
of one's proceduresand results. It is the duty of the taxonomistqua archaeologist
to selectmeaningful OTU's forclassificationand to bringtheproperdata to bearupon
temporal and spatial problems(see Sokal and Sneath, 1963 : 174). As Spaulding
points out :
(1953 313), the entiretaxonomic systemis only as good as the input.
It is up to the archaeologistto determinerelevanceand meaning; in the idiomof
data processing," Garbage in, garbage out ".

IV. Summary
I have discussedtherelationship oftraditionalarchaeological
typologyto modern
methodsof numericaltaxonomyin the beliefthat nobodywilluse a methodif he is
unconvincedof its validity(Sokal and Sneath,1963 : 2 ; Chang,1967a : 12). On
the otherhand, I would scornthe cheap mentalityof any who would advocate use
of a methodsuch as numericaltaxonomysimplybecause it is new, quantitative,
computer-oriented or esoteric. I share Service's disenchantment with Mouthtalk
(1968). In exploringthenatureofthearchaeological typeconcept,biologicalanalogy
has beenemployed. But I have avoidedslavishadherenceto biology,foran archaeo-
logicaltypeis a culturalproduct,not merelyan imperfect bacteriumor a misguided
Drosophila. I have distinguishedtwo levels of analysis,pheneticsand phyletics.
Numericaltaxonomicaids are advocated only forthe pheneticlevel, i.e. morpho-
logicaltypes. Pheneticsare viewedas an operationalized versionoffeel,upon which
initial artifactsortingshave traditionallybeen based. Historicaland functional
typesare goal-directed interpretations of pheneticgroupings. This discussiondoes
not deal specifically withoperationalmethodsof archaeologicalphyletics,although
the workof Camin and Sokal (1965) and Edwards and Cavalli-Sforza(1964) may
have relevance.
Let me further pointout that one cannotjustifiablycriticizea new methodfor
inadequacieswhichorthodoxy itselffailsto answer. The newproposalcan be judged
successfulif it answersmore meaningfulquestionsor solves some of the previous
difficultiesof its predecessor. I am fullyaware of the fallacyin the quest forthe
" "
perfectsystem (Stebbins,1969) and make no such claims. One windfallof
dissectingfundamentalconcepts,like flow-charting a computeralgorithm,is the
unavoidable appraisal of one's basic operations and assumptions. As Sneath
"
(1964 : 43) observes, Indeed, the necessityto think out clearly the aims of
systematics may be amongthe greatestbenefitsof numericalstudies".

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 47
References
" "
Adams, William Y. (1068) :" Invasions, Diffusion,Evolution ? ' Antiquity,Vol. 42, pp. 104-215.
Bayard, Donn T. (1969) : Science, Theory and Reality in the New Archaeology'" American
Antiquity,Vol. 34, pp. 376-84.
Beckner, Morton (1959) : The Biological Way of Thought. New York : Columbia University
Press.
Binford,Lewis R. (1963) : "A Proposed AttributeList for the Description and Classificationof
Projectile Points." Anthropological Papers of the Museum of Anthropology,Vol. 19,
pp. 193-221.
Binford, Lewis R. (1965) : Archaeological Systematics and the Study of Cultural Process."
American Antiquity,Vol. 31, pp. 203-10.
Binford,Lewis R. (1967) : Commentsto ' Major Aspects of the Interrelationshipof Archaeology
"
'
and Ethnology by K. C. "Chang." CurrentAnthropology, Vol. 8, pp. 234-35.
"
Blackwelder, R. E. (1964) : Phyletic and Phenetic versus Omnispective Classification in
Pheneticand PhylogeneticClassification(ed. by V. H. Hey wood and J. McNeill). London :
The Systematics Association, pp. 17-28.
Blackwelder, R. E. (1967) : "A Critique of Numerical Taxonomy." SystematicZoology,Vol. 16,
pp. 64-73. '
Blackwelder,R. E., and Blackwelder,A. (1962) : Ammal Taxonomy and the New Systematics."
Survey of Biological Progress, Vol. 4, pp. 1-57.
"
Brew, J. O. (1946) : Archaeologyof Alkali Ridge, Southeastern Utah." Papers of the Peabody
Museum of AmericanArchaeology and Ethnology,21. Cambridge : Harvard University.
" The Old Cordilleran Culture in the Pacific Northwest." Occasional
Butler, B. Robert (1961) :
Papers of the Idaho State College Museum, No. " 9.
Camin, Joseph E., and Sokal, Robert R. (1965) : A Method forDeducing Branching Sequences
in Phvlogeny." Evolution, Vol. 19, pp. 311-26. "
Carmichael, J. W., George, J. Alan, and Julius, R. S. (1968) : Finding Natural Clusters."
SystematicZoology, Vol. 17, pp. 144-50.
Chang, K. C. (1967a) : "RethinkingArchaeology. New York : Random House, Inc.
Chang, K. C. (19676) : Major Aspects of the Interrelationshipof Archaeologyand Ethnology.
CurrentAnthropology, Vol.
" Toward8,a pp. 227-43.
Science of Prehistoric Society. In SettlementArchaeology
Chang, K. C. (1968) :
(ed. by K. C. Chang). Palo Alto, California: National Press Books, pp. 1-9.
Clarke, David L. (1968) : Analytical Archaeology. London : Methuen and Co. Ltd.
"
Colton, H. S. (1939) : PrehistoricCulture Units and Their Relationships in NorthernArizona.
Museum of NorthernArizona, Bulletin 17.
Cole, A. T· (Ed.). (1969) : Numerical
" Taxonomy. London and JNewYork: Academic ¿Tess.
Constance, Lincoln (1964) : Systematic Botany- an Unending Synthesis. Taxon, Vol. 13,
pp· 257-73· __ . . . . _
Cowgill,GeorgeL. (1967) : " Commentsto Major Aspects of the Interrelationshipsof Archaeology
'
and Ethnoloerv bv K. C." Chane." CurrentAnthropology, Vol. 8, pp. 236-37.
'
Cowgill, George L. (1970) : Review of F. T. Cloak, Jr., The Computer in Archaeology'."
Computersand the Humanities, Vol. 4, P. 221.
Davis, Wilbur A. (1966) : "Theoretical Problems in Western " Prehistory." In the "Current
Status of AnthropologicalResearch in the Great Basin (1964), Warren L. d'Azevedo and
others (eds.). Social Sciences and Humanities Publications,No. 1. Reno : Desert Research
Institute, pp. 147-66.
"
Deetz, James (1965) : The Dynamics of StylisticChange in Arikara Ceramics." Illinois Studies
in Anthroboloev , No. 4.
Deetz, James (1967) : Invitationto Archaeology. Garden City, New York : The Natural History
Press. "
Deetz, James (1968) : Cultural Patterningof Behavior as Reflectedby ArchaeologicalMaterials."
In SettlementArchaeology(ed. by K. C. Chang). Palo Alto, California: National Press
Books, pp. 31-42. "
Dumond, D. E. (1969) : Toward a Prehistoryof the Na-Dene, with a General Comment on
Population Movements among Nomadic Hunters." American Anthropologist,Vol. 71,
pp. 8 «57-63. " Reconstructionof
Edwards, A. W. F., and Cavalli-Sforza, L. L. (1964) : Evolutionary Trees."
In Pheneticand PhylogeneticClassification(ed. by V. H. Heywood and J. McNeill). London :
The Svstematics Association,
" pp. 67-76.
Ehrlich, P. R. (1961) : Systematics in 1970 : Some Unpopular Predictions." Systematic
Zoology, Vol. 10, pp. 157-58.
Flannery, Kent V. (1967) : Culture History vs. Cultural Process : a Debate in American
Archaeology." ScientificAmerican, Vol. 217, pp. 119-21.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
48 USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY
" The
Ford, James A. (1954e) : Type Concept Revisited." American Anthropologist, Vol. 56,
pp. 42-54. "
Ford, James A. (1954&) : Spaulding's Review of Ford." American Anthropologist, Vol. 56,
pp. 109-12. " The
Frake, Charles O. (1962) : Ethnographic Study of Cognitive Systems." In Anthropology
and Human Behavior (ed. by Thomas Gladwin and William C. Sturtevant). Washington,
D.C. : Anthropological of Washington, pp. 72-85.
" FourSociety
Codes forthe Description of Artifacts: an Essay in Archaeological
Gardin, J.-C. (1958) :
Technique and Theory." American Vol. 60, pp. λ^-^τ.
" On a Possible Anthropologist,
Gardin, J.-C. (1965) : Interpretationof Componential Analysis in Archaeology."
American Anthropologist, Vol. Pt. 2, pp. 9-22.
" Methods 67,for the Descriptive Analysis of Archaeological Material."
Gardin, J.-C. (1967) :
American Antiquity,Vol. 32, pp. 13-30.
Gifford,James C. (i960) : The Type-Variety Method of Ceramic Classificationas an Indicator
of Cultural Phenomena." American Antiauitv. Vol. 2 ς. pp. 34.1-47.
"
Gilmour, J. S. L. (1940) : Taxonomy and Philosophy." In The New Systematics(ed. by J. S.
Huxley). Oxford : Clarendon Press, pp. "461-74.
Gilmour, J. S. L., and Turrill, W. Β. (1941) : The Aim and Scope of Taxonomy." Chronica
Botanica, Vol. 6, pp. 217-19.
Goodall, D. W. (1966) : "A New SimilarityIndex Based on Probability." Biometrics,Vol. 22,
pp. 882-Q07. "
Gould, Richard A. (1968) : Living Archaeology: the Ngatatjara of WesternAustralia." South-
western Tournai of Anthropology . Vol. 24, PP. 101-22.
Harris, Marvin (1964) : The Nature of Cultural Things. New York : Random House.
"
Harris, Marvin (1966) : The Cultural Ecology of India's Sacred Cattle." CurrentAnthropology,
Vol. 7. pp. «si-66. "
Harris, Marvin (1968a) : Comments." In New Perspectivesin Archeology(ed. by Sally R.
Binford and Lewis R. Binford). Chicago : Aldine Company, pp. *«50-61.
Harris, Marvin (19686) : The Rise of AnthropologicalTheory. New York : Thomas Y. Crowell
Company. "
Heider, Karl G. (1967) : Archaeological Assumptionsand Ethnographical Facts : A Cautionary
Tale fromNew Guinea." SouthwesternTournaiof Anthropology, Vol. 23, pp. 52-64.
Heizer, Robert F. (Ed.) (1950) : A Manual ofArchaeologicalField Methods. Millbrae,California:
The National Press. "
Hodson, F. R., Sneath, P. Η. Α., and Doran, J. E. (1966) : Some Experimentsin the Numerical
Analysis of Archaeological Data." Biometrika, Vol. 53, pp. 311-24.
"
Hull, David L. (1968) : The Operational Imperative : Sense and Nonsense in Operationalism."
SystematicZoology, "Vol. 17, PP. 4^8-57.
Johnson,Leroy (1968) : Item Seriation as an Aid for Elementary Scale and Cluster Analysis."
Universityof OregonMuseum of Natural HistoryBulletin, No. 15.
Kemeny, John G. (1959) : A PhilosopherLooks at Science. New York : Van Nostrand Reinhold
Company. " The
Krieger, A. D. (1944) : Typological Concept." American Antiquity,Vol. 9, pp. 271-88.
"
Krieger, A. D. (i960) : ArcheologicalTypology in Theory and Practice. In Men and Cultures
(ed. by AnthonyWallace). Philadelphia : Universityof Pennsylvania Press, pp. 141-51.
Maisel, Herbert and Hill, L. R. (1969) : A Kwik Index of Publications in Numerical Taxonomy
in the Period 1948-1968. Washington, D.C. : Computation Center,GeorgetownUniversity.
(Mimeographed.) "
Mayr, Ernst (1965) : Numerical Phenetics and Taxonomic Theory." Systematic Zoology,
Vol. 14, pp. 73-91. "
Meeuse, A. D. J. (1964) : A Critique of Numerical Taxonomy. In Phenehc and Phylogenettc
Classification(ed. by V. H. Heywood and J. McNeill). London : The Systematics Associa-
tion, pp. ΐΐς-122.
" The Effectson Classificationof
Minkoff,E. C. (1965) : Slight Alterations in Numerical Tech-
nique." SystematicZoology, Vol. 14, pp. 196-213.
Mulvaney, D. J. (1969) : The Prehistoryof Australia. London : Thames and Hudson.
Olson, E. C, and Miller, R. L. (1958) : Morphological Integration. Chicago : University of
Chicago Press. "
Phillips, Philip (1958) : Application of the Wheat-Gifford-WasleyTaxonomy to Eastern
Ceramics." American" Antiauity,Vol. 24. PP. 117-λο.
Piddocke, Stuart (1965) : The Potlach System of the South Kwakiutl : A New Perspective."
SouthwesternJournal of Anthropology,Vol. 21, pp. 244-64.
Rappaport, Roy A. (1967) : Ritual Regulation of EnvironmentalRelations among a New Guinea
People." Ethnology,Vol. 6, pp. 17-30.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions
USE AND ABUSE OF NUMERICAL TAXONOMY IN ARCHAEOLOGY 49
" Test oftheNon-Specificity
Rohlf,F. James(1965): A Randomization Hypothesisin Numerical
TaYnnnmv." Taxnn. Vol. τλ. τνη. 262- 67.
" in Haiti,a Studyin Method." Yale University
Rouse,Irving(1939): Prehistory Publications
in Anthropology, No. 21.
Rouse, Irving(i960) : "The Classification of Artifacts in Archaeology."AmericanAntiquity,
Vol. 25, pp. 313-23."
Rouse, Irving (1968): Prehistory,Typology,and the Study of Society." In Settlement
Archaeology (ed. by"K. C. Chang).
' Palo Alto,California
' : NationalPressBooks,pp. 11-30.
Sanday,Peggy(1968): The Psychological Reality of American-English KinshipTerms: An
Information-Processing Approach."
" AmericanAnthropologist, Vol. 70, pp. 508-23.
Schneider,Harold K. (1957): The SubsistenceRole of CattleAmongthe Pakot and in East
Africa." AmericanAnthropologist,
" Vol. 59, pp. 278-300.
Sears,WilliamH. (i960) : CeramicSystemsand EasternArchaeology."AmericanAntiquity,
Vol. 25, pp. 324-29.
Semenov,S. A. (1964): PrehistoricTechnology,English translationby M. W. Thompson.
London: Cory,Adams " and MacKay.
Service,Elman R. (1968): ModelsfortheMethodology ofMouthtalk." Southwestern Journal
of Anthropology, Vol. "2ς, pp. 68-80.
Shepard, Anna O. (1956) : Ceramics for the Archaeologist."Carnegie InstitutionofWashington
Publication609. Washington, " Current D.C.
G.
Simpson,George (1965) : Issues in Taxonomic Theory." Science, Vol. 148,p. 1078.
Smith,MarianW. (1954): Attributes and theDiscoveryofProjectilePoint Types." American
Antiquity, Vol. 20, pp. 15-26. "
Smith,RobertE., Willey,GordonR., and Gifford, JamesC. (i960) : The Type-Variety Concept
as a Basis forthe Analysis of Maya Pottery." American Antiquity,Vol. 25, pp. 330-40.
P. H. A. : " Introduction." In Pheneticand
Sneath, (1964) PhylogeneticClassification(ed. by
V. H. Heywoodand T. McNeill). London: The Systematics Association, pp. 43-46.
Sneath.P. H. A. (iq6q) : " Recent
"
Trendsin NumericalTaxonomy." Taxon,Vol. 18,pp. 14-20.
Sokal, RobertR. (1962): Typologyand Empiricismin Taxonomy." Journalof Theoretical
Bioloev.Vol. λ. dp. 230-67.
"
Sokal,RobertR. (1969): The SecondAnnualConference on NumericalTaxonomy." Systematic
ZoolOËV. Vol. l8. DD. IO3-IO4.
"
Sokal, RobertR., and Camin,JosephH. (1965): The Two Taxonomies: Areasof Agreement
and of Conflict." Systematic Zoology,Vol. 14, pp. 176-95.
Sokal, RobertR., and Sneath,Peter H. A. (1963): Principlesof NumericalTaxonomy. San
Francisco: W. H. Freeman and Company.
" Statistical
Spaulding,AlbertC. (1953): Techniquesforthe Discoveryof ArtifactTypes."
AmericanAntiquity, Vol."18, pp. 305-13.
Spaulding,AlbertC. (1954)· "Replyto Ford." AmericanAntiquity, Vol. 19, pp. 391-93·
Stebbins,LedyardC. (1969): Commentson the Search fora Perfectbystem. Taxon,
Vol. 18. do. ^^-ςο. "
Steward,JulianΗ. (1941): ReviewofColton,1939." AmericanAntiquity, Vol. 6, pp. 366-67.
H.
Steward,Julian (1954) : Typesof Types." AmericanAnthropologist, Vol. 56, pp. 54~57·
" NeededforSignificant Numerical
Steyskal,G. C. (1968): The Numberand Kind ofCharacters
Taxonomy." Systematic " Zoology,Vol. 17, pp. 474-77·
Suttles,Wayne (i960) : AffinalTies, Subsistenceand Prestigeamong the Coast Salish.
AmericanAnthropologist, Vol. 62, pp. 296-305.
Taylor,WalterW. (1948): A Studyof Archeology.Carbondale: SouthernIllinoisUniversity
Press. ÍIQ67 ReorintEdition.) "
Vayda, Andrew,Leeds, Anthony,and Smith,D. (1961): The Place of Pigs in Melanesian
Subsistence." Proceedings of the 1961 Annual SpringMeetingoftheAmericanEthnological
Society,pp. 69-77. Seattle: Universityof WashingtonPress.
Vayda,AndrewP., and Rappaport,Roy A. (1968): Ecology,Culturaland Non-Cultural. In
Introduction to CulturalAnthropology (ed. by James A. Clifton). Boston: Houghton,
Mifflin Company,pp. 477-97. "
Wheat, Joe Ben (1953): An Archaeological Surveyof the AddicksDam Basin, Southeast
Texas." River Basin SurveysPapers, No. 4. Bureau of AmericanEthnology, Bulletin
No. I«>4. PP. Ι4λ~2 52.
"
White,J. Peter (1967): Ethno-Archaeologyin New Guinea: Two Examples." Mankind,
Vol. 6, pp. 409-14. _. _
Willey,GordonR., and Phillips,Philip (1958): Methodand Theoryin AmericanArchaeology.
:
Chicago University Chicagoof Press.

David H. Thomas.

This content downloaded from 86.29.223.245 on Fri, 27 Jun 2014 18:53:50 PM


All use subject to JSTOR Terms and Conditions

You might also like