binford (1964) a consideration of archaeological

17
SAMERICAN AN VOI.UME 29 APRIL 1964 e, FACULTAD DeFILosoF / AyL MUSEO ETNOGRÁFICO BIBLIOTECA Nl'MBKR 4 Oc4oí;ists n íntion of A CONSIDERATION OF ARCHAEOLOGíCAL RESEARCH DESIGN* LE\VIS R AiiMR.Mrr It is argucd that the methodology mo.st appropriate or the task of ¡solating and studyinu processes of cul- ural change and evolution is one which is regional in ¡cope and cxecnted with the aid of research dcsigns '. on the principies of probability sampling. The anous types of observational populations which archae- must study are discusscd, toyethcr with an cvalu- the methodological diffcrences attendant u pon -fndequatc and reliable investigación of each. Two basic IsamplinK' universes are discusscd, the región and the site, "ftogcthcr with their mcthodoloyical and rescarch-dcsign Jpcciiliarities. Thesc are uscd as a basis for discussiofl and .Ipast and currcnt rcscnrch programs are evalunted in tcrms ' fof \vhnt nrc believcd to be major limitations in obtainins; ithe "facts" pcrtinent to studies of cuhural processes. ~. , :\ TT SEEMS FA1R to generalize that archaeolo- •'á JL gists are becoming more interested in the ex- íplanatory potential which studies of paleoccol- I«"V> paleodemography, and evolution offer for . f incrcasing oür understanding of formal and Estructural chnnfe in cultural systems. Severa! f amhropologists have recognized a growing inter- |est in questions dealing with the isolation of con- ¡ditions and mechanisms by which cultural Ichanges are brought about (Adams 1960; Draid- fwood 1959; Haag 1959; Steward 1960). In short, "¿jwe scek answers to some "how and why" qucs- " Itions in additionto the "what, where, and when" questions so characteristically asked by archae- ologists. This paper is concerned with present- ing certain methodological suggestions, some of which must be adopted if we are to. make prog- ress in the study of processes and move archae- ology into the "explanarory level" of develop- ment (Willey and Phillips 1958: 4-5). In any general discussion of method and theory there is inevitably an argumentativa bias ¡¿X¡¿ on the part of the writer. It should be pointed out that I believe the isolation and study of cul- tural systems, rather rhan aggregates of culture- traits, is the only mcaningful approach to under- standing cultural processes (Steward 1960: 173- •~'- * This pnpcr was presented at the Annual Mectinj; of the Society for American Archaeology at the University of Colorado, Boulder. Colorado, May, 1963. 4). A cuhural system is a set of constant or cyclically repetitive"a?tTculations betwecn the so- cial, technological, and idcological extrasomatic, adaptive means available to a human population (White 1959: 8). The intimatc systemic articu- lation of localities, facilitics, and tools with specific tasks pcrformcd by social scgments re- sults in a strticturcd sct of spatial-formal rcla- tionships in the archaeological record. People do not co-opcrate in exactly the same \vay when pcrforming different tasks. Similarly, different tasks are not uniformly carried on at the same locations. As tasks and co-opcrating grotips vary, so do the implcments and facilities (Wagncr 1 1960: 88-117) of task performance. The loss, brcakage, and abandonmcnt of implcments and facilitics at different locations, where groups of variable structure performed different tasks, leaves a "fossil" record of the actual operntion of an extinct society. This fossil record may be read in the quantiíaliveiy variable spatial clus- terings of formal classes of artifacts. \Ve may not always be able to state or determine what spé- cific activities resulted in observed difFerenti?! distributions, but wc can recognire that activities were diffcrentiated and determine the formal na- ture of the observable variability. I have arguecí elsewhere (Binford'1962: 219) that we can re- cover, both from the nature of the populations of artifacts and ffom their spatial associations, the fossilhed structure of the tota! cultural system. The archaeological structure of a culture should, and in my opinión does, reflcct al! other struc- tures, for cxamplc, kinship, economic, and politi- cal. All are abstracted from the events which- occur as part of the normal functioning of a cul- tural system. The archaeological structure results \. from these same events. The definition of this^ 1 structure and the isolation of the archaeological remains of a cultural system are viewed as re- search objectives. Such an isolation can be made by the demonstration of consistent between-class correlations and mutual co-variations among classes of artifacts and othcr phenomcna. 1 he . •í:5

Upload: catedraehta

Post on 27-Apr-2015

539 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Binford (1964) A Consideration of Archaeological

S A M E R I C A N A NVOI.UME 29 APRIL 1964

e,

FACULTAD DeFILosoF/AyL

MUSEO ETNOGRÁFICO

B I B L I O T E C A

Nl'MBKR 4

• Oc4oí;ists níntion of

A CONSIDERATION OF ARCHAEOLOGíCAL RESEARCH DESIGN*

LE\VIS R

AiiMR.Mrr

It is argucd that the methodology mo.st appropriateor the task of ¡solating and studyinu processes of cul-ural change and evolution is one which is regional in¡cope and cxecnted with the aid of research dcsigns

'. on the principies of probability sampling. Theanous types of observational populations which archae-

must study a re discusscd, toyethcr with an cvalu-the methodological diffcrences a t tendant u pon

-fndequatc and reliable investigación of each. Two basicIsamplinK' universes are discusscd, the región and the site,

"ftogcthcr with their mcthodoloyical and rescarch-dcsignJpcciiliarities. Thesc are uscd as a basis for discussiofl and.Ipast and currcnt rcscnrch programs are eva lunted in tcrms

' fof \vhnt nrc believcd to be major l imitat ions in obtainins;i the "facts" pcrt inent to studies of cuhural processes.

~. ~Í, :\ TT SEEMS FA1R to generalize that archaeolo-

•'á JL gists are becoming more interested in the ex-íplanatory potential which studies of paleoccol-I«"V> paleodemography, and evolution offer for

. f incrcasing oür understanding of formal andEstructural chnnfe in cultural systems. Severa!f amhropologists have recognized a growing inter-|est in questions dealing with the isolation of con-¡ditions and mechanisms by which culturalIchanges are brought about (Adams 1960; Draid-fwood 1959; Haag 1959; Steward 1960). In short,

"¿jwe scek answers to some "how and why" qucs-" Itions in additionto the "what, where, and when"

questions so characteristically asked by archae-ologists. This paper is concerned with present-ing certain methodological suggestions, some ofwhich must be adopted if we are to. make prog-ress in the study of processes and move archae-ology into the "explanarory level" of develop-ment (Willey and Phillips 1958: 4-5).

In any general discussion of method andtheory there is inevitably an argumentativa bias

¡¿X¡¿ on the part of the writer. It should be pointedout that I believe the isolation and study of cul-tural systems, rather rhan aggregates of culture-traits, is the only mcaningful approach to under-standing cultural processes (Steward 1960: 173-

•~'-

* This pnpcr was presented at the Annual Mectinj; ofthe Society for American Archaeology at the Univers i tyof Colorado, Boulder. Colorado, May, 1963.

4). A cuhural system is a set of constant orcyclically repetitive"a?tTculations betwecn the so-cial, technological, and idcological extrasomatic,adaptive means available to a human population(White 1959: 8). The intimatc systemic articu-

lation of localities, facil i t ics, and tools withspecific tasks pcrformcd by social scgments re-sults in a strticturcd sct of spatial-formal rcla-tionships in the archaeological record. Peopledo not co-opcrate in exactly the same \vay whenpcrforming different tasks. Similarly, differenttasks are not uniformly carried on at the samelocations. As tasks and co-opcrating grotips vary,so do the implcments and facilities (Wagncr

1 1960: 88-117) of task performance. The loss,brcakage, and abandonmcnt of implcments andfacilitics at different locations, where groups ofvariable s t ruc ture performed different tasks,leaves a "fossil" record of the actual operntionof an extinct society. This fossil record may beread in the quan t i í a l i ve iy variable spatial clus-terings of formal classes of artifacts. \Ve may notalways be able to state or determine what spé-cific activities resulted in observed difFerenti?!distributions, but wc can recognire that activitieswere diffcrentiated and determine the formal na-ture of the observable variability. I have arguecíelsewhere (Binford'1962: 219) that we can re-cover, both from the nature of the populations ofartifacts and ffom their spatial associations, thefossilhed structure of the tota! cultural system.The archaeological structure of a culture should,and in my opinión does, reflcct al! other struc-tures, for cxamplc, kinship, economic, and politi-cal. All are abstracted from the events which-occur as part of the normal functioning of a cul-tural system. The archaeological structure results \.from these same events. The definition of this^1

structure and the isolation of the archaeologicalremains of a cul tural system are viewed as re-search objectives. Such an isolation can be madeby the demonstration of consistent between-classcorrelations and mutual co-variations amongclasses of artifacts and othcr phenomcna. 1 he .

•í:5

Page 2: Binford (1964) A Consideration of Archaeological

AM/:KICANT ANTÍQI//TV [ \\<t.. >). No. 4, 1964

: isolation and dcfinition of cxtinct c u l t u r a l .sys-. tems, both in tcrms of content and demonstra-' ble pattcrns 'of m u t u a l formal-spatial co-varia-- tion, can be accomplished. Once accomplishcd,such an archaeologica! s t ruc ture is amenable toanalysis in tcrms of fonn and complcxity; insliort, we can spcak of cul ture types. Methoclsfor correlating archaeologically defined c u l t u r etypes with s t ructural forms defined ¡n terms ofbehavioral attributes can be developed. Whenthis is accomplishcd, archaeologists and "socialanthropologists" will be in the position to makejoint contributions to the solution of commonnnthropological problems, a condftion thnt hard-ly obtains today.

In addition to m a i n t n i n i n g the position tha twe should strive to bolate the archaeologicals t r u c t u r e of extinct c u l t u r a l systems, i r is arguedt h a t chances in c u l t u r a l systems must be investi-ga ted with regard to the adapnvc or coping situ-ations which are presented to h u m a n popula-tions. If we are profitably to study process, wcmust be able to isolate c u l t u r a l systems ands tudy them in thei r adap t ive mi l i eu conccivcdin terms of physical, biological, and social cli-mcnsions. The physical and biological dimen-.sions neecl l i t r l e cxp lana t ion because anthropolo-gists are f a m i l i a r w i th the problems of the na-ture and stability of na tura l environment and inthe physical and demogrnphic human basis ofc u l t u r a l systems. Hovvever, íhe social dimen-sión is f rcquent ly excluded from considerationsof adaptación. The re is hu le neecl to belabor thepoint tha t as the densi ty and1 comple.xity of sep-árate socio-cultural sysiems increase wi th in amajor geographic roñe, the cul tura l means fora r t i c u l a t i n g and "adjusting" one society wi thanother become more complex. Certainly thecoping situations and.hence the adaptive stressesassociatccl with a changing pattcrn of socio-political distr ibution within a major zone mustbe considered when a t tempt ing to understandchanges in any given system (Gearing 1962).

í~Xs long as "cultures" are defined in terms ofstylis_tic similarity, and the qucstion of possible

j Bifferences in the material inventory of func-í tional classes and in the internal structure of thei assemblage is unanswered, there is l i t t l e possi-; bili ty of dealing realistically with questions of( process. It is a system that is th'e seat of process.

Because of these convictions I will frequentlymention the "regional approach" or the detailedand systematic study of regions that can be ex-pected to have supported cultural systems. The

.'C

cxtent of such regions wi l l vary because it isrecogni:ed t h a t c u l t u r a l systems diífer grcatly inthe l imi t s of their adaptive range and milieu.As cultural systems become more complex, theygencrally span g r c a t e r ecológica! ranges andenter into more complex, widespread, extraso-cietal intcraction. The isolation ancl definición'"!of the contení, the stnicnirc, and the range of ¡a cu l tu ra l system, together w i th its ecológica! irelationships, may be viewed as a rosearen oh- .

J_ecti_v£. Admittedly it is an obje"cfive~\vrnch_J

may or may ñor be successfully accomplishedunder any given research design. The researchdesign should be aimed at the accomplishmentof this isolation which, I beüeve, is most profit-ably prosecuted wi th in a regional un i r ot invcsti-gation. Undcr cu r r en t programs of salvage ar-chaeology and increased íoundat ion support forarchaeological research, we are being given theopportunity to study major regions intensively.In spire of the opportunit ies cu r r en t l y ava i lab le ,it is my impression that very l i t t l e thought hasbeen given to research clesign. Methods a n d -approaches u f i l i z e t l in such inves t igat ions scemto be l i t t l e more than expanded or grea t ly en-larged field sessions of the type t b a t has t r ad i -t i o n a l l y character i red Amer ican nrchai -o lo i r icp!data collection. Fo be sure, the work may beneater, more atrention may be given to stratig-raphy, more classes of phenomena may be ob-servecl and collected than in the field work o!years past; yet the general methocls of data coi-lection and observa tion remain unchanged. Iwish to argüe tha t current lack of concern wi ththe development of planned research dcsignsgenerally obvia tes the recovery of data perrinentto questions which derive from current theo-retical intcrcsts. Investigatory tools must fit thejob; current field procedures werc developed t;1

provide data re levanr to a l imitcd number oíproblems. Concern has bcen wi th problems ofstylist ic chronological placement and histórica!continuity bctwecn and among archaeologicallydefined units. The methodological tools devel-oped fvir the investigación of such problems areinappropriate for supplying información rele-vant to our broadening research intereses incul tura l processes.

MHTHODOLOGICAI. PROBLHM ÁREAS

Laymen frequent ly ask: "What are you dig-ging for?" I think most of us will agree that weare digging to recover facts for the elucidation oípast cultures. In the absence of explicit state-

Page 3: Binford (1964) A Consideration of Archaeological

BlNFORD ] ARCHAEOLOG/CAL RESEARCH DESKW 417

ments conccrning the kinds oí facts which ar-chaeologists hope to obtain, we can only assumethat we know how to recoVer the "pertinent"ones. Most of us will agrec that this is not true._I not only have becn unable to use other investi-gators' data, but I have also frequently foundmy own data lacking in many important "facts"— facts which could have been collected had Ibeen aware of the questions to which the givenobservation was rclevant. For instance, I re-cently wanted to demónstrate that most of thesites in a part icular área were located adjacentto streams. This was impossihle beca u se I hadno data as to where the archaeologist reportingon the arca had concentrated his survey efforts.Was the fa i lure to report sites in áreas not adja-cent to streams the result of sites being absent,or was it simply a lack of investigación in thoseáreas not adjacent to streams? In another in-stance, I wanted to compare the relative densityof Middle Woodland sites in two majar riverbasins in order to make statements about therelntive occupational intensi ty in the two áreas.This was impossible because I was unable todetermine whether or not the reponed differen-t i a l densities were the result of diffcrences in theintensity of survey, presencc or nbsencc of forestcover ; if l°ccting ¡lie l ikchhood of silo.-- being recog-ni:ed, or di fFerent ia l aboriginal use of the valleysunder consiclcration. Such unccr tnint ies makeit obvious that we are concerned with answeringquestions for whicli our resenrch desipns, fieldmethods, anci reporting procedures are not adc-qua te to supply the "pertinent facts." Sucha situation cannot be prevented entirely, but wecan strive to devise techniques for gathering thefacts which are pertinent to questions cur ren t lybeing asked of our data. As the general theo-retical developmem within archaeology goes for-ward, more and more fncts previotisly ignoredw i l l be recogniíed as important and per t inent .We can look forward to continucd concern withkeeping our investigatory tools sufñcient to thetnsk; in short, we wi l l be increasingly involvedin the development of new and improved re-search designs and methodologies by means ofwhich they may be opcrationalized.

One clue to a "methoclological problem área"in our current prácticos can be found in anotherqucstion which laymcn are apt to nsk: "How doyon know where to dig?" My answer to such aquestion is that we dig where there are surfaccindications of past occupations or cul tura l ac-t i v i t y - — a secmingly accurnte answer, yet we do

not excávate every location which yields surfaceindications. What are the rneans whereby we /select certain sites for excavation and not others? j_A quick revicw of 37 regional reports spanningthe period from 1954 to 1963 failed to revcal asingle exposition by the authors as to the criteriathcy utilized in selecting sites for excavation. Atypical statement may read as follows:

Durinií this t ime a total of 51 sites was located in sur-vey \vork and 13 of the more important sites were exca-vatcJ to some extent.

It is my impression that there is no single setof criteria for selecting "important" sites. Somearchaeologists seicct sites because they representa time period. Others are selected because theyare largc and productive. Certainly some havebeen excavated because they were accessible toinodern roads. Despitc the lack of systematicstatement, it is rcpeatedly mentioned that sitesare representative or that they are large andyield much material. Less frequently, economyis cited as a reason for selecting a particular site.Although it is not commonly cxpressed, wc maygenerali:e that archaeologists want representa-tire and reliable data u'irrun the bmmds of tlieirrcstrictcd time and mnnctary rcsoitrces. Thisis p rac t i cn l ly the def in i t ion of the aims of mod-ern sampl ing procedures. Sampling, ¡is iisedhere, does not mean the mere substitution ofa par t i a l coverage for a total coverage. It is thescience of controll ing and measuring the relia-bil i ty of Information through the theory of prob-abi l i ty (Deming 1950: 2). Certainly we are a l laware t h n t we must substi tuto par t ia l coveragefor complete coverage in our investigador!. Giv-en this s i tuat ion, there is only onc currentlyknown means for accomplishing coverage sothat the result s can be evaluated as to their re-l i a b i l i t y in renresentation of the population in-vestigated. This is through the application ofsampling theory in the development and execu-tion of t lata-collecting programs.

!n this discussion of sampling we shall incro-cluce certain terms tha t are used by writers onsampling and a t t endan t statistica! problcms. Anrm-er5c is the isolated field of stucly. In mostcases archacological field work is conductedwi th in a universo of tcrritory, a univcrse spa-t i n l l y defined. A ¡>o¡ntlation consists of an aggre-gate of analytica! units within the universe sothat , p.t least in principie, each unit may be as-

Page 4: Binford (1964) A Consideration of Archaeological

AMER/CAN ANTíQU/TY [ Voi.. 29, No. -í, 1964

jgned a dcfinite location for n given unir of¡me. The population has a distribution in spacepnsisting of the aggrcgate of individual locn-¡ons (Duncan and others 1961: 21). In addi-jon to a distribution, we can speak of the sf>a-ial stntctiirc of a population. S.tructure sug-íests a pattern of interrelationships among dis-¡inguishable parts of an organi:cd whole (Dun-pn and others 1961: 2), and for our purposcs[he spatial structure of archaeological popula-tions derives from the complex interrelation-ships between people, activities, and material¡Ítems within a cultural system. In addition, we¡níay speak of the /orm o/ a population, which isjthe nature and quantitatively variable consti-jtution of subclasses and the relative frequency¡of analyt ical units.'( The application of the method of probabilityi snmpling presupposes that a universe can be sub-| divided into distinct and identifiable units calledI sample units. These units may be natural units,i such as sites or individual projectile points, or! thcy may be arbitrary units, such as 6-inch levéis

in an excavation, or surface áreas deftned by aprid system. Regardless of the basis for defini-tion, the application of the method of probabili tysampling presupposes the availability of a list ofall the potential sample units within the uni-verse. This list is called the jrame and providesthe basis for the actual selection of the sampleunits to be investigated. The frame varíes withthe nature of the archaeological populationundcr investigation. When a population of sitesis sampled, the frame is normally a list of siteswithin a stated universe, such as the alluvialbottoms of the Rock River between two specifiedpoints. When partial coverage of a populationwithin a stated universe is attempted, the sampleunits are selected from the frame so that allunits of the frame have an equal chance of bc-ing choscn for investigation; the selection is gov-erned by the "laws of chance" alone, maximir-ing the reliability of the sample.

Before we approach the subject of differentmethods of probability sampling and their rangeof application in archaeological research, cer-tain principies which underlie and guide the re-search design aimed at the proper and emcientexecution of sampling techniques will be men-tioned. This presentation is adapted from a sec-tion entitled "Types of Sampling" by Parten(i95oy:

(1) The population to be sampled and the uni ts com-posing it must be c lcnr ly clefincd so tha t thcre will he noquestion as to whnt the sample represems.

(i) A universe pnr t i t ioned by 3 frame composed ofmany small un i í s is p rc fc rab lc to one composed of fewerbut larger units. This is a safegiiard against accidentalinclusión of nn unreprcsentative amount of "hetcro-gcneity" in anv given sample.

(3) The u n i t s of the f r ame should be approximatc lycqual in si:e. This c l imina tes bias which could resultfrom a systematic relationship between the s tructure andthe si:e of íhe population.

(4) All uni ts should be indepcndent of cach otherso that if one is drawn for sampling, it will in no wayaffect the choice of another.

(5) The same units should be used in sampling, tabu-lation, and analysis. A sample of mounds is of no use ifgeneralirations abotit general site distributions are bein.eattempted.

(6) The universe musr be present or cataloíied so tha tcvery unir in it is listed or can be given an ident i fyingsymbol.to be used durin<,' the drawing of a sample. Forinstance, a grid system is estabiished with 12 ten-footsquares on a side, and 20 of the 144 squares nre chosenfor excavatiún by a random method. Later, ir is decidedto extend the Rricl system six more squares in one direc-tion. fhe enlarged system does not havc the randomcharacrer of íhe sample drawn under the f r ame deftnedhy the original grid system.

(7) The inctb^c! of (!r; iwin>? the snmple .should becomplctcly indcpendcnt of the character is t ics to be ex-amined.

(S) In order for the sample to remain random, evíryu n i t drawn must be accessible. For instance, in the casewhere a site has been selected for sampling and the prop-er ty owncr refuses permission to dig, inaccessibility biasesthe sample. This is particularly true because refusa! ofpcrmission may be rclatcd to ideas of the "valué" ofmaterials on his property relative to those on otlurrs'property.

With these ¡írinciples serving as a background,we can turn to a discussion of types of samplingand their ranges of applicability to archaeologi-cal inves t iga t ion . Althcugh there are manytypes of .sampling, only two will be discussed:simple random sampling and stratiñed sampling.

SIMPLE RANDOM SAMPLING

This is by far the simplest of the methods ofprobability sampling. It implies that an equalprobability of selection is assigned to each unitof the frame at the time of sample selection.The term random refers to the method of select^ling the sample units to be investigated rather jthan to the method of investigating any given ¡unit. A practical procedure for selecting a ran- (dom sample is by utilizing a table of randomnumbers (Arkin and Colton 1957: 142). Thel

Page 5: Binford (1964) A Consideration of Archaeological

BlNFORU ' ARCHAEOLOG/CAL RESEARCH DES/GN

procedure is to (1) determine the number of'units in the frame, identifying them seriallyfrom one to (n); (2) determine the desired sizeof the sample, that is, actual number of units tobe investigated; (3) select the required series ofnumbers from the table of random numbers(the required series is the number determinedsufficient to constitute a representative sample ofthe population); and (4) investígate those unitsin thc frame that corrcspond to the numbersdrawn from the table of random 'numbers. Itcannot be overemphasized that this is a tcch-nique for selccting the units to be investigatedand does not refer tp the procedures used ingathering the data.

STRATIKIED SAMPLINGIt can be shown that the precisión of a sam-

ple depcnds upon two factors: (1) the size ofthc sample and (2) the variability or hetero-geneity of the population being sampled. If wcdesire to increase the precisión, aside from in-creasing thc sample size, wc may devise meanswhich will effectively reduce the heterogeneityof the population. One such procedure is knownas the method of stratified sampling. The pro-ccdure io to p a r t i u ü n thc u n i v e i s e or d iv ide iíinto classes, each of which is treated as an inde-pendent sampling viniverse from which simplerandom samples are drawn, following thc meth-qds outlined above. This procedure has a num-ber of advantages. Classes may be establishedwith regare! to different variables that one wishcsto control, which makes possible the reliablecvaluation of variability in other phenomcnawith respect to the class-defining variables.

As has bcen suggested, the size of the sampleis an important factor for consideration in striv-ing for reliabil i ty in the sample as representativeof the population. An optimum sample is onewhich is efficient, representative, reliable, andflexible. The sample size should be small enoughto avoid unnecessary expense and large enoughto avoid cxcessive error. To arrive at a samplesize which is considered optimum in terms ofthe above criteria, there are a numbcr of factorswhich must be considered, each largely inte-grated with thc other rather than independent.Each wi l l vary, in different sampling situations,as to thcir relative importance in influencing de-cisions about thc appropriate sample size. Ofparticular importance to archaeologists is thcrcalization t h n t the size of the sample neccssaryto mect normal requirements of r e l i a b i l i t y and

^ veconomy is greatly affecto^by fkfi^purnber ofsubclasses into which the recoverea data will bedivided. For instance, in the case of a sampleof ceramics composed of 100 sherds, it is likelythat if only two "types" are represented, thesample may be sufficient to give a fairly reliableestímate of the relative proportions of the twotypes in the population. On the other hand, ifwithin the sample there are 15 "types,". then thereliability of the sample as an estímate of therelative proportions of the recognized types inthe parent population is very low. A furthercaution applies to samples drawn from multi-component sites where there are clearly severalrecognizable populations of sepárate historicalorigin. In thc latter case, some of the subclassesare independent of one another, and any givensubclass may be representative of only one of themúltiple historical populations. Also, in the lat-tcr case, sample size must be determined by therelative frequencies of the smallest independentsubclass. In other words, sample size should belarge enough to give reliable measures of thesmallest important brcakdowns made within thesample. If the population is relatively homo- ]gencous, thcn sample size may be relatively'smnl i nnd s t i l ! rel inblc. On thc othrr hnnd, ifthe population is heterogencous, more observa-tions are needed to yicld reliable data.

This discussion of sampling is very elemen-tary, but enough has bccn presented to serve asa bnsis for evaluation of the types of observa-tional populations which archaeologists investí-gate and thcir invcstigatory pcculiarities as abasis for fur ther discussion of research clesign.

BASIC UNITS OF ARCHAEOI.OGICAL OBSHRVATION'

Albcrt Spaulding has provided us wíth aclassic statement of what we as archaeologistsare doing, thereby setting forth an operationaldefinition of thc field of archacology. Spauld-ing's introductory statement is reproduced hereas a point of departure for fur ther discussionof archacological data collection.

A scicnce deals with some class of objects or events intcrms of some spcciiicd dimensions of the objects orevctus. Tile simplest (and most elegant) of the sciences,mechanics, has all physicnl objects as its center of at-tent ion, and the dimensions of thesc objecrs as studied areIcngth, mass, and t ime; ¡.e., roughly speaking mechanicshas thrce kinds of mcasuring instruments: a yardstick, aser of scnles, and a clock. The imerrelaüonships and trans-format ions of mcasuremcnts with thcsc scales is the bus¡-ncss of mechanics. It is clcar tha t prchistoric archaeology¡tlso has a class of objects, a r t i f n c t s , as its center of atten-

Page 6: Binford (1964) A Consideration of Archaeological

rAMÍiK/CA.Y A.YT/yU/TV \'oi.. :y, N'o. 4, 1964

lion. The conccpt "ani fnc t" prcsupposcs tile idea of cul-iurc, which I wil l trc;i t as a «¡ven. Thus wc can define anar t i f ac t as any mater ia l resille of cu l tu ra l behavior. SÍIKCcul tura l behavior is our u l t íma te referent , it follows thatwc are interestcd in only ihosc properries, ch.iracteristics,a^pccts, and a t t r i hu re s of a r t i f ac t s which are the rcsult ofor havc a M^nif icam relat iou to c u l t u r a l behavior. \Vhatare the dimensions of a r t i f ac t s whosc interrclationshlp.sare the spccial busincss of archaeolocy? Pl'ainly there aretwo in ihc stricr sense of dimensión: time and spacc. \\'cwant to know where and \vhcn ar t i íacts werc made, uscJ,and deposited. Plainly there is nnothcr class of dimen-sions fundamental to archaeo!of:¡cal study; tlie many di-mensions which are sets of physio-chcmical propcrties ofthe artifacts. Wc can firoup them for convenicnt rcfer-cnce undcr the Inbcl, formal propcrtic?, and collectivelyas the formal dimensión. \Ve are now ¡n a position to de-fine archaco'iocy ns the study of the imerrelationships andtransformations of artifacts with rcspcct ro the formnl,temporal, and .spatial dimensions. As a footnotc, forma!and spatial attributes can be observe*.! directly, but tem-poral attributes are ahvays infcrred from formal andsp.itial attributes. Indócil strictly spcaking ar t i facts areobjccts and do not have temporal attributes — they mere-ly exist. But a r t i fac t? do imply cvents, and events do havethe property of occurriní,' at a definitc t ime, so whcn wcspealc of the t e m p o r a l a t t r i h u t e s of an a r t i í a c t , \vcrca l ly refer to an in le rencc abolir some event or proco--implicJ by the fo rma l and/or spatial a t t r ibutes of the a r t i -fact. This lenvcs us with describíng and orderiní: formaland spatial a t t r ibu tes as the p r imary task. These are thecmpir ical data of archaeolony and this describing ando r J e r i n ^ a re prcrccj ' j i s i tc to the chronoloijical i n f e r c n c c ^(Spaulding, mimcoyrapheJ versión, rcvised 196C: 437-9).

Artifacts as a class of phenomena rcpresenta number of different typcs of populations whichare definable in terms of their spatial, formal, orspatial and formal attrihutes takcn in combina-ción. Because of the different nature of ar t i -fnctual populations, investigation of the severalrecognizable populations must differ as regareisthe appropriatc mcnns to provicle the nccessaryinformation for their formal, spatial, and tem-poral analysis. The several classes of art ifactunlpopulations recogni:ed hcre will be termcd ty/x'soj observational ¡w^nlations, since it is arguedthat different sampling techniques, and heneeresearch strategy, are necessary for the investiga-tion of each.

POPULATIONS oi: CULTURAL ITF.MS

Cultural items are discrete cntities, the formalcharactenstics of which are at least partially theresult of cultural activity or events. A furtherqualiíication is that the formal charactenstics ofthe ítem are not altered through removal fromtheir matrix; they are transportable and maybe formally analyzed without recourse to infor-mation aboiu their provenience. This is not to

r-ay that such in tornia t ion is not crucia l to intcr-pretation of the formal propcrties, only thnt theformal properties thcmselvcs are independent oftho matrix and provenience associations.

The form of the cu l tura l itcm may vary inicrms of the function of the itcm as an elementin the cultural system, for cxample, technologi-cn l l y (manufac tu r ing techniques and raw ma-terials) or stylistically.

The sampling universo for the investigation ofpopulations of cul tural items is necessarily thesite. The sampling and field-observation proce-dures utilired do not alfect our ability to analyíeitems formaliy, but they greatly affecc our abilityto study the tlistribution, form, and s t ructure ofa ¡'o^nlaiion oj cnlturaí ¿tenis. It will be re-membered that a population necessarily has spa-tial attribute.s both in its distribution and itss t ruc tu re . Sampling control is therefore neces-sary to próvido data for the description of pop-ulations of cultural items. Wc as anthropolo-uists hope to be able to assess rhe range of formalv a r i a b i l i t y in classes oí cul tural Ítems and tostudy their distribution and populación struc-ture in terms of spatial clusters of quantitatively »»v a r i a n t class nssociations. It is necessary to de- \l ino n c c u r a t e l y tTTé7 range of formal variation\vi thin classes ot cu l tura l ítems as woll asj'ela-tíve frcqu¿!icjes among recognized classes with-TrrThe~population. Whilc excavating, we haveno precise knowledge of the boundarics of thepopulation of cultural ítems being investigated,and we generally sample in terms of areal unitsdesigned to cover the tcrri tory defined by thepresence of a r t i f n c t s . We are sampling "artífactspace" as a means to both definición and segre-gation of populations of cultural ítems. We canonly accomplish this through cxcrcising tightspatial controls for gaining information neces-sary to the analytic determinación of what cul-tu ra l items are, spatially and temporally clust-ercd one with another and with other arti-factual materials. Such insights are a clue to the"role playcd" by various items in the operationof cxtinct cultural systems. Similarly, the samecontrol is necessary to the definition of the spa-tial structure and form of the population. Wewant to utilire techniques which will insure re-liable and representative data regarding therange of formal variability within a given sub-class of cultural items, the content of the popu-lation of cultural items defined by the relativefrequencies of the recognized classes, and thestructure of the populatíon defined by the spa-

Page 7: Binford (1964) A Consideration of Archaeological

BlNFORD ] ARCHAEOLOGICAL RESEARCH DESIGN 431

tial structure of between-class associations ofÍtems and other classes of artifacts.lError arising in the sampling of populations

of cultural ítems normally results from (1) in-complete and nonrepresentative coverage of theuniverse; (2) failure to partition the universeso that a single undifferentiated collection ismade, thereby excluding the possibility of invcs-tigating the homogeneity or heterogeneity of thepopulation; or (3) samples are far too smallto allow the adequate evaluation of the varia-bility represented in a givcn class of culturalÍtems or to yield a reliable estímate of the be-tween-class relative frequencies.

POPULATIONS OF CULTURAL FEATURES

Cultural features are bounded and qualita-tively isolated units that cxhibit a s t ruc tura l as-sociation between two or more, cultural Ítemsand types of nonrecovcrable or composite mat-rices. The cultural feature cannot be formallyanalyzed or at least formally observed after itsdissection in the field. Many of the formal ol>servations must be made while the fen tu re is be-ing excavated. Features include such classes ofremains as burials, mouncls, structurcs, pits, andhcartbs. Formal vanauons aniong cul tura! íca-tures are depcndent upoii (1) thcir funct ionswithin the represented cultural system, (2) tecli-nology in terms of the raw matcrials uti l ired intheir production, (3) alterations occurring as arcsult of their participation in other natural sys-tenis, for example, organic decay, (4) their cul-tural history (how óften thcy were parts of suc-cessive cul tural events, rcsulting in their repair,secondary modification, and dcstruction), and

. (5) stylistic variation. Ünlike cultural Ítems,the cultural feature cannot be formally dcfincdwithout precisé and detailed obscrvntion and"analysis" in the field. The field investigatormust at least make decisions as to what attri-butes are cul tura l ly rclcvant and meaningfu lprior to beginning field obscrvation and record'ing. This adds an additional field burdcn to thenormal cxercise of sampling control character-istic of the investiíjation of cultural Ítems.

As in the case of cu l tura l Ítems, the samplinguniverse for populations of c u l t u r a l features isthe site. Similarly, our sampling proceduresshould insure reliable and rcprcsentative dataregarding (1) the variability within any class offeatures, (2) the formal content of the popula-tion of features, and (3) the s t ructure of thepopulation of c u l t u r a l features.

Sources of error which frequently arise insampling such populations are (1) incompletaand nonrepresentative coverage of the "artifactspace" so that, while the number of recoveredfeatures may be large, there is no way of demon-strating or determining whether the between-class frequencies are representative of the popu-lation present; (2) samples are far too small to

• allow adequate evaluation of the variability rep-resented in a given class of features or to yieldreliable estimatcs of the between-class relativefrequencies.

POPULATIONS OF CULTURAL ACTIVITYLocí: SITES

The site is a spatial cluster of cultural featuresor ítems, or both. The formal characteristicsof a site are defined by its formal content andthe spatial and associational structure of thepopulations of cu l tu ra l Ítems and features pres-ent.

(1) Nitc's víiry IR their <¡<.'/)<«¡n<ma¡ contc'.vr.Sites e.xhibiting primary depositionaj^contexthave not been altcrccTTn thcir fóYínal propcrtiescxcept through the natural proccsses of the de-cay of or^anic material, or the physico-chcmicala l t e ra t ion of fea tures and ítems since the penoüof occupnncy. Sites e.xhibiting secondary dcposi-tional context are those whose formal character-istics, defined in terms of soils, features, andítems, have been spatially altered through physi-cal movement or delction from the locí. Someor possihly all of the original associations be-twcen the various classes of a r t i f ac t s have beenchangecl. This disruption in the structure of thesite may havc occurrcd through the agency oferosión, geophysical changcs, or through dcstruc-tion as a result of later cul tural activity. Siteswith primary depositional context yield the mostcomplete archaeological record. Howcvcr, siteswith secondary depositíonal context must fre-quen t ly be studied in order to understand theregional distributíon of act ivi ty loci. -*

(2) Siri'i" vflry in thcir tíc'/iosírioiuií history.The culturally dependent characteristics of asite may have been the result of a single short-term occtipation, a single long-term occupation,múl t ip le occupation over a rathcr limited tem-poral span, múltiple occupations over an ex-tended period of time, or combinations of allof these.

(3) Sites vary in thcir culture history. Sitesexhibiting a complex depositional history may

Page 8: Binford (1964) A Consideration of Archaeological

H32 AMERICAN ANT/QÜ/TY \'o[.. 29, Xo. 4, 1964

iar may ñor exhibit a complex cultural history.

site could be repcatedly occupied by repre-intatives of the same stable socio-cultural sys-

[tcm for the same purposes, in which case therejmay be a complex depositional history with a[simple cultural history. Similarly, a site with ajsimple depositional history may exhibit a com-jplex cultural history, for example, an extendedI long-term occupation spanning a period of majorEstructural changes in the cultural system. A¡single locus may be sequentially occupied by¡ social units of different socio-cultural or socio-> polítical units, adding to the complexity of the

cultural history of the site.

(4) Sites and arcas irithm sites vary /imcnon-ally. Since sites are the result of cultural activi-ties performed by social units within restrictedspatial bounds, we would expect them to van-formally as a function of the activities of thesocial units represented. It is a known and de-monstrablc fact that socio-cultural systems varyin the degree to which social segments per formspecialized tasks, as well as in the cyclical pat-tern of task performance at any given location.These differences have spatial correlates withrenard to the loci of tnsk performance; henee we

"c.xpect sites to vary íormally and spatially withregard to the nature of the tasks performed atcach, and the social composition of the unitsperforming the tasks.

^ All possible combinations of the above-men-tioned basic forms of variation may occur atsites which archaeologists investígate. Archae-ologists must be prcparecl to make the pertinentobscrvations necded to define the form andstructure of the populations of artifacts and cul-turally relevant nonartifactual material present,and to isolate the form and structure of his-torically different archaeological assemblagesrepresented. Unlike populations of culturalÍtems or features where the normal universe \is the site, the sampling universe for populations >.of sites is of necessity a región. Once the archae-^jíologist has determined the relative homogeneityof the sampled site population as regards thehistorical and functional nature of the archae-ological assemblage present, he is in a positionto consider the nature of the site as a whole andto classify it within a typology of sites (based onthe attributes of both the form of the artifactualelements present and the structure of their spa-tial and formal associations). Such an approach

is the methodological aim of sampling a uni-verse of sites regionally clefined.

Two major sources of error arisc in the inves-tigation of site pópulatioñsTThe first source of er-¡ror is incompleto and nonrepresentative cover-¡age of the range of variation represented amongisites within the universe.This ariscs inevitably asa result of the "selection" of sites for investi-gation on the basis of criterin othcr than thosc ofthe method of probabílity sampling. For instancersites are frequently selected because of a highdensity of cultural ítems almost to the exclusiónof sites with low density. The density of culturalítems at a site is a formal attribute of the specificactivity loci and is only relevant to the selectionof sites for excavation as an attribute in a pro-visional site typology. A given universe mayhave very few sites with dense conccntrationsof cultural Ítems, while the number of sites ex-hibiting less-densc concentrations may be quitehigh. In this case, the sample of sites for investi-gation must he composed of a proportionallyhighcr number of sites exhibiting low dcnsitiesof cultural Ítems. The second source of error isfailure to sample with sufficient intensity toyield a reliable mcasure of the variability presentin the population. Inadequate sample size meas-ured by the number of invcstigated sites is uneof the major sources of error. Ideally, a sam-ple of sites shoultl be adequate to represent theformal range of variability in site form, the rela-tive frequencies of recognized site types, andtheir spatial structuring within the universe.

POPULATIONS OF ECOFACTS

In addition to the investigation of culturalÍtems, features, and activity loci, we must sam-ple populations of ecofacts. Ecofact is the termapplicd to all culturally relevant nonartifactualdata. Cultural systems are adaptive systems,and in order to understand their operation andthe processes of their modification, we must bein a position to define their adaptive milieü. Allj:hose.£lement«-w.m'chjepres£ntj^Jniorrn aboutthe £OÍ04:s<.pX articulation'between the cultural"systém and otlie?vriátu'rat^y1^eTft^rrrasrbé^píélS^Thls^^Ti^exTrlíi^ély^archaeological data collection and is accom-panied by many field complications in terms ofmethods of observation and sampling. The gen-eral class of ecofacts can be broken down intomany subclasses representing different popula-tions, such as pollen,; soil, and animal bone,each with specific attendant sampling problems.

Page 9: Binford (1964) A Consideration of Archaeological

BlNFORD ] ARCHAEOLOG/CAL RESEARCH DES1CN 433

However, for the purpose of this presentation,we will consider ecofacts as a single populationwhich, in general, requires certain methodologi-cal considerations distinct from the problems as-sociated with sampling artifact populations.

BASIC SAMPLING UNIVERSES

Although we have recognized four majortypes of observational populations, each differ'ing in the way it must be observed and sampled,there are only twobasic sampling universes in

jxcavation or_fiélcrwor}c, the reglQh~añ'd'the'site."Pópulations of sites must" be'iñvestigatecTwithíñ"a universe defined.in spatial terms, the región.Populations of cultural Ítems and features musthe investigated within a universe defined by thebounds of artifactual distribution at a given lo-cation, the site. Ecofactual populations may besampled wirhin both universes, depending onthe types of information desired. If culture typesare to be defined, it is cssential thnt we isolate areliable and representative sample of the popu-lation of sites charactcristic of a given culture.For adequate definition of types of activity lociwe need a reliable sample of populations of cul-tural Ítems and fearurcs assignable to any givenoccupat ion. In ordcr ;o obta in such in for rna-tion, we must have well-planncd research de-signs roored in the application of probabilitysampling procedures.

LlMITATIOXS OF CuRRCN'T PROCEDURES

It is my impression thnt archacologists havenot consciously aimed at sampling populationsof sites. They have concentrated on collecting"samples" of cul tural ítems within regionallydefined universes. The sites have been treatedlargely as "mines" for such Ítems. In excep-tional cases, wherc sites have becn intensivclyinvcstigated and populntions of cul tura l featuresstudied, there,. is little attempt to analyíe thejíopulntion of cu l tu rn l Ítems wijrh regard to itsspat ia l_strúcture or form, while inordinate at-tention is frequently given to describing the"norms" of recognized formal subclasses withinthe population. When cul tura l features are in-vestigated, they are usually reported cartographi-cally, with li t t le attempt to conduct a detailedformal analysis aimed at the description of typesof features. One rarely finds a report in whichcorrelntions between the spatial structure of pop-ulations of cu l tura l features and Ítems havebeen attempted as a matter of "standard" pro-cedure. For the most part , nrchaeologists have

concerned themselves with vertical spatialanalysis, and the search has_been,/or stratified

_sites.Jn_which ajjmited^!test,pit". wül yield a^stylistic sequence.that.may be..used..to^deyelop^a regional chronolqgy. This results in "culturesbeing isolared on a regional ba~sTs"líñcr"cTeríñed

^argelyln^terrns oí tEe" Stylistic cRaraHeristÍ5§,.p¿j^rnorjjjitems. The resulting information is in-sufficient for the structural definition of arti-fact assemblages and site typologies in preciseterms. Cultural, "tel usi velyir^srylisjric .

'OrTtHe other hand, there has been inordinateinterest in certain classes of cultural features,such as burial mounds and platform mounds.It was carly recognized that such mounds wereexcellent "mines" for exotic and artisticallyplcasíng objects and were therefore attractiveto untraíned investigators and relie hunters.Work in mounds, whether prompted throughhumanistic intcrests or through the "salvagemotive," has contributed inordinately to our"sample" of art ifactual data which serves as thetaxonomic basís of many archaeologically de-fined cultures. This lack of representative dataplus inadcquate information on the form of thefea tu res is a real and l ími t ing bias in the datacurrently available for study. Data have beengathered in terms of problems which concén-trate investigations on populations of culturalÍtems at the expense of and to the exclusión ofcultural features. Investigaron has also beenconcentrated on par t icular types of obvious fea-tures, such as mounds, and there has been verylittle awareness that the aim of archaeological

> investígation is the definition of the structure ofan archaeological assemblage in addition to itscon ten t. These factors have contributed greatlyto our current. inabi l i ty to deal systematicallywith archaeologícal data.

Current interests demand that we do not per-petúate these limitations in our methodology.We must approach our work wíth the methodo-logical ideal of sampling a spatiarunivérsé^ re-gardless of whether it is cónducteH underTarge-scale regional research programs or over an ex-tended period of time through a series of small-scale investigations. Such samplingys aimed ato!üí? 'n Jüo_íLJl£ÜAÍ!Í£».áí3S"représen ta ti ye sa mplé""of the range of variation in formal-structuraíjerrns"orsites^vjfTnnTO£rgiveri~7egion. Selectionof sites for excavation shouT3 be m'ade on thebasis of some method of probability sampling asthe best means of insuring that the expenditure

Page 10: Binford (1964) A Consideration of Archaeological

AMERICAN' ANT/QI 'ITV YOL. 2-5, No. 4, 1964

of time and moncy in cxcavntion will yield thedesired information. Sites selected for cxcava-tion must be investigated so that thcy can beformally defined from the standpoint of thenature of the populations of cultural Ítems pres-

[ ent, but equal attention must also be given tothe population of cultural features. This Js theonly waj^ jo_.ap_proacl\jhe necessary taslc of de-velofjnTgji_site typology in functional and struc-

jural terms) an absolute necessity for the defini-lion and isolation of trie archaeologícaTstrüctü'fé

^jj.nct^cul.yral^yste.ins. "The latter is judgeda^necessary,stcp toward tlíeTscientific investiga-'¿ion qf cultural processes.

A HYPOTHKTICAL RHSEARCH DESIGN

In an initial attempt to think through someoí the practical problenis associated with the de-sign and execution of a research projjram \vhichattempts to operationalhe some of the sugges-tions advanced thus far, I will present n "hypo-thetical" research program. Hypothe t ica l isplaced in quotes because many of the ideas andproblems discussed are the result of work cur-rently being under tnken in tlie southern pnr t ofIllinois, speciñcnlly in the Carlyle Reservoir. Re-gardless oí dio projcctcd i r n p l c m e n t n t i o n ofmany of the ideas set forth, the program remainshypothetical because the suggestions are un-tried and undcmonstratecl. It is hopee! that bypresenting these ideas in the form of a research"model," othcrs may gain a clearer understancl-in» of what is intcndcd by the application ofprobability sampling approaches in field work.It is fur ther hoped that this model can serve asa "whipping boy" for the improvement and fur-thcr development of field methods and the exe-cution of well-planned research designs.

Let us assume that we are given the task ofinvestigating the prehistoric remains within aregión. Our aim is to determine with the great-est degree of precisión and reliability the iiaturcof the extinct cultural systems represented forthe cntire range of human occupation. \X/e mustface the problcm of isolating the variable cul-tural Ítems, cultural features, and sites of ac-tivity for the cultural systems represented. Inaddition, we must gather ecofactual data as abasis for understanding the way in which theextinct cultural systems participated in the re-gional ecosystems of the past. We want to knowthe internal structure of the systems, the degreeof structural differentiation and functional spe-cialization of the social segments, as well as how

thcsc segments werc articulated into a func-tional cultural system. Wc want to know thedemographic basis and how it varíes with respectto isolated structural changes in the cultural sys-tems. In short, we want to know all wc canabout the structure and functioning of the ex-tinct cu l tura l systems and how thcy relate oneto another as regards processes oí chance andevoliuion.

The initial problem is the location of thevarious loci of past cu l tura l activity within theregión. This phase of the work should be di-rccted toward tletennining the dcnsity and dis-tribution of activity loci with respect to classesof ecofactual phenomena, such as plant com-munities, physiographic features, and soil types.In order to accomplish this task, there is only oneappropriate procedure short of complete cover-a^e, a procedure rooted in some form of probabil-ity sampliny. One suggested approach is to strar-ify the regional universe on the basis of ecofactu-al cri ter ia judged desirable to control, such as soiltypes. If we assume for purposes of presentationthat soil types have been decidcd upon for theareal stratification, in most cases the bounds ofthe vnrious soils will be defined fair ly accuratelyon n soil mnp, and we can simply determine theextent of each in square miles, acres, or otherappropriate units. Having accomplished this,we can impose a frame within each samplingstratum (áreas of common soil typc). It will beremembercc! tha t n universe partit ioned bymany small sample units is preferable to onewith fewcr but larger units, and that the unítsof the frame should be approximatcly equal insi~e. Using these guides, we can impose a gridsystem over the áreas of the various soil types.The actual size of a given unit in the framewould be determined by considerations of sur-vey logistics and the need to have múltiple butalso practicable units for investigation. For pur-poses of presentation, it is assumed that the gridis composed of squarcs equaling one-half squaremile. We would then count and enumérateeach unit in the sepárate frames for each sam-pling stratum (soil type). The next methodo-logical consideration is arriving at a "samplesize." This can be quite complicated. For pur-poses of argument, it will be dismissed and wewill assume that a 20% areal coverage withineach sampling stratum has been judged sufficient.The next step is to draw the sample for eachsampling stratum, and this may be accomplishedby use of a table of random numbers. The sam-

Page 11: Binford (1964) A Consideration of Archaeological

DlNFOKD ] ARCHAEOLOG/CAL RESEARCH DES1GN •435

plíng units within each frame will then be com-pletely surveyed for purposes of locating sites.

What are the advantages of such a proce-dure? If executed under ideal conditions, it willpermit the objective evaluation of site densityin terms of ecofactual controls and also providedata relevant to summary statements about theintensity of past activities in the región as awhole within definable limits of error. The pro-cedure will also permit the concentratl'on of ef-forts on intensive study áreas, making the logisti-cal expenditure less than if the entire regiónwere surveyed in a haphazard fashion. In addi-tion to these advantages, it elimínales "hiddenbias" in the" form of differential attention paidto ecological situations which the invcstigator"feels" were preferred by prehistoric inhahitants.By following such a plan it is possiblc to dem-ónstrate the ecological prcfercnces of past occu-pants of the región.

On the other hand, there are_ certain prob-lems which arise with any attempt to sample inthis manner. Of primary importnnce are theconditions of the área i tself in terms of the typeof cover, prcsence or absence of modern com-munities, and dis tr ibut ion of ag r i cu l tu ra l lancl.Such factors could var ionsly affect one's accessto the land for site locational survey as well asthe relntive efTiciency of observation. These convplication? are not ncw ñor are they stumblingblocks to the suggested procedures. They arepresent no mnt tc r what type of research designwc attempt to execute. The advantage of thisparticular approach is that it próvidos a method-olopical frame of reference for documenting andcvaluatinj: such bias. There are many ways tocorrect complicatecl sampling conditions. \X;henapproaching the problem of locating sites as out-lined, it becomes imperative that the investi-ga tor concern himself with the control of biasresulting from different ia l su rvey cond i t ions ,somethiriK not generally considercd under nor-mal haphazard survey procedures.

Assuming tha t wc have executed a researchplan as outlined, the next step is to ilt-jine 5/>u-ri«!!}' and sample i n i t i a l l y the populations ofcu l tu ra l ítems present at each of the identif iedloci of cu l tura l nctivity. This is prerequisite tothe evaluat ion of the formal characteristics ofthe sites themselvcs; t he -u l t ima t c aim is a classi-fication of ac t iv i ty loci as to their degrecs of simi-l a r i t y and diffcrence. A working taxonomy ofsites is a necessary prerequisite to the selectionof sites for cxcavation and inves t iga t ion of the

populations of cultural Ítems and features pres-ent. Limited data are obtainable through sam-pling populations of cultural Ítems present onthe surface of a site as well as through exercís-ing spatial control over the "context" of the arti'factual populations as regards ecofactual data.This information provides the classes of attri-butes utilized in classifying sites. There are threemain attribute classes which can be normallycontrolled through the use of surface-sampledata: the size and density of the.cluster of cul-tural items, the formal constitution of the popu-lation of cultural items, and the degree of sty-listic and functional homogeneity of the popu-lation.

The methods utilized to control the attributesof size and density will also allow us to partitionthe population of cultural items and speak ofthe relativo densities and of its formal classes,that is, the spatial s tructurc of the population.A fur ther class of data, largely ecofactual, canand must be controlled. This is the topographicand physiographic na ture of the location. Suchinformation can be obtaincd at the same t imethat the spatial controls are established for sam-pling the population of c u l t u r a l items.

In order to control the re levant variables andobtain the necessary data, we must have a num-ber of sample units tlistributed over the áreaof the site and its immediate environs. Thesesample units must be rather evenly distributedwithin and beyond the suspccted bounds of thesite. In addition to establishing areal l imi ts ofthe frame, we n jsr determine the appropriatesi:e of each sampling uni t to insure the recoveryof an adequatc sample. A normal topographicsurvey of the site will provide a basis for thenotation of ecofactual data as well as for thespatial control- of the sampling frame and thelocation of sampling units.

Somc methocl of "systematic sampling" is sug-gested as being the most appropriate to surfacesampling. Only the first uni t is selcctcd at ran-tlom and then others are selected in terms of apre-established interval (Yescelius 1960: 463).Systematic sampling ensures an cqual dispersiónof sample units, a desirable condition when den-sities and assregational analysis are attempted.There is a fu r the r advantage in tha t spatial con-trol on the placement of sampling units is easierto ma in ta in with an cqual spatial un i t betweenthem; thus, it is easier to lay out and iden t i fythe selected sampling uni t s in the field.

Page 12: Binford (1964) A Consideration of Archaeological

•136 AMER/CAN ANT/QÜÍTV [ \'oi.. 29, No. 4, 1964

A typical example of the cxccution of such aprogram is givcn below:

(1) Imposc a frnmc ovcr the arca of the site in thetorm of a <:rid systcm composcd of sampling uni ts of <ip-propriatc si;c.

(2) Enumérate the sampling units in the framc fromonc to n. Determine the necessary samplc siic and thendetermine the appropriate sampling interval.

(3) Consult a table of randam numbcrs and drav theinitial sample unit. Then draxv cach sample unit sepárate(rom the initial one by the dcsifinated samplins* interval.

(4) Lócate on the site the selected sampling units andcollect all cultural Ítems within the bounds of that unit.

This procedure can be speeded considerablyin open or cultivatecl áreas by use of a "dog-leash" technique. Each person who collectsítems has attachcd to his belt a cord of predc-termined length to which is attached a stake.The stake is placed in the ground at the appro-priate location, and the person collects all thecultural ítems within the radius of the circle de-ñned by the "dog leash." The location of sam-pling units can be detcrmined quickly by meansof a tape and compass or with a transit. Such amethod is considerably faster than setting up agrid and collecting Ítems from a square unit, allfour corners of which must be defined.

Rcgardless of the particular procedure fol-lowed, applícation of the principies of probn-bility sampling to the collection of cultural Ítemsfrom the surface makes possible the objcctivcdefinition of the site in terms of density clines.This permits objective comparison of sites internis of site si:e and ítem density, in additionto the form, homogeneity, and structure of thepopulation of cultural Ítems present. On thebasis of such comparisons, we can arrive at aprovisional typology of the range of variabilityin the population of sites within the regionaluniverse. Workiñg hypotheses can be generatedto account for the observable dífferences andsimilarities in form, density, and spatial struc-ture, and these hypotheses can be tested by ex-cavation.

A comparative study of information collectedthrough the applícation of sampling techniquesprovides a basic set of data for the constructionof a stratified sampling frame of provisional sitetypes within which selection of sites for excava-tion can be made. This brings us to one of themajor questions considered: how do we knowwhere to dig? I think that the answer to thisquestion logically tests-with a methodologywhich attempts to test working hypotheses con-

cerning the nature of variation that is observ-ablc in populations of surface-collected culturalítems and with techniques of probability sam-pling. Are sites that exhibit similar si:e, density,and composition of cultural ítem populationssimilar with rcspect to populations of culturalfeaturcs, depositional and cultural history, andgeneral function within the cultural systems rep-resented? If such a hypothesis were to be con-firmcd,«we would be in a position to generalírefar beyond the data deríved from direct excava-tion and could make statements about settle-ment systems based largely on surface-collecteddata. This is not possible whcn sites are nottreated within a sampling universe or when sur-face data are not used in the generation of struc-tural hypotheses.

How do we nctually go about the selection ofsites for excavation? The following procedure issuggcstcd:

(1) Devclop a taxonomy of sites based on formal at-i r ihu tcs invcsti<:ated durinu the surface survcy.

(-) Determine the relativo frcquendes and distrihu-tions of site types according lo the original sampliní.;strata, for cxamplc, soil types.

(3) Strat i fy the population of sites into samplinystrata based on the typology further stratified in termsof the original areal strata, ihat is, soil types.

(-í) Dcícrrr.inc in tcrms of the t ime nnd funds avail-ablc what proportion of the total numbcr of cach site typecan be e.xcavatcd to yield rcliablc information on theirintcrnal composition.

(5) Enumérate cach site in cach samplins stratumfrom onc to n.

(6) Consult a tablc of rnndom numbers and draw thenppropriate sampling units dcsignated by the randomnumbers.

(7) Proceed to excávate all those sites whosc unit-dcsignator numbcr was drawn from the table of randomnumbers.

The use of such a procedure can be justifiedin a number of ways. First, it will be rcmem-bered that the initial taxonomy based on sur-face-collected materials grouped sites judged tobe similar or different. Within each taxonomicclass, sites are excavated to test the reliability ofthis judgment and to further explícate the natureof the variability through more detailed investi-gation of populations of cultural ítems and fea-tures. Only by the use of such a procedure canwe explícate the meaning of observed differencesin the surface-collected material and therebyprovide the necessary information for confirm-ing the validity of gcneralizations based on suchdata. Secondly, the procedure insures an ade-quate and representative sample of the popula-

Page 13: Binford (1964) A Consideration of Archaeological

BlNFORD ] ARCHAEOLOGÍCAL RESEARCH DES1GN ,„^ J l

tion of cultural activity loci within the defineduniverse. It is a complete and unbiased across-the-board investigation of the full range of for-mal variability within the population.

The next phase of research planning is tomany the most importanr, and it is the phasethat has received most attention under the ru-bric of "field methods." Initially, it must be rec-ognized that in excavation w^are not samplingactivity loci; vve are sampling popuiations ofcultural ítems, cultural features, and ecofactsat an activity locus that may or may not have acomplex cultural and depositional history. Wewant data which will allow us to understandthe histórica! aspects of the various occupations,as well as the functions of the occupations in thetotal cultural system represented.

If we view excavation as having a part icularrole in the schcmc of data collectíon, it is reason-able to think in terms of an excavational strate-gy. First, it must be kept in mind thn t a l l exca-vation is exploratory in addition to being a meth-od for securing samples. Some phases of exca-vation may be parnmetric in the sense of it be-ing possible to enumérate a sampling frame priorto data collection, while other phases are e.x-ploríitory an:"! snmpling must be on!y provision-al ly paramctric. Spatial control on a horironralaxis makes possible the parametric definición ofa sampling f r ame in terms of spatial units. Thistype of f rame is ideal for investigation of thehomogeneity or heterogeneity of the popula-tion of c u l t u r a l Ítems, and it is appropriate fore: ploratory v/ork that secks the solution of prob-lems of depositional and cultural history. How-ever, it is with excavation that we hope to ac-complish the máximum correlational controland thereby obtain the data that will allow re-liablc interpretaron of the interna! variation inthe spatial s t ructurc of functional and stylisticclasses of cu l tu ra l itcms. Excavation will fu r the rprovide well-documentcd and correlated sam-ples of ecofacts, the basic data relevant to thenaturc of the local environment, and the way inwhich the represented social units were adaptedto it. It is by correlating the distributions of cul-tural itcms with cliffercnt funct ional classes ofcultural features that insight into the "causes"of different ia l d i s t r i b u t i o n s is o b t a i n e d , andhenee understanding of the range, location, andnaturc of the various activities conducted at thesite. Sampling frames designcd solely to ob-tain cul tural Ítems and provide information con-ceming depositional and cultural history are

generally inadequate as a sampling frame forcultural features. Ideally, we should have anX-ray machine which would allow us to lócateand formally evalúate the range of variationmanifest in cultural features. Given such infor-mation, we could construct a frame and excá-vate features within each recognized formal classin proportion to their relative frequency. Sucha*procedure would be analogous to excavatingsites selected on the basis of a previously de-fined frame of site types. Unfortunately, no suchX-ray machine exists, and we must attempt toobtain the desired sample by opening up áreasof the site in such a way as to (1) allow therecognition of the presencc of cul tural features,(2) provide a representative spatial coverage ofthe universe in order to define the spatial distri-but ion and s t ruc tu re of the features, and(3) provide the necessary contextúa! data forthe forma! analysis of the recognized features.

This discourse is not intended as a discussionof excavational techniques. Howcver, a limiteddiscussion of some widely utilizecl approaches asthcy relate to the general problems of researchstrategy seem to be in order.

Ideally, once a site is selected as a unit in asampling frame oí sites, it should be completelycxcavated. In such a case, títere will be no ques-tion abotit one's abi l i ty to give parametric defmi-tion to a sampling frame for cul tural ítems com-mensurate with the cfficient investigarion of cul-tural features. Complete excavation wi l l insurecomplete recovery of the entirc record of pastactivit ies at the given location.

ln most cases it is impossible to undertakecomplete excavation of a site; only rarely arefunds and pcrsonncl available for such an un-dertaking, particularly when sites are largc. Onemerhocl frequently resorted to when faced witha largc site, or when the invcstigator is interestedin obtaining information concerning the consti-tu t ion of the population of cul tural features, isto opcn up large "block" áreas such as was donea tKinca id (Colé 1951). A relatively large num-ber of contiguous excavation units were opened,and this resulted in the complete excavation ofa largo "block" of the site área. This method in-sures recovery of the formal range of culturalfeatures present in any given block but, as nor-mal ly implcmented, does not insure that the ex-cavatcd block is representative of the range offeatures and act ivi ty loci present at the site. Asnormal ly practiced, the block or blocks selectedfor cxcavation are in the "core" área of the site

Page 14: Binford (1964) A Consideration of Archaeological

43S AMERÍCAN' AN'TIQUITV [ Voi.. :9. No. 4, 19<?4

and thcrefore bias the sample toward fea turesand activi ty áreas that wcre centrally located.As previously noted, our aim should be for ade-quate, reliable, and representative data. Blockexcavations as normally utili:ed do not supplythis type of Siformation.

Test pits and test trenches are appropriateunits of excavation whcn one is investigating cer-tain limited, formal properries of the site', butthey are inappropriate to investigation of the siteas a whole. Test pits are by defmition small , non-contiguous units. Such units are useful in pre-liminary investigations of depositional problemsand as a means of solving site Cultural historyproblems. They can also be profitably employedin the collection of a dispersed number of sam-ples of cultural Ítems, but they do not normallyexpose áreas large enough to define and samplepopulations of cultural features.

Test trenches are excellent means of investí-fjating and defming problems of cul tural anddepositional history, but they have most of thelimitations of test pits when sampling cul turalfeatures. However, they f requent ly providemore information on the different iaj distribu-tion of feature types if they happen to be openedin sufficient density to "cross cut" major arensof the site. Data gathered from test trenchesalso have the advantage of being pa r t i cu la r lyuseful for analysis of itcm densities on a l inearaxis. The l imitat ions of test trenches are thoseof any tcchnique which does not open up a largecomiguous área, and does not cover, in a repre-sentative manner, the entire site.

As test pits and trenches are normally utili:ed,they do not provide aclequate data regarding thepopulation of cu l tura l 'Ítems because they arenot normal ly distributcd at random in sufficientnumbers over the site. A greater limitation is thefai lure to expose large contiguous áreas, a neces-sary condition for adequate sampling of culturalfeatures.

Phase excavations seems to be the most appro-priate term. to apply to the procedures whichwill be suggested as a means of overcoming someof the diffkulties inherent in sampling the differ-ent types of observational populations at a site.The term implies that the excavation of a sitemay involve several different excavational steps,each largely dependent upon the results of theearlier "phase" for the details necessary to theproper planning and execurion of the succeed-ing phase. Each phase is designed to answer cer-

tain spcciftc questions with the most economicalancl cxpcdient means.

As an example, in i t ia l discussion will centeraround a relatively small, single-componcnt sitethat lacks primary archaeological context belowthe plow roñe (except, of course, cu l tu ra l fea-tures). Init ially, we want to know whether ornot there are clusters of differential density incultural items, a clue to the possible location ofcultural features. In addition, we want a com-plete and unbiased sample of the population ofcul tura l items in order to makc judgments as tothe "meaning" of demonstrable differential dis-tributions of recognized stylistic and functionalclasses. Such a sample could be obtained by theexcavation of a series of "test pits," the size anddensity of which would be determined by theestimated density of cul tura l items present. Thedistribution of the pits would be deterrrL.ied bysome technique of probability sampling normal-ly executed within a grid frame. The plow ¿onewould be excavated and sifted for each of theselected excavation or sample units. This meth-odology could he fur ther implemented by com-bining controllecl surface collection from selectedsample units. The data collected in this wayshould yield the desired in format ion concerningpopulation form, structure, and contení. Oncethis is accomplished rhe next phase of excava-tion should be planned to yield the sufficientcontrolled data on the population of cul turalfeatures present. In this particular case, wherethere is no pr imary archaeological context be-low the plow roñe, \ve can most efficiently ac-complish our task by complete removal of theplow roñe with the aid of power equipment.The result would be the exposure of cul turalfeatures which could be mapped and excavated,uti l izing techniques designed to yield máximumcorrelational control.

The suggested excavation program wouldamount to a two-phase sequence. The firstphase is designed to yield information aboutthe population of cu l tura l items present, where-as the second phase would yield the desired in-formation concerning the population of culturalfeatures.

In many field situations the sites are morecomplex, having múltiple occupations with pri-mary archaeological deposits below the dis-turbed plow zone. In such a case, a three-phaseexcavation program may be more appropriate.The initial phase would consist of opening up aseries of test pits selected for excavation on the

Page 15: Binford (1964) A Consideration of Archaeological

DlN'FORD ] AKCHAEOLOG/CAL RESEARCH DESIGN

basis of a random or systematic sampling patternwithin a grid frame. This procedure, as in theearlier case, should provide the data necessaryfor the reliable definition and isolation of differ-ent stylistic and functional áreas within the totalsite área. Next is the problem of sampling popu-lations of cultural features. Since the site hasseveral components, tight correlational controlmust be exercised to ensure the possibility ofcorrelating feature forms with forms of culturalÍtems representative of discrete occupational epí-sodes. An appropriate procedure is to employblock excavations as the second phase. Since themulticomponent nature of the site presumablywould have been recognized during the earlystages of the test-pitting phase, such a procedurealso ensures that the initial exploratory expo-sures will yield information obtained under con-ditions of máximum correlational control. Thishopefully makes possible the correlafion of fea-ture types with oíd soil surfaces, which in turncan be isolated and investigated in terms ofcultural-item contont. With this type of infor-mation we should, assuming that the block ex-posures have becn succcssful, be able to developn formal taxonomy of features which can becorrelnted wirh the variable populations of cul-tural Ítems representativo of the sepárate occu-pations. Once such corrclations are establishcd,the third phase of excavation can begin, the"stripping phase." This is rcmoval, by meansof power equipment or by hand, without at-tempting to recoven cul tural Ítems, of the com-plete cu l tu ra l deposit down to the level wherec u l t u r a l fea tures can be observed in t rud ing in tothe natural . Once this is nccomplished, thesefeatures can be jnappcd and excavated, usingtechniques that will ensure máximum correla-t iona l control. The clistributional data thus ob-tained will supplemcm that already coliectedand mnkc possible the definition of activity áreasand the general i n f e rna l community structurerepresentative of the sepárate occupations.

If we have been careful in planning and suc-cessful in the execution of the three phascs ofexcavation, we should have the data necessaryfor the dcmoiistration of clifferences and simi-larities between occupations in terms of the for-mal, spntial, and s t ruc tura l composition of thesepárate populations defined by both c u l t u r a lÍtems and features .

Enough has been saicl to suggest what is in-tended by phase excavations. As the complex-i t y of a site is compounded in deposit ional and

culture historical aspects, more and more atten-tion must be given to maintaining máximumcorrelational control. The intensity of samplingor the sample sire needed to ensure adequatedata increases wirh the heterogeneity of the uni-verse under investigation. This means that datanecessary for justifying the step from one phaseto another become expanded, and in general thenature of the appropriate phases changes withthe complexity and form of the universe. Themore complex the site, the more complex theexcavational procedures, and the larger must bethe recovered samples for any given phase ofexcavation.

This recognition provides the justification forwhat I cali the planning of an excavational se-quence. In áreas where a number of sites havebeen selected for excavation, the temporal se-quence of excavation can be very important inpromoting the efficicnt use of resources in bothlabor and funds. The ini t ia l sites excavatedshould be the least complex, so that the chancesof making false correlations are dirninished andthe máximum conditions obtain for observingthe formal spatial structure of features and cul-tu ra l Ítems. Once an understanding is gainedof the formal and s t ructura l characteristicswhich may be encountered, one is in a muchbetter position to investígate a complex sitewhere the nature of the variabi l i ty may not beso clearly depicted. Informed excavation of acomplex site can often great ly expedite its cfli-cicnt investigation.

It is hopee) that by following this "hypotheti-cal" research program, the reader has gained aclearer understanding of what is intended bythe argumcnt that methods of probability sam-pling are applicable on all levéis of field investi-gation. By poínting out some of the complica-tions, I trust an apprcciation can be gained as tothe potential which rhe application of proba-bi l i ty sampling methods holds for improvingourclata-collection methods. Such methods fu r the rprovide a basis for a grcatly expanded analysisof nrchaeological data directed toward the defi-nition of archaeological assemblages in struc-tural terms, ultimately with a view toward theisolation and definition of cxtinct cu l tura l sys-tems.

Of equnl importance is the recognition thatfield work must not be conducted scparatelyfrom analysis. Running analysis is a necessaryp;irt of f ea tu re description, and of even greaterimportance is the recognition that the results of

Page 16: Binford (1964) A Consideration of Archaeological

AMER/CAN AM/QUJTV VOL. 29, No. 4,1964

í».

running analysis largely serve as the bnsis for theplanning and decision-making regarding succes-sive methodological stcps taken in the exccutionof a field program. Much ink has been spilt onthe argument that the archaeologist as sucli isa technician (Taylor 194S: 43). Only in a veryrestgicted sense can such a position be defendedbecause the field archaeologist is forcver makingdecisions as to what are pertinent and relcvant"facts." Such decisions can only be made withknowledge and understanding of the questionsbeing asked of the data. The field archaeolo-gist must also be an anthropologist to make suchdecisions efficiently and effectively. As Brew(1946: 65) has argued that there is no single oreven adequate taxonomy sufficient for "bringingout all the evidence," so I also argüe that thercis no sufficient set of field tcchniques. Field workmust be conducted in terms of a running analy-sis and against a backdrop of the \videst possibleset of questions to which the data are potentiallyrelevant. This is no technician's Job. This is thejob of an anthropologist specialircd in the collec-tion and analysis of data concermng e.xtinct cul-tural systems. Only after the myth of simplicitywhich surrounds the training of field archaeolo-gists is dispelled, and af ter more a t t en t i on isgiven to recovering information concerning theoperation of extinct cultural systems as opposedto the recovery of things, will archaeologistsmake significant advances in studies of cul turalprocess.

SUMMARY AND CoNCLL'SIOXS

\ It has been argued that as archaeologists weare faced with the methodological task of iso-lating extinct socio-cultural systems as the mostappropriate unit for the study of the evolution-ary )orocesses which result in cultural similaritiesand differences. If we view culture as man'sextrasomatic means of adapta tion, we must iso-late and define the ecological setting of any givensocio-cultural system, not only \vith respect tothe points of articulation with the physical andbiological environment, but also \vid~i points ofarticulation with the socio-culrural environ-ment. It is suggested that changesin the ecologi-cal setting of any given system are the primecausative situations activating processes of cul-tural change.

It is argued that the methodology most ap-propriate to the study of cultural process is aregional approach in which we srtempt to gain

•jreliable.-. and representative information con-

cerning the internal structure and ecologicalsetting of successivc cultural systems. It is ob-served that undcr current programs of salvagearchaeology and greater founda t ion supportfor archaeological research, archaeologists areactually being given the opportunity to studysuch regions. It is argued that, in spite of suchopportunitics, our current practices largely ob-viate-the recovery of data necessary to the studyof cul tural process. The development of tech-niques for the recovery of data in structuralterms is believed to be crucial, for it is the struc-ture of archaeological remains that informsabout the cultural system, and it is the culturalsystem which is the seat of process.

Probability sampling is suggested as a majormethodological improvement which, if executedon all levéis of data collection in fu l l recognitionof the .nherent differences in the nature of ob-servational populations which archaeologists in-vestígate, can result in the production of ade-quate and representative data useful in the studyof cultural process.

Observational populations of cul tural Ítems,features, and activity loci are recognized as hav-ing certain characteristics which demand differ-ent t rea tmcnt in both field obscrvation and sam-pling methodology. On the other hand, onlytwo major sampling universes, regions and sites,are recognized as appropriate to field investiga-tions. Many of the limitations of currentlyavailablc data are believed to derive from thefailure to sample populations of activity lociwithin a regional universe. Emphasis has beenon sampling populations of cultural ítems withina regional rathcr than a site universe. This pro-cedure has made impossible the structural defi-nition of populations of cultural Ítems or thestudy of activity loci from a structural point ofview. Consequently, our current understandingof the prehistoric past is largely in terms of styledistributions and cultures defined in terms ofdiscrete traits and stylistic characteristics; this iscertainly not a situatíon conducíve to studies ofcultural process.

The argument for planned and well-pacedexecution of research design has been presentedin the form of a "hypothetical" research pro-gram, along with a limited discussion of the tech-niques and levéis of applicability of probabilitysampling procedures. Problems attendant uponthe recovery of structural information withinboth the regional and site universe have beenmade explicit in a number of examples of types

i

I

.-.t-harae

Page 17: Binford (1964) A Consideration of Archaeological

>fis>f

•d:d

h-ty>nin

:nes

BlNFORO ] ARCHAEOLOGICAL RESEARCH DES1ON 441

of sampling problem and excavational situation. ~DEMING, WILLIAM EDWARDSIt ÍS concluded that the design and execution of 1950 Soma Thcory o¡ Snm/>lm£. John Wiley anda research program is the Job of an anthropolo- SoI1S| Inc- New York'pist, and only in a limited way can the ficld ar- DUNCAX, O. T., RAY P. CIV./ORT, AM> HEVKRI.Y DUNCANchaeologist be considered a "technician." Field 1961 Suamical GcoSrat-hy. The Frec Press of Glen-work is an on-going process demanding mcthod- C11C' Glencoe.ologícal decisions bascd on a running analysis of GKARING, I-"RI;I>the data recovcred from prior field work. It is ^ IVÍCMS an., Warriors. Mt.moir5 „/ thc Amcri.concluded that if we are to be succcssful in thccollection of data relevant to studies of cu l tu ra lprocess, field work must be conducted withinthe framework of a well-planncd research de-sign which provides for thc application of proba-

1 bility sampling techniques at all levéis of investi-° catión. The fielcl strategy cxccutcd within the

framework of the research design must be di-rectcd by a well-trained anthropologist capableof making interpretat ions and decisions in termsof the widest possib'c factual ancl theoreticalknowlcdge of general anthropology, ancl thetypes of questions must be drawn up which bist i n t a mny be useful in solving. It is bclieved tha tmoJification oí currcnt prácticos along tlieselines is n necessary prercquisilc ior movinu ar-chaeology on to the level of dcvelopment which ,sn . \ \ -M ( l , jrn\s IIWilley and Phi l l ips (1958: 4-5) have callcd the"expíanatory level."

cim Am/iro;>olo£¡ai¡ Associiition, No. 93. Mc-n:isha.

, WILLIAM G.1959 Tne Status of Evolmionary Thcory in Ameri-

i'an History. In íiYoliitiim and Antl\rof>olnx\:A CcMcn

QTL ̂ ;^"isnA

!- «"«d by Hct ty J. Mep-i;crs, pp. 90-103. The Anthropolosical Socicty ofWashiii^ton, Washington.

" '

Mll-D!!l:u

i;, Ai .ni .m1 C.I he nnncii.--ii.ins of Aivhje.'Ki^y. In í:Siii\s ¡n

f /u" .SVii'r.fi.' '>/ ( ^ ü í f i n t 1 : ín Honor oj Lt:<lic A.\\"/:ilc, ccliicil hy ÍJertniJc !:. I.Xilc nnJ Rol'crtI. Ca inc i r i ' , pi\ -1 >7-5d. Tilomas Y. C'rowcllC 'ornpiuiy, NVu' ^ « ,'k.

19('0 l i v o l i i i i o n n r y l ' r inciples an. i Social Typcs. InT/ic' ¡ ; t<i lui íon o/ Míin: \liiu¡. C^ilínu' iuu¡ So-i 'ii-iv, c J i t e i l hy Sol T:ix, pp. l(<9-Sí\ The l 'n i -

I i i T/it1

N.K-I 'I-[ \ .\'crí-ity o

,MS, Roiu.nr M. ! v ' ' " "' ' ' 'l'^oO The U v o h i i i o n a r y Process in lia.-ly Civili::i;ion.s. TAYLOR, W A I . I L R NX'.

194S A SuiJy of Archaeoloi;v. \\cmoirs < . f t/ioAiiu-nVun A n i / i r . » ; > . < l i ) i ; i V ( ! l Associci i ion, No. 69.Mennsha .

A K K I N . I I . % N I > I \ A V M H S M > {..'OI.TOS1057 _

19(V0 Archa to lo t ic i i l Sampl i i ic : A í ' roblcm of Slaiis-ucal I n l c r c n c c . ln iissayf in thc Ncifiur i>¡ C^ul-nnc. ln Honor u\ ¡.alie A. \\"/uu', cd i ted by Ger-tr t idc H. Dole and Robert L. Carneiro, pp. 457-

i f í í i i r i t m (ij Xían: \ i int i , (Jiiínnc'. íiruiinei l l'y Sol Tnx, pp. 155-dS. The l ' n i -C^hio;i;,-v> i'Vess, Chic:!i;v>.

í. l ín rncs : u i < ! N'oMc Inc. ,

70. Thomas Y. Crowcll Compnny, New York.

\ew York.

BIM'OKP, l.i \\ is R.I l íd2 Arc.h:ieolo¡:y as Anthropoloev. AnuTiVíin An-

l ú / i c / t v , \"ol. '2S, No. 2, pp. 217-25. Snlt I-akcC:ny.

WACÍSI- .R, P I I I L I I -l ÍKMimiKm. Roii i 'Ri 'J . lq(;C T/u- Hu.nmi Use- ,;/ r/U' í;,,u¡,. The !;rec Press

1959 Archneoloyy aiul ihc r .volut ionnry Thcory. In of Glencoc, GIcncoc.

;i7-ii is<i! , cdiieJ hy Bct ty J. Meg'^er?, pp. 76-S9. -Wimr. , I . I .S I .SK A.The Anlhropolo.uical Socieiy of Washington, ,959 T/lc. í-',.,,;,,,;,,,, „/ CH/(,.U-. McCíraw-Hill IlookNVns:1"lí;Ion- . Companv, New York.

H u h U , J o l l N Ol l s\X l l . l . l - . Y , (JOIU)OS' [\. ASI! I l l l l . l l ' l ' l l l l . l . l P j

195S .Vfi-í/K.uí dntí Tíu'orv in Amejictin Archticolugy.1946 Archaeolo^y of A l k a l i RiJce, Soinheastern

Utah. I'íi/it'rs u¡ t!\c l'cabody Miisci i i i i o¡ Anii-rí-

r, i-'.SY-Cnoi'i.Kll¡51 K'iiu-.u,!. A í ' r . ' l i is t .niu I l l i .u . is Mi-ü

vers i ly oí C 'h icaf jo Press, Chicado.

The Univc r s i ty of Chicago Press, Chicato.

U s ' i N i K s n v oí- CMII:AWICliicafio, I l l ino is

May, 10(>3