EXPERIMENTAL . COLESM , F.S.A.SCOT. THE term experimental archaeolog conveniena s yi describinf o y twa collectioe gth n of facts, theories and fictions that has been assembled through a century of interest in the reconstruction and function of ancient remains. This paper is an attempt to provide an explanation of some of the theory and principles involved in experimental archaeology briea d f an ,descriptio somf no e experiments don thin ei s countrd yan abroad which can serve as examples to show the advantages and disadvantages of various approaches to the subject.

A. THEORY Experimental archaeolog dividee b n yca d sectionsinto otw concernee ,on d with imitative th e aspect duplicatioe th , productf no industre th f so f earlyo y mane th , other utilising such product determininn si g their functional capabilities. Basio ct both aspect productioe th s si copief no prehistorif so c objects woodf ,o , stone, bonr eo metal, either to be studied as replicas in themselves, or to be tested in some way. By making these copies archaeologise th , gain ca tn some insight thingintw oho s were produce ancienn di t times ofted an , n these experiments yield e hintth sn i tha d ai t interpretatio otherwisf no e incomprehensible part objectf so r marko s s upon them. Casting seams and hammer scars on metal artifacts, for example, are most easily understood by attempting to duplicate them experimentally. This imitative aspect involve considerablsa e amoun specialisef o t d experienc artcraftsn d ei san , ands a , usual, the archaeologist must draw heavily upon the knowledge of others. Often the entire experiment will depend upon the ability of the craftsman to make accurate copie f ancieno s t objects, copie t onlsno y thae identicaar t appearancn i l e th o t e origina als e t togethear o pu t bu l r wit same hth e view towards their eventual function. Aexamplen a s closel y woodea , ma yd resemblar n copn a f yanciene o eth t object (fig. 2) but the type of wood used, its age and its condition, and its grain, all must be considered carefull lighe th tywoodn e i bot th f ho - worker's knowledg stressef eo d san of the condition of the ancient material. Here the value of examination and analysis of the material will be important, not only to yield information about the marks of productio objece wead th t alsnan n rdetermino tbu o t exace eth t compositioe th f no basic material. The obvious example of this is the reproduction of ancient metal artifacts, where spectrographi othed can r analyse yieln sca d essential data aboue th t composition, and therefore the properties, of the material. Specialists, present during the actual experiment, will also be valuable to the archaeologist to help him under- stand what is happening during the progress of his attempt to reproduce ancient objects. The conduct of an experiment on archaeological material may be compared with an excavation, where the archaeologist has set a problem and draws upon the specialised informatio otherf no s to heldeterminm phi courss ehi o actiof t eo d nan 2 PROCEEDINGS OF THE , 1966-67 draw conclusions abou finishee tth d productnecessaryt no s i t possibler I . e no , th r ,fo archaeologist to possess all the details of the scientific processes and theories involved, appreciato t necessars i t m i hi t scope r bu potentialitieseth d yfo ean , and, more important, the limitations of such scientific aids as he may be able to bring to bear upon the problem. I believe this statement applies equally to an archaeological excavation or an experiment. In the copying of an obj'ect, the primary difficulty facing the archaeologist is the determination of the processes to be employed. It may fairly be said that in theory only those methods shoul usee db d thae presumear t havo dt e been availablo et prehistoric man. Most experimenter practicn si e have tende avoio dt d this problem and have made a small part of the object by using simple methods, such as cutting a piec woof structuraa eo r dfo l reconstruction wit hstona e axethereafted an , r using modern equipment. For the purposes of making a quick copy, such a procedure is self-evident amoune th t informatiof bu ,to gain ca n e abounw finishee tth d producs ti severely reduced estimato N tim.e th ef eo involve manufacture th n di madee b n ,eca as extrapolations are notoriously misleading in such cases, and the essential factor of 'prehistoric need' is unknown (see p. 7). In addition to the loss of information about the time involved in the production of objects, through the use of modern methods, there is also a danger that the modern product will possess characteristics different from those of the ancient . For example, the crushing and compression of surface areas of wood and bone through the use of high quality steel tools have not been considered, nor perhaps the effects of sophisticated methods of heating and casting of metals. Similarly, the breaking-up and shovelling of chalk with modern equipment may have a different effect upon the compaction of an experimental earthwork, when the latter is compared with a structure built entirely by methods more within the reach of prehistoric man. Another disadvantag moderf o e us ne methodth f eo completeo st copy an e y th lie n si total lac informatiof ko n abou processeeffecte e th th t f o s s upo toole nth s employed, a corollary which could be of the greatest interest. Here the amount of information obtained through microscopic examination of edge wear on prehistoric tools could

be profitably compared with work traces on material used in experiments,1 ' for example, the ancient use of blades as whittling knives has been determined both through examinatio f striationo n n theio s r edges y experimenta(figb .d ian ) l reproduction. In determining the methods to be employed in making duplicates, the rule of thumb use experimentery db generalls si y tha inefficientf o t , because inexperienced, us primitivf eo e unspecialised objects suc flins ha t flakes, stone axes, wooden punches and mallets, bone shovels and scoops. These in many cases are second-best, because they too easily replace intelligent consideration of the degree of technological ad- vances that prehistoric peopl maded eha . Block-and-tackle, shear legs, sledged san rollers, all were without doubt available to many prehistoric groups, to aid in transportatio d erectionan f largno e objects, jus s specialisea t d potter d metayan l equipment was available to comparable people making small artifacts. With such 1 e.g., S. A. Semenov, Prehistoric Technology (1964). EXPERIMENTAL ARCHAEOLOGY

flinFIGf o te . bladi.Us whittlins ea g knife, with retouched edg finger efo r support stipple Th . e indicatee sth area of wear-striations, and lines show their direction. (After Semenov, 1964, fig. 45)

ranga equipmentf eo , allie centurieo dt experimenf so experienced an t t wouli , e db easy to devise and carry out relatively sophisticated methods of handling and manu- facture e experimentaTh . l archaeologist, therefore, should hav detaileea d know- technologicae ledgth f eo l methods available r presumeo , availablee b o e dt th o t , prehistoric peoples with who concerneds i e mh ; such knowledge wil basee b l d upon estimates of prehistoric time, surviving materials, economy and environment. A major problem in experimental archaeology is the subjective element, the fact tha experimentee tth r comes fro mdifferena t cultural body wit hdifferena t approach to existence cannoe h ; t hop reproduco et mentalitye eth spirituad an , l background, obtainin prehistorin gi c time. Everythin experimene th n gi t shoul aimee db t da neutralisin eliminatinr go experimenter'e gth s personal influenc experimente th n eo , throug e incorporatioth h s mana f o ny natural factor s possiblea s e severaTh . l experiments in recent years with ploughing devices are cases in point. Although the natural conditions, soil structure, seaso weatherd chosene an b n thao n,ca s t they yma be presume reproduco dt prehistorie eth c condition closels sa possibles ya othee th , r subjective elements plouge ,th ardr hchoico e ,th animaf eo l traction ploughmane th , , reproduclesy e ar sma reliable d ar e e exactlth ; yprehistoria c specimens i t i t bu , necessary to consider carefully the need to train animals and men to pull and guide it (fig. 2, left). Tractors, and probably horses, are not suitable for such experiments if the soil marks, or wear on the ard, are to be examined. Clearly, such an experiment involve greasa t dea researcf o l h int moduse oth operandi. e seconTh d aspec f experimentao t l archaeology e functioe studth th ,f yo f o n ancient materials, requires that these materials shoul accuratele db y duplicatedn I . theory, this functional approach should assis e archaeologisth t n deducini t e gth precise use of artifacts in prehistoric times, and this approach, allied to that of 4 PROCEEDINGS OF THE SOCIETY, 1966-67 microscopic examination for wear traces, may allow us, in time, to describe material in terms of actual function rather than shape alone. Metal axes, flint scrapers, bone point they seee ma nsb havo nt e serve s chiselsda , knive d drillsan s , respectively, functions whic n fachi t have already been establishe r somdfo f theso e e general classes through both methods noted above. In experimenting with actual ancient materials, or their copies, it is essential to determin purpose eexperimentth e th f eo tha o suitabilite s , th t environmente th f yo , and the fitness of the material can be assessed. In ploughing experiments, for instance e soilth , , crop, seaso weathed nan r mus e assesseb t d accuratel f valiyi d comparisons are to be made, and the human elements strictly controlled. In the case of experiments in the destruction of houses, the natural factors must be allowed to act without human and subjective interference, so far as is possible. If the house is to be burn , thent2) (PI, , II providin. structure gth s beeha e n built accordinn a o gt assessment of prehistoric building methods, the fire once started will act entirely naturally, dependent only upon natural factor winf o sdampd dan .

FIG . (right2 . e DonneruplanTh ) d ard. Lengt foresharf ho . . 13(Afteec 0cm r Glob, 1951, ) fig29 . (left) Reconstructio f cross-ploughingno . (After Skalk, 1958)

decae Th structuref yo s through other natural agencie st induce whicno e har s d i eminently suitable for experiment, but such processes as rainfall, snow, frost and win generalldare long-teryof m effect. Provided tha experimenthe t carrieis t dout a suitabl n i e environmental setting, where such agencie f scomparabl o wil e b l e strength and effect to those of early times (bearing in mind any long-term alterations climate)n i humae th , theoryn i , elemenbe , n entirelca t y eliminated maie Th .n problem is that of the long-term nature of the experiment, because it is not possible accelerato t effecte eth weathef o s r without distortion. Alread Overtoe yth n Down experimental earthwor yieldes kha d interesting result rapif so d siltin slumpingd gan , totae th t l significancbu structure th f e o onl n revealee ca y b termn di decadef so f so years. Almost entirely oppose example th do t housf eo e deca naturay yb l agencie thas i t of the effects of human occupation upon a similar house. In this case, no useful experiment can be carried out, as there are too many human, too few natural, factors EXPERIMENTAL ARCHAEOLOGY 5 involved e e occupanthabitth Th f e .experimenta o sth f o s l house l woulal n di probability bear no resemblance to those of the original occupiers, and any decay of the structure or deposition of debris through occupation would not necessarily be applicabl traceo et decaf so depositior yo prehistorin i c structures. sourcA informatioeof n abou productiothe t functionand anciennof t objectsis available through the study of existing primitive . Such recourse to living events has tended to be ignored by the archaeologist, rightly in the case of society structure and religion, but the material culture of primitive groups may yield im- portant informatio r understandinou r nfo f prehistorigo c artifacts e superficiaTh . l resemblance of special ovens for the extraction of iron in the Sudan and adjacent area Africf so thoso t a e recovered from Irositee experiDenmaro n st Ag i d le -s kha ment reconstructen so d Iroovene nAg s which indicate tha similarite tth shapn yi s ei measura similarite th methof e eo th n yi d use thesn di e unconnected areas.a n O 1 different line, the observation of a specialised percussion method for flint flaking in Angola toda suggestes yha d tha tere tth m 'pressure-flaking' ofteo s , n appliee th o dt products of many peoples, is misleading.2 A clear guide is provided in this case for careful experimental duplication of 'pressure-flaked' objects through both controlled percussion and pressure; the significance of such work might be of far-reaching consequence indicatinn si g that technological developmentse neeb t dno considered to mark cultural change. It is important to realise that environment plays a part in this functional approach to archaeology. By this I mean that the testing must be carried out with materials d undean r conditions approximatin o thost g f prehistorio e c times n obviouA . s exampl portablr efo e materia thas i l bronzf o t leathed ean r shields, which muse b t 'tested exacy b ' t copie actuar so l specimen Bronzf so speae sword eAg ran d (PI , i)II .. Examples for outside activities such as house building or ploughing are immediately apparent f coursO . reproductioe eth f originano l circumstance overdonee b n ca s , t considereno s i t i d necessaran o drest y n animai s l skins before flaking , although such a garb seems to be generally popular among some 'experimenters'. Historisk-Arkaeologise Th k Fors0gscente t Lejra r Denmarn ei k represente th t sa moment the best example of this environmental approach to experimental archaeo- logy. The Center was established in 1964 on approximately 75 acres of land which

include3 s arable land, moor smala , l lake whicd an , boundes hi l side al hill y n sb d o s or forest. The aim of the Center is threefold, to conduct experiments, to collect data from other groups and individuals for an archive of experiments, and to make available the results both to science and to the public. The public are allowed to view the experiments in progress, and thereby contribute their interest and finance to the undertaking. More than 50,000 visitors attend the Center each year. The experi- ments planned include cultivation and animal husbandry as well as house building and a host of smaller experiments. Such large scale organisation of experimental

1 O. Voss, 'Gammelt Jern', Skalk (1963) no. 2, 4; O. Voss, 'Jernudvinding i Danmark i forhistorisk Tid', Kuml (1962), 7. 2 . JClark.D , Prehistoric cultures of northeast Angola theird an significance tropicaln i Africa (1963) ,171. 3 S. Nielsen, 'Eksperiment', Skalk (1966) no. 3, 13. 6 PROCEEDINGS OF THE SOCIETY, 1966-67 archaeology is unmatched elsewhere; in the British Isles, experiments have generally been individual efforts.1 lase tTh poinnotee b o probablt s d i moste yth importan concernes i alld f to an , d with the limitations of the subject. The value of experimental archaeology must not over-emphasisede b experimentee tase th f th k o s i t I . obtaio t r knowledgw nne o et o that d t gainead d through other archaeological disciplines observatioy B . f o n materials and sites, theories about functions and past events may be formulated, and experiment made b teso n et stca som sucf eo h theories naturae th n I .l sciencese th , experimental metho proof do vitals fi scientise ;th dealins ti g with existing thingd san events which are observable and which can be repeated and measured. In archaeo- logy evente th , paste sar , gone forever cannod an , observee b t r measureddo e Th . experiment in the last resort is based upon conjecture, the idea of the archaeologist who believes tha a specifit c objec performed mads a certaian wa tn y ei nwa a d certain function provo t s i disprovr e.o o d Whan i o t s theory s i sett e et hi th i s ou t ,bu the natursituatioe th f eo n tha cannoe h t easo certaie to resultn e b y ti s i th f t nI o . experimental archaeolog falo yt l withi compase nth thaf so t damning definitiof no prehistory unwarrantable th , e deduced fro unverifiablee mth onln ca y e assurH . e himself that certain object e verar s y unlikel o havt y e successfully served certain specified purposes cannoe certaih s a t e b bu ,t n that they were used solel another yfo r function. An example of this is the beaten bronze shield which has been shown by experiment to be physically unsuitable as a defensive weapon; it is unlikely, therefore, that warriors, if there were such people, would carry such objects into armed combat. They might however have served as ornamental gear up to a time immediately preceding a battle, or they might have been made as gifts of ostentation, or as ritual or ceremonial objects. We cannot know their precise function, but we have probably eliminate possibilite don y fro liste mth . Another example will serv amplifo et y this point prehistoriA . c object suca s ha Irise hor th r Scandinavia hf o n o n experimeny Bronzb e showe e b eAg b n o nt ca t capabl yieldinf eo twelvgo t fro e morr meon o e notes, dependin instrumene th n go t (fig. 10)t moderbu , n analogue primitivn si e societies show tha potentialite tth n a f yo instrument need not indicate its precise range when in use. We cannot know if prehistoric man was capable of producing as many as 12 notes from one instrument or if he could do so, whether or not he used all or any of them, or for what purpose. Other prehistoric objects coul e treatee samb d th en di way , potentially serving severa experimentale purposesth l al d an , l archaeologis eliminats i o d n eca t somf eo these. This is perhaps an over-gloomy assessment of the value of experimental archaeology, but I believe that the limitations inherent in the material and the method mus constantle b t mindn yi . Although experimental reproduction archaeologicaf so l material were first made a century ago, it is only in the past two decades or so that serious attempt has been mad tesfunctiono e et th t prehistorif o s c tools. Thes othed ean r experimenty sma convenientl e divideyb d into three groups dealine on , g wit productioe hth d nan consumption of food, another with the manufacture and decay of buildings and 1. ColesJ.M , 'Flere Forsog', Skalk (1967. . i26 , no ) EXPERIMENTAL ARCHAEOLOGY 7 structures, and a third with the production and function of tools, weapons and other artifacts. It is not possible to describe here more than a few examples in each of these groups.

B. EXPERIMENTS i. Food Production Clearance and Cultivation Most experiments on the processes from forest clearance to cultivation have been carried out in Scandinavia. Jorgensen and Iversen have recorded experiments on clearance th f foreseo stony b d flint ean t axes, hafted accordin survivino gt g pre- historic examples experimente on 1n i ; , 200 fores. f ydso 0sq . t were cleared, leaving treek oa ovef so r . diamete14in r girdled. Oak lesf o s s than . diamete12in r were felled in half an hour, and three men cleared 500 sq. yds. in four hours. Comparisons have been made between the felling of trees by these stone axes and by iron axes. Usin formere beegs th ha n t i ,estimate couln weema de e thacleaon k on n ti r haln fa acr foresf eo t compared wit acre clearancf hon eo metae th ef o lthroug axee us . e hth Yet in Upper Canada, where the latter rate of clearance was recorded,2 the essential point is that although on this reckoning one man could clear 52 acres per year, he in fact averaged only 4 to 5 acres annually, leaving stumps in the ground. In addition to factors of seasonal weather, and basic subsistence, the 'need factor' is surely worthy of more consideration tha generalls i nt i y given believI . e thawrons i t ti considego t r thatimthe t e availabl prehistori eto virtuall was ycman almos limitlessis it t and , certain that he would be constantly attempting to reduce his activity load by stream- linin planninactions y gb hi d san g ahead. Prehistoric communities must have been well enough organise mako dt e provisio r futurnfo e arable land unde e lanth r d exhaustion proces cultivationf so . Tree suitabla n s i are w eane could easil ringee yb d and left to wither well in advance of the season and year of first cultivation. In addition, time-saving devices such as windrow-felling on slopes would cut the time by almost one half. Following the experiments on forest clearance in south Jutland, the felled trees and brushwood were burnt so that the ground was thoroughly heated.3 In western Europe the only suitable method of forest clearance was probably burning because of the heavy undergrowth which would inhibit other forms of clearance, yet musn ma t also have been awarneee th kee o df t e o p felled areas smal alloo t l w forest regeneration quickly, and to avoid the development of heathland. The area was then sown with Neolithic varieties of wheat and barley, by sprinkling and raking with sticka unburnn A . t are clearef ao d fores alss owa t sow controla s na botd an ,h areas were weeded and hoed. A strong crop of grain grew on the burnt area in the first year, hardly any on the unburnt control area, but this primitive slash and burn

. Jorgensen1S , 'Skovrydnin d Flint0kse'me g , Fra Nationalmuseets Arbejdemark (1953). J Iversen ; 36 , , 'Forest Clearanc Stone th n eei Age', Scientific American, 194 (1956). 36 , . GuilletC . E , Early LifeUppern i Canada (1933). 32 A. Steensberg, 'Mit Braggender Flamme', Kuml (1955), 63. PROCEEDING E SOCIETYTH F O S , 1966-67

VIII

PLANTAIN ELM

Vllb

Vila

VI

FIG. 3. Extract from pollen diagrams for Dalnaglar. Depth in metres below surface (left) Plantain: percentage of total non-tree pollen (right) Elm: percentag tref eo e pollen (After Durno, 1961—2)

FIG. 4. Cultivation scenes in Bronze Age rock art (left) Finntorp, Bohuslan, Sweden. Scale £. (right) (After Glob, 1951) EXPERIMENTA9 L ARCHAEOLOGY method was only effective for one year. In the second year, the crop was much smaller. The experiments also demonstrated that the weeds returning to the area differed both on the unburnt and burnt ground; colonising plants in the burnt area included dandelion, thistl plantaind juss i ean t ti sucd ,han plants tha recordee ar t pollen di n diagrams fro e thirmth d millennium B.C Scotlann i . diagrae d (figTh . m.3) from Dalnaglar in Perthshire1 must indicate Neolithic occupation in the area at this date, hely explaio pma t d situatioe an nth groupind nan Neolithif go c barrow partn si f so Perthshire wels a ,pointins a l furtheo gt r fieldwor excavationd kan . The Danish experiments continued with cattle grazing, an activity also recorded in pollen diagrams through the decline of elm pollen, the leaves and shoots of which tree were use wintes da r fodder diagrae Th . m from Dalnaglar neatly illustrates this correlation appearance betweedeclinm th el d e an enth f weedeo f cultivatioo s n (fig-3). The essential fact emerging from these experiments is the short-lived nature of the successful grain cro newln pi y cleared ground faca , t which must indicat generaea l mobility of Neolithic peasantry in cultivated fields if not in settlements. Ploughing experiments have been carried out in several countries, but those in Britain and Czechoslovakia have been hampered by the absence of trained traction animals. Near Reading, a full scale model of the Donnerupland ard (fig. 2), of late first-millennium date, was pulled over various soils by horse and tractor.2 Two fragments of wooden ards are known from south-west Scotland.3 Earlier evidence of ards or ploughs comes from rock art in both northern and southern Europe, and these tend to show that paired oxen were used (fig. 4) .4 At Lejre in Denmark, experiments are now being carried out using trained oxen, and the results both in terms of the d shoul f greatear o soie e db le weath rmarkth n valuo rd an s e than r thosfa o es achieved. An interesting aspect of ploughing experiments would be the measurement of the time necessary for traces of plough marks on subsoil to disappear if they were t covereno barrowa y db t woulI . essentiae db carro t l t suc yexperimenn ou ha n i t areas similar in local environmental conditions to those where barrows with preserved ploughing marks have been recognised resulte Th . s would hel settlo pt debate eth e abou rituae th t l natur sucf eo h prehistoric ploughing. A number of experiments have been carried out on the harvesting of grain, but thos Steensberf eo Denmarn gi mose th te objectivk ar wel e ar l d recorded.ean 8 Most experimente ofth s have been concerned wit reproductioe hth glose sicklesth n sf o n o , now know producee b reapine o nt th y d b corf go other no r siliceous grasses. Steens- berg attempte shodo t w which prehistori earld can y historic implements serves da reaping tools (fig. 5). Flint blades, unretouched or with serrated edge, bronze and iron sickles and iron scythes were used to cut two crops, one drill-sown barley and

. DurnoE . S , 'Pollen Analysi Peat'f so , PSAS, (1961-2)v xc , 154. . BowenC . . AberH A , d . 'PloughinF g21an g Experiments wit hreconstructea d Donnerupland Ard', Antiquity, xxxiv (1960), 144. 3 A. Fenton, 'Early and Traditional Cultivating Implements in Scotland', PSAS, xcvi (1962-3), 264. . GlobV og d . P 4Plov Ar , i Hardens Oldtid (1951). . Steensberg5A , Ancient Harvesting Implements (1943). 10 PROCEEDINGS OF THE SOCIETY, 1966-67 oats on a clay soil, the other hand-sown barley and oats on sandy loam. A number of small plots were sow thesn i maturee e th way d san d grippin graiy theb s t nnwa cu g a bunch of straws below the head and cutting with the implement, then gripping anothe untin ro bunco handfula s d collecteds han wa l numbee Th . cutf ro s required obtaio t nhandfua recordeds wa l numbee wels th a , s a l uprootef ro d straws takes na percentaga time totalth e . emd th take.sq thesf f plotsan e,o 0 o l 5 reao net e Al .p th

FIG. 5. Harvesting implements (left) Bronze knob sickle, (centre) Flint saw-edged sickle, (right) Iron Viking scythe (After Steensberg, 1943) represent function efficience th variouf e so th f yo s implements experimente Th . s also indicate bese dth t metho usinf d o implement e gth , some cuttin sweea gn i p neae rth ground, others cuttin inclinen a n gi d sweep. Takin percentage gth f uprooteeo d strawcriterioe th s sa efficiencyf no Vikina , g scythe (cf. fig , right).Roma5 a d an , n scythe were the best, followed by a modern, sickle, also of iron. But an unhafted unretouched flint blad nexts ewa , demonstrating once agai impossibilie nth - ties of deducing the function and efficiency of material by its typology alone. Bronze sickles (fig , left.5 ) were also suitable implements when measure thiy db s method, followe lesy db s satisfactory flints (fig , centre).5 . Experiment storage th n corf so eo n have recently been carrie Britain,n i t dou 1 Bowen. C . 1H , 'Corn Storag Antiquity'n ei , Antiquity, XLI (1967), 214. EXPERIMENTAL ARCHAEOLOGY II base pitn do s which occur frequentl south e Irositeyn e i th t whic nn sAg i bu , als e har o known from Neolithic times in and Scotland. The experimental pits had narrow mouth werd san e lined with basketry largA . e storage pit. deepft ,5 , would require a basketry strip of the order of 15 ft. long by 5 ft. wide, and would hold over bushel0 4 cornf e spito Th s. wer eweek x airesi r sdfo before filling afted an , r filling were sealed to exclude air and to allow carbon dioxide to develop. After six months, corne th , whether drie notr do beed ,ha n successfully preserved apart from some loss edgee topd th t san a . Corunsealen completels a n i wa t dpi y ruined. More recent attempts to provide adequate sealing with puddled clay and chalk on wicker have indicated that hav y thisema bee primare nth y difficult thin yi s metho preservinf do g corn. In Denmark, some experiments have recently been carried out on the preparation of corn, involving quern clad san y ovens, perhapf o t 1bu s greater interes publia s ti c experiment carried out by the B.B.C. some years ago involving the preparation and consumptio 'prehistoria f no c meal'.2 Examinatio contente th stomace f nth o f so h stranglee oth f fron mdma Tollun Denmarn di k revealed thalass hi tt mead ha l consisted of a soup containing barley, linseed, camelina seed (mustard), persicaria (a pale pink flower), with traces of other plants; sphagnum moss was also present, as wel soms a l e fin experimentae e th sand r Fo . l meal, mos thesf o t e ingredients were boiled slowl mako yt egrea y soup which tasted strangel thoso yt e archaeologists accustome o mort d e 'civilised' dishes n factI . , Pliny records thae grueth tf o l peasants consisted of barley, linseed, coriander seed and salt, a comparable recipe.

Animal husbandry and cooking The experimental reconstitutions of beasts externally resembling Bos primigenius through selective breeding is well-known to archaeologists,3 and the building-up of small flocks and herds at Lejre, for purposes of husbandry practices and for training of beast r tractionsfo s alsha o, been noted experimento Tw . Britain si n have been concerned with the preparation of meat for consumption. Recently an attempt has been mad reproduco et emethoa cookinf do g tha attestes ti sixteente th n di h century A.D Irelandn i . , involvin boiline gth meaf animan go a n i t l skin ove open a r n fire.4 Scottish troops on the retreat in 1327 are reported to have left behind 'more than four hundred cauldrons made of hide with the hair left on, full of meat and water hung over the fire to boil'. The experiments were performed with a 5 Ib. sheepskin holdin galloe gon watef no whicn ri meae hth t layfiree .Th , however, coul t boidno l the water and by shrinkage of the skin through heat and loss by evaporation, most watee olosts th f repea A wa r. thif o t s experiment, usin glargea r skin perhapr o , s pot-boilers such as are attested in certain societies, would be valuable. More successful from the standpoint of the finished product was an experiment carrie t somdou eIreland. n yeari o sag 5 Excavation t Ballyvournesa . CorCo k n yi 1 Nielsen . citop ,. Johnstone. "P , Buried Treasure (1957), 97- . ZeunerE . 3F History,A of Domesticated Animals (1963), 204. 4 M. L. Ryder, 'Can One Cook in a Skin?', Antiquity, XL (1966), 225. 5 M. J. O'Kelly, 'Excavations and Experiments in ancient Irish Cooking-Places', JRSAI, LXXXIV (1954), 105. B 12 PROCEEDING SOCIETYE TH F O S , 1966-67 revealed trace structura f so e comparabl plan ei thao n t historically-know f to n deer- roasting place mid-secon e sitee th sf Th (fig,o ) 6 . d millennium B.C. reconstructes ,wa d to for mwoodea t containinnhu gmeat-raca tabld kbutcheran r eo y block boilina , g troug int g peae surrounded hodu th an t majoo tw ry db hearths roastina d an , g pit. experimentemployee s th wa butchern e t I th hu s de a th , y area, whert e gamcu s ewa priop u cooking o roastinre t th d gan , pit whicn ,i h mea coveres corbelletwa a y db d

FIG. 6. Plan of cooking-place at Ballyvourney. Note post-holes of hut and internal furniture, and vertical slabs enclosing hearths and roasting pit; the trough is formed of stone and wood. Scale i in. to 7 ft. (After O'Kelly, 1954; most secondary structures omitted) structure of hot stones, served as an oven. The trough, dug into the peat below the water-line, held 100 gallon waterf so , whic broughs boie hwa th l o quicklt poty yb - boilers fro e adjacenmth t hearths, only thirty minutes being require heao dt e th t water; few stones were neede keeo d t wate e pth onct join placee rho s e th th twa dn i trough. After nearly four hours, a leg of mutton was cooked to perfection, and was consumed by the archaeologists.

2. Structures Houses numbeA f experimentao r l reconstruction f prehistorio s c buildings have been carried out, again mostl Denmarkn yi ,Loug e althougth r f o hGu e hcopa on f yo house Irelann si B.B.C. e buils th r dwa fo t 1 Danish Televisio alss nha o participated in the rebuilding of an Iron Age house at Roskilde, in this case so that it could be 1 Johnstone cit. .op , EXPERIMENTAL ARCHAEOLOGY 13 burnt down (plan, fig. 7, left). All of the other Danish reconstructions have been

attempted to allow study to be 1 made of the techniques of building and the processes of decay. One of the first tasks of the Historisk-Arkaeologisk Fors0gscenter has been reconstructioe th numbea f no r of Neolithi Irohousesd e can n Ag , using plans supplied by excavators of prehistoric sites. An earlier rebuilding of a small Neolithic settlement has provided important information about the quantity of material required for even a small dwelling.2 One of the houses measured 20 by 15 ft., and was made of wattle and daub with supporting posts. Nine tons of clay were required, as well as a load of hay, over 200 bundles of reeds, 48 posts up to 8 ft. long, 28 rafters 13 ft. long, 100 lath tie-rodssand , 2000 haze osieand l r twig wattlesfor stonethe and , sfor

FIG . (left7 . ) Sketch pla f Irohousne o nAg t Sk0rbaekea . Note wall-post foundation i s n trench, doublw ero of roof-posts, opposed doors with paving, hearth at centre of (stippled) clay floor, and stall partitions (upper right). (After Skalk, 1966) (right) Pla f burnno t reconstructed Neolithic hous t Allerslevea . Note lin walf ehearthso o ltw basd e,an single line of roof-posts, and charred wood of walls and roof. (After Hansen, 1961, fig. 11). Plans are not to uniform scale floor. Afte materiae th r beed ha l n collected toon , 12kme 10 day builo st d this small house; this figure can be compared with a recorded 4 man-days in Upper Canada to cut and erect a log cabin measuring 20 ft. by 12 ft. Although stone axes would probably double this Canadian figure log-buile th , t houses were certainl ymora e economical proposition in prehistoric Europe if the wood and equipment were avail- ablebuildine Th . larga f go e mortuary hous logf havd eo sen emighe alloweth n ti d mora e imposing barro erectee b wo t d wit hsavina timen gi . Danise Th h wattl daud ean b hous occupies ewa timea recordr d dfo an , s kepf o t rising dam frosd pan t damage smokd an , e blackenin walle e roofd th f an sTh g o . experimenter reckoned that the latter fact was so well marked that even if the house falled ha n down, archaeologists might have considere havo t t di e been burntn I . actual fact house accidentalls th , ewa alight yse t durin s kindline gth wa fird a f egan o thereafter excavated (fig , right).7 t woul ;i interesf o e db comparo t t deposite eth s of destruction from this 'occupied' house with those from houses built and burnt before drying and weathering had occurred. e DanisTh h T.V. experimental burnin s Iroe house th wa nAg f ) geo 2 (PI , II .

1 Nielsen, op. cit. 2 H.-O. Hansen, 'Ungdommelige Oldtidshuse', Kuml (1961), 128. 14 PROCEEDING SOCIETYE TH F O S , 1966-67 recorde markiny db l timbergal s with metal placinplatesy b d gan , thermocouplen so floore house th Th . e collapsed after twenty minutes thirty-fivn i d an , e minutes wa t si almost totally destroyed excep r heavfo t y postse temperaturTh . e centrth t a e reache dsufficiens 9oo°Csidee wa wallth d bake d o st t a an ,tTh mu e. e somth f eo site was excavated after two days, and the plan of the collapsed timbers could be compare prehistorie thao th d t f o t c house upo plae nwhicth f no reconstructioe hth n beed ha n made. Walls Banksand In Britain, a number of attempts have been made to reconstruct defences of hill- fortothed an s r sites, mainl obtaio yt n som etime ideth ef ao involve buildinn di g operations smalA . ltimbee are th earthed f ao ran n rampar t Bindoa t n Hill, Dorset, was reconstructed,1 as well as part of the rampart revetment-wall at Stanwick, Yorkshire.2 A more recent experiment in Britain has been the construction of an earthwork t Overtoa n Dow Wiltshire.n ni maio Tw n purpose behine sli d this attempt, first to recor methode time dth th e d requiresan builo dt earthworn d3a k comparabln ei siz thao et man f o t y prehistoric monument soute th Englandf hn o si a bana , d kan pitch separated by a slight berm, second to observe over a period of years the exact brocesses of decay and collapse that are at work on this structure. The earthwork was duilt in 1960, so that the observations are only just beginning, but already some important facts have emerged aboudecae siltinth e ditce variouf td th yth o f han go s samples buried in the bank.4 Durin constructioe gth diggine ditchbankth e e th th d f f gnan o ,o , some observa- tions were functionamade th primitivf f eo o e us l e diggin collectind gan g implements. Among these red-deee th , r antler picks, quite commonly foun prehistorin do c sites soute th Englandf hn o i , were founefficiene b o dt looseninr fo t chale gth k blockst bu , shovels made of the scapulae of horses were not at all successful, even as scraping tools; wicker baskets were employed for transporting the loads of chalk up to the bank. These British experiments have prompted some estimates of the time involved erectioe th n i monumentf no prehistorin si c times. smalA l ditched . barroft 0 4 f wo

diameter typa f o e, found over muc southerf ho 5 n Britain, would have involven dte weeke peoplon nortr n i ; efo h Britain comparable th , e cairn would have taken less time if sources of material were close at hand. Such estimates as these are useful as a general guide, but it is not quite so easy to visualise the time allotted to the buildin largef go r monuments Pitnacree Th . e barrow might havn e takeme n nte fifty days to build on this reckoning, but for monuments of much greater size, such as Avebur Silburd yan y Hill, such estimate realle sar y quite meaningless. Another experiment on building structures was carried out in Scotland a number

1 M. Wheeler, 'An Early Iron Age "beach-head" at Lulworth, Dorset', Ant. J., xxxm (1953), 13.

. WheelerM 2 Stanwicke Th , Fortifications (1954.). . JewelA . P l3 (ed.) Experimentale Th , Earthwork Overtonn o Down Wiltshire 1960 (1963). 4 P. A. Jewell and G. W. Dimbleby (eds.), 'The Experimental Earthwork on Overton Down, Wiltshire, England: the first four years', PPS, xxxn (1966), 313. . Cornwall AshbeW . , P I d ean , 'An Experimen Fieln i t d Archaeology', Antiquity, xxxv (1961), 129. 5 EXPERIMENTAL ARCHAEOLOGY 15 of years ago, involving the erection of a timber-laced wall with a core of rock (fig. 8).1 purpose experimene Th th f edetermino o t s burnine wa t th f ei sucf go hstructura e would produc phenomenoe eth n know vitrifactions na vitrifiee Th . d . ScotfortE f so - well-knowo to lan e dar requiro nt e description hereexperimentae Th . . ft l6 wals lwa . highft 6 , wid d witan e h horizontal timbers interlaced with stone slabs. After ignition through brushwood fires aroun wale dth l face wale th , l bega buro nt d nan after three hour t collapsedsi e cor basalf Th eo . t rubble becam hotd ere , probably

FIG. 8. Timber-laced wall at Plean, Stirlingshire. Note ends of horizontal timbers in elevation (left) and ends and sides of timbers in section (right). Scale c. -^s (After Childe and Thorneycroft, 1937-8, fig.i) reachin o I2OO°Ct 0 d afteg80 an , r excavatio e bottonth m rubble parth s f o tewa founvitrifiede b o dt , with rock droplet castd stimbef an so r preserved experie Th . - ment proved that a timber-laced wall of this character could become vitrified through explanatioe th firet reasonbu e , th f nsucr o sfo h widespread treatmen thesf o t e Iron Age forts remains uncertain. Stone Erections Another monument thabees subjece ha t nth experimenf o t Britain i t Stones ni - . in 6 bluestones e . henge th ft iy f b o . replic ft e A . 2 on ,y f ab measurino . in 6 . ft g7 weighind an g abou toneasils 2 t swa y transporte waten do overland.d an r 2 Three canoes wit hcrea fouf wo r were use draughe floa o dstonet e th th td an ,t requires dwa only of the order of 9 inches. On land a sledge proved less effective than rollers, and only a dozen men would have been required to pull and push the stone, with another grou movo pt rollere eth turnn si experimento N . s hav t beeeye e nth carrien o t dou transportatio bluestonee th f no s from south-western Wale Stonehengeo st . Even more impressive experiments have been made involvin movemene gth f o t large stone n Easteo s r Island. e statueTh thin o s s islan tons0 d6 weigo ,t p hu

although experiments hav3 e been tried only with 12-ton massesfouns wa dt I tha. t 180 people could mov stone eth e without much difficulty thad t coulan ,ti raisee db d into a vertical position with a small labour force in about two weeks. The problems of difficult terrain and the superposition of up to lo-ton stones on top of the erect 1 V. G. Childe and W. Thorneycroft, 'The Experimental Production of the Phenomena distinctive of Vitrified Forts', PSAS, LXXII (1937-8). ,44 R. J. C. Atkinson, (1956), 95.

2 3 T. Heyerdahl, Aku-Aku (1958). l6 PROCEEDINGS OF THE SOCIETY, 1966-67 statues would have placed enormous strains on the strength and ingenuity of the original statue builders. This experiment, although not perhaps as rigorously observe soms da e others, nevertheless represent impressivn a s e contributioe th o nt whole problem of experimentation with original materials. Scale-model experiment generalle sar lesf yo s value, particularly when roped san leverexperimentae th usede t sar bu , lerectioe worth n ko stonef no t Stonehengsa e with a Y2 scale model has been accepted by both engineers and archaeologists.1 At Old Keig, Aberdeenshire, the pointed asymmetrical bases of a number of the large standing stones have been explained in mathematical terms as aiding the thrust necessary to raise the stone into its erect position, but no attempt seems to have been made to duplicate the proposed method by experiment.2

3. Artifacts Among recent experiments has been the firing of pottery vessels under controlled condition reconstructen si d Romano-British kilns understandine ,th partld ai o yt f go archaeomagnetic dating, but also to gain knowledge of the techniques. A kiln was 3 Wate e basie th buil th f so n r o tNewto ndiameten i kilns . ft turf-supportedd 4 , an r . pottere Th stackes ywa d inside layerd an , greef so n stack e grasth f so . placep to n do After firing up to rooo°C, the kiln was cooled for 30 hours. The whole procedure indicated that there were many difficultie overcome b pottere o t sth y eb . Among difficulte thesth s emovinwa n yi g unfired pots, suggesting tha pottere tth y workshops must have been situated nea kilnse th r . Observation amoune th n stimf o d o t ean materials required for a kiln load were also made, and suggested that about 2 tons woof o d were require successfua r dfo l firing pottee On . r plu assistann sa t could fill kile th n with unfired pot aboun si weeke on t . kilne Mosth f s o tteste d were severely damage actuae th y dlb firing would an , d have required substantial repairs beforea second load could be fired. Other experimental work on coarse pottery was carried out on ware from Glenshee, Perthshire, but was primarily concerned with analysis of the content rather than wit technique hth firing.f eo Nevertheless numbea , triaf ro l

firings woon i , kilnsn di fired , an swer e recorded carriean t dou detailn di e Th . 4 conclusions were prehistori e tha clae th f th ty o c war locals thad pote wa an ,th t s beed ha n firewooa n di pear do t temperatura fir t ea e slightly over 500 . Compar°C - able wor f moro k e recent dats beeha en intimately connected with thin-section examination of pottery. Bone and Antler

Organic materials have also been teste experimenty db thesf o e e testOn s . swa . StoneH Stones e . E Th ,1 of Stonehenge (1924). 2 H. E. Kilbride-Jones,' Stone Circles: a new theory of the erection of the monoliths', PSAS, LXVIII (1933-4), 81. 3 P. Mayes et al., 'The firing of a pottery kiln of a Romano-British type at Boston, Lines.', Archaeometry, iv (1961), 4. W. Thorneycroft, 'Observations on hut-circles near the eastern border of Perthshire, north of Blair- gowrie'4 , PSAS, LXVII (1932-3), 187. EXPERIMENTAL ARCHAEOLOGY IJ designe discoveo dt functionae rth l efficienc Mesolithif yo c harpoons mad antler.f eo 1 Reproduction thesf so e perforated harpoon-heads unperforated an , d head earlief so r date, were thrown at leather-covered pillows. As a result, it was possible to suggest that the reasons for the appearance of the perforation, and its movement from near the base to near the centre of the harpoon-head, were entirely functional (fig. 9). In

?"" " i' '"""I

FIG. g. Harpoons. Hatched line represents animal skin; arrows show direction of thrust and drag. (left) Magdalenian harpoon susceptible to withdrawal when shaft drags, (right) Azilian harpoon with central perforation swing blockd san s withdrawal when shaft drags. (After Thompson, 1954)

early post-glacial time spreae s th growtd dan foresf ho t vegetatio northern i n Spain southerd an n France forced Mesolithi adapo t huntins n thi cma g weapon thao e s , tth harpoon-head, once lodged inside the animal's body, would not pull out if the shaft of the harpoon became entangled in the undergrowth. The central perforation of the harpoon-head functioned as a toggle, swivelling the head around against the animal's skin so that it could not pull out. A more grisly experiment was that carried out many years ago on the trepanning of skulls.2 Skulls from which a roundel of bone has been removed are known from many prehistoric cultures experimene th d an ,devises wa t timo dt operatioe eth n no child'adula n a d tan s skull. Usin gshara p flak glassf eo experimente e th , r scrapeda hole in the adult skull in 50 minutes, but the child's skull only took 4 minutes. The conclusion was that only a child could have survived the operation, and that no adult could have manage remaio dt n alive after nearl hour'n ya sheads worhi n .ko The theory has been proved wrong by clear evidence of the healing of certain adult trepanned skulls, the record apparently being survival after seven such operations. Nevertheless, the experiment indicated that such individuals must have possessed strong constitutions.

Metal There have been many experiments mad metalworn eo Bronze th Irof d ko ean n Ages, bot thin hi s countr d abroad yan e recen Th . t iron-smelting experimenn i t

1 M. W. Thompson, 'Azilian Harpoons', PPS, xx (1954), 204. 2 Munro, Prehistoric Problems (1897), 220. l8 PROCEEDING SOCIETYE TH F O S , 1966-67 Denmark, and comparable work in England, have both indicated that reduction of 1 2 ore could be done in a relatively simple bowl furnace. Casting and beating of copper and bronze objects such as axes and shields have also been carried out experimentally on a number of occasions. Few of these productions have been put to the test for function. One element in bronze working has however been subjected to testing on many occasions, and this is the Bronze Age horn; two forms are known, the Scandinavian lur and the Irish bronze horn (fig. 10). One Scottish example is known. Musical

FIG. 10. Scandinavia (after nlu r Skalk, 1961 d Irisan ) h horn. experiments with lurer have been mad n severao e l occasions e rangth f o d e an , accessible notes has been shown to be quite long ; we can make intelligent guesses 3 about whic thesf ho e notes would have been availabl Bronzo et musicianse eAg t bu , we cannot know which wer facn ei t usedsmallee Th . r Irish end-blow horns present a different problem, in that they are only on rare occasions found with their mouth- certaipiecese noteb e th o f T sn.o playe prehistoriy db c necessars manwa e t i ,th n yi musical experiments to play these instruments without mouthpieces, as any improvi- sations might have falsifie e evidence.th d e rangTh 4f thes eo e instrument- re s i s

1 Voss cit. .op , 2 E. J. Wynne and R. F. Tylecote, 'An experimental investigation into primitive iron-smelting technique', . SteelIronJ d an Inst., 190 (1958), 339. 3 H. C. Broholm el al., The Lures of the Bronze Age (194.9). . ColesM 1. J , 'Irish Bronz Horne theid eAg san r relations with northern Europe', PPS, xxrx (1963), 337. EXPERIMENTAL ARCHAEOLOGY ig stricted to only three or four notes, possibly five or six with proper mouthpieces. Another grou Irisf po h horns consist side-blof so w horns, wher narroe s eth i d wen closed fore ;th comparablm s i instrumente thao eth t f to ivorf so anima d yan l horf no part Africaf so musd an , t have been based upo same nth e idea, perforatioe th f no horns of cattle, leaving the solid tip intact. Translated into bronze, the Irish side- blow horn is basically a one-note instrument, blown only with great difficulty because of the absence of any form of built-up mouthpiece. In the nineteenth century in Ireland attempts to blow these instruments proved to be both unsuccessful instancee andon n i , , fatal Roberr D ; t Bal Dublif lo n apparently burs blooa t d vessel died an d while tryin wreso gt nota t thesf e o fro e em on side-blo w horns. Hide Leatherand Few experiment hid n leathed so ean r have been made, although ancient methods of preparing and tanning hides appear to be well understood. Analysis of surviving material s alsha so contribute thiso dt recen A . t experimen leather-workinn o t g involved the production of a shield comparable in appearance to a surviving Irish specimen. shiele Th d presented special problems ribbee b , d o becaust d an d ha t ei 1 then hardened and made waterproof. After a cold soak to soften the leather, it was scrubbed to remove any excess tannin, then beaten into a wooden mould (PI. I). e dryinTh g process took three days, with regular hammerin preveno gt t shrinkage. After thishiele functionals th dwa t woulbu , d soften under damp conditionsn I . Ireland, and Scotland, it would without doubt be necessary to make some pre- caution against the damp, so that the shield would not collapse. Various methods of hardenin waterproofind gan g were attempted mose th td practicaan , f o e l werus e eth boiling water eithe shielre poureth employer do o t n do dipa immersiod s da , an f no the shield in hot wax (beeswax could be used) for approximately 30 seconds, to allow penetration of the wax through the leather. Both methods were successful in producing stiffened shields capabl f withstandino e g prolonged damp conditions, particularly the wax-impregnation method. The experiment continued with the production of a metal shield of comparable thinnes hardnesd san Lat e th eo st Bronz shielde eAg s from Achmaleddie, Aberdeen- shire, and Yetholm, Roxburghshire. This and one of the leather shields were selected for the final experiment to test their capabilities in standing up to blows and thrusts of sword and spear (PI. II, i).2 The metal shield was cut and slashed to pieces by the first blows. The leather shield withstood repeated heavy blows of a copy of a Bronze Age sword specially sharpene occasione th r conclusioe dfo Th . tha s metae nth wa t l shields were not suitable for actual physical conflict, but the leather shields could have served a useful function in battle.

It is the aim of archaeology to obtain more information and new sources of information to add to the foundations of knowledge upon which our culture is based. Experimental archaeology, when employed scientificall wit awarenesyand han sof

1J. M. Coles, 'European Bronze Age Shields', PPS, xxvm (1962), 175. 2 J. M. Coles, 'Archaeology by Experiment', ///. Land. News, 2 March 1963, 299. 0 2 PROCEEDING SOCIETYE TH F O S , 1966-67 its limitations materialld ad sourcer n ou ca , o yknowledgf t s o e abou paste d th t an , ma consideree yb importann a s whicdn a i y htwa archaeolog comn y ca grip o et s with mentae th l processes, plan activitied san prehistorif so c man.

ACKNOWLEDGMENT gratefur m H.-Oa M I o t l. Hansen, Directo e Historisk-Arkaeologisth f o r k Fors0gscenter, Lejre, Denmark, for comments on this paper.