9th ICAZ Conf''erence, Durharn 2002

Colr.tnisntion, Migrcttir.tn, ctnd Mctrginal Areas, (ed. M. Mondini, S. Muiioz & S. Wickler) pp. 71-85

10. Identifying Dietary Stress in Marginal Environments:Bone Fats, Optimal Foraging Theory and the Seasonal

Round

Alan K. Owtram

The importctnce o.f .fat in the cliet is outlined ancl the importance of bones as a reliable source of fctt is explained-

Diflerent p(ttterns o.f bctne trarrow^ ancl grease exploitation are discttssed with porticular reference to marginal

envirtnments antl how Levels of exploitation will be related to levels of dietary s/ress. The possible role of Optimal

Forcrging Theor.v in a1dressing this issue is outlinecl and adaptations of Marginal Value Theorent and Diet Breadth

,1rugiyi, to bone .fat expptitation are put forward and described. The methodologies Jbr studying patterns of bone fat

exploitcttisn within archaeological assernblages are outlined and four example applications relating to Norse and

pctleo-Eskimo Greenland, lr,lorse lceland and Miclclle lr{eolithic Gotland are used to illustrate whcrt these methods

can show. These case studies are cliscussecl with specific reference to identifying dietary stress in marginal

environrnents ancl the role o.f seasonality to this issue.

Introduction

The importance of fat has often been underestimated in

zooarchaeological studies, with more attention being paid

to the acquisit ion of meat. This is not because information

on the essential role that fats play in the diet has not been

documentecl or because the subject has lacked academic

champions, but is probably more related to Western

society's subconsciously negative view of fat in terms of

both health and body image. This negative view of fat is,

of course, a relatively modern phenomenon that really

only gained momentum within the 1960's (Beardsworth

and Keil 1991. 176).In terms of energy, fat can provide 2257o the number

of calories compared to equal quantit ies of either carbo-

hydrate or protein (Mead et al. 1986; Erasmus 1986).

Within subsistence economies, therefore, sources of fat

are l ikely to be very highly valued. Within economically

marginal environments operating marginal subsistence

economies the full exploitation of available fat resources

may make the difference between the viability of that

society and starvation. [n dietary terms, fat becomes an

even more important resource when reliable sources of

carbohydrate are absent. Speth (1983; 1981 1991) and

Speth and Spielmann (1983) made a highly significant

contribution to archaeologists' understanding of this issue

with their work on the body's reaction to high protein

diets. If the body is reliant upon protein for almost all i ts

energy, amino acids are broken down to meet energy

needs rather than fulf i l l ing their normal function of

replenishing body protein. In severe cases, existing muscle

protein wil l also be broken down. Very high protein diets

are, therefore, very dangerous. Carbohydrates are best at

averting this metabolic problem (Speth and Spielmann

1983, 14), but in many environmental and economic

regimes, sources of carbohydrate are almost totally absent

and, therefore, an adequate supply of dietary fat becomes

essent ia l . Speth and Spielmann (1983, 4) c i te several

ethnographic and historical examples of hunters' practical

awareness of this problem. Another important dietary

issue is that small quantities of certain types of fat are

required within the diet for the proper functioning of the

body. These are referred to as 'essential fatty acids'(Erasmus 1986; Mead et al. 1986)' Furthermore, fats can

be important in the supply of l ipid-soluble vitamins such

as A, D, E, and K (Mead et al. 1986, 459). So, it can be

clearly demonstrated that the levels of fat exploitation

within subsistence economies can be crit ical and all the

more cr i t ical in economies dependant upon animal

products (hunting or pastoral economies) due to the limited

availability of carbohydrate under such regimes. The more

marginal the economy, in terms of meeting calorific needs,

the more essential fat becomes.

Identfuing Dietary Stress in Marginal Environments t5

Dietary fat can be obtained from a number of sources.Plant foods, particularly nuts, can provide dietary fats,but much larger quantit ies are available from animalproducts. Fat can be found within meat and in quite richquantit ies in dairy products, but also in almost pure formin the depot fats ro be found under the skin (adipose fat)and within bones (marrow fat and bone grease). If one isto study the exploitation of animal fats archaeologically,the best option is to concentrate upon bone fats. Thereare a number of reasons fbr this.

Firstly, bones frequently survive archaeologically andin order to exploit bone fats the bones need to bedeliberately broken into. The exploitation of bone fats,therefore, leaves direct archaeological evidence. Toexploit bone marrow, one needs to access the medullarycavity to be found in long-bone shafts and a few otherelements. To exploit bone grease, one must break upbones into pieces and render them. The detailed patternscreated by bone marrow and grease exploitation and howto recognise them are discussed in more detail below.

It is true that the presence of the bones also impliesthe presence of quantit ies of meat and adipose fats aswell, but there is no direct evidence for their exploitation.In particular, it wil l not be clear just how much fat wouldhave been present. The amount of adipose and meat fatsto be found on an animal varies wildly in relation toseason, health and life cycle. If one takes recl deer (Cervuselaphus) as an example, though this point applies tomost mammals, levels of , rump fat ' can be almostnegligible for large parts of the year but there can bequite substantial reserves built up in particular seasons.A further complication is that the pattern is very differentfor males and females (Mitchell, Mccowan and Nicholson1916; clutton-Brock and Albon l9g9). Archaeologicalanimal bone assemblages might also indicate the ex-ploitation of animals for dairy products. This is a proxyindicator of the exploitation of animal fats, but a iatherindirect and diff icult to quantify one. The exploitation ofbone fats themselves can be far more readily quantifiedby direct analysis.

The second reason for concentrating upon bone fatsrelates back to extreme variation in quantities of deporfats that can be seen in animals. Animals store depot fatsas an energy reserve. when they need to call upon thisreserve, they call upon some depots in preference to others.Within this 'fat-mobil ization

sequence,, bone fats areusually the last reserve to be called upon (cheatum 1949,Brookes et al. l9l7; peterson et al. l9g2; Davis et al.1987). In fact, in many cases, the animal could have diedfrom starvation but sti l l have significant levels of fatwithin its bones (Peterson et al. 1992,550). Bones are.therefore, a very reliable source of fat when animals havelitt le other fat in their bodies. Levels of fat within bonescan change with the condition of animals, but not asdrastically as other depots and only in a serious way whenthe animal is in a very poor state. As well as being a goodsource of fat to target, amounts of bone fat available can

be more easily quantified than other fat sources. This canbe done by employing indices calculated using modernspecimens (see below for further discussion) or simply byquantification of amounts of archaeological bone present.Medullary cavities, that would have been fi l led withmarrow fat, could, in theory, actually be measured. Brink(1997) found, during experimental studies of bison. thatdry bone mass was an accurate predictor of an element'sbone grease yield.

So far we have established that bones would be sortafter as a reliable source of fat, that the amount of fatavailable is more quantifiable than other sources and thatthe exploitation of bone fats is more directly demonstrablewithin archaeological assemblages. The final, and key,reason for studying bone fat exploitation is that there is aclear rank order of utility within the resource. Differentelements contain different amounts of marrow and grease.Marrow is easily exploited, whilst bone grease exploitationrequires far more effort. There is a range of utility frombones with large medullary cavities, f i l led with fat. thatcan be accessed and consumed within seconds. to bonesthat contain very little fat but require much effort to obtainit. If one can successfully assess the extent to which bonefats are being exploited, one might have a very goodindicator of the relative needs of people within differenteconomies and environments. Given that bone fats aresuch an important resource, and their exploitation couldbe a matter of l i fe and death, this type of study courd bekey in identifying economic and dietary stress and help usunderstand more about the palaeoeconomics of marginalenvironments.

Patterns of Bone Marrow and Grease Exproitation

The exploitation of bone marrow as a food is very wellknown and accounts of its use can be found within ethno-graphic accounts of a very wide range of peoples in veryvaried environments. For example, good descriptions ofthe use of bone marrow can be found in accounts of Inuitpeoples (e.9. Binford 1978), Native Americans (e..q. Litt leBear 1982), the peoples of Siberia (Levin and potapov1964), the Hadza of Tanzania (O'Connell and Hawkes1988), the Kalahar i San (yel len 1991, Kenr 1993) andAlyawara of Austral ia (o 'connel l and Marshal l l9g9).one could probably obtain a decent account of the use ofbone marrow from an elderly person in almost any culturein any part of the world. Bone marrow is found principallyin the long bones, but also in very small cavities in otherappendicular elements l ike phalanges and in largerquantit ies in the mandible. The pattern associated withits exploitation is simple; the bone's medullary cavitywil l have been deliberately broken into, whilst the boneis sti l l in a relatively fresh condition (for detailed dis-cussion see Outram 1998: 2002\.

'Bone grease' refers to bone fat which is chemicallymuch like the fat in medullary cavities but is trapped

I

16 Alan K. Outram

Fig. 1. Cctribou shcft cylinders .from tibia (rigttt) anclmetutarsals (left).f'ront the paleo-Eskimo .site of ltivnera,We.stern Greenlund (.scule in lun divisions).

within the structure of the bone. The articular ends oflong bones and almost all the axial skeleton are made upof spongy or 'cancellous' bone that contains much fatwithin its structure. This fat can only be easily extractedby breaking up cancellous bone into small pieces andboil ing the pieces to melt the fat out. The fat wil l come rothe surface of the water where it can be solidified as itcools and removed. A very detailed description of thisprocess can be found in Binford's (1g7g) account of theNunamiut, but there are also good accounts of very similarpractices by the Loucheux Indians (Leechman 195 l; 1954)and the Hidatsa (Wilson 1924). According ro Binfbrd( 1978), the processing of bone for grease can be on a fairlyindustrial scale. Long bone ends and axial elements aresaved up in pi les unt i l there are suf f ic ient to makerendering worthwhile. Long bones are often broken nextto the articulation at both the proximal and distal ends sothat the marrow can be poked out tbr consumption and theends saved up. This can produce many very deliberatelycreated long bone cylinders. Such cylinders are present invery large numbers at the Paleo-Eskimo site of It ivnera inGreenland (Mghl l9j2; Ourram 199g; lggg), which canbe seen pictured in Fig. l. The axial and articular unitsare then comminuted into small pieces and boiled. Thisprocess is incredibly labour intensive, especially whencompared to the amount of tat that can be obtainecl frommarrow cavities fbllowing a single blow. The patterncreated by this activity is fairly distinctive, in terms ofwhat gets broken up into small pieces and what remainsrelat ively unfragmented. Detai led discussion of thesepatterns can be found in Outram (1998; 2001a).

An issue of great importance is that there is a definiterank order to the exploitation of bone fat resources. withregard to marrow bones, it is clear that some bones havelarger marrow cavities than others and the degree towhich one bothers wi th the lesser resources is verydependent upon people's needs at the time. There is atraditional saying amongst the Nunamiut that says that"the wolf moves when he hears the Eskimo breakins

mandible for marrow" (Binford 1978, 150). This is. infact, well backed up by observations of the Nunamiutbone fat exploitation habits. Using Binford's records ofNunamiut marrow processing decisions and processingefficiencies, Jones and Metcalfe (1988, 422) mad.e thecalculation that it appears that the Nunamiut were, undernormal conditions, unwill ing to process elements thaty ie lded less thar 500 kcal /hr . This just excludes themandible. The yield rate of elements and this cut-offpoint are displayed in Fig. 2.

with regard to bone grease, it is clear that differentelements will contain different levels of grease, but thereare other issues related to the rank order of grease pro_duction. Blood is created within bone marrow and bonemarrow that contains a lot of blood is often referred to asred marrow. Red marrow contains less fat and moreprotein. In adult animals red marrow is concentratedwithin the cancel lous t issues of the skul l , r ibs andvertebrae, whilst most of the marrow in long bones isprincipally fat (Rixson 2000, I I ). This makes the longbone ends better for bone grease production than axialelements. This distinction can be seen in the ethnographicaccounts. Binford (1918,32) notes that appendicular andaxial elements are stored and processed separately andthere is a preference for the use of the appendicular bonegrease. This distinction is backed up by Wilson (1924),with reference to the Hidatsa. Binford (197g, 146) alsonotes that in times of desperation, the Nunamiut alsoresort to the rendering of dense shaft fragments, whichyield very l itt le indeed.

The implication of these patterns is clear. The choicespeople make about which types of bone to process andhow much of them to process can tell us much about thestate of their subsistence economy and indicate whetherthey are suffering from dietary stress or not. If evidenceof such activit ies is preserved archaeologically and canbe identified and quantified, then there is great potentialfor understanding the extent of subsistence stress amongstpeople apparently l iving in marginal environments.

Using Optimal Foraging Theory as a Framework

The above discussion of resource ranking implies thatsomething akin to optimal foraging theory (OFT) needsto be applied to the study of bone grease exploitation.Below is a discussion of how bone fat exploi tat iondecisions might f lt in with existent OFT models.

There are two selective processes that need to beconsidered with regard to bone fat exploitation strategies.Firstly, if the animal or animals to be exploited have beenkilled some distance from the site where f-at processingwill take place, one has to consider element transportchoices. one needs to understand which f-at resourcesarrived at the site in order to understand the selection forf-at processing. Also, the transport decisions themselvesmight relate to economic strategy (see Binford l97g).

TIj

Identifying Dietarv stress in Marginctr Environments 77

.c

?o

J

1 750

1 500

1250

1000

750

500

250

0

. = ( 6 - J- o ( / ) ( U ( d= k * =

_.:.Y s o-

E T N

* e g d t . e =" - = = J O . > =E H - P ? F T E( s g ^ \ ! - o* F T U F

Skeletal Part

Fig' 2' A graph to show marrow extraction efficiencies.frctm different caribouand the processing threshold (values deriverl.frorn Jones ancl Metcatfb I9gB,

This is l ikely to apply much more to hunter rgatherersocieties than to farming ones, where animals ui. -or.likely to be slaughtered where they are needed. There hasbeen much discussion in the riterature of differentialskeletal part transport and the very many considerationsthat need to be taken into account when considering it(e.g. Binford 1978; Metcal fe and Jones lggg; O,Connel l ,Hawks and Blurton Jones 1990; outram 2001b). Below,element transport choices wil l be modelled within theframework of an adaptation of Marginar value Theorem(MVT).

Marginal value Theorem (charnov rgr6) was origin-al ly created within the f ie ld of ecology to model anorganism's use of different patches of resource, but it caneasily be applied to human behaviour as well. Anyonewith experience of collecting berries from wild hedgeiowswill understand this model very well. A berry collectorwil l stop at a bush (or patch) and collect berries for aperiod of time. They are likely to leave that patch beforeall the berries on that bush are used up. This is becauseit gets harder and harder to flnd the berries on that bushand the rate of return gets rower and lower. At somepoint, one makes the judgement that it is probably betterto go and find another bush. If there are rots of goodbushes about, this point is l ikely to come sooner ratherthan later, but if bushes themselves are hard to find, onemay stay longer at the present bush. MVT is ail aboutpredicting the optimal point at which one should movefrom one patch of resource to another.

A graphical solution for MVT can be seen in Fig. 3(after Bettinger 199r, f ig. a.3). one can see thar theenergy acquired in the patch (expressed as a rate of return.usually energy/time) drops off as more time is spent atthe patch. The dotted l ine represents the overall averagerate of energy return of the environment, takins into

skeletal perts by the Nunutniut peopletable 3 ).

account travel t ime, search time and handling time. Ifthe overall return rate is higher than that from the presentpatch, then one would be better off leaving ancl lookingfor another patch. where the straight overall return l inetouches the curved patch l ine is the opt imum t ime toleave. The distance along the x-axis from the origin tothe departure t ime is the opt imal foraging t ime. thedistance in the opposite direction along the x-axis fromthe origin, to the intersection of the axis by overall returnIine, represents the time that it is worth travell ing to findother resources. The better the environment is, the steeperthe overall l ine wil l be. It wil l touch the patch curveearlier and create a shorter optimal fbraging time ancl ashorter acceptable t ravel t ime. I f the environment ispoorer, the reverse is true.

This model can easi ly be adapted to apply to thehunting of an animal and the exploitation of its carcass.A graphical solution for the adapted MVT can be seen inFig. 4' This works in a very sirnilar way. In this case thecurved line represents a particular kil l ancl the rate ofreturn one can get from it as one field-butchers it fortransport, selecting to take the highest return erementsfirst (l ike picking the easiest to find berries). The dortedline represents the overall return fbr the environment interms of f inding, kil l ing and transporting another anirnal.The same rules apply. The relationship between thesetwo l ines gives one the acceptable hunt ing t ime andoptimal butchery/transport t ime. This time the modelwil l dictate how many and which bones get transportedback to the s i te. This mocrel , however, arso sees theintroduction of the notion of total need. Infinite need isassumed within the standard MVT model, as the type ofresource being dealt with is small and slowly accrued.Hunting can be different if a very rarge prey is captured.Total need might be met from the one carcass, if storase

Iutility cut-off?]

l 8 AIan K. Outram

otrJ(E5oO

(t(ruJz]U

Energy Acquiredin patch

f- Optimal Departure Time

Travel ForagingT IME

Fig. -i. A visttctl representation of Marginol Value Theorem (after Bettinger I99l,.fig. a.3)

Total Need (A)

otuglao

CIIEIUztrJ

fdr?silb-;lme I

.y'= optimal cut-off (A)

Total Need (B)

i acceotable-hunting time

Travel/Hunting Butchery/TransportTIME (and number of elements transported, starting with highest yielding elements,l

Fig. 4. A vi,surtl representation o.l' Marginal Vttlue Theorem as adapted to the consideration of hunters' elementtransport choice.s.

is not possible. If total need is reached before the inter-section of the two lines, as in the case of total need B, theoptimal cut-off wil l be earlier (cut-off point B), but iftotal need is higher (A) then normal rules apply.

Bettinger ( l99l , 108) suggested that element transportchoices could be dictated by the theory of Diet Breadth.

This is certainly true, but the adapted MVT model pre-sented in Fig. 4 is possibly a more powerful tool becauseit effectively combines MVT with Diet Breadth. The modelassumes that higher-ranking elements, in terms of energ-etic intake, wil l be exploited before lower ones. Thisassumption underlies the Diet Breadth Model (see below),

\

Identiff ing Dietary stress in Marginal Envirorunent.g t9

Overall Foraging Time(search + handling)

Optimal Point

Search Time

Handling Time

Overall Ti

Optimal Point

Procurement Time

Processing Time

to.=ETo

E locUJ

o=c3

oooE

!to5q(,

o)ocIIJ

o.=cf,

oo.oE't=

2 3 4 5 6 7 8

Diet Breadth(as measured by lowest ranking item in diet)

Fig. 5. A visual representation of the Diet Breadth Moclel(cfter Bettinger I99I , .fig. a.l ).

but is not normally present in MVT because the unitsbeing collected are normally assumed to be equal in value.In the adapted MVT model, both the rank order ofexploitation and the optimal cut-off point are predicted.

Diet Breadth comes into its own in considering thechoices people must make regarding which elements toprocess for their fat content. The Diet Breadth theory(MacArthur and Pianka 1966) can be seen graphicallydisplayed in Fig. 5 (afier Bettinger rggr, f ig. 4.1). Dietaryitems are arranged along the x-axis in decreasing order ofthe food value of that item. This food value represents theenergy yield divided by the handling rime. A high-rankingitem would be something l ike a large soft fruit, r ich insugar and easy to handle, whilst a low-ranking item mightbe something l ike a very small nut in a very hard shell. Itwould, in an ideal world, be best if one could just exploitthe very high-ranking items, but this is where the searchtime comes in. The more items there are in one's diet, thegreater the chance of encountering them whilst searchingfor fbod. Diet Breadth suggests the best balance betweenease of handling and ease of searching.

In the graph (Fig. 5), it can be seen that as items areadded to the diet (lower and lower ranking ones) theamount of handling time goes up, but at the same timethe time needed to procure items goes down. The optimalnumber of items to include in the diet is then determinedby where the two lines cross. Alternatively, if the twofactors are combined into one line representing total timespent per unit of energy gained. the optimal point iswhere the time spent per unit of energy is lowest. Ifprocurement is a problem (i.e. a low general yield from

3 4 5 6 7Number of Different Marrow Bones Exploited(as measured by lowest ranking bone exploited)

Fig. 6. A visual representation oJ'the Diet Brertrtth Moctetas adapted to the consideration of'bone .f'at exploitutionchoices.

the environment), then this point wil l be reached laterand more low-ranking items wil l be included. If procure-ment is not a problem then the optimal point wil l bereached early and only highly-ranked food items wil l beincluded in the diet.

This model can easily be adapted to consider bone fatexploitation more explicit ly (see Fig. 6). This rime rhedifferent fat-bearing bones are put in rank order accordingto processing ef f ic iency. As suggested by the aoovediscussion of bone fat exploitation patterns, this wil l notsimply be a l ist of which bones contain the most fat.Highest ranking resources wil l be large, easily accessedmarrow cavities, then smaller cavities wil l follow. onethen has to move down to much less efficient greaseprocessing and start with large appendicular ends, thensmaller ones, then lower quality greasy elements fiomthe axial skeleton and, perhaps, f inally attempts to extracttiny amounts of fat from Iong bone shaft fragments.clearly the more different types of bone f-at resources oneexploits, the easier it wil l be to procure the raw materials.The important question is how easy is it to procure f 'atresources. The harder it is to procure resources of fat(and one should perhaps be thinking about sources of fatholistically, rather than just bone fat), the later the cut-off point wil l be in terms of the exploitation of low-ranking bone fat resources.

From the above discussion of oFT models, it is crearto see how peop les ' an ima l p rocess ing dec is ions ,particularly for fat resources, can be very clearly l inked tothe environment around them. optimal behaviour wil lchange depending on just how marginal the environment

80 Alan K. Outrant

is. By apply ing OFT in th is way, is one being ,environ_

mentally determinist', as some would charge? correctlyused, OFT is not determinist ic, but rather provides ameasuring stick against which actual human behaviourcan be compared (Foley I 985, 222; Bettinger 199 I , 106).without such a measuring stick it is very diff icult toascribe meaning to different human activit ies. How canone discuss the issue of marginality at all without referenceto models that deal with resource availabil ity within anenvironment and how people can best l ive in such anenvironment ' l

It is certainly true that optimal models do not deter-mine human action, but having said that it can be arguedthat they can l imi t i t . Higgs and Jarman (r975,2) statedthat " . . .u l t imately al l human cul ture and society is basedupon and is only made possible by biological and econ-omic viabil ity...and however unfashionable the terms andideas behind determinism may be, the very existence ofnatural laws presupposes a degree of determinism.,, Thisstatement is extremely diff icult to refute. people have tobe able biologically to survive to have any sort of cultureat all! optimal models give us an idea of the best thar canbe done in a given environment, in terms of efficiency ofresource exploitation. This is an upper l imit that peoplemay not necessarily adhere to. However, there is also alower l imit. People neecl so many calories and certainnutr ients to survive. Almost by def in i t ion, in aneconomically marginal environment, these two limits wil lbe close together and at the very edge of viabil ity they areone and the same thing. one has to exploit efficiently allavailable resources to l ive. In a marginal environment,the chosen mode of economy wlLL be heavily determinedby that environment. As such, optimar models becomehighly relevant to understanding how that economyfunctions. within this type of theoretical framework, thestudy of bone fat exploitation patterns could represent apowerful tool for assessing marginality and levels ofdietary stress.

Methods

In order to understand the exploitation of bone fats, withina theoretical framework l ike the one presented above,one has to have some understanding of the rank order ofelements, and portions of elements, in terms of their faty ie ld and processing ef f ic iency. Lewis Binford (r97g)revolutionised this field of study in two ways. Firstly, hecreated indices of fat uti l i ty (both for marrow and grease)for the different elements of two species, sheep andcaribou. secondly, he recorded efficiencies fbr greaseprocessing by the Nunamiut using traditional techniques.Jones and Metcalf.e ( 1988) further refined and discu.ssedthese data. whilst many food uti l i ty indices ancl meatuti l i ty indices have been produced since lgig. manv donot ref-er specifically to bone fats. Some exception, ur"Br ink and Dawe's (1989) work on bison, Brumenschine

and Madrigal's (1993) work on East African ungulatesand outram and Rowley-Conwy's ( 199g) work on horses.Despite the relative lack of attention given to bone fatuti l i ty in the production of indices, we now have a fairrange of information on the relative yields of bone fatsfrom sample species with different stature and locomotorcharacteristics. It is also extremely useful that Brink's(1991) experimental work showed that fat content is veryclosely correlated to dry bone mass.

In order to take this line of enquiry further, one needsto be able to assess exactly which bone resources havebeen exploited for their fat content, and quantify thisexploitation. The development of just such an analyticalprotocol, for application to archaeological bone assem-blages, has been the subject of much of this author'srecent work. Detailed explanations of this methodologycan be found in Ourram (1998;2001a; 2002) and detaileddiscussion of example applications of the methodologycan be found in Outram (1998; 1999, in press; for th_coming). what follows is a very brief summary and shouldbe treated as such. The method relies upon the threestrands of evidence, the fragmentation levels of differenttypes of bone, the fracture patterns within the assemblageand the holistic consideration of a wide range of taphon-omic indicators.

with regard to fragmentation rever, one clearly needsto consider what types of bone have been broken up withregard to their potential value as a fat resource. Themethod used for recording the levels of fragmentation isas fo l lows. Al l f ragments are included, whetheridentif iable or not. whilst identif ication to species andelement may not be possible, such fragments sti l l carryvaluable information in the form of size, fracture patternsand bone type. It is possible to tell cancellous bone fromdiaphysis bone on even very smail fragments and suchinformation is very important in the context of a studylike this. The entire assemblage is divided into size classes(by maximum dimension). The size c lasses used are<20mm, 20-30mm, 30-40mm, 40-50mm, 50_60mm,60-80mm, 80-100mm, 100+mm and part and wholebones. whole bones clearly have not been exploited forgrease at all. Part bones include bones that are not whole,but represent whole units that could have been exploitedfor grease but were not broken up. part bones includeentire epiphyses and complete vertebral centra.

Quantif ication of the size classes is by number andmass. whilst numerical data is collected, mass data tendsto be more useful because it represents actual amounts ofbone present. clearly one unbroken large bone representsthe same amount of potential fat as a similar elementbroken into many pieces, yet the latter would be repre_sented by many hundreds on a numerical count. By mass,bo th wou ld be su i tab ly equa l . A lso , as p rev ious lymentioned, Brink (1991) concluded that dry bone masswas an accurate predictor of elements' bone grease uti l i ty.

For each size class, a distinction is made betweenwhether the bone is cancellous or cortical in nature. For

\

Identifving Dietary S/ress in Marginal Environments 8 l

large size c lasses, dist inct ion is made between axialcancellous bone (other than ribs), ribs, articular bonefrom appendicular elements and diaphysis bone. Thisenables one to see, in terms of bone fat uti l i ty, whichtypes of bone had been fragmented and to what level.

The study of fracture patterns is essential in establishingwhether the fragmentation was the result of human agencyand, if it was, to quantify the extent of deliberate com_minution. when dynamically fracturecl in a fresh state,dense diaphysis bone creates a very distinctive fracturepattern. Such fractures are characterised by helical fracturelines radiating out from the point of impact. The fracturesurface will form either an acute or obtuse angle to thecortical surface of the bone. This fracture surface will alsotend to be smooth in texture (Johnson l9g5; Morlan t9g4;Outram 1998, 2001a, 2002). As bones dry out they developsmall cracks that interfere with the fracture line, creatingroughness or steps, hence affecting the fracture shape andsurface texture. As bones lose their organic content theyreact differently to force. Loss of elasticity results in bonessnapping in straight l ines that tend to be perpendicular tothe cortical surface. A targely mineralised bone wil l breakwith a straight, rough edge that is close to being at right-angles to the cortical surface.

The three criteria of fracture outline (shape), fractureangle (to cortical surface) and fracture texture (smoothor rough) can be used to characterise large assemblagesof fragments in terms of the amount of deliberatefracturing of fresh bones versus levels of post-depositionalbreakage of dry bones (Vil la and Mahieu r99l: outram1998; 2001a;2002), An indexing sysrem has been de_veloped and tesred exper imental ly by Outram (199g;2002).

All diaphysis fragments of 30mm or more in lengthare studied fbr fracture type providing that preservationis good enough. For each of the three criteria, a score of0, I or 2 is awarded. In broad terms, a score of 0 denotesthat that criterion is consistent with fracture of a freshbone, a score of one denotes a mixture of fresh and unfieshfeatures (but with fresh still dominating) and 2 denotesthat unfresh f'eatures are dominant. Much greater detailis available in Outram (2002). Shape, angle and rough_ness are all estimated by eye. This is essential to makeassessments of large samples practical. Individual mis-judgements wil l be irrelevant as the method is beingemployed to characterise the assemblage in general andsample sizes are large. The angle and outl ine charac-ter ist ics are fa i r ly easi ly def ined, but assessment ofroughness is more subjective and relies upon the analysthaving a good mental template (l ike much zooarchaeo_logical analysis) of the possible range.

When the scores are added together one ends up witha score from 0 to 6 for each fragment, called the FractureFreshness Index (FFI) score (Outram 20}la; 2002).Scores of 0, I and 2 wil l represent bones broken in arelatively fresh state. Scores of 3, 4,5 represent eitherbones that were broken when becoming fairly dry (un_

likely to be for fat extraction) or bones which had somefresh fracture on them but were further fragmented whenunfresh. A score of 6 represents a bone with no evidenceof fresh fracture. The profile of scores and overall averagefor a sample can be displayed.

The FFI score is a very good indicator of the taph_onomic history of the assemblage, but other indicatorscan be recorded that provide more detail and help to dealwith potential problems of equifinality within the FFI. Ifone records whether or not a fragment has an example ofan indiv idual mineral ised break i t is possible todistinguish between bones that had fresh features, butthen got broken when mineralized, and bones that showedno fresh or completely mineralized,features, but were drywhen broken ( i .e. scores 4, 5, 6) . Such completelymineralised breaks are easy to spot (on their own theywould score 6). New breaks (caused by excavation orstorage) can also be recorded, as the fracture surf'aces willbe an obviously different colour. Dynamic impact scars,created at the point of impact on a fresh bone, can also berecorded as evidence of deliberate fracture, much likebulbs of percussion on fl ints. If the bone cylinder wasbroken on an anvil, there may be a rebound scar due to theopposing force of the anvil (see Outram 2002, fig. 6.g).

several other criteria are important for understandingthe taphonomic history of the assemblage. Shaft frag_ments are also studied for evidence of animal gnawingand butchery (chops, cuts, polish and sawing). Numbersof burnt fragments are counted for the entire sample atthe level of size class and bone type. Butchery can clearlyadd to the overall level of deliberate breakage, however,the breakage of bones for butchery purposes wil l berestricted to particular elements for a particular purpose,such as an alternative to disarticulating a diff icult joint.There is also likely to be a difference between the fracturesproduced by chopping through meat and bone and thosecreated by direct impact to the bone. [t is essential thatall indicators (fragmentation level, bone types, fracturepatterns, gnawing, butchery and burning) are consideredholistically to effect a successful interpretation and avoidpitfalls of equifinaliry.

Case Studies

Below, four case studies of the analysis of bone fatexploi tat ion patterns are out l ined. The detai ls of theresults for each case study wil l not be displayed as thesehave been or wil l be published at some length elsewhere(Outram 1998; 1999; in press; forthcoming). The purposeof this section is to present the principal points of interestemerging from the results of each study, so that they canbe discussed with reference to the question of marginality.All case studies relate to economies that were highlydependent upon animal products and would have hadlitt le access to large quantit ies of dietary carbohydrate.As such, fat becomes a key resource.

82 Alan K. Outram

Fig. 7. Helically-frac:tured shaft splinter,s frorn the Norsesettlement rt' Sandne,s, We,ste rn Greenland, resulting fromnrorrow extraction (scule in lcm divisions).

Fig. B. Heavily comminuted cancellous bone from theN o r s e s e tt I ement of' s undne.c, w e st e rn G re e nl and, re s urtin g.from bone grease rendering (scale in lcm clivisionsl.

The flrst two case studies are set in Western Greenland(for detailed results see Outram 1999). Greenland, beingphysically diff icult to reach, being at high latitude with aharsh cl imate and having di f f icul t terrain wi th l i t t lesuitable land for settlement, is perhaps a classic exampleof somewhere people think of as being marginal. Thefirst study looked at two medieval Norse farmsteads calledSandnes (V5l) and Niaquussat (V48). The Norse semledon Greenland in around AD 985 (Buckland et al. 1996)and operated a principally pastoral economy using cattle,sheep and goats. The diet was augmented with huntingof caribou, seals and wild birds, but f ishing does not

Fig. 9. Large, unprocessed rih fragrnent.s from the Norsesettlement of Sandnes, Western Greenland (scole in lcntdivision,s).

appear to have occurred much at all (for detailed dis-cussion of economy see McGovern 1985; McGovern eral. 1996, Buckland et al. 1996). It is clear thar l i fe wasvery harsh indeed and one of the key limiting f-actors inthe economy was the short growing season and limitedavailabil ity of winter fodder for animals (McGovern1985). With the climatic downturn at the start of theLittle Ice Age, matters became worse and the Norse finallyabandoned Greenland by the end of the l5th Century(Buckland et al. 1996).

Before the study by this author, Bucklancl et al. (1996)had already suggested from entomological evidence thatthe Norse were so stressed that they needed to renderbones extensively for fat before depositing them in theirmiddens. The study of the bone fracture and fragment-at ion patterns very much conf i rmed this. The landmammal bones had been very heavily processed indeed,at both farms, and the pattern was very clear and con-sistent. Almost all long bone shafts had been cleliberatelyfractured to extract marrow (see Fig. 7) and almost allcancellous bone (axial and appendicular) had been com-minuted into very small pieces for rendering (see Fig. g).The only exceptions to this were the ribs (see Fig. 9),which are quite poor in terms of their grease value andcontain largely red marrow (see above). Seal bone tendednot to be fragmented. It was also clear that this patternwas almost entirely the result of deliberate human action,when considering all the taphonomic data holistically.The conclusion is that the processing of land mammalbone extends a very long way along the x-axis of a dietbreadth model, which suggests that procurement ofsufficient fat resources, and probably food in general,was very di f f icul t , forc ing the exploi tat ion of verymarginal resources of bone fat. There wil l be morediscussion of seal below.

The other case study involved the examination of two

Identifying Dietary Stress in Marginal Environrnents 83

Paleo-Eskimo sites of the Saqqaq culture, which dates toapproximately 2400*1000 BC (Gr@nnow lgBB, 24).Qeqertasussuk (see Gr@nnow 1988; Bdcher and Fredskild1993) is a specialist seal hunting site and Itivnera (seeMghl 1912) is a specialist caribou hunting site. Just l ikeon the Norse sites, the seal bone was not fragmentedmuch at all. There was no evidence for the extraction offat from them. At Itivnera, however, it was clear that thecaribou bones were being processed for both grease andmarrow. The patterning and method of processing seemedto closely follow that described by Binford ( 1978) for theNunamiut. Long bone ends had been deliberately removedto leave shaft cylinders (Fig. 1). There was good evidencefor the grease rendering, but certainly not in the quantitiesencountered at the Norse sites. The Saqqaq people leftquite a few axial and appendicular cancellous portionsunexploited, leading to the conclusion that they were lessstressed and did not need to extend their diet breadth. infat terms, as far as the Norse.

The third case study continues the theme by examiningearly medieval Norse settlements on Iceland at two farmscalled Hofstadir and Sveigakot in Northern Iceland (seeOutram in press fbr detailed study). Iceland has somesimilar features to Greenland, in environmental terms,but it does not have quite such a harsh climate and hasbetter land availabil ity, in other words is less marginal.The economy of Norse Iceland was also primarily pastoraland based upon cattle, sheep and goat with additionalexploitation of pig and seals (Amorosi 1992; Tinsleyforthcoming). Caribou were not available as a huntedresource (Amorosi 1992. 123). but on Iceland there wasmuch fishing of both fresh water and sea species (Amorosi1992, Byock 2001). The results of bone fracture andfragmentation analysis on these sites were in stark contrastto the Greenlandic results. It was immediately apparentthat, whilst there was evidence for a certain amount ofmarrow extraction, there was no evidence at all for bonegrease rendering. Fragmentation levels were very low andmany bones survived whole. This suggests that the pro-curement of sufficient fat was not at all difficult within theIcelandic economy and that perhaps their economy wasindeed less marginal than that on Greenland.

The final study relates to a Middle Neolithic site onthe Swedish Island of Gotland called Ajvide (for deraileddiscussion of resul ts see Outram forthcoming). TheMiddle Neolithic culture on the island is referred to asthe 'Pitted Ware' culture, which spans from about 2100-2300 Cal. BC (Burenhult 1991). Gotland had seen rheintroduction of domestic animals including cattle, sheepand pigs in the Early Neolithic, but, during the pittedWare phase, it seems that the inhabitants of the islandreverted to hunting and gathering, exploit ing variousspecies of seal, wild/feral pigs and fish (Rowley-Conwyand Stori 1991). Just like on Iceland, the bone fractureand fragmentation study uncovered very little evidencefor systematic bone grease rendering, but it was clearthat marrow was regularly exploited.

Discussion

It is clear from the above results that the patterns of bonefat exploitation were very diff.erent within the economiesdiscussed. On a simple level, one can argue the following,with regard to relationship between bone fat exploitationand marginality. The Norse settlers and paleo-Eskimos

on Greenland shared a very similar environment, but theNorse economy was a cultural import from elsewhere thatwas not ideally adapted to the environment of Greenland.The result was significant stress upon the subsistenceeconomy. The difficulty in procuring suftlcient fat re-sources forced an increase in the diet breadth for bone fatexploitation. The clear evidence that this pattern of bonefat exploitation left is detectable archaeologically in a waythat other evidence of stress might not be. The Saqqaqpeoples clearly needed to exploit some bone grease, but itis equally clear that they were not in such a desperatesituation, and probably had an economy that was far betteradapted to the environment.

It appears that the Norse Greenlanders were doingtheir best to optimise returns from their chosen economy,but it also seems that they had culturally chosen to operatean economy that fell some way short of optimal for theirenvironment. This underlines the point that OFT modelsdo not determine actual behaviour. People do makecultural choices about how they wil l l ive within a givenenvironment. However, if we do not have optimal models,or at least some set measuring stick, how would we knowwhether cultural decisions were divergent from practicaladaptations to environment and of particular interest.Having said this, one must return to the point that l i fe isonly viable within certain l imits. Cultural variabil ity isonly possible within the l imits of optimal models ar thetop end and economic and biological viabil ity at the other.On Greenland, the Norse did optimise their economy asfar as they could without fundamentally changing it, butthey did not truly optimise it to the environment. Withthe onset of climatic downturn, the l imits of optimalityand viabil ity came closer together and their culturallychosen economy began to fa l l outs ide those l imi ts.Environment demanded adaptation, but culture refusedand the result was the abandonment of Greenland.

The Icelandic and Got landic examples show twoeconomies that do not appear to have been so dietarilystressed. In both cases it appears that only marrow wasbeing exploited, which is relatively highly ranked in adiet breadth model indicating that fat and food procure-ment in general was not that diff icult. The four studiestaken together show that the fracture and fragmentationmethods can pick up the full range of different fatexploitation patterns and that these patterns, once identi-f ied, can be very useful in interpreting levels of economicstress.

Looking in greater detail at some of the complexitiesof these case studies, it is necessary to draw attention tosome important interpretative issues, however. The first

84 Alan K. Outram

Y

and most important issue relates to the seasonal roundwithin subsistence economies. It is extremely importantto realise that economies are not static throughout theyear and some times of year are likely to be far morestressed than others. It is highly l ikely that extremepatterns of bone fat exploitation will relate to particularseasonal dearths in food resources. This does not in anyway negate the result for the settlement as a whole,however, as a subsistence economy is only as good as itsweakest part. If one cannot weather the bad times throughoptimising resources (including use of storage whereapplicable), the good times are irrelevant.

In the case of the Greenlandic peoples, in both casestudies the season of dearth was probably winter. In theNorse example there was food available from the pastoraleconomy during the summer, which would be nutrit iousand contain a good source of fat. The sealing appears tohave taken place dur ing spr ing (McGovern 1985, 101).Seals have vast amounts of fat in their blubber, so thebone fats would have been, in relative terms, much furtherdown a diet breadth ranking l ist. This is no doubt whyseal bones were not fragmented by either the Norse orSaqqaq peoples. The hunting of caribou would probablybe best in Autumn, when they are at their f ittest anddomestic stock may well also have been slaughtered atthis time, before over-wintering in the byres. This leaveswinter, with only stored food available, including manybones from the autumn kil ls that could be rendered. InIceland and at Ajvide there is not such a significant gapin the provisioning of food. On Iceland better environ-mental conditions could have led to the possibil i ty ofstoring up more dairy products for winter, but, perhapsmore importantly, the large amount of fishing carriedout would have allowed stockfish to be kept for the winterperiod. At Ajvide there was also much fishing, but it wasalso clear that sealing was not so seasonally l imited withdifferent species being exploited through from autumn tothe spring (Rowley-Conwy and Stori 1991). lt can beseen that the seasonal availabil ity of fat could be thecrit ical factor in the study of these economies.

One f inal issue for discussion is a s l ight warning.Bone fats can be valued for industrial processes, forexample waterproofing skins and treating bowstrings(Binford 1978, 24), tanning (Levin and Potapov 1964,636) and l ight ing (Burch 1972, 362). As such, theconsideration of bone fat exploitation as an indicator ofdietary stress should not be taken out of the overalleconomic and archaeological context of the case study.Arguments about marginality and stress must use the fullrange of economic and environmental data holistically.

Conclusion

The study of patterns of bone fat exploitation in thearchaeological record can be a powerful tool for under-standing dietary stress wi th in economical ly marginal

environments. Our understanding of such environmentscan be enhanced by the use of appropriate models ofoptimal behaviour for such environments, when suchmodels are correctly used as measuring sticks, not asdetermining laws. In considering marginal economies,special attention needs to be given to the identification ofperiods of seasonal stress. When using bone marrow andgrease exploitation patterns as an indicator of dietarystress, this evidence should be fully integrated with otheravailable economic and environmental data. and nottreated in isolation.

AcknowledgementsI would like to thank Helen and Derek Outram for theircomments on earlier drafts and Mariana Mondini andSebasti6n Mufloz for organising the conference sessionin which I presented a paper on which this paper isbased.

References

Amorosi , T.1992. Cl imat ic impact and humarn response in NortheastIceland: archaeological invest igat ions at Svalbard, | 986-1988,pp. 103-38 in Morr is , C. D. and Rackharn, D. J. (eds), Norsermd Later Settlentent und Subsistenc'e in the North Atlantic.Glasgow: Univers i ty of Glasgow, Departrnent of Archaeology.

Beardsworth, A. and Keil, T. 1991. Sociologt on tlte Menu. London:Routledge.

Bett inger, R. L. 1991. Hunter-Gutherers: Arc:haeologi t :a l undEvolutionary Theory,. New York: Plenum Press.

B lumesch ine , R . J . and Madr iga l , T . G . 1993 . Var iab i l i t y i n l ongbone marrow yields of East Afiican ungulates and its zooarchaeo-logical i mplication s. J o u rnu I o.f ' A rc h u e o I o g i t ' u I S r: i e nc: e 20, 5 5 5-8 1 .

Bdcher, J. and Fredski ld, B. 1993. Plant and ar thropod remainsfrom the Paleo-Eskimcl s i t .e on Qeqertasussuk, West Greenland.Meddelelser om GrQnland; Geo,science 30.

Binfbrd, L. R. 19J8. Nunumiut Ethnourchtreologt ' . New York:Academic Press.

Br ink, J. W. 1997. Fat content in leg bones of Bison bison, andapplications to archaeology . J o u nru I of A rc hrt e o I o g i r: tt I S c i e n c e2 4 . 2 5 9 - 1 4 .

Br ink, J. and Dawe, B. 1989. Final Report o. f ' the 1985 and I986Field Season at Heud-Smu,shed- ln Bu. f f 'u lo Jump, Aberto.Edmonton: Archaeological Survcy of Alberta.

Brookes, P. M., Hanks, J. and Ludbrook, J. V. I 917 . Bone marrowas an index of condition in African ungulatcs . South Afric'unJournal of Wildlife Research 7, 6l-6.

Buckland, P. C. , Amorosi , T. , Bar low, L. K. , Dugmore, A. J. ,Mayewsk i , P . A . , McGovern , T . H . , Og i l v ie , A . E . J . , Sad le r , J .P. and Skidmore, P. 1996. Bioarchaeological and c l imatologicalevidence for the fate of Norse farmers in Mcdieval Greenland.Ant iqui t v ' 70, 88-96.

Burch, E. S. l912.The car ibou/wi ld re indeer as human resource.American Ant iqui t t , 37, 339-68.

Burenhul t , G. 1997. Introduct ion, pp. IX-XXI in Burenhul t , G.(ed.), Remote Sensing, Volume 1 (Theses and Papers in North-European Archaeology | 3a). Stockholm: Stockholm Univers i ty .

Byock, J. L. 2001. Vik ing Age lcelctnd. London: Penguin.Charnov, E. L. 1976. Optirnal fbraging: the marginal value theorem.

Theoreticul Popttlation Biologt 9, 129-36.Cheatum, E. L. 1949. Bone marrow as an index of malnutr i t ion in

deer. ly'ew York Stote Conservation 3. 19-22.

Identifiing Dietary Stress in Mar7inal Environtnents 85

Clut ton-Brock, T. H. and Albon, S. D. 1989. Rert Deer in theHi,qhlonds. Oxfbrd: BSP Professional Books.

Dav is , J . L . , Va lkenburg , P . and Reed , S . J . 1987 . Cor re la t i on anddeplet ion pat terns of marrow fat in car ibou bones. Journal ofWiltl l i .fe Munugement 51, 365-71.

E,rasmus, U. 1986. Futs ctndOil,s: the Complete Guide to Fats andOiLs in Heul th unt l Nutr i t ion. Vancouver: Al ive Books.

Fo lcy , R . 1985 . Op t ima l i t y rheory i n an th ropo logy . Man20 ,222-42.

Gronnow, B. 1988. Prehistory and permafrost : invest igat ions at theSaqqaq s i te, Qeqcrtasussuk, Disco Bay, West Greenland. . /or . r r -nal o.f Dani.sh Archueolog ,- 7 , 24-39.

H iggs , E . S . and Ja rman , M. R . 1975 . Pa laeoeconomy, pp . l -7 i n E .S. Higgs (cd.) , Pulueoeconomy. Cambridge: Cambridge Uni-ve rs i t y P rcss .

Johnson, E. 1985. Current developments in bone technology, pp.l5 l -235, in Schi f l 'er , M. B. (ed.) , Advunces in ArchoeolocicalMethod uncl Thertrr- Vol. B. New York: Academic press.

Jones, K. T. and Metcal fe, D. 1988. Bare bones archaeology: bonemarrow indices and ef f ic iency. Journt t l o. f ' ArchaeologictL lSt:ience 15, 415-423.

Kent, S. 1993. Var iabi l i ty in faunal assemblages: the inf luence ofhunt ing ski l l , shar ing, dogs, and mode of cooking on faunalremains at a scdentary Kalahar i community . Journrt l ofAnthropoktgicul Art:hueolog,v 10, l-26.

Leechman, D. 195 l . Bone grease. Americun Ant iqui t1,16, 355-56.Leechman, D. 1954. The Vanta Kutchin. Nat ional Museutn of

Cunada, Bul ler in 130. l -17.Levin, M. G. and Potapov, L. P. 1964. The peoples of Siber ia.

Chicago: Chicago Univers i ty Press.Li t t le Bear, R. 1982. Grandmother in her house, in Hungry Wolf , B

(ed.), The Wa,,-,s o.f 'M v- Grandmother.r. New York: euil l.MacArthur, R. H. and Pianka, E. R. 1966. On opr imal use of a

patchy environment. Americ 'an Natural is t 100, 603-09.McGovern, T. H. 1985. Contributions to the palaeoeconomy of Norse

Grcenf and. Actu Archueologict 54, j3- lZZ.

McGovern, T. H. , Amorosi , T. , Perdikar is, S. and Wool let t , J . 1996.Vertebrate zooarchacology of Sandnes V-51 : economic change ata chief ia in 's larm in West Greenland. Arct ic Anthropolog) 33,94 -121 .

Mead, J. F. , Al f in-Slater , R. B. , Howton, D. R. and popjak, G. 19g6.Lipid,s: Chemistry, Bior:hemistry und Nutrit ion. New york:

Plenum Prcss.Metcal f -e, D. and Jones, K. T. 1988. A reconsiderat ion of animal

body-part ut i l i ty indices. Ameri t :ur t Ant iqui t l t 53, 496-504.M i t che l l , 8 . , McCowan, D . and N icho lson , I . A . 1976 . Annua lcyc les

of body weight and condi t ion in Scot t ish red deer. Ceryuseluphus. Journal o.f Zoology 180, 107-2i.

MOnt, U. 1912. Animal bones f rom l t ivnera, West Greenland: areindeer hunting site of the Sarqaq Culture. Meddelelser omGrLnlund l9 l .

Morlan, R. E. 1984. Toward the definition of criteria forthe recognirionof artif icial bone alteratit.,ns. Quuternury Research 22, 160-7l.

O'Connel l , J . F. and Hawkes, K. 1988. Had.zahunt ing, butcher ing,and bone transport and their archaeological implications. Jour-nal o.f Anthropological Reseurch 44, 113-61.

O'Connel l , J . F. , Hawkes, K. and Blur ton Jones, N. 1990. Reanalysisof large mammal body part transport among the Hadza. JournaloJ Archueological Sc' ience 17, 310-16.

O'Connel l , J . F. and Marshal l , B. 1989. Analysis of kangaroo bodypart t ransport amongst the Alyawara of Central Austra l ia.Journal o.f Archaeologir:ul Science 16, 393-405.

Outram, A. K. 1998. The Ident i f icat ion und paLcteoeconont ic

Context o.f'Prehistorit' Bone Murrow und Crease Exploitation.Unpub l i shed Ph .D . The s i s , Un ive rs i t y o f Durham.

Outram, A. K. 1999. A compar ison of paleo-Eskimo and medievalNorse bone fa t exp lo i ta t i on in Wes te rn Green land . A r r : t i tAnthropolog,r," 36, 103- 17.

Outram, A. K.2001a. A new approach to ident i ty ing bone nlarrowand grease exploi tat ion: why the " indeterminate" f ragntentsshould not be ignored. Journul o. f Art 'hueologicul Science 28,4 0 1 - 1 0 .

Outram, A. K.2001b. FOCUS: The scapula reprcsentat ion could bethe key: a fur ther contr ibut ton to the 'Klasies pat tern ' debate.Journal of Arc'haeologir:ul Science 28, 1259-63.

Outram, A. K. 2002. Bone l racture and wi th in-bonc nutr ients: Anexper intental ly based method fbr invest igat ing levels o l 'marrowex t rac t i on , pp . 51*64 in M i rac le , P . and M i lne r , N . (eds ) ,C ons um in g P o s s io n s antl P ott e rn s o.f ' C o n s u nt pt i o rt. Cantbridge:McDonald Inst i tute for Archaeological Research.

Outram, A. K. In Press. Compar in-e levels of subsistence stressamongst Norse set t lers in lceland and Greenland using lcvels ofbone fat explo i tat ion as an indicator . Environmentul Art .hue-olog-v-.

Outram, A. K. Forthcoming. The ident i f icat ion of levels o1'bonemarrow and grease exploitation and the place of bone marrow asa resource w i th in seasona l rounds : case s tud ies f romGreenland and Sweden, in Mutv i l le , J and Outram, A. K. (cr is) ,The Zooarchaeologt,of Animal Fats und Duirt,ing (proceedingsof |CAZ2002). Oxfbrd: Oxbow.

Outram, A. and Rowley-Conwy, P. 1998. Meat and marrow ut i l i tyindices for horse (Equu,s). Journul rlf Art'hcteologir:ul Scient:e25 ,839-49 .

Pe te rson , R . O . , A l l cn , D . L . and D ie tz , J . M . 1982 . Dep le t i on o fbone marrow fat in moose and a correlation fbr dehydration.Jounrul o.f WildliJe Munttgement 46, 541-51.

Rixson, D. 2000. The Histor t , o. f ' Meut Truding. Nott ingham:Nott i ngham Univers i ty Press.

Rowley-Conwy, P. and Stor i , J . l99i . Pi t ted Ware seals and pigsfrom Ajv ide, Got land: methods of study and f i rs t resul ts, pp.I l3-30 in Burenhul t , G. (ed.) , Rentote Sen,s ing, Volune 1 (Thesesand Papers in North-European Archaeology l3a). Stockholm:S t o c k h o l m U n i v e r s i t y .

Speth, J. D. 1983. Bison Ki l ls ancl Bone Counts: Deci , t ic tn Mukingbv- Ancient Hunters. Chicago: Univers i ty of Chicago press.

Speth, J. D. 1987. Ear ly hominid subsistence strategies in seasonalhabitats. Journul of Archrteological St' ience 14, 13-29.

Speth, J. D. 1991. Nutr i t ional constra ints ancl late g lacral aclaptrvctransformat ions: the importance of non-protein cnergy sources,pp. 169-78 in Barton, N. , Roberts, A. J. and Roe, D. A. (cc ls) ,The Late Glacial in North-West Europe: Huntun Adapttttiorrand Environmentul Chunge at the End ofthe pleistor-ene. york:

Counci l fbr Br i t ish Archaeology.Speth, J. D. and Spielmann, K. A. 1983. Energy source, protein

metabol ism, and hunter-gathercr subsistence strategies. Journalof Anthropological Research 2, 1-31.

Tinsley, C. M. Forthcoming. The Vik ing sct t lemcnt of NorthernIceland: A prel iminary zooarchaeological analysis, in Houslcy, R.A. and Coles, G. (eds), Atluntic Connectirrts untl Atluptutittrts:Economies, Environments untl Subsisten.e in Lantls Borderii luthe North Atluntic. Oxfbrd: Oxbow Monographs.

Vi l la, P. and Mahieu, E. 1991. Breakage pat terns of human longbones. JournuL o. f 'Humun Evolut ion 21,21-18.

Wi lson, G. L. 1924. The Horse unt l Dog in Hidutsu Cul tnre(Anthropological Papers ol ' the American Muscum of NaturalHistory 15,2) . New York: American Museum of NaturalHistory.

Ye l l en , J . E . 1991 . S rna l l mammals : !Kung San u t i l i za t i on and theproduct ion of faunal assemblages. Journal of AnthroplogiculArchaeoLog.y' 10, I -26.

Alan K. OutramDepartment of ArchaeologyUniversity of ExeterNorth Park RoadExeter, EX4 4QEE-mail: a.k.outram @ exeter.ac.uk