cribra orbitalia in two temporally disjunct population samples from the dakhleh oasis, egypt

Cribra Orbitalia in Two Temporally Disjunct Population SamplesFrom the Dakhleh Oasis, Egypt

SCOTT I. FAIRGRIEVE1* AND J.E. MOLTO2

1Anthropology Program and Department of Biology, Laurentian University,Sudbury, Ontario, Canada P3E 2C62Department of Anthropology, Lakehead University, Thunder Bay, Ontario,Canada P7B 5E1

KEY WORDS cribra orbitalia; porotic hyperostosis; anemia;palaeonutrition; parasitism; third-intermediate; Romano-Christian;Egypt; Dakhleh Oasis

ABSTRACT Cribra orbitalia (CO), an osseous sign of anemic stress,occurs in 67% (n 5 296) of the pre-Roman (n 5 153) and Roman (n 5 143)period crania from the Dakhleh Oasis, Egypt. CO is primarily a childhoodcondition in these samples, and its prevalence is significantly higher invirtually all cohorts in the pre-Roman group, including among females, whodisplay higher rates of active lesions. This temporal trend suggests that theunderlying causative factors (i.e., synergism between disease and nutrition)were less pervasive in the Roman period. In both population samples, anemicstress develops in some perinates prior to the expected minimum age for thedevelopment of iron deficiency anemia. This suggests additional causes ofanemic stress in the Dakhleh population. A strong candidate is folic aciddeficiency and its concomitant, megaloblastic anemia, which results fromweaning of infants on goat’s milk, a known practice in ancient Egypt. Theputative incorporation of other food items in the weanling diet, particularlyhoney, a confirmed source of C. botulinum, represents yet another retrospec-tive data source to help understand the epidemiological profile of cribraorbitalia in this population. Comparative data from other Egyptian popula-tions, though limited, show similar patterns, however, they display a lowerprevalence than the data from Dakhleh. Am J Phys Anthropol 111:319–331,2000. r 2000 Wiley-Liss, Inc.

Cribra orbitalia (CO) is a bone lesion ofthe orbital roof purported to be an indicatorof an anemic stress (e.g., Mensforth et al.,1978; Ortner and Putschar, 1985; Mittlerand Van Gerven, 1994). This condition hastraditionally been linked to an ectocranialvault lesion known as porotic hyperostosis.The etiological association of these two bonepathologies has been confirmed by Stuart-Macadam (1989). Currently, references inthe literature to the orbital lesions are foundin discussions of porotic hyperostosis. Thesupport for this classification has a physi-ological basis; an anemic stress suffered inearly childhood results in the expansion of

hemopoietic bone marrow to increase eryth-rocyte (red blood cell) production (Moseley,1974; Stuart-Macadam, 1985, 1987).

Recording the prevalence of cribra orbita-lia in osteological analyses is commonplace(e.g., Cohen and Armelagos, 1984). Studies

Grant sponsor: Natural Sciences and Engineering Council ofCanada; University of Toronto Open Doctoral Fellowship; Laur-entian University Research Fund; Social Sciences and Humani-ties Research Council; Grant number: 410-94-0857; Grant spon-sor: Lakehead University Research Grants.

*Correspondence to: Dr. Scott I. Fairgrieve, Forensic OsteologyLaboratory, Subdepartment of Anthropology, Laurentian Univer-sity, Sudbury, Ontario, Canada P3E 2C6.E-mail: [email protected]

Received 2 September 1997; accepted 10 October 1999.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 111:319–331 (2000)

r 2000 WILEY-LISS, INC.

tend to go beyond mere description by provid-ing an epidemiological profile (age and sexdistribution) of the condition. Generally,these lesions show greater severity andprevalence in subadult remains (e.g., El-Najjar et al., 1976; El-Najjar, 1976). Genderdifferences are not a consistent finding, al-though interpopulation variation is consider-able. Various explanations have been in-voked to the latter. For example, Angel(1964, 1971), in his classic studies of porotichyperostosis in populations from Cyprusand Greece, suggested that these lesionsreflect thalassemia genotypes. Using theHardy-Weinberg model he attributed thevariability to the allelic distributions of ahypothetical mating population. Underlyingthis model is the concept that heterozy-gotes for the thalassemia allele will survivelonger and will be less severely affected thanhomozygotes who died before reproductivematurity. However, Angel (1971) cites theprevalence of porotic hyperostosis as approxi-mately 20% regardless of severity, which issignificantly less than the prevalence pre-dicted from the Hardy-Weinberg equilib-rium. El-Najjar et al. (1976), however, posea dietary hypothesis, relating the high preva-lence of CO in their American southwestsamples to a dependence on maize. Walker(1986) found that porotic hyperostosis in hissouthern California Island populations var-ied according to the availability of freshwater supplies, with more stagnant condi-tions being highly correlated with increasedprevalence of cribra orbitalia. He suggeststhat the condition is most likely of an infec-tious origin. Clearly, a multifactorial etiol-ogy would seem the most appropriate modelfor palaeoepidemiological studies of porotichyperostosis, although in each environmen-tal/population niche certain specific synergis-tic factors will predominate. Addressing po-tential etiological factors in specific niches isan elusive goal of current research.

The often quoted prevalence of these le-sions in the literature, and its importance topalaeoepidemiological research, warrantsthe reporting of any situation where theproportion of affected individuals deviatesgreatly from other published values. Thissituation occurs in samples from the Da-khleh Oasis of Egypt’s Western Desert. In

the population samples from two cemeteriesin this Oasis, a combined prevalence of over60% of the children show CO to varyingdegrees. Also, there is a temporal decreasein the prevalence of CO, which suggests thatthe etiological factors were changing. Thispaper provides a detailed account of CO andpossible etiological factors in the DakhlehOasis populations.

MATERIALS AND METHODSSkeletal sample

The Dakhleh Oasis (Fig. 1) is one of fivemajor depressions in Egypt’s Western De-sert, measuring 100 km long from east towest, and 25 km at its widest point fromnorth to south (Cook et al., 1988). It islocated approximately 800 km south-south-west of Cairo, with a land area of 2,000–3,000 km2 (Mills, 1979). The flat clay plainsof the Oasis are made up of soil with a highiron oxide content (Churcher, 1983). Pres-ently, the climate is extremely arid with anannual rainfall of 0.7 mm (Mills, 1979). Thehumidity rarely exceeds 50%, and the tem-perature ranges from a daily maximum aver-age in January of 21.5°C to a July average of39°C (Mills, 1979). The current populationof the Oasis (approximately 35,000) is agri-culturally based, and grows vegetables, fruits(dates, apricots, oranges, and olives) andcereals (rice, sorghum, and wheat). The veg-etables are grown for local consumption andare supplemented from the Nile Valley inthe hot weather (Mills, 1979). Agriculture,and indeed human existence, is made pos-sible by the exceptional nature of the Nu-bian Sandstone and shale series, which con-tains one of the largest ground waterreserves in the world. Today, as in the past,some water seeps to the surface, althoughthe main access for human use is throughartesian wells. The history of settlementpatterns strongly suggest that the condi-tions in the Oasis were as arid in the past asthey are today (Schwarcz et al., 1999). Thereare no contemporary records of past cli-mates; however, it is generally known thatthe eastern Sahara was about as arid todayas it was throughout late Pharaonic andRoman times (Butzer, 1976). Establishingthe amount of surface water and rainfall inthe past is a valuable research pursuit as it

320 S.I. FAIRGRIEVE AND J.E. MOLTO

has the potential to alter understanding ofthe epidemiological patterns of a population.It is known, however, that past agriculturalsubsistence pursuits were dominantthroughout the Pharaonic Periods and coa-lesced during the Roman Period (Mills,1984).

Since 1977, the Oasis has been the focus ofthe Dakhleh Oasis Project (DOP). The bioar-chaeological component of the DOP has fo-cused on the analysis of human remainsfrom two cemeteries; ’Ein Tirghi (ET) andKellis 2 (K2). ’Ein Tirghi (31/435-D5-2) hasyielded remains of over 700 individuals from58 tombs that date to the Late Period. Thecemetery was used over a period of severalhundred years from ca 900 BC to possiblyearly Ptolemaic times. Skeletons from onlyone tomb, E31, have been radiocarbon dated.These skeletons, 25 and 36, yielded respec-tive calibrated dates of 795 6 70 BC and800 6 60 BC (Isotrace TO-4476), which areboth in the Late Third Intermediate Period(Molto, 2000). Hope posits that the majorityof the ET samples would be in the Late

Period (C.A. Hope, personal communication,January, 1997), a view that requires furthertesting. This study involves crania fromeight ET tombs, E31, E33, E34, E36, E37,E40, E45, and E52. The ancient town ofKellis, has two major cemeteries associatedwith it. The western cemetery, Kellis 1 (31/420-C5-1), predates the Christian Period (c.60 BC–100 AD), while Kellis 2 (31/420C5-2),the focus of this study, is clearly within theChristian Period (c 100 AD–400 AD) asillustrated by the classic Christian mortu-ary position (i.e., single inhumations withthe body oriented east-west with the head inthe latter position). Five calibrated radiocar-bon dates on human skeletons from K2(burials 5, 6, 94, 95, and 116) support theChristian Period affiliation of this sample.The sample of 278 burials excavated so far,date between 300 and 450 AD or during thefinal phase before the town of Kellis wasabandoned. Therefore, the chronologicalseparation of the cemeteries in this studypotentially span some 1300 years. Geo-

Fig. 1. A map of Egypt showing the position of the Dakhleh Oasis in the Western Desert.

321CRIBRA ORBITALIA IN THE DAKHLEH OASIS

graphically these cemeteries are only ap-proximately 15 km from one another.

The preservational quality of the bone inall samples is excellent. Extraction of Type Icollagen from bone tissue and tooth dentinreflects the preservational quality through ahigh percentage yield (ca. 15–20%) by weight.This conclusion is also supported by anamino acid residue analysis (Fairgrieve,1993). Additionally, the remains were com-plete in most cases, affording the opportu-nity to analyze the skeletons completely forvital statistics and pathology. These factsare important in order to present an accu-rate differential diagnosis of any conditionthat may be present.

The combined demographic profile of the’Ein Tirghi and Kellis samples is presentedin Figure 2. A composite sample of 385skeletons were included in this study. Ofthat total, eight could not be ascribed eitherto an age or sex category due to damagesuffered largely during episodes of tomb-robbing. A total of 377 were preserved wellenough to be aged and sexed using conven-tional osteological methodologies (Bass,1987). Specifically, age at death assessmentswere based on dental development (Ubelaker,1978), and pubic symphysis metamorphosis(Katz and Suchey, 1986; and Brooks andSuchey, 1990). Sex determination of adultremains was based on cranial morphology

(Bass, 1987) and pubic symphysis morphol-ogy (Phenice, 1969). One member of theDakhleh Oasis Project (M. Marlow) has de-veloped a refined age sequence for the in-fants based on dental calcification. Thesedata were used in conjunction with longbone length to age neonates. All of the agingmethods utilized have their own standarddeviations. An appreciation of this fact isparticularly important due to the signifi-cance of infants to this study. AlthoughUbelaker’s dental development standardsfor infants have a standard deviation of 3months, it should be recognized that with amean age of 6 months remains could be 3, 6,or 9 months of age. The importance of suchan early onset of an anemic stress is notnecessarily diminished by these factors.When combined with other aging methodsour analysis of lesion distribution is stillpresented using age ranges.

Assessment of cribra orbitalia

Although a general assessment of all pa-thology present on the remains was con-ducted (Molto, 1986), a more detailed studyof the prevalence and severity of CO wasconducted by SIF. It is well established thatthere are as many methods for visuallyassessing these lesions as there are investi-gators (e.g., Nathan and Haas, 1966; Stein-bock, 1976; Mensforth et al., 1978; Guidotti,

Fig. 2. Demographic profile ofthe ’Ein Tirghi and Kellis 2 cemeter-ies.


1984; Stuart-Macadam, 1982, 1985). How-ever, for comparative purposes and expedi-ency, the method outlined by Steinbock(1976), adopted from Nathan and Haas(1966) is used herein. The categories aredefined as follows (Steinbock, 1976:239): theporotic type is ‘‘characterized by scatteredfine openings affecting the roof of the orbit’’;the cribrotic type shows ‘‘openings largerand more numerous, tending to coalesce intolarger apertures’’; and finally, the trabeculartype, exhibits small openings coalescing into‘‘large, irregular apertures often arranged inradiating patterns from one or more centersin the orbital roof.’’ In addition, the lesionswere recorded for their status at time ofdeath [i.e., healed (Fig. 3) and reactive (Fig.4)] following Mensforth et al. (1978). Thisrecording system is consistent with themethod reported by Mittler and Van Gerven(1994) in their study of the Medieval Kulubn-arti remains. Both orbits were assessed ineach case.

In order to interpret the epidemiologicalprofile of this condition, statistical compari-

sons (x2) on the basis of age, sex and theassessed state of lesions were conducted.

RESULTS

The combined data of cribra orbitalia rela-tive to age, sex, and lesion status are summa-rized for both cemeteries in Table 1. Figure 5depicts the percentage of active and healedlesions for each cohort of affected individu-als including females and males over 18years of age. The trend clearly indicates thatactive lesions are more prevalent in youngerindividuals, declining gradually in the oldercohorts. This pattern characterizes both ETand K2 (Figs. 6 and 7, respectively). Thesedata support the argument that cribra or-bitalia is indicative of a childhood anemia(Stuart-Macadam, 1985). However, the over-all prevalence of 66.9% (198/296) obscures asignificant diachronic decrease.

Tables 2 and 3 summarize the prevalenceof cribra orbitalia according to age, sex, andlesion status for the ‘Ein Tirghi and Kellis 2cemeteries, respectively.

Fig. 3. Typical ‘‘healed’’ or inactive CO found in an adult from the Dakhleh Oasis, Egypt.


The prevalence at ‘Ein Tirghi is 78.4%(120/152) compared to 54.6% (78/152) atKellis 2; a statistically significant difference(x2 5 19.88, degree of freedom (df) 5 1,

P 5 0.001). It follows that whatever caus-ative factors were in operation they weremore pervasive in the earlier population.This assertion, however, requires a detailedspecific investigation of these data, particu-larly in the younger age cohorts.

Individuals were included in the perinatalcategory if the remains were developmen-tally consistent with birth to 6 months inage. It is germane to note a discrepancy insample size; 3 for ET and 33 for K2. All theneonates were affected in the former sitecompared with 21.2% (7/33) at K2. Thoughthis result is statistically significant(x2 5 5.035, df 5 1, P 5 0.0248 with Yates’correction for continuity) the sample sizedisparity suggests a degree of caution.

The second cohort (6 months–3 years)duplicates the sample size differential(n 5 31 for K2 versus n 5 17 for ET) and thehigher and statistically significant preva-lence pattern (100% for ET and 70.6% forK2; x2 5 4.318, df 5 1, P 5 0.0377). Clearly,the probability of experiencing anemia in

Fig. 4. Typical ‘‘active’’ CO found in a subadult from the Dakhleh Oasis, Egypt.

TABLE 1. Prevalence of cribra orbitalia and activelesions by age and sex from the ‘Ein Tirghi and Kellis 2

cemeteries combined

Age cohortTotal

NAssess-

able

Cribraorbitalia

Activelesions

n1 % n2 % of n1

0–6 months 57 36 10 27.77 9 90.006 months–3

years 70 48 39 81.25 27 69.234 years–12

years 41 33 31 93.94 18 58.0613 years–18

years 7 5 5 100.00 1 20.00Females .18

years 112 96 58 60.42 9 15.52Males .18

years 90 74 52 70.27 5 9.62Sex ? .18

years3 8 4 3 75.00 0 0.00Total 385 296 198 66.89 69 34.851 Refers to the number of individuals with CO.2 Refers to the number of individuals with ‘‘active’’ lesions of CO.3 Skeletons of indeterminant sex over the age of 18 years.


these early cohorts exceeds the probabilityof having normal hemoglobin levels.

In the final childhood cohort (4–12 years)no significant difference between ET and K2occurs (100.0% and 90.0%, respectively). Asis normal of paleodemographic profiles, thereare fewer individuals in the next cohort(13–18 year olds), although the number islower than expected (see Henneberg andSteyn, 1993). In total, only seven individualsrecovered, while only five (four from ET andone from K2) could be analyzed for CO.While all five demonstrate CO, it is onlyactive in one (from ET). The prevalence ofCO in adult females (.18 years) is the samefor ET and K2 (60.7% and 60.0%, respec-tively). In adult males (.18 years), thosefrom ET have a significantly higher preva-lence than those from K2 (84.3% (43/51)

versus 39.1% (9/23); x2 5 13.403, df 5 1,p 5 0.0003 with continuity correction). Ofnote is the fact that the prevalence of activelesions is higher in females in both popula-tion samples, although the males had ahigher overall prevalence in ET and lowerin K2.

Epidemiological trends for cribra orbitaliawere also tested within each cemetery sam-ple. In the case of ET, the prevalence of COand age of the individual demonstrated asignificant relationship ( p 5 0.0031). Re-lated to this trend is a significant relation-ship between the age and level of lesionseverity ( p 5 0.0001). However, there wasno significant relationship with CO severityand prevalence with the sex of the individual.For K2, prevalence and age ( p 5 0.002), andage and severity ( p 5 0.0001) were statisti-

Fig. 5. Prevalence of active andhealed lesions by age and sex forthe ’Ein Tirghi and Kellis 2 Cem-eteries combined.

Fig. 6. Prevalence of active andhealed lesions by age and sex forthe ’Ein Tirghi Cemetery.


cally significant. Similarly, the K2 sampledemonstrated no relationship among the sexof the individual, severity and occurrence ofCO.

Thus, similarities and differences in COcharacterize these two temporally distinctpopulation samples. Individuals from ETshow rates of 100% for the occurrence of COin all childhood age cohorts. Although theprevalence of CO in K2 is also high, it issignificantly lower than in ET. Upon reach-ing adulthood, females were similarly af-fected at both cemeteries whereas maleswere more highly affected at ET. Females inboth samples show a high prevalence ofactive lesions, especially in ET.

DISCUSSION

The human skeletal remains from theDakhleh Oasis, Egypt are a unique series ofsuperbly preserved skeletons showing excep-tionally high rates of CO. Given the abovetrends, the interpretation of the rather obvi-ous stress on these people is problematic formany reasons. In part this may be due to thegeographic position of the Dakhleh Oasis. Asthe individuals from both cemeteries livedin northeastern Africa, this made them po-tentially susceptible to various environmen-tal stressors that could favor the develop-ment of some sort of anemia. The possiblefactors include parasitic infestations (e.g.,schistosomiasis), infectious diseases (e.g.,one of the malarias), potentially low bioavail-

Fig. 7. Prevalence of active andhealed lesions by age and sex forthe Kellis 2 Cemetery.

TABLE 2. Prevalence of cribra orbitalia and activelesions by age and sex from the ‘Ein Tirghi cemetery

Age cohortTotal

NAssess-

able

Cribraorbitalia

Activelesions

n1 % n2 % of n1

0–6 months 3 3 3 100.00 3 100.006 months–3

years 17 17 17 100.00 12 70.594 years–12

years 13 13 13 100.00 11 84.6213 years–18

years 4 4 4 100.00 1 25.00Females .18

years 62 61 37 60.66 6 16.22Males .18

years 54 51 43 84.31 4 9.30Sex ? .18

years3 4 4 3 75.00 0 0.00Total 157 153 120 78.43 37 30.831 Refers to the number of individuals with CO.2 Refers to the number of individuals with ‘‘active’’ lesions of CO.3 Skeletons of indeterminant sex over the age of 18 years.

TABLE 3. Prevalence of cribra orbitalia and activelesions by age and sex from the Kellis cemetery

Age cohortTotal

NAssess-

able

CO Active lesions

n1 % n2 % of n1

0–6 months 54 33 7 21.21 6 85.716 months–3

years 53 31 22 70.97 15 68.184 years–12

years 28 20 18 90.00 7 38.8913 years–18

years 3 1 1 100.00 0 0.00Females .18

years 50 35 21 60.00 3 14.29Males .18

years 36 23 9 39.13 1 11.11Sex ? .18

years 4 0 — — — —Total 228 143 78 54.55 32 41.031 Refers to the number of individuals with CO.2 Refers to the number of individuals with ‘‘active’’ lesions of CO.3 Skeletons of indeterminant sex over the age of 18 years.


ability of iron in the diet, and even geneticanemias. Connected to the possibility of aparasite induced anemia is the model posedby Stuart-Macadam (1992) that such ananemic response by the host is actually anadaptive response. In short, a mild irondeficiency (hypoferremia) would be benefi-cial as this condition deprives the pathogenof the much needed iron that could bewrested from the host. Thus, according tothis model, the presence of CO should actu-ally be considered as an indicator of thisadaptation.

Recently, this parasite model has beenchallenged by Holland and O’Brien (1997).Although they list many concerns in theirreview of the topic, they raise the point thathypoferremia may provide some benefit;however, the immune system would still becompromised by the chronic anemia. In otherwords, this ‘‘adaptation’’ to the presence ofpathogens does so at the expense of a vastarray of physiological reactions involved inmaintaining homeostasis, such as collagenbiosynthesis. It is important to consider thatiron deficiency still exacts a physiologicalcost (for a review see Holland and O’Brien,1997:185).

In this study, it is germane to discuss theprevalence of porotic hyperostosis (includingCO) from areas adjacent to the study region.Unfortunately, there are no systematic sur-veys of the condition from other sites withinthe Oasis. Other areas of Egypt and Sudanhave yielded studies of skeletal remainswith variable rates of porotic hyperostosis(Batrawi, 1935; Satinoff, 1972a,b; Carlson etal., 1974; Strouhal and Jungwirth, 1980;Hummert and Van Gerven, 1983; Mittlerand Van Gerven, 1994). None of these re-ports would serve as an appropriate compari-son due to the geographic context of theOasis. However, they do bring the Dakhlehremains into context for the region.

Mittler and Van Gerven (1994) updated aprevious study of the early and late Chris-tian cemeteries at Kulubnarti (Hummertand Van Gerven, 1983) by more closelyexamining the age specific trends of cribraorbitalia in remains spanning several timeperiods. The childhood specificity of CO issupported in their study, as in all studies.Additionally, they found that life expectancy

is reduced in individuals with the condition.Of significance in their study is that 100% ofthe individuals in the first year of life exhibitactive lesions. This rate drops to 59% inchildren between 1 and 3 years, with afurther decline in older children until theage of 12 years; beyond which all affectedindividuals exhibit healed lesions. They posethe question of whether this is evidence ofan amelioration of the condition or whetherthe skeletal response has changed (Mittlerand Van Gerven, 1994). An additional factormay be that the more intensive period ofblood element formation has passed. Thus,the same conditions persist that caused theoriginal anemia, however, the individual isnot anemic to the point of affecting themarrow to the same extent as in childhood.They conclude that the pattern is consistentwith the clinical pattern of iron deficiencyanemia. In this case, the low bioavailabilityof iron in cereal grains (the largest portion ofthe diet), particularly those containingphytates, which act as iron chelators, andintensified parasitic infections causing gas-trointestinal bleeding are all purported toresult in the iron deficiency for this popula-tion.

The high prevalence of CO in young chil-dren characterizes both the Dakhleh andthe Kulubnarti samples. However, the differ-ence with the Dakhleh sample lies in theprevalence and severity for adults. It ispossible then that there are some differencesin the etiological factors underlying the con-dition in these populations. What evidencecan be used to interpret the causes of CO inthe Dakhleh population?

In an attempt to address the parasiticmodel, soil samples from the pelvic cavitiesof several individuals were examined forany signs of helminthic infestation. Al-though evidence of indigenous parasites werenot found, this alone does not negate thepossibility. Nonetheless, the possibility ex-ists that the isolation of Dakhleh from theNile Valley may be acting as a filter keepingparasitic infections to a minimum. Infantdiarrheal diseases are still the primaryhealth risk today in underdeveloped coun-tries (Mensforth et al., 1978). Hrdlicka (1912)notes that gastroenteritis was the majorcause of death in the Kharga Oasis (adjacent


to the Dakhleh Oasis) in 1908. The etiologyof infant gastrointestinal disorders is com-plex and multifactorial, involving the syner-gistic effects of malnutrition and infection.The infectious agents are often endogenousalthough, in some environments (e.g., en-demic malaria) exogenous sources could playa significant role. Malaria must be consid-ered a real factor in this region as it wasdocumented in the Kharga Oasis by bothHoskins (1837, as cited by Hrdlicka, 1912)and Hrdlicka (1912). However, Hrdlicka(1912:8) noted that ‘‘Malaria is not veryfrequent, except in the date season (Septem-ber–October) when there are extraordinarynumbers of flies and mosquitoes.’’ Hoskins(1837 as cited by Hrdlicka, 1912:8) furtherprovided clues as to the general health of theKhargan inhabitants during his visit bydescribing them as ‘‘chiefly remarkable forthe pallid and unhealthy hue of their counte-nances . . .’’ and having ‘‘. . . a languid andsickly appearance; a listlessness in theirmanner; a sluggishness in their movements;a total want of energy and vivacity.’’ Hefurther remarked on the ‘‘pallid hue’’ asbeing ‘‘most remarkable in their childrenand women.’’ These descriptions indicatethat the population of the Kharga Oasis wascertainly of compromised health, particu-larly the children and women. But there isno way of knowing if their pallid constitu-tions reflect an underlying anemia.

As noted above, skeletal evidence of ge-netic anemias is absent on the Dakhlehskeletons (see Ortner and Putschar, 1985;Resnick and Niwayama, 1988; Hershkovitzet al., 1997; Aufderheide and Rodrıguez-Martın, 1998). As a potential cause of the COin the Dakhleh samples it is important toconsider not only iron deficiency, but alsofolic acid and vitamin C deficiencies. Whenone of these nutrients is lacking, it directlyaffects the physiological status of the othertwo. For example, folic acid deficiency oftenaccompanies a vitamin C deficiency. VitaminC acts as an electron donor in the metabo-lism of folic acid (Stokes et al., 1975). A folicacid deficiency may lead to vitamin C andiron deficiencies. It is of interest to note thatif goat’s milk was being fed to infants, theywould be getting a substantial proportion ofascorbic acid. It is the folic acid that would

be lacking in such a scenario (Forman, 1974as cited by Janssens, 1983). Thus, evidenceof compromised hemostasis in the form of asubperiosteal reaction in the orbits may bepresent without any of the other lesionsindicative of scurvy. Two of the three nutri-ents, vitamin C and iron (Fe21), are requiredin the hydroxylation of proline. The interre-latedness of these three nutrients dictatesexamination of the pathological changes todiscern the likelihood of any or all three ofthese nutrients being deficient in the diet. Inno instance have any of the infracranialremains from Dakhleh shown any evidenceof the hemorrhaging, typically associatedwith vitamin C deficiency, regardless of ageat death (Molto, 1986). Although Greenfield(1986) states that the pathological changescan be resorbed upon the amelioration of thedeficiency, possibly yielding infracranial ele-ments with no signs of ever having thecondition, it is unlikely (given the intensityof the orbital lesions and their pervasivenature) that no individuals would show infra-cranial involvement.

In order to consider folic acid deficiency asan etiological agent for megaloblastic ane-mia, and thus the development of CO ininfants, one must consider infant feedingpractices. To that end one of the DakhlehOasis researchers (T. Dupras, McMaster Uni-versity) is currently completing carbon andnitrogen stable isotope analysis of infantand juvenile remains from K2 in order todetermine the timing of the weaning processand food(s) infants were weaned onto. Heranalysis, with the addition of the dietaryinformation, should further refine our under-standing of the Dakhleh data relative to thefolic acid hypothesis.

The cultural context (Roman-Byzantine)of these remains must be is taken intoconsideration. Soranus (AD 98–117) favoredmaternal breastfeeding, however, he alsobelieved that this should not begin before 3weeks after birth of the infant in order toallow the mother time to regain her strength(Jackson, 1989). The recommended substi-tute for breastfeeding by the mother was awet-nurse or even honey diluted with wateror goat’s milk. Both Galen and Soranusrecommended a gradual weaning process;bread crumbled and softened with either


milk, hydromel (a mixture of honey andwater), sweet wine or honey wine (all avail-able during this time in the Dakhleh Oasis).

The introduction of both goat’s milk andhoney can have dire consequences for in-fants. Goat’s milk is relatively low in bothcobalamin (Cbl) and folate (0.1 µg/L and 6µg/L, respectively) compared to human milk(4 µg/L and 52 µg/L, respectively) (Chana-rin, 1990). Human infants that are startedon goat’s milk develop a severe megaloblas-tic anemia at about 3 to 5 months due to folicacid deficiency (Becroft and Holland, 1966).Normally the transfer of folate across theplacenta is most active in the last weeks ofpregnancy and fetal stores build-up is mostevident after the 37th week (Chanarin,1990). After birth there is normally a fall inred cell folate with the lowest levels beingreached at 11–12 weeks. After 12 weeks, thefetal folate stores are exhausted and thenewborn is dependent on dietary sourcesalone. A reduction in serum folate can resultin iron malabsorption, hemopoietic marrowexpansion and reduced levels of plateletsand fibrinogen in the blood. The folate defi-ciency would have an effect almost immedi-ately because of the low storage of folate inthe infant. Chanarin (1990) notes that theWHO recommends that folate intake for thefirst 3 months of life should be 16 µg daily,between 3 and 6 months 24 µg daily andbetween 6 and 12 months 32 µg per day. Theoccurrence of active CO in these samplessupports a case for the supplementation ordependence on goat’s milk in infants lessthan 6 months of age. Such a situationwould have likely resulted in a megaloblas-tic anemia.

As previously mentioned, honey has beenimplicated by the cultural context and docu-mentary evidence from Soranus and Galen.Honey is a commodity that is listed withinthe Kellis Agricultural Accounts Book (Bag-nall, 1997). This document lists foods grownand traded in ancient Kellis and is roughlycontemporary with the K2 remains in thisstudy. Honey, when fed to infants, is aconfirmed source of Clostridium botulinumspores. This saprophyte colonizes the in-fant’s intestinal tract and forms a botulinaltoxin resulting in botulism (Aron et al.,1979; Merenstein et al., 1991). Botulism is a

severe and often fatal form of food poisoning.A lethal dose for mammals is less than 1µg/kg (Passmore and Eastwood, 1986). Botu-linal toxin blocks transmission at neuromus-cular junctions. Symptoms of onset withinthe first 6 months of life include apathy,weakness, constipation, floppiness, difficul-tyin swallowing, sudden apnea (occasion-ally), ocular palsies, and, if unresolved, pa-ralysis of the muscles of respiration leadingdirectly to death (Passmore and Eastwood,1986; Merenstein et al., 1991). With honey’sconfirmed presence in the Oasis at the timeof the K2 sample and the possibility thatweaning practices may have roughly fol-lowed those outlined by Galen and Soranus,it is reasonable to hypothesize that some, ifnot most, infants in the sample may havebeen compromised by botulism as well asmegaloblastic anemia. Merenstein et al.(1991) note that most infants recover fromthe botulism after an illness that may lastseveral weeks to months in modern popula-tions. In the context of the inhabitants ofancient Kellis, recovery may not have beenas well assured as in modern times.

In the case of an iron deficiency in adults,it is expected that young females will bemore severely affected than males due tophysiological and reproductive demands.There is no evidence for a difference in theprevalence or level of severity of the lesionsbetween males and females at either cem-etery. However, more females have activelesions than males in both samples, thoughit is more accentuated at ET. The pattern ofan iron deficiency anemia being accompa-nied by a higher incidence of infection, asreported by Mensforth et al. (1978), is notwell supported in this sample (see Cook etal., 1989; Molto, 1986, 2000). The apparentanemic reaction does not necessarily have tobe exclusively attributed to an iron defi-ciency as other nutritionally-related deficien-cies can also cause an expansion of marrow.

In contrast to the potential depletion rateof fetal folate stores, iron could take up to 4or 5 months to deplete due to infant storagein the liver and the reticuloendothelial sys-tem of a full-term, healthy infant. Therefore,if infants less than 4 months were irondeficient they would not necessarily be ex-


pected to demonstrate any lesions. However,individual DK-31/39 (aged birth to 2 months)of the Dakhleh sample exhibits CO lesionsin both orbits characterized by large andsmall foramina penetrating the bone cortex.It would seem, in this instance, a physiologi-cal demand for iron is not being met, produc-ing an anemic reaction. As noted by Mens-forth et al. (1978) clinical data indicate thatiron deficiency is rare before 6 months ofage, mainly because the newborn is affordedmaternal protection. Yet, in the ET sample100% (3/3) of the infants in the birth to 6month age cohort exhibit active CO; whereasthe rate for K2 is 21% (7/33). This indicatesthat, at the very least, these children werenot born with suitable stores to protect themfrom the anemic stress.

CONCLUSIONS

The ET cemetery provides evidence of apopulation that was more severely compro-mised than the later K2 group. The ethno-graphic material concerning the inhabitantsof the adjacent Kharga Oasis by Hoskins(1837) and Hrdlicka (1912) provide tantaliz-ing support for the argument that the anemiccondition of these people was contributed toby gastroenteritis possibly exacerbated bymalaria. However, the awaited stable iso-tope data may resolve the question as towhether or not the infants were beingweaned before the age of 6 months on goat’sor cow’s milk. Given the early onset oflesions, it is more likely that it was goat’smilk that was being used, and this leddirectly to the development of megaloblasticanemia in the infants. This not withstand-ing, malaria and even gastroenteritis cannotbe eliminated as possible sources of anemicstress. Further, the introduction of honey asa dietary supplement possibly resulted incases of botulism in infants.

Cultural differences in the diet cannot beruled out as a possible explanation for thedifference seen in the infant remains fromboth cemeteries. Such cultural differencescould include differences in weaning prac-tices and a change in dietary components orproportions of dietary constituents. Otherfactors such as malaria and gastroenteritismay have also changed in their prevalence

in the Oasis over time. For example, today itis reported that there has not been a case ofmalaria since the 1940’s (A.J. Mills, per-sonal communication, 1990).

The inhabitants of ancient Dakhleh pre-sent a remarkable opportunity to study theirdietary habits in light of direct documenta-tion of foods grown in the Oasis, as well asexceptional preservational quality of plant,animal and human remains for palaeopatho-logical and palaeonutritional analyses.

ACKNOWLEDGMENTS

The authors express their appreciation toMr. Tony Mills, Director of the DakhlehOasis Project, Dr. Peter Sheldrick, Ms.Megan Cook, and Mr. Alan Hollett for alltheir assistance in the field. The authorsalso thank Ms. Tracy S. Oost for producingthe map in this publication. Dr. Fairgrievewas supported by a two-year postgraduatedoctoral scholarship from the Natural Sci-ences and Engineering Research Council ofCanada (NSERC), a University of TorontoOpen Doctoral Fellowship, and further travelfunds by the Laurentian University Re-search Fund. This research was also facili-tated in part a grant held with Dr. HenrySchwarcz from the Social Sciences and Hu-manities Research Council (No. 410-94-0857). Dr. Molto’s research in the Oasis hasbeen supported by several small grants fromthe Lakehead University Senate ResearchCommittee. Helpful editorial comments weremade by Dr. Henry Schwarcz and Ms. ToshaDupras, to whom we are grateful.

LITERATURE CITED

Angel JL. 1964. Osteoporosis: Thalassemia? Am J PhysAnthropol 22:369–374.

Angel JL. 1971. The People of Lerna. Washington, D.C.:Smithsonian Institution Press.

Arnon SS, Midura TF, Damus K, Thompson B, WoodRM, Chin J. 1979. Honey and other environmentalrisk factors for infant botulism. J Pediatrics 94:331–336.

Aufderheide AC, Rodrıguez-Martın C. 1998. The Cam-bridge encyclopedia of human paleopathology. Cam-bridge: Cambridge University Press.

Bagnall RS. 1997. The Kellis agricultural account book(P. Kell. IV. Gr. 96) Dakhleh Oasis Project Monograph7. Oxford: Oxbow.

Bass WM. 1987. Human osteology: A laboratory andfield manual. 3rd ed. Columbia MO.: Missouri Archaeo-logical Society.

Batrawi AM. 1935. Report on the human remains,mission Archeologie de Nubie 1929–1934. Cairo: Ser-vice des Antiquites du L’Egypte, Government Press.


Becroft DMO, Holland JT. 1966. Goat’s milk and megalo-blastic anaemia of infancy: a report of three cases anda survey of the folic acid activity of some New Zealandmilks. New Zealand Med J 65:303–307.

Brooks S, Suchey JM. 1990. Skeletal age determinationbased on the os pubis: a comparison of the Acsadi-Nemeskeri and Suchey-Brooks methods. Hum Evol5:227–238.

Butzer KW. 1976. Early hydraulic civilization in Egypt.Chicago: University of Chicago Press.

Carlson DS, Armelagos GJ, van Gerven DP. 1974.Factors influencing the etiology of cribra orbitalia inprehistoric Nubia. J Hum Evol 3:405–410.

Chanarin I. 1990. The megaloblastic anemias, 3rd ed.Oxford: Blackwell Scientific Publications.

Churcher CS. 1983. Dakhleh Oasis Project palaeontol-ogy: Interim report on the 1982 field season. J SocStud Egypt Antiq 13:178–187.

Cohen MN, Armelagos GJ, editors. 1984. Paleopathol-ogy at the origins of food production. Toronto: Aca-demic Press.

Cook M, Molto E, Anderson C. 1988. Possible case ofhyperparathyroidism in a Roman period skeletonfrom the Dakhleh oasis, Egypt, diagnosed using bonehistomorphometry. Am J Phys Anthropol 75:23–30.

El-Najjar MY, Ryan DJ, Turner CG, Lozoff B. 1976. Theetiology of porotic hyperostosis among the prehistoricand historic Anasazi Indians of southwestern UnitedStates. Am J Phys Anthropol 20:329–337.

El-Najjar MY. 1976. Maize, malaria, and the anemias inthe pre-Columbian New World. Ybk Phys Anthropol9:75–78.

Fairgrieve SI. 1993. Amino acid residue analysis of typeI collagen in human hard tissue: an assessment ofcribra orbitalia in an ancient skeletal sample fromtomb 31, site 31/435-D5-2, Dakhleh Oasis, Egypt. PhDThesis, Department of Anthropology, University ofToronto.

Greenfield GB. 1986. Radiology of bone diseases, 4th ed.Philadelphia: JB Lippincott.

Guidotti A. 1984. Frequencies of cribra orbitalia incentral Italy (19th century) under special consider-ation of their degrees of expression. Anthropol Anz42:11–16.

Henneberg M, Steyn M. 1994. Preliminary report on thepaleodemography of K2 and Mapungubwe popula-tions (South Africa). Hum Biol 66:105–120.

Hershkovitz J, Rothschild BM, Latimer B, Dutour O,Leonetti G, Greenwald CM, Rothschild C, JellemaLM. 1997. Recognition of sickle cell anemia in skeletalremains of children. Am J Phys Anthropol 104:213–226.

Holland TD, O’Brien MJ. 1997. Parasites, porotic hyper-ostosis, and the implications of changing perspectives.Amer Antiq 62:183–193.

Hoskins GA. 1837. Visit to the Great Oasis of the LibyanDesert. 8°, London.

Hrdlicka A. 1912. The Natives of Kharga Oasis, Egypt.Smithsonian Misc Collect 59:1–153.

Hummert JR, Van Gerven DP. 1983. Skeletal growth ina medieval population from Sudanese Nubia. Am JPhys Anthropol 60:471–478.

Jackson R. 1989. Doctors and diseases in the RomanEmpire. Norman, OK: University of Oklahoma Press.

Janssens PA. 1983. Porotic hyperostosis and goat’s milkanaemia: A theory (more). OSSA 8:101–108.

Katz DM, Suchey JM. 1986. Age determination of themale os pubis. Am J Phys Anthropol 69:427–435.

Mensforth R, Lovejoy CO, Lallo J, Armelagos GJ. 1978.The role of constitutional factors, diet and infectiousdisease in the etiology of porotic hyperostosis andperiosteal reactions in prehistoric infants and chil-dren. Med Anthropol 2:1–59.

Merenstein GB, Kaplan DW, Rosenberg AA. 1991. Sil-ver, Kempe, Bruyn and Fulginiti’s handbook of pediat-rics. Norwalk, CT: Appleton and Lange.

Mills AJ. 1979. Dakhleh Oasis Project: report on the firstseason of survey, October–December, 1978. J Soc StudEgypt Antiq 9:163–185.

Mittler DM, Van Gerven DP. 1994. Developmental,diachronic, and demographic analysis of cribra orbita-lia in the Medieval Christian population of Kulub-narti. Am J Phys Anthropol 93:287–297.

Molto JE. 1986. Dakhleh Oasis Project: Human skeletalremains from the Dakhleh Oasis, Egypt. J Soc StudEgypt Antiq 16:119–127.

Molto JE. 2000. The comparative skeletal biology andpalaeoepidemiology of the people from ’Ein Tirghi andKellis, Dakhleh Oasis, Egypt. In: Mills T, Marlow M,editors. Oxford: Oxbow Press.

Moseley JE. 1974. Skeletal changes in the anemias. SemRoentgenol 9:169–184.

Nathan H, Haas N. 1966. On the presence of cribraorbitalia in apes and monkeys. Am J Phys Anthropol24:351–360.

Ortner DJ, Putschar WGJ. 1985. Identification of patho-logical conditions in human skeletal remains, Smith-sonian Contributions to Anthropology 28. Washing-ton, DC.: Smithsonian Institution.

Palkovich AM. 1987. Endemic disease patterns in paleo-pathology: Porotic hyperostosis. Am J Phys Anthropol74:527–537.

Passmore R, Eastwood MA. 1986. Davidson and Pass-more human nutrition and dietetics. Edinburgh:Churchill Livingstone.

Phenice TW. 1969. A newly developed visual method ofsexing the os pubis. Am J Phys Anthropol 30:297–302.

Resnick D, Niwayama G. 1988. Diagnosis of bone andjoint disorders. Philadelphia: W.B. Saunders.

Satinoff MI. 1972a. The origins and geographical spreadof the thalassemias and abnormal haemoglobins. JHum Evol 1:79–82.

Satinoff MI. 1972b. The medical biology of the earlyEgyptian populations from Asswan, Assyut, and Ge-belen. J Hum Evol 1:247–257.

Schwarcz HP, Dupras TL, Fairgrieve SI. 1999. 15Nenrichment in the Sahara: in search of a globalrelationship. J Archaeol Sci 26:629–636.

Steinbock RT. 1976. Paleopathological diagnosis andinterpretations. Springfield, IL: Charles C. Thomas.

Stokes PL, Melikian V, Leeming RL, Portman GH, BlairJA, Cooke WT. 1975. Folate metabolism in scurvy. AmJ Clin Nutr 28:126–129.

Strouhal E, Jungwirth J. 1980. Paleopathology of theLate Roman-Early Byzantine cemeteries at Sayala,Egyptian Nubia. J Hum Evol 9:61–70.

Stuart-Macadam P. 1982. A correlative study of a palaeo-pathology of the skull. PhD thesis, Department ofPhysical Anthropology, University of Cambridge.

Stuart-Macadam P. 1985. Porotic hyperostosis: represen-tative of a childhood condition. Am J Phys Anthropol66:391–398.

Stuart-Macadam P. 1987. Porotic hyperostosis: Newevidence to support the anemia theory. Am J PhysAnthropol 74:521–526.

Stuart-Macadam P. 1989. Porotic hyperostosis: Relation-ship between orbital and vault lesions. Am J PhysAnthropol 80:187–193.

Stuart-Macadam P. 1992. Porotic hyperostosis: A newperspective. Am J Phys Anthropol 87:39–47.

Ubelaker DH. 1978. Human skeletal remains: Excava-tion, analysis, interpretation. Washington, D.C.: Ta-raxacum.

Walker PL. 1986. Porotic hyperostosis in a marine-dependent California Indian population. Am J PhysAnthropol 69:345–354.


cribra orbitalia in two temporally disjunct population samples from the dakhleh oasis, egypt

Documents