Mirila - Slaus

8/22/2019 Mirila - Slaus

1/10

155

MEASURING BODIES OR SOULS AND THE MIRILA

CONUNDRUM A PHYSICAL ANTHROPOLOGY

ASSESSMENT BASED ON STATURE ANALYSIS

8

In June o 2009 I was approached by dr. Andrej Pleterski rom the Scientic Research Centero the Slovenian Academy o Sciences and Arts to do an analysis o the mirila data that have beencollected rom several sites on the southern Velebit mountain range in Croatia. Te reason or this isthat the rationale o the mirila is, at present, not clear. Tese measurements were recorded during analmost two and hal century time period (rom the 18th to the middle o the 20th century) and vary inlength rom 52 cm to 210 cm. Te ollowing theories are currently used to interpret their unction:

1) they correspond to the lengths o human cadavers on their way to burial,2) they symbolize the dimensions o an individuals soul, and3) they represent a combination o the two.Available or analyses were the dimensions o over 250 mirila. What could not be done was to

correlate any o these data with either specic individuals o known stature, or even with sex, or age-at-death.

As a orensic anthropologist and bioarchaeologist I requently collaborate with colleagues romdierent scientic elds in an attempt to answer various questions. So ar, all o these contributionshave depended on my analyses o human osteological material recovered rom either orensic orarchaeological contexts. Tis is the rst time I have been asked to contribute to the solution o aproblem that does not involve my examination o human remains rom specic sites.

On the assumption that mirila represent an accurate measurement o the lengths o humancadavers on their way to burial, the range o the recorded values (rom 52 cm to 210 cm) suggeststhat both subadults and adults were measured. Te rst step to solving the problem is, thereore, to

calculate a cut-o point that would separate between these two groups. Te second is to determinewhich adult measurements corresponded to males, and which corresponded to emales. I these datacan be obtained, the mean statures o males and emales in the series can be calculated and comparedto dierent contemporary historical populations, as well as to the modern Croatian population. Ithe mirila values signicantly deviate rom these values it would be a strong argument against thehypothesis that they represent cadaver lengths. Additionally, tests that analyze whether male andemale statures (as represented by the mirila values) are normally distributed in the series can bemade. Given the relatively large number o mirila available, a signicant departure rom a normaldistribution would, once again, be a strong argument against the hypothesis that mirila reected thestature o males and emales.

As a more detailed examination o this puzzle shows that it relates to the problem o statureestimation in historical populations I accepted the challenge and devised the ollowing strategy to

resolve it.By now, the estimation o adult human stature rom skeletal remains has been researched orover a century (Rollet, 1889; Dwight, 1894; Pearson, 1899). As noted by Lundy (1985), two approach-es have been employed in this endeavor: the anatomical approach, which involves the summationo superoinerior measurements o contributory skeletal elements to determine stature as directly aspossible; and the mathematical approach, which involves the extrapolation o living stature romindividual skeletal measurements by the utilization o ratios or regression ormulae. Te ormermethod, most oen attributed to Georges Fully (1956), has recently been reexamined in detail byRaxter et al. (2006, 2007). Relatively ew studies have, however, employed this approach as regressionormulae or long bones (usually the emur and tibia) are much easier to apply. Indeed, a substantial

Mario

LAUS


2/10

156

literature has been devoted to the mathematical method. Researchers have developed techniquesto estimate living statures rom a variety o skeletal elements, including the major long bones o thelimbs (e.g., rotter, 1970), crania (e.g., Ryan and Bidmos, 2007), and other whole bones, as well asragmentary remains (e.g., Simmons et al., 1990). Te vast majority o equations have been devel-oped or long bones o the upper and lower limbs to estimate statures o Arican, Asian and Europeanpopulations (rotter and Gleser, 1952, 1958; Fujii, 1960; Allbrook, 1961; Olivier, 1976; Feldesmanand Lundy, 1988; De Mendonca, 2000). Tese studies, as well as studies by numerous other authors

have shown that adult stature is both populations specic, and changes through time. o determine,thereore, which mirila correspond to males, and which to emales, it is necessary to approximate thestature o males and emales rom a reerence collection o temporally congruent Croatian historicalpopulations. o this end the Osteological collection o the Croatian Academy o Sciences and Artswas checked or the presence o 18 th to 19th century historical populations.

Te collection, ounded in 1993 currently contains slightly more than 5500 skeletons rom 37archaeological sites in Croatia that cover the time period between approximately 9000 BC to the 19thcentury AD. wo archaeological sites ullling the necessary qualications were ound: Sisak, andRijeka. Bicondylar lengths o le emurs rom well preserved male and emale skeletons in whomsex could be unequivocally assigned based on pelvic (Bass 1987), and cranial morphology (Krogmanand Ican, 1986), were measured according to the criteria o Martin (Martin, 1928, FBL #2). Tesemeasurements were used to calculate adult stature using regression ormulae or the emur calculatedby Pearson (1899). Pearsons ormulae were used because a recent study o modern Croatian adultstature estimation based on radiographic measurements o cadaver long bone lengths (Petroveki,2001) showed that these ormulae were most applicable to Croatian populations.

Once adult statures are successully calculated, a discriminant unction analysis can be per-ormed on them to determine whether stature can be utilized to dierentiate between males andemales. Sex determination is amenable to discriminant unction analysis based on the assumptionthat the two sexes will produce a bimodal curve. Numerous researchers, including researchers omodern and archaeological populations rom Croatia were able to obtain discriminant unction ac-curacy rates o 92.2% to 94.4% using various dimensions o the emur (laus, 1997; laus et. al.,2003), and tibia (laus and omii, 2005). As both o these long bones are highly correlated withadult stature it is reasonable to expect that discriminant unction analysis o stature will also result

in a high accuracy rate.o separate between subadults and adults mean emoral lengths o modern subadults measuredby Anderson et al. (1964) are used to (once again using Pearsons regression ormula) calculate thecut-o point or subadult stature.

MATERIALS AND METHODS

A total o 350 mirila measurements rom two sites: Vukii (n=266), and Kruica-Korita (n=84)were available or analysis. Te mirila were categorized into one o three groups mirila taken pre-dominantly during the 18th century (category 1), those taken predominantly during the 19 th century(category 2), and mirila taken during the 20th century (category 3). Because there is some ethno-

graphic evidence that mirila taken during the 20th

century may represent con lengths it was decidedto leave these measurements out. Te total number o measurements in this analysis is thereore 328,with Vukii contributing 256 and Kruica-Korita 72 measurements.

Te two Croatian historical populations used to approximate the stature o males and emalesbased on emoral lengths are Sisak and Rijeka. In geographic terms both sites are relatively near toKruica-Korita and Vukii. Rijeka is approximately 160 km north-west o the two analyzed sites,while Sisak is approximately 130 km north-east o them. Both sites are also temporally congru-ent. Te osteological material rom Rijeka was recovered during archaeological excavations on theRijeka rg pul vele crikve site carried out rom 2008 to 2009. Tis is a multicomponent site that,among other eatures, contained a late historic cemetery. A total o 115 skeletons were recovered dur-

8.1


3/10

157

ing the excavation. Unortunately only 38 skeletons (17 male and 21 emale) in which sex could beunambiguously assigned based on pelvic and cranial morphology had suciently preserved emora.Archaeological and historical sources indicate that the Rijeka cemetery was used rom the 17th to the19th centuries (Vinji, personal communication).

Te osteological material rom Sisak was recovered during archaeological excavations o theSquare o Ban Josip Jelai in Sisak carried rom 1997 to 2001. Excavations revealed the presence othe Late Historic town cemetery that was located adjacent to the church o the Assumption o the

Holy Cross. Te cemetery was in use rom the 17th to the 19th century. A total o 147 graves were ex-cavated (Loli, 2001). Slightly less than a third (n=57 o which 33 were males, and 24 were emales)contained skeletons with suciently preserved elements to measure the emora and accurately de-termine sex. Te total number o skeletons available or discriminant unction sexing based on adultstature was thereore 95.

Pearsons regression ormulae or estimating stature rom the dimensions o the emur wereutilized to calculate the stature o males and emales in this sample. Te ormula or estimating malestature is:

male stature = 81.306 + 1.88 x length of femur;

the ormula or estimating emale stature is:

female stature = 72.844 + 1.945 x length of femur

As previously noted, Pearsons regression ormulae were used because a recent study (Petroveki,2001) o modern Croatian adult stature estimation based on radiographic measurements o cadaverlong bone lengths demonstrated that these ormulae had the best predictive values or the mod-ern Croatian population. Several regression ormulae including those developed by elkka (1950),rotter and Gleser (1958), Jantz (1992) and De Mendonca (2000) were also tested, but the margin oerror in all o these was greater than the one noted or Pearsons regression ormulae.

Pearsons regression ormulae or the emur were also used to determine the cut-o point be-tween subadults and adults. Stature estimates or children are not usually attempted in bioarchaeol-ogy because in lie their standing height is derived rom both the diaphyses and epiphyses o long

bones, joined by a cartilage growth plate. Our inability to assess the contribution o the cartilagi-nous growth plate to length and standing height (because it varies in thickness at dierent timeso the childs development, as well as between individuals) prevents accurate estimates o stature inskeletonized remains. Tis having been said, some data are available that allow or the approximatethickness o the growth cartilage and epiphysis to be calculated. Maresh (1955) collected both di-aphyseal and long-bone length data or male and emale French Canadian White children. His dataoverlaps in the 10 to12-year age groups or boys and girls revealing that emoral epiphyses accountor 3.8% o bone length in boys and 4.6% o bone length in girls. Fortunately this analysis is notconcerned with accurate stature determination o subadults o known age based on the dimension otheir long bones. What I am looking or is a simple cut-o point that separates subadults rom adultsin the Vukii and Kruica-Korita series. In bioarhaeological studies subadults are usually denotedas individuals younger than 15 years. Te reason or this is that the three elements that comprise the

adult innominate: the ilium, ischium and pubis, usually use by 15 years allowing accurate deter-mination o sex. However, as this analysis deals with the stature o individuals that lived during the18th and 19th centuries (and all o our reliable published data on subadult emur lengths comes rommodern populations) it is necessary to take into account actors such as secular growth, and the sig-nicantly improved nutrition o subadults living during the 20 th century. Tereore, to be absolutelysure that the analysis dierentiates between subadults and adults - and not between subadults alongwith smallish adults, and taller adults, the cut-o point between subadults and adults was calculatedusing the mean length o the subadult emale emur aged 12 years minus one standard deviation asmeasured by Anderson et al. (1964).


4/10

158

RESULTS

Te descriptive statistics or the mirila are presented in able 1. As is evident rom this table themean lengths o the mirila in Krucica-Korita (mean length = 155.24; sd = 37.03; n = 72), are consid-erably greater than those recorded in Vukii (mean length = 138.35; sd = 39.96; n = 256). Analysis ovariance indicates that this dierence is statistically signicant (F = 10.356, P = 0.001).

8.2

Table 1.Descriptivestatistics for themirila by siteand category

able 2 shows the descriptive statistics o adult statures or both sexes in the reerence Croatianhistorical series. Tese statures were calculated rom the bicondylar lengths o the le emurs usingPearsons regression ormulae or the emur or males and emales. Te index o dimorphism is great-er than 100 indicating that males in the series were, as expected, higher than emales. Te F-ratioindicates that the dierence is signicant (P < 0.001) while the standard deviations indicate that malesexhibited more variation than emales. Tis result indicates strong sexual dimorphism in the analyzedsample and presupposes that adult stature is useul in dierentiating between the sexes.

Table 2. Sexual dimorphism and unifactorial statistics of adult stature1 in the temporally congruent skeletal seriesfrom Sisak and Rijeka

Once the existence o a strong sexual dimorphism was determined, a discriminant unction was

generated. Te unstandardized discriminant unction coecient, sectioning point, and cut-o pointare presented in able 3. Unstandardized discriminant unction coecients are used or calculating

Table 3. The unstandardized 0 discriminantfunction coefficient, constant, sectioning point, andcut-off point calculated by the discriminant functionanalysis of stature in the Rijeka and Sisak skeletalseries

Vukii18th century(n=119)


Vukiiotal(n=256)

Kruica-Korita18th century(n=22)


Krucica-Koritaotal(n=72)

DescriptivestatisticsMean 135.76 140.59 138.35 166.55 150.26 155.24Median 140.00 154.00 146.00 166.00 164.00 165.00Std. Deviation 38.78 40.97 39.96 19.88 41.65 37.03Range 153 156 156 68 146 146Minimum 54 52 52 138 64 64Maximum 207 208 208 206 210 210

Males Females Sexual dimorphism

Variable N Mean Std. Dev. N Mean Std. Dev. Index2 F3

Adult stature 50 168.19 4.07 45 155.31 2.90 108.29 308.79

1 Stature was calculated with Pearsons regression ormulae or estimating stature rom the dimension o theemur2 Index = male mean/emale mean x 1003 Signicant at P < 0.001

Variable Adult stature

Unstandardized coe. 0.280Constant -45.428Sectioning point -0.095Cut-o point males > 162.75 > emales


5/10

159

discriminant unction scores rom the raw data. A discriminant score is obtained by multiplying thevariable (adult stature) with the unstandardized coecient and then adding the constant. I the scoreis greater than the sectioning point the individual is considered male, while a lower score indicates aemale. An alternative, and simpler, approach in discriminant unctions where only one variable is usedis to compare the dimension o the analyzed specimen to a cut-o point. Te cut-o point is in this casethe simple average o the means or each sex or in this case 162.75 cm. A value higher than this indicatesa male, a lower value indicates a emale.

Reclassication o the cases used to develop the unction shows that the overall accuracy or bothsexes is a very high 97.9% (able 4). A slightly lower accuracy was achieved or males (96.0% or 48/50)while all 45 emales in the series were accurately sexed based on their stature as approximated withPearsons regression ormula.

Table 4. Sexing accuracy forthe discriminant function

As the cut-o point between male and emale stature in the temporally congruent skeletal seriesrom Croatia is 162.75 cm, all mirila equal to, or longer, than 164.0 cm are considered to have beenmales. As previously noted, to dierentiate between subadult and adult mirila, the mean length othe modern subadult emale emur aged 12 years minus one standard deviation was used. Accordingto the results o Anderson et al. (1964) this gives a emoral length o 38.23 cm resulting in a statureo 147.1 cm. Te cut-o point between subadults and adults in the series is thereore determined at148.0 cm.

All together this means that all mirila smaller than 148.0 cm are judged to have been subadults,mirila between 149.0 and 163.0 cm are considered to have been emales, and all those higher than164.0 cm are deemed to have been males.

Table 5. The sex distributions in the Vukii and Kruica-Korita series

Applying these criteria to the Vukii and Kruica-Korita mirila produces the ollowing sex dis-tributions (able 5). Slightly more than hal (52.0%) o the Vukii series were subadults while malesin the sample (80/256 or 31.2%) outnumber emales (43/256 or 16.8%) by a ratio o 1.86 : 1.00. Inthe Kruica-Korita series males (37/72 or 51.4%) are most requent, ollowed by subadults (22/72 or30.6%), and emales (13/72 or 18.0%). Te male to emale ratio in this series is 2.85 : 1.00.

An additional point o interest concerns the distribution o subadult mirila values. As previous-ly mentioned, subadults comprise a large proportion o both the Krucica-Korita (22/72 or 30.6%)and Vukii (133/256 or 52.0%) series. On the assumption that the mirila represent the lengths o

Males Females

Variable N % N % N Average

Adult stature 95 96.0 48/50 100.0 45/45 97.9%

Vukii

18th

century(n=119)

Vukii

19th

century(n=137)

Vukii

otal(n=256)

Kruica-

Korita18th century(n=22)

Kruica-

Korita19th century(n=50)

Krucica-

Koritaotal(n=72)

Sexdistribution

N % N % N % N % N % N %

Males 32 26.9 48 35.0 80 31.2 12 54.5 25 50.0 37 51.4Females 17 14.3 26 19.0 43 16.8 6 27.3 7 14.0 13 18.0

Subadults 70 58.8 63 46.0 133 52.0 4 18.2 18 36.0 22 30.6

otal 119 100.0 137 100.0 256 100.0 22 100.0 50 100.0 72 100.0


6/10

160

cadavers being taken to burial this would suggest a high subadult mortality ratio. By itsel, this isnot problematic. What is, however, interesting is that the majority o the subadult mirila are longer

than 100.0 cm. According to modern global standards the average height o a three years old child isapproximately 100.0 cm. Applying this criteria to the Vukii and Kruica-Korita series results is asubadult age distribution in which only 6/22 or 27.3% o subadults in Kruica-Korita, and 55/133 or41.3% o subadults in Vukii died beore reaching three years o age.

Te descriptive statistics or male and emale mirila in Vukii and Kruica-Korita are presentedin able 6. As is evident rom this table male mirila values are consistently, in all subcategories, at thelevel o complete Vukii and Kruica-Korita samples, as well as the level o the complete analyzedsample - higher than emale values. O interest is the act that the, considerably smaller, Kruica-Korita series (n= 50) exhibits evidence o secular growth average male mirila values increase romthe 18th to the 19th century by 3.45 cm (rom 180.67 cm to 184.12 cm), and average emale mirila

values increase by 2.53cm (rom 153.33 cm to 155.86 cm), while the larger Vukii series (n= 123)exhibits a considerably smaller increase o 1.76 cm in average emale mirila values (rom 155.35 cm

in the 18th

century to 157.11 cm in the 19th

century), while average male mirila values decrease by1.01 cm (rom 182.03 in the 18th century to 181.02 in the 19th century).Kolmogorov-Smirnov tests o normality (able 7) indicate that in the total Kruica-Korita se-

ries neither male nor emale mirila values have a normal distribution (Kolmogorov-Smirnov statistic= 0.147; P = 0.041 or males, and Kolmogorov-Smirnov statistic = 0.240; P = 0.039 or emales). In thetotal Vukii series male mirila values, once again, do not exhibit a normal distribution (Kolmogorov-Smirnov statistic = 0.130; P = 0.002), while emale values do (Kolmogorov-Smirnov statistic = 0.120;P = 0.127). Te same trend is noted at the level o the compete analyzed series - emale mirila valuesexhibit a normal distribution (Kolmogorov-Smirnov statistic = 0.104; P = 0.200), while male do not(Kolmogorov-Smirnov statistic = 0.107; P = 0.002).

Table 6. The descriptive statistics for male and female mirila by site and category



Vukiiotal(n=123)



Krucica-Koritaotal(n=50)

Male

N 32 48 80 12 25 37Mean 182.03 181.02 181.42 180.66 184.12 183.00Median 180.00 180.00 180.00 175.00 182.00 179.00Std. Dev. 12.90 11.83 12.20 15.79 13.83 14.37Range 43.00 44.00 44.00 41.00 45.00 45.00Minimum 164.00 164.00 164.00 165.00 165.00 165.00

Maximum 207.00 208.00 208.00 206.00 210.00 210.00Skewness 0.424 0.626 0.533 0.999 0.555 0.639

FemaleN 17 26 43 6 7 13Mean 155.35 157.11 156.41 153.33 155.85 154.69

Median 155.00 157.00 157.00 153.00 155.00 155.00Std. Dev. 4.52 4.36 4.46 2.25 5.27 4.21Range 13.00 13.00 13.00 5.00 14.00 14.00Minimum 150.00 150.00 150.00 151.00 149.00 149.00Maximum 163.00 163.00 163.00 156.00 163.00 163.00Skewness 0.202 -0.191 -0.047 0.012 0.570 1.101


7/10

161

DISCUSSION

Our perception o our own or another individuals stature is inuenced by numerous actors.

Bias in sel-reporting o stature has previously been documented (Willey and Falsetti, 1991), as haveproblems in the perception o stature by others. For instance, Komar (2003) reported that less thana third (29.4%) o the positively identied individuals recovered rom mass graves in the Srebrenicaregion in Bosnia and Herzegovina had a height that ell within the estimated ranges provided by closerelatives. Similarly, the identication team commissioned by the Croatian Government to identiyvictims recovered rom individual and mass graves related to the 1991 War in Croatia used a largemargin o error (10 cm) or stature estimates o missing individuals provided by close relatives (lauset al., 2007).

Analyses described in the preceding part o the paper were undertaken to determine whethermirila values correspond to the lengths o human cadavers that were being transported to burial.At rst glance that is a reasonable assumption. Te mirila are clustered around a road in a mannerreminiscent o a cemetery. Some mirila are grouped together and appear to represent amily units.

Additionally, the morphology o some o the mirila is uncannily similar to that o an osteometricboard an instrument used to measure the lengths o long bones in anthropological studies.Te results o the analyses carried out in this investigation, however, strongly suggest that mirila

do notcorrespond to cadaver lengths. Tis assertion is based on the ollowing ve arguments.Te rst concerns the demographic proles o the Vukii and Kruica-Korita series deter-

mined through the cut-o points calculated by discriminant unction analysis o male and emalestature in the temporally congruent historical populations o Rijeka and Sisak. An exceptional eatureo both the Vukii and Kruica-Korita series is their markedly uneven male to emale ratio. In theVukii series this ratio is 1.86 : 1.00, and in the Kruica-Korita series it is an even larger 2.85 : 1.00.Te total male to emale ratio, thereore, in both series is 2.10 : 1.00.

In Croatian archaeological and historical cemetery series the male to emale ratio is usuallyclose to 1.00 : 1.00. For instance, the male to emale ratio in Rijeka is 1.31 : 1.00, while in the Sisak se-

ries it is 1.16 : 1.00. Comparisons with other, larger composite series rom Croatia support this trend.Te male to emale ratio in a large (n= 201) historic period series rom continental Croatia is 1.04: 1.00 (laus, 2002), in an equally large (n= 175) Late medieval series rom continental Croatia theratio is 0.97 : 1.00 (laus, 2002), and in an even larger (n= 331) Early medieval series rom Croatiaseastern Adriatic coast the ratio is 1.04 : 1.00 (laus 2006). Te absence o emales in the Vukii serieswhere they account or only 16.8% o the total sample, and in the Kruica-Korita series - where theyaccount or just 18.0% o the total sample is, thereore both exceptional and bafing.

A second important demographic inconsistency concerns the age distribution o subadults inthe two series. As previously noted, subadults comprise a large raction o both series: in Kruica-Korita they account or 30.6% o the total sample, while in Vukii they represent more than hal

Table 7. Kolmogorov-Smirnov tests ofnormality for Krucica-Korita, Vukii,and the total mirila sample

8.3

Kolmogorov-Smirnov

Statistic d Signicance

Kruica-KoritaMale 0.147 37 0.041Female 0.240 13 0.039

Vukii

Male 0.130 80 0.002Female 0.120 43 0.127

otal mirila sampleMale 0.107 117 0.002Female 0.104 56 0.200


8/10

162

(52.0%) o the total sample. By todays standards such requencies o subadult mortality appear stag-geringly high. However, and unortunately, i anything these values probably underestimate the realmortality o subadults in pre-industrial populations. Tese populations had a limited knowledgeo the dangers that lurk in the microscopic world o microbes, bacteria and viruses, and even moreimportantly had limited access to adequate nutrition and no access to antibiotics. Consequentlysubadult mortality, particularly during the rst three years o lie, was extremely high. In some ar-chaeological and historical series children younger than one year account or up to a third o the

total sample (Coale and Demeny, 1966; Ledermann, 1969; Acsadi and Nemeskeri, 1970). Tis is theproblem with the Vukii and Kruica-Korita series. While general subadult mortality in these seriesappears to have been high, the mortality o children younger than three years o age is unexplainablylow. Employing a airly conservative criterion that connes subadults younger than three years o ageto a stature less than 100.0 cm results in a subadult age distribution in which only 39.3% o the sub-adult sample or 18.6% o the complete sample died beore reaching three years o age. Tis suggestsa highly unlikely scenario in which mortality rates rom our to twelve years were higher than thoseduring the rst three years.

Inant mortality gures or many past populations are uncertain, but it is estimated (Orme,2001) that in sixteenth-century England around 27% o children died beore the age o one year.Lewis (2007) summarized child mortality patterns derived rom archaeological skeletal collectionsin the published literature. Despite the almost ubiquitous under-representation o inants (individu-als aged between birth and the rst year) in archaeologically derived skeletal series in the 42 samples(containing 9658 individuals) included in this review, the percentage o subadults who died betweenbirth and the rst year was 29.5%, while the percentage o subadults who died between one and ouryears was 27.5% thus giving a total child mortality rate o 57.0% or the period between birth to ouryears o age. Te considerably lower (39.3%) mortality rate o subadults aged between birth to threeyears in the Kruica-Korita, and Vukii series is, thereore, hard to explain.

Tree inconsistencies are also noted when the actual mirila values are analyzed. Te rst con-cerns the absence o secular growth in males rom the Vukii series coupled with the simultane-ous presence o secular growth in both males and emales rom Kruica-Korita, and emales romVukii.

Secular changes in growth and maturation have been well documented in various world popula-

tions, with secular increase especially noticeable in the developed countries. For instance, Ohyamaet al. (1987) analyzed secular growth in a sample o 738 students rom Kyushu in Japan rom 1961to 1981 and noted an increase in means o standing height, leg length, and ratio o leg length tostanding height. Loesch et al. (2000) compared data on stature and body weight obtained during1992-1993 rom 1804 Melbourne school students aged 5 to 17 years, with historical data collectedrom white Australians during the last 100 years and noted a secular increase in adult stature withthe rate o increase varying between 0.4 and 2.1 cm/decade in males, and 0.01 and 1.6 cm/decade inemales. In contrast to these results, Pretty et al. (1998) analyzed the adult stature o Aboriginal SouthAustralians rom prehistoric to recent times and noted no signicant increase in height in either sex.

No systematic analyses o secular changes in Croats have, so ar, been undertaken but some ten-tative approximations can be made. Te oldest available data on adult Croat stature is available romthe rst Croatian manual or orensic medicine published in 1889 by dr. Niko Selak. Among other

data Selak (1889) gives the mean cadaver lengths o adult males (168.5 cm, range rom 157.0 180.0cm), and emales (156.5 cm, range rom 153.0 166.0 cm) without, unortunately, giving the numbero cadavers these values were calculated rom or their provenance. As a matter o interest, the valuesthemselves are (taking into account secular growth as these values were collected presumably duringthe end o the 19th century) very similar to the cut-o points calculated rom the Rijeka and Sisakseries to dierentiate between male and emale mirila (emale range rom 149.0 163.0 cm, malerange 164.0 cm and higher).

Data on modern Croat stature is supplied by Petroveki (2001) and Mustajbegovi (1992).Petroveki (2001) analyzed 41 modern cadavers and calculated that the mean cadaver length omales in Croatia is 173.6 cm, and that o emales is 162.4 cm. Based on the analysis o the stature o


9/10

163

4844 individuals living in rural and urban communities in Croatia Mustajbegovi (1992) calculatedthe mean height o urban males at 174.0 cm, rural males at 173.3 cm, urban emales at 161.9 cm, andrural emales at 161.5 cm.

Comparing these data to Selaks suggests a secular increase in stature o 0.54 cm/decade in e-males, and 0.46 cm/decade in males. As these data indicate a noticeable secular increase in statureor both males and emales, and the same trend is noted in males and emales rom Kruica-Korita,along with emales rom Vukii, it is unclear why Vukii males exhibit a decrease in height o 1.01 cm

during the same time period.A urther inconsistency related to the mirila values concerns their distribution. A normal

or Gaussian distribution is a continuous probability distribution that describes data that clusteraround the mean. Te graph o the associated probability density unction is bell-shaped, with a peakat the mean, and is known as the Gaussian unction or bell curve. Human stature, particularly i thesample is large enough, usually exhibits a normal distribution. A point in act is that Wikipedia theree internet encyclopedia, currently uses adult stature o males in the USA to demonstrate what anormal distribution is.

Summary statistics o the Vukii and Kruica-Korita mirila data that include measures o cen-tral tendency such as the mean and median, measures o dispersion such as the standard deviation,and measures o distribution such as skewness and kurtosis, as well as Kolmogorov-Smirnov testso normality, show that in the total Kruica-Korita series neither male (n = 37) nor emale (n = 13)mirila values have a normal distribution. Tis can perhaps be explained by the act that these are, byar, the smallest o the analyzed samples. In the Vukii series emale mirila (n = 43) have a normaldistribution, while male (n = 80) do not. Tis is harder to explain as is the act that at the level o thetotal mirila sample rom both sites emale values (n = 56) once again exhibit a normal distribution,while male (n = 117) do not. Tese results are, thereore, consistent with the theory that actors otherthan human stature were being incorporated into mirila values.

Finally, comparing mirila values with the calculated statures o males and emales rom othertemporally congruent populations, as well as with the stature o modern Croats strongly suggeststhat mirila are not an accurate representation o the stature o the past inhabitants o Vukii andKruica-Korita.

Te mean value o male mirila in the total Kruica-Korita and Vukii series is 181.9 cm (s.d. =

12.9 cm), the mean value o emale mirila is 156.0 cm (s.d. = 4.4 cm). By itsel, this is a ascinating actthat suggests that the average dierence in height between a man and a woman in these communitieswas 25.9 cm. o put this into context, the average dierence between modern Croat males and emalesis 12.1 cm, while the dierence between mean male and emale stature in the Rijeka and Sisak historicalseries was 12.9 cm.

Te mean male stature in the temporally congruent historical series rom Rijeka and Sisak is168.2 cm (s.d. = 4.1 cm), while the mean emale stature is 155.3 cm (s.d. = 2.9 cm). Tis suggests thatmales in Vukii and Kruica-Korita were, on average, 13.7 cm taller than those in Rijeka and Sisak.A Kruskal-Wallis test indicates this dierence is signicant (c2 = 48.20, d = 1, P < 0.001). At the sametime, the dierence between emale statures is almost negligible. Females in Vukii and Kruica-Koritawere, on average, just 0.7 cm taller than those in Rijeka and Sisak and a Kruskal-Wallis test indicates thisdierence is, as expected, not signicant (c2 = 0.48, d = 1, P = 0.49).

Comparing mirila values with the stature o modern Croats is also illuminating. Te averagemale stature in Croatia today is 174.0 cm while the average emale stature is 161.9 cm. Tis meansthat males in Vukii and Kruica-Korita were on average 7.9 cm taller than modern Croats, whileemales were 5.9 cm shorter. Tese values are simply not realistic.

o summarize, the preponderance o the collected evidence suggests that mirila represent actualcadaver dimensions o the 18th and 19th century inhabitants o Vukii and Kruica-Korita only ithese people: a) lived in communities in which males outnumbered emales by a margin o morethan 2 : 1, b) i subadult mortality rates in these communities were dierent rom all other publishedarchaeological and historical series with relatively low mortality rates rom birth to the third year,and high mortality rates during the period between the ourth and twelh year, c) i males in Vukii


10/10

164

exhibit not only an absence o the secular growth noted both in the emales who lived with them, andin males and emale rom Kruica-Korita but, in act, decreased in height rom the 18th to the 19thcentury, d) i males rom Vukii and Kruica-Korita did not have a normal distribution o heightwhile emales did, and additionally were on average almost 26 cm taller than the women in theircommunities, and e) i males in the 18th and 19th century Vukii and Kruica-Korita communitieswere signicantly taller than their contemporaries in Rijeka and Sisak, and modern Croats.

I any o these assumptions are not true - and I can think o no valid reason why they should

be - than mirila measured something other than simple cadaver lengths.

Documents

Mirila - Slaus