emilia arcaleni - secular trend and regional differences in thestature of italians, 1854–1980

Secular trend and regional differences in the

stature of Italians, 1854–1980

Emilia Arcaleni *,1

Istituto Nazionale di Statistica (ISTAT), Direzione Centrale per le Indagini su

Condizioni e Qualita della Vita, via Adolfo Rava 150, I-00142 Rome, Italy

Received 16 June 2005; accepted 16 June 2005

Abstract

We present 127 years of data on the physical stature of military conscripts born in Italy during

1854–1980, as well as an analysis of regional variations in height (for birth cohorts born during 1927–

1980). The height of young men has increased in all regions of Italy. The secular trend and the

regional changes in stature are correlated with economic growth and a general improvement of living

conditions. This is suggested by the relationship between height and various socio-economic

indicators. A comparison of the 1927 birth cohort with the 1980 birth cohort shows that the mean

heights for populations in Italy’s southern areas, which were shorter than the national average in

1927, underwent the largest increases. In recent years, mean height has been gradually approaching

an upper bound for all regions in Italy, with the exception of some central and southern regions. A

multiple regression analysis evaluates the impact of living conditions on the convergence of regional

heights.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Height; Anthropometrics; Secular trend; Biological standard of living; Italy; Physical stature

1. Introduction

Secular changes in stature have been well documented in populations around the globe

(Eveleth and Tanner, 1990). In Italy, anthropometric studies were conducted for the first

http://www.elsevier.com/locate/ehb

Economics and Human Biology 4 (2006) 24–38

* Tel.: +39 06 46734598; fax: +39 06 46734596.

E-mail address: [email protected] ISTAT is not responsible for the opinions herein expressed, which must be attributed to the author only.

1570-677X/$ – see front matter # 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.ehb.2005.06.003

time in the 1870s, more than 40 years after Belgian and French investigators performed

their pioneering studies (Villerme, 1829; Quetelet, 1832). Initially, Italian studies of

anthropometric traits were local in nature; the secular height trend had not been recognised.

Instead, attention was focused on the various factors that could influence height, such as

climate, altitude, nutrition and ethnic origin (Lombroso, 1877; Perozzo, 1878; Pagliani,

1879; Riccardi, 1887; Livi, 1896a,b, 1898, 1905). The increasing height of Italians was first

recognised at the beginning of the 20th century (De Rossi, 1903) and was attributed to

improvements in the country’s economic and health conditions.

Costanzo (1948) was the first to collect data on stature of Italian conscripts, beginning

with the 1854 birth cohort until the 1920 birth cohort, on the basis of original military

archives. These data were published by the Istituto Nazionale di Statistica (ISTAT, 1958) in

the Sommario di Statistiche Storiche, the official collection of Italian historical statistics,

and have been updated in subsequent editions. Since 1949 (the 1927 birth cohort), in fact,

the Ministry of Defence has been providing ISTAT with semi-aggregated data every year

and the mean stature is calculated and published in ISTAT’s (1949–2003) Annuario

Statistico Italiano.

Since the 1980s, analysis of the stature of Italians has focused on conscription data,

often in collaboration with the Ministry of Defence (Distaso et al., 1984; Terrenato and

Ulizzi, 1983; Golini et al., 1985; Ulizzi and Terrenato, 1985). More recently, the emphasis

is on the additional information available in the military records of conscripts, such as

profession, education, region of birth, family name and blood group (Corsini, 2004).

2. Data quality

National compulsory recruitment of young men for military service was established in

1863 (birth cohort 1842) but abolished in 2005, having been replaced by a voluntary

enrolment program. Consequently, medical examination of all young men is no longer

performed. Although the 1985 birth cohort was the last to experience compulsory

recruitment, data for the 1980 birth cohort were the last to have been released by the

Ministry of Defence. Therefore, the present article analyses data for male Italian conscripts

born during 1854–1980. Recruitment was compulsory for these cohorts, and medical

examinations were performed on all young men of conscription age to check their physical

fitness for military service. Several other data were also recorded during the examination,

including health status and anthropometric and socio-economic characteristics. ISTAT data

on stature used in this study pertain to all of the conscripts whose height was measured,

even if they did not meet the minimum height required for military service.

Although recruitment was compulsory, for several reasons, the height of many young

men was not measured. In the second half of the 19th century, information about height was

lacking for up to 30–35% of young men nationwide and for up to 60% in particular

districts, notably areas in southern Italy (Ilari, 1989–1990). In the first period of the Italian

conscription history, in fact, there was a large percentage of draft dodging in the Italian

army, mainly because of the high rate of emigration to other countries among young men

from southern Italy (for example, the draft-dodging rate for the 1842 birth cohort, the first

cohort that underwent compulsory recruitment, was 11.5% nationally but 57% in Naples,

E. Arcaleni / Economics and Human Biology 4 (2006) 24–38 25

45% in Catania and 44% in Palermo). Because the mean height of young men from Italy’s

southern regions was lower than the national average, there was probably a systematic

overestimation of the mean national height, especially for birth cohorts drafted during the

first decades of Italian conscription. During the 20th century, the national rate of draft-

dodging decreased from 8% for the 1906 birth cohort to 4 and 0.8% for the 1923 and 1972

birth cohorts, respectively (Ilari, 1989–1990).

Although medical examination was most often performed when the young men reached

the age of 19–20 years, it was occasionally performed at a younger age (down to 17 years)

or an older age (up to 22 years), depending on the particular needs of the military. However,

from the 1960 birth cohort onward, all medical examinations were performed when recruits

were 18 years of age. For that reasons ISTAT also calculated a standardized height, centred

at 20 years of age, by adding growth increments (positive or negative) to the actual heights

measured at younger/older than 20 years ages2 (Costanzo, 1948) (Fig. 1). Adoption of the

standardized estimates has an appreciable effect on data for the 1896–1900 birth cohorts,

who were drafted earlier (even as young as 17 years of age) because of World War I,

correcting for an apparent height decrease of more than 2 cm. The correction for the rest of

the period under study is generally moderate. Beginning with the 1930 birth cohort, the

actual and standardized estimates essentially coincide, and ISTAT ceased to correct for

age effects, beginning with the 1954 birth cohort. Therefore, we use standardized

E. Arcaleni / Economics and Human Biology 4 (2006) 24–3826

Fig. 1. Stature measured during medical check-up (black dotted line) and stature standardized at the age of 20 (red

thicker line) for the 1854–1980 birth cohorts; Italian per capita Gross Domestic Product for years 1850–1980

[source: ISTAT and OECD (2003)].

2 Because of the lack of Italian growth rates, Costanzo used rates calculated for Danish males conscripts by

Mackeprang (1907–1911). Correcting for age-based differences reduces the ‘perspective error’ (Schreider, 1968)

typical of military anthropometric data due to two factors: the young age of the conscripts who were still growing

at the time they were measured and the secular growth anticipation trend (that means that young men reach earlier

their final adult height) (Tanner, 1968). This allows us to infer that the increase in Italian observed height data is a

real gain of definitive stature (except for the residual growth coming after 20 years of age) and that anticipation of

growth is a modest component of the secular height trend (as supposed by Damon, 1965).

(i.e., age-corrected) data for the 1854–1953 birth cohorts and observed (i.e., uncorrected)

data for the 1954–1980 birth cohorts3.

Data for only few birth cohorts are incomplete, because of events (i.e., loss of

documents and archives) associated with World War II (birth cohorts 1921–1926) and data

delivery irregularities from the Ministry of Defence to ISTAT (birth cohorts 1966, 1968,

1970 and 1971). The Ministry of Defence has only recently understood the importance of

statistical accuracy during data collection. This explains why past military anthropometric

data are affected by problems, such as inaccuracy and heterogeneity. Many height

measurements (in centimetres) ended in 0 or 5, because of rounding, and the problem still

persist in the most recent data to minor extent4. In first decades of Italian recruitment,

accurate measurement of height was recommended exclusively for recruits whose stature

was close to the minimum threshold for enlistment5, because many conscripts used tricks in

order to be declared unfit (Grippa, 1863).

Before the 1927 birth cohort, there were no data for Navy conscripts, although such

individuals accounted for 2–3% of all conscripts. Nevertheless, there is no reason to believe

that height data before this period would be significantly different if they included data for

Navy conscripts.

Regional data on mean stature are available beginning with the 1927 birth cohort. The

regional averages are comprised of data obtained during medical examination and are not

standardized to adjust for age-based differences. Fortunately, as mentioned above, actual

and standardized data have been approximately coincident from the 1930 birth cohort

onward, at least at the national level.

Some minor interventions have been necessary to correct trivial summarization errors,

corrigenda and editing errors present in the official ISTAT data.

3. Trend analysis

3.1. Trend at the national level

Italian military data demonstrate that secular height increased by 12.19 cm between the

1854 and 1980 birth cohorts of conscripts. Overall, stature increased from 162.39 to

174.58 cm, indicating an average growth of 0.97 cm per decade (Fig. 1). The national trend

followed an S-shaped pattern: after a long initial period in which the mean height increased

slowly, a phase of accelerated increases started with the generations born in the 1940s,

followed by a change in the concavity of the stature curve for the cohorts born in the early

1950s, and a final phase of diminishing rates of increase. Nevertheless, the Italian trend is

not completely regular; periods of steady increase are followed by sudden decreases and


3 From the 1954 birth cohort on, only observed stature is available (varying likely around 18–20) but from the

1960 birth cohort on, the examinations is fixed at the age of 18. An eventual dis-homogeneity in 1954 or in 1960

seems to be excluded, looking at the graph and considering that the correction due to the standardization reduces

progressively, so that in 1953 is imperceptible.4 The rounding was first analysed by Livi (1883), who concluded that it is not a problem for the calculation of

the mean height, and does not affect sensibly the proportions of shorter and taller than average heights.5 The height threshold for enlistment gradually shifted from 150 to 156 cm (Arcaleni, 1998).

subsequent recoveries. For example, the slight decrease in mean stature between birth

cohorts 1880 and 1895 could be ascribed to a slowdown in Italian economic growth

(Federico, 2003), and the decrease of height of the birth cohort born in 1945, an

exceptionally bad year, was brought about by the war. In fact, a dramatic worsening of

living standards took place during World War II, when per capita intake of protein, fats,

carbohydrates and calories declined. Moreover, the evidence of a relationship between

fluctuations in secular trend of heights and political-economic cycles is pointed out in

recent studies (Martınez Carrion, 2002; A’Hearn, 2003).

Despite worsened economic and health conditions during World War II, cohorts of

young men recruited for military service during the war (birth cohorts 1919–1925) and

young men born immediately before and during the war (birth cohorts 1930–1945) show

increasing mean statures.6 Thus, war-related deprivation seems not to have adversely

affected the height. On the contrary, throughout this period, the highest rates of growth

were recorded for the generations born in the 1940s and 1950s (Fig. 2). In Italy, severe

deprivations caused by World War II, such as malnutrition and disease, were immediately

followed by a long cycle of economic growth, during which children were nearly able to

reach their physical potential for growth. This evidence corroborates the auxological

theory about the recovery capacity of children after periods of severe deprivation. The

achievement of genetically based growth potential depends on the rapid restoration of

favourable nutritional conditions after a period of malnutrition and deprivation, as well as

on the duration of the crisis (Tanner, 1989).

Between the second half of the 19th century and the present, the distribution curve for

data on the stature has undergone changes in shape and position: the distribution of stature

shifted to the right and, after a long period of hyper-kurtosis with negative skewness (Gini,

1949), the height distribution curve progressively became Gaussian (Ulizzi and Terrenato,


Fig. 2. Growth rates of average stature of Italian conscripts (% yearly increase): 1854–1980 birth cohorts [source:

ISTAT].

6 The statures measured during medical examination of recruits from the 1919–1925 birth cohorts suggest a

slight decrease. However, these young men were recruited at a younger age, and standardization of their heights

confirms positive rates.

1985). Recent data suggest a moderate shift toward positive skewness caused by the

increased proportion of recruits with high and very high statures (the percentage of people

who were at least 180 cm tall increased from 0.6% in the 1854 birth cohort to 20.1% in the

1980 birth cohort) and the decreased proportion with short and very short statures (the

percentage of people who were shorter than 160 cm decreased from 33.5% in the 1854

birth cohort to 1.4% in the 1980 birth cohort). In particular, the percentage of recruits with

very short statures (less than 150 cm) in the 1980 birth cohort is now negligible (0.1%),

compared with 4.3% in the 1854 birth cohort. In the Alpine areas, at the end of the 19th

century because of the lack of iodine, very short statures were widespread among recruits

and were associated with hypothyroidism, physical stunting and mental retardation; in

Aosta Valley and Lombardy, the proportion of young men who were declared to be unfit for

military service because of goitre even reached 30% (Livi, 1883). In the first decades of the

20th century, cretinism was endemic in Italy so that the government designed specific

policies to deal with this problem: in 1931, the distribution of iodine to children produced

excellent results, with rapid recovery of mental and somatic development.

3.2. Territorial analysis

Data for the 1927–1980 birth cohorts reveal that stature increased in all areas of Italy

during this period, albeit with remarkable regional differences in both percentage and

absolute increases in height (Fig. 3). Such increases were higher in southern areas, which

were initially characterized by statures that were lower than the national average (Table 1;

Fig. 4). For example, the mean height increased by 10.8 cm in the southern region of

Basilicata. However, an increase of only 5.4 cm was observed in Liguria, which, similar to

other northern and central regions, has always had more industrialization and a higher

standard of living than southern regions, with respect to the most important socio-

economic indicators (ISTAT, 1986). This evidence confirms the existence of a

compensation effect in populations in which socio-economic conditions were initially

substandard: ‘‘secular changes usually are particularly intense in the most retarded or least

developed populations and/or social strata (. . .). The long-term changes are of

compensatory nature; namely the lower the initial value of the trait in a population,

the more intense its compensation (in the presence of similar external conditions)’’

(Wolanski, 1985; 7). The rank of regions according to height for the 1980 birth cohort is

almost the same as that for the 1927 birth cohort: recruits from Friuli and Sardinia remain

the tallest and shortest, respectively (Fig. 5). That depends on the fact that in Italy still there

are substantial differences in genetic endowments at regional level (Cappieri, 1971).

Italy can be divided into 4 areas (the Northwest, Northeast, Centre and South),7 enabling

us to observe a wide gap in height between the northern and central regions and the

southern areas (Fig. 6). Despite growth rates in southern Italy that were higher than those in

the other regions, the regional variability has not yet vanished (the difference has decreased

from 8.94 to 6.42 cm). This moderate process of convergence between height in the


7 Northwest: Piedmont, Aosta Valley, Lombardy and Liguria. Northeast: Trentino Alto Adige, Veneto, Friuli

Venezia Giulia and Emilia Romagna. Centre: Tuscany, Umbria, Marches and Latium. South: Abruzzo, Molise,

Campania, Apulia, Basilicata, Calabria, Sicily and Sardinia.

southern region and height in the northern and central regions is the result of two main

factors, the accelerated rate of height increases observed in the South and a decreased (but

positive) rate of growth in northern and central Italy, starting from cohorts born in the

1950s. This second fact is strongly related to the massive emigration from southern Italy to

northern and central Italy, which started in the early 1950s and reached its peak during the

second half of the 1960s (Golini, 1978). Immigrants from the South moved primarily with

the purpose of matching the increased labour demands of the northern industrial triangle

(Piedmont, Lombardy and Liguria),8 of the increasing number of small enterprises

emerging in Tuscany and Emilia Romagna and of the administrative bodies in Latium

(where the capital of Italy is located) (Pellicciari, 1970). The migratory flow of people to

northern and central Italy greatly slowed the rate of growth of mean height in many of the

latter regions (Liguria, Piedmont, Lombardy, Emilia Romagna, Trentino, Tuscany and

Latium), some of which even experienced slight decreases in mean human stature. The

immigrant populations caused the average stature for the regions where they eventually

settled to decrease: immigrant fathers and their sons were, on average, shorter than the

local residents. However, the northern and central regions registered a small recovery in the

mean height for cohorts born during the second half of the 1970s. The populations of

immigrants from southern Italy was mainly comprised of young men who took up their first

regular job at the age of 18–35 years and whose sons belonged to birth cohorts of the 1950s,


Fig. 3. Regional mean stature of the 1927 birth cohort [source: ISTAT].

8 To understand how migrations have changed the demographic characteristics of the destination areas, consider

the following. In 1951 in Turin (the main city of the industrial triangle), there were 77,000 children aged 0–4

years, but 20 years later, in 1971, the corresponding cohort of young men aged 20–24 years had more than doubled

to 174,000—an increase of 126% (Golini, 1978).

1960s and 1970s. Children of immigrants, usually tend to be taller then their parents

because of the more favourable conditions in which they were raised and because of

heterosis (Damon, 1965; Schreider, 1968), but for the first waves of immigrants at least, we

can exclude the possibility that a significant increase in the mean stature of the second

generation was due to exogamy. In effect, endogamous marriages were much more

frequent among the first wave of immigrants from southern Italy (Rosoli, 1978). Given the

size of the migratory flow and the variety of ethnic backgrounds in Italy, it would be useful

to have data based on the region of birth of the recruits (not only on the region of residence),

as well as some basic indicators of socio-economic status, to better understand the

dynamics of stature in Italy. Moreover, for immigrants in the northern and central regions,

it would be interesting to know the age at which conscripts moved to their new residence, to

understand how changes in the standard of living affected their growth.

The extent to which the decreasing growth rates characterizing the second half of the

20th century could be explained by modifications in the percent regional weight is also

analysed (Table 2). During the 20th century, the percentages of conscripts resident in

Veneto, Friuli and Trentino (regions characterized by very high statures) decreased,

whereas the percentages of conscripts resident in Campania and Apulia (southern regions

characterized by higher-than-average birth rates) increased. The progressive increase in the


Table 1

Italian Regions by mean height and growth rates between the 1927 and the 1980 birth cohorts

Regions 1927 1980 Variation (cm) Increase (cm/year) Increase (%)

Basilicata 162.55 173.36 10.81 0.20 6.7

Abruzzo 165.84(*) 175.20 9.36 0.19 5.6

Sardinia 162.15 171.58 9.43 0.18 5.8

Molise 164.45(*) 173.12 8.67 0.17 5.3

Apulia 164.29 173.36 9.07 0.17 5.5

Calabria 163.35 172.40 9.05 0.17 5.5

Umbria 167.16 175.82 8.66 0.16 5.2

Campania 164.90 173.12 8.22 0.16 5.0

Sicily 164.67 172.70 8.03 0.15 4.9

Trentino Alto Adige 169.17 177.12 7.95 0.15 4.7

Veneto 169.01 176.96 7.95 0.15 4.7

Latium 167.62 175.52 7.90 0.15 4.7

Marches 167.40 175.29 7.89 0.15 4.7

Lombardy 168.06 175.21 7.15 0.13 4.3

Friuli Venezia Giulia 171.09 178.00 6.91 0.13 4.0

Aosta Valley 168.98 175.73 6.75 0.13 4.0

Emilia Romagna 169.23 175.45 6.22 0.12 3.7

Piedmont 168.84 174.97 6.13 0.12 3.6

Tuscany 169.82 175.81 5.99 0.11 3.5

Liguria 169.73 175.10 5.37 0.10 3.2

Northwest 168.47 175.14 6.67 0.13 4.0

Northeast 169.30 176.48 7.17 0.14 4.2

Centre 168.32 175.60 7.29 0.14 4.3

South 164.29 173.02 8.73 0.16 5.3

Italy 166.95 174.58 7.63 0.14 4.6

Ordered by cm/year increase. Data flagged by (*) refer to the 1931 birth cohort [source: ISTAT].

proportion of Italians resident in southern regions has produced a composition effect

resulting in reduced national average growth rates. Although the percentages of conscripts

resident in Latium and Lombardy also increased during the same period, these regions

were characterized by lower birth rates and, therefore, these increases were mostly due to


Fig. 5. Regional mean stature of the 1980 birth cohort [source: ISTAT].

Fig. 4. Mean variation (cm per year) between the 1927 and the 1980 birth cohorts [source: ISTAT].

the flow of immigrants from the South. This last composition effect too contributed to

decreases in the national average growth rates.

The extent to which genetic factors and living conditions cause the persisting

differences in mean height between the northern and central regions and the southern

regions is not clear. Even though a progressively decreasing difference in basic socio-

economic and health-related conditions (including nutritional status and the prevalence of

infectious diseases) has been observed, regional indicators show that a difference still


Table 2

Proportion of conscripts by Region of residence and birth cohorts

Regions 1918 1938 1960 1980

Piedmont, Aosta Valley 7.4 6.5 6.5 6.9

Lombardy 12.9 13.7 14.8 15.7

Trentino A.A.,Veneto, Friuli V.G. 11.2 12.3 12.3 9.6

Liguria 2.5 2.7 2.0 1.3

Emilia Romagna 9.0 7.5 5.4 6.4

Tuscany 7.1 6.4 5.0 5.6

Umbria 2.1 1.8 1.6 1.3

Marches 3.3 3.0 2.4 2.1

Latium 6.0 7.4 8.4 9.2

Abruzzo, Molise 4.1 3.2 3.0 2.9

Campania 9.1 9.8 10.6 12.9

Apulia 6.6 7.3 7.2 8.4

Basilicata 1.5 1.5 1.6 1.3

Calabria 4.7 4.3 4.9 4.3

Sicily 9.6 9.6 11.0 9.1

Sardinia 2.9 3.0 3.3 2.9

Northwest 22.8 22.9 23.4 23.9

Northeast 20.2 19.8 17.7 16.1

Center 18.5 18.6 17.4 18.2

South 38.5 38.7 41.5 41.8

Italy 100.0 100.0 100.0 100.0

Source: Costanzo (1948) for the 1918 birth cohort and ISTAT for all the other birth cohorts.

Fig. 6. Mean stature of regions grouped in geographical areas: 1927–1980 birth cohorts [source: ISTAT].

exists. It can be easily demonstrated that the main available indicators of social and

economic disadvantage are negatively correlated with height (Fig. 7). This evidence

strongly suggests that, in southern Italy, the potential for increased height due to improving

living standards has not yet been fully exploited and that the height difference between the

South and the northern and central regions cannot be entirely attributed to differences in

genetic endowments. At the same time, available data also show that heights for most

regions are gradually converging towards their limiting levels, which are related to the

theoretical genetic endowments of the residents. This could be the case for Sardinia, where

mean height is the shortest despite many socio-economic indicators that are not reliably

different from those of northern regions. At present, the difference between Sardinia and

the other regions of Italy is not diminishing, despite the socio-economic progress in this

area. On the contrary, the difference between Sardinia and the region with the second

shortest height has increased from 0.40 to 0.82 cm.

The increases in average statures followed a common S-shaped pattern in almost all

regions; that is, an initial period marked by increasing rates of growth was followed by a

period of decreasing, but still positive, rates. For almost all regions, rates of growth began

to decrease for the cohorts born during 1948–1954, immediately after World War II. At

present, the majority of Italian regions are approaching growth rates close to 0 cm/year;

only some northern regions (Piedmont and Lombardy), central regions (Marches, Latium

and Umbria) and southern regions (Sicily and Sardinia) still report increasing growth rates.

This finding can be further analysed by looking at the fit of curves on a regional level. In

some regions, during the years in which an inversion of the curve’s concavity was observed,

rates at which heights increased appeared to be very rapid, suggesting that the changes

involved an upward shift of the general trend (see Emilia Romagna, Liguria, Lombardy,


Fig. 7. Stature and unemployment by region. Stature of the 1980 birth cohort; unemployment rate 15–64 (year

1980) [source: CNR-FGB-ISTAT research (http://www.aiel.it/bacheca/base_dati2.htm)].

http://www.aiel.it/bacheca/base_dati2.htm

Piedmont, Sicily and Sardinia); this seems to indicate that the socio-economic progress that

started in the post-war period, had an immediate impact on stature.

Multiple regression analysis was performed, with use of SAS (Statistical Analysis

System; version 8.01), to determine the impact of living conditions on the differences

between regional heights. Mean regional height for the 1980 birth cohort of conscripts was

regressed on the following social, economic and health-related indicators: unemployment

rate, per capita GDP, illiteracy rate, mortality rate among boys 0–14 years of age and a

dummy variable for the South.9 The data on unemployment (1980), per capita GDP (1980),

illiteracy (1981) and mortality (1990) tend to represent the socio-economic environment

where the generation was born and spent its childhood (Table 3). Although the value of R2

is large, the outcome of the full model is unsatisfactory, because the high standard error and

low value of the t statistic suggest the existence of multicollinear variables. It is also

noteworthy that in the full model the parameter estimate for per capita GDP is negative.

Forward insertion of variables in the model10 indicated that only unemployment rate and

the dummy variable were statistically significant (P = .01 and .05, respectively). Table 4

shows the restricted model for the 2 variables (unemployment and dummy) that are the

most correlated with stature.11 The statistical results of this model are decidedly better:

both variables are significant (P = .01). Regional unemployment rate and a dichotomous


Table 3

Multiple linear regression of height: forward selection of explanatory variables

Variable Parameter Standard error t-Statistic P > jtj Partial R-square

Constant 182.94 2.60 65.15 <0.0001

Unemployment �0.27 0.08 �3.20 0.0062 0.744

Dummy for South �1.77 0.87 �1.96 0.0556 0.077

GDP �0.91 0.51 �1.81 0.0809 0.028

Illiteracy �0.06 0.10 �1.13 0.2407 0.010

Mortality �0.12 0.14 �0.81 0.4729 0.006

R-square 0.87

N 20

F 18.04

P > F <0.0001

9 The ratio of population living in municipalities of up to 10,000 inhabitant size (an indicator of the urbanization

process) was not included in the analysis, because no correlation has been found between urbanization and

regional heights (r = 0.04). The indicator does not appear to be characterised by areas; small and very small

municipalities are disseminated throughout Italy. The little villages in the central and northern regions (such as in

Tuscany and the Alps, respectively) usually have quite good standards of living, whereas the little villages in the

southern region are usually poor and isolated.10 The forward-selection technique begins with no variables in the model and adds variables. For each of the

independent variables, the forward method calculates F statistics that reflect the variable’s contribution to the

model if it is included. The P-values for these F statistics are compared to a specified value (.50). If no F statistic

has a significance level greater than the specified value, the forward selection stops. Otherwise, the forward

method adds the variable that has the largest F statistic to the model. Then the evaluation process is repeated for

the variables still remaining outside the model, until no remaining variable produces a significant F statistic. Once

a variable is in the model, it stays.11 Values of the simple correlation coefficients are quite high. The coefficients are ranked as follows:

unemployment (�0.86), dummy (0.86), illiteracy (�0.80), GDP (0.76) and mortality (�0.56).

variable for the South accounted for approximately 82% of the regional variability in

height.12

4. Conclusions

Italian data support the hypothesis that height, although limited by genetic factors, is a

sensitive indicator of the quality of the socio-economic environment (Fogel, 1986; Komlos,

1994; Komlos and Baten, 1998). During the 127-year study period—especially the period

after World War II (i.e., during the economic boom of the 1950s and 1960s)—Italy

gradually changed from a country in which severe, diffuse poverty was prevalent to one

undergoing substantial economic development, characterised by improved nutrition,

improved health status and economic growth (Fig. 1). Changes in the living conditions after

World War II contributed to the remarkable secular increase in height: a significant increase

was registered among individuals born (and/or raised) in the decades most favourable to

economic development. Moreover, data for the generations born during times of war seem

to confirm the auxological theory about the capacity of the human body to recover after a

limited period of severe malnutrition.

In Italy, the secular trend in height can be studied only in terms of the regional

characteristics of height, otherwise the concept of secular trend ‘‘can conceal more than it

reveals. It tends to imply that all those peoples which have experienced an upward trend in

heights have done so at roughly the same rate or for the same reasons’’ (Floud et al., 1990;

20). Indeed, there is a clear difference between the heights observed in the South and those

observed in central and northern Italy, although the difference has decreased as the rate of

economic development has increased. A similar difference is also revealed by some of the

main socio-economic indicators (per capita GDP, unemployment, illiteracy and mortality).

The multiple regression analysis shows a strong correlation between average height and the

main socio-economic indicators. It seems to confirm that the current difference between

heights in southern Italy and those in central and northern Italy is correlated with socio-

economic disparities, in addition to possible differences in genetic endowments, and that


Table 4

Multiple linear regression of height: restricted model with two explanatory variables

Variable Parameter t-Statistic P > jtjConstant 177.05 375.01 <0.0001

Unemployment �0.21 �2.72 0.0144

Dummy for South �1.61 �2.70 0.0152

R-square 0.82

N 20

F 38.95

P > F <0.0001

12 In the full model, the unemployment rate is more significant (and properly signed) than per capita GDP: that

seems to indicate that effects on heights depend more on the inequality in income distribution (proxied by

unemployment rate) than on the mean value of income (thanks to M. Di Marco for comments on this topic).

suggests that a potential for growth in the southern regions does remain to be realized in the

future.

Acknowledgements

Thanks to Silvia Bruzzone, Carlo A. Corsini, Marco Di Marco, John Komlos and

Giovanni Vecchi for useful comments and suggestions. I want also thank Scott O’Brien for

editing the manuscript.

References

A’Hearn, B., 2003. Anthropometric evidence on living standard in Northern Italy, 1730–1860. J. Econ. History 63/

2, 351–381.

Arcaleni, E., 1998. La statura dei coscritti italiani delle generazioni 1854–1976. Bollettino di Demografia Storica.

Societa Italiana di Demografia Storica 29, 23–59.

Cappieri, M., 1971. Le basi etniche della statura degli italiani. Genus 28 (4), 267–324.

Corsini, C.A., 2004. Introduction to the Workshop ‘‘Per una ricerca su statura, salute e migrazioni: le leve militari

italiane e altro’’. Dipartimento di Statistica, Universita degli Studi di Firenze, 3 December (proceedings being

printed).

Costanzo, A., 1948. La statura degli italiani ventenni nati dal 1854 al 1920. Annali di Statistica, ISTAT, Roma, 8, 2,

59–123.

Damon, A., 1965. Stature increase among Italian-Americans: Environment, Genetic, or Both? Am. J. Phys.

Anthropol., Philadelphia 23 (4), 401–408.

De Rossi, G., 1903. La statura degli italiani e l’incremento in essa verificatosi nel periodo 1874–98. Archivio per

l’Antropologia e l’Etnologia, 33.

Distaso, S., Da Molin, G., Dell’Atti, A., 1984. Caratteristiche antropometriche e socio-economiche dei coscritti

della provincia di Bari, 1867–1917. Primi risultati, Universita degli Studi di Bari, Cacucci, Bari.

Eveleth, P.B., Tanner, J.M., 1990. Worldwide Variation in Human Growth. Cambridge University Press, Cam-

bridge.

Federico, G., 2003. Height, calories and welfare: a new perspective on Italian industrialization, 1854–1913. Econ.

Human Biol. 1 (3), 289–308.

Floud, R., Wachter, K., Gregory, A., 1990. Height, health and history. Nutritional status in the United Kingdom,

1750–1980. Cambridge University Press, Cambridge.

Fogel, R.W., 1986. Physical growth as a measure of the economic well-being of populations: the eighteenth and

nineteenth centuries. In: Falkner, F., Tanner, J.M. (Eds.), Human Growth. A comprehensive Treatise.

Methodology, Ecological, Genetic and Nutritional Effects on Growth. Plenum Press, pp. 263–281.

Gini, C., 1949. La distribuzione delle stature e ipernormale. Atti della VIII Riunione Scientifica SIS, Roma, 23–42.

Golini, A., 1978. Migrazioni interne, distribuzione della popolazione e urbanizzazione in Italia. In: Rosoli, G.

(Ed.), Un secolo di emigrazione italiana 1876–1976. Centro Studi Emigrazione, Roma, 153–187.

Golini, A., Damiani, S., Distaso, S., Stornelli, R., 1985. Trends in Height and Weight for Males from Southern

Italy. Acta Medica Auxologica Centro Auxologico di Piancavallo 17 (1/2), 35–48.

Grippa, V., 1863. Catechismo della leva (ovvero la legge, i decreti, i regolamenti, le appendici, le modifiche e le

istruzioni ministeriali emanatesi a tutto agosto 1863 sul reclutamento dell’esercito nazionale attivo).

Stamperia del Vaglio, Napoli.

Ilari, V., 1989–1990. Storia del servizio militare in Italia: 1506–1870 (I), 1871–1918 (II), 1919–1989 (III), Rivista

militare, Roma.

ISTAT, 1949–2003. Annuario statistico italiano, Roma.

ISTAT, 1958. Sommario di statistiche storiche italiane 1861–1965, and following editions 1968, 1976 and 1986.


Komlos, J. (Ed.), 1994. Stature, Living Standard and Economic Development: Essays in Anthropometric History.

Chicago University Press.

Komlos, J., Baten, J. (Eds.), 1998. The biological standard of living in comparative perspective. Franz Steiner

Verlag, Stuttgart.

Livi, R., 1883. Sulla statura degli italiani. Studio statistico antropologico. Archivio per l’Antropologia e

l’Etnologia, Firenze, 13.

Livi, R., 1896a, Geografia ed orografia della statura e del colore dei capelli e degli occhi in Italia, Archivio per

l’Antropologia e l’Etnologia, Firenze, 26.

Livi, R., 1896b, Antropometria Militare. Risultati ottenuti dallo spoglio dei fogli sanitari dei militari delle classi

1859–63, eseguito dall’Ispettorato di Sanita militare per ordine del Ministero della Guerra, Giornale medico

del Regio Esercito, Roma, 2 voll and atlas.

Livi, R., 1898, Antropometria Militare. Risultati ottenuti dallo spoglio dei fogli sanitari dei militari delle classi



Livi, R., 1905. Antropometria Militare. Risultati ottenuti dallo spoglio dei fogli sanitari dei militari delle classi



Lombroso, C., 1877. Dell’influenza dell’orografia sulle stature. Archivio di Statistica, Roma 2, folio 3.

Mackeprang, E. Ph., 1907–1911, Height of the Danish Conscripts. Meddelelser om Danmarks Antropologie,

Copenhagen, pp. 1–3 and 153–158.

Martınez Carrion, J.M., 2002. Well-Being and Biological Standards of Living: The Spanish Case. Presentation at

the XIII Congress of the International Economic History Association, Buenos Aires, 22–26 July.

OECD, 2003. The World Economy: Historical Statistics. OECD Development Centre, Paris.

Pagliani, L., 1879. Lo sviluppo umano per eta, sesso, condizione sociale ed etnica, studiato nel peso, statura,

circonferenza toracica, capacita vitale e forza muscolare’. Giornale della Societa italiana di Igiene, 1, pp. 357–

376, 453–491 and 589–610.

Pellicciari, G. (Ed.), 1970. L’immigrazione nel triangolo industriale. Franco Angeli, Milano.

Perozzo, L., 1878. Sulle curve della statura degli iscritti misurati in Italia. Nota dell’ingegner L. Perozzo. Annali di

Statistica 2 (2), 238–240.

Quetelet, A., 1832. Recherches sur la loi de la croissance de l’homme, Nouveaux Memoires de l’Academie Royale

des Sciences et Belles Lettres, Brussels, Hayez, 1–32.

Riccardi, P., 1887. La statura in citta e campagna. Note antropometriche. Archivio per l’Antropologia e

l’Etnologia.

Rosoli, G. (Ed.), 1978. Un secolo di emigrazione italiana 1876–1976. Centro Studi Emigrazione, Roma.

Schreider, E., 1968. L’influence de l’heterosis sur les variations staturales. L’Anthropometrie 72 (3/4), 279–296.

Tanner, J.M., 1968. Piu precoci e piu alti. Le Scienze 1, 68–75.

Tanner, J.M., 1989. Foetus into Man: Physical Growth from Conception to Maturity. Harvard University Press.

Terrenato, L., Ulizzi, L., 1983. Genotype and environment relationship: an analysis of the stature distribution

curves during the last century in Italy. Ann. Human Biol. 10, 335–346.

Ulizzi, L., Terrenato, L., 1985. Nature-nurture relationships: the secular trend of stature is reaching in Italy its final

phase. Acta Medica Auxologica 17 (1/2), 119–124.

Villerme, L.R., 1829. Memoire sur la taille de l’homme en France. Ann. d’hygiene publique 1, 551–559.

Wolanski, N., 1985. Secular trend, secular changes, or long-term adaptational fluctuations? Acta Medica

Auxologica 17 (1/2), 7–19.


emilia arcaleni - secular trend and regional differences in thestature of italians, 1854–1980

Documents