East-West Orientation of Historical Empires
Peter Turchin, Jonathan M. Adams, and Thomas D. Hall
November 2004
Does environment affect the ability of states to project power? If state expansion is more easily
accomplished by staying within the same ecological zone, then state territories should be oriented
in the east-west direction, mirroring the orientation of major ecological zones of the world. Our
analysis of 62 largest empires in history supports this conjecture.
In a chapter entitled âSpacious Skies and Tilted Axesâ Jared Diamond (1997) argued that
food production spreads more easily latitudinally (in the East-West direction) rather than
longitudinally. A latitudinal shift is easier because similar climates and soil types tend to be
arranged in east-west oriented bands. This geographic pattern is best illustrated by a map of the
global distribution of biomes (Figure 1). A
biome
is a major type of ecological community such
as the grassland, desert, or temperate seasonal forest (Ricklefs 2001).
Although Diamond focused primarily on the spread of crop cultivars and domesticated
animals, the same principle should influence the military/political, demographic, and cultural
dynamics of societies. Consider the greatest empire ever (in terms of territorial extent), conquered
by the Mongols under Chinggis Khan and his immediate successors. The core of the Mongolian
Empire was the Great Eurasian Steppe that stretches for many thousands of kilometers from the
Khingan Mountains in the east to the Carpathians in the west (McNeill 1964). The Mongols were
steppe warriors, and they were able to rapidly extend their influence over this whole region
(Barfield 1994). The regions inhabited by settled agriculturalists adjacent to the steppe were
incorporated more slowly and to a lesser degree than the steppe. For example, the Russian
principalities of the forest zone were not occupied by the steppe-dwellers, being instead subjected
to tribute. As a result, the Mongolian Empire, with its core based on the steppe, was much wider
in the latitudinal rather than longitudinal direction. The ease of conquest was not the only factor
promoting the latitudinal spread of the great empires. Societies inhabiting similar ecological
zones tend to be more similar to each other than societies located in very different zones.
Techniques developed for integrating and controlling a certain type of society should, therefore,
be easier to extend latitudinally. Note that this âecologic factorâ should be detectable only at large
geographic scalesâa small state, as long as it stays within the same biome, will find it equally
easy (or equally hard) to expand in any direction, longitudinal or latitudinal.
Territorial expansion by states is a complex macrosociological process, influenced by
many factors other than the environment. Does the ecological factor have a detectable effect on
the projection of military/political power, or is its influence lost in the ânoiseâ of complex
interactions? To answer this question, we compiled a list of all large historical empires with peak
territories exceeding 1 Mm
2
(= 1,000,000 km
2
), and measured the distances from their eastern to
western extremes, as well as from the northern to southern extreme (Endnote 1). Our measure of
the tendency to expand in the latitudinal direction is the log-transformed ratio of the east-west
distance to north-south distance (Endnote 2) the âlatitudinal index.â
The frequency distribution of the latitudinal index in our sample of 62 historic empires is
strongly skewed to the right (Figure 2), and the mean index is significantly greater than zero (
t
=
4.83,
P
< 0.001). The great majority of empires, nearly 80%, have a positive latitidinal indexâ
that is, they are wider in the east-west compared to the north-south direction. There are only three
empires that have a strong north-south orientation, and these are the proverbial exceptions that
prove the rule. The New Kingdom of Egypt had at its core the valley of a major river running
south-north, the Nile. The Inca empire was located on the west coast of South America, where
ecological zones run longitudinally (see Figure 1) due to a major mountain chain, the Andes.
Finally, the Khmer empire was located entirely within the wet tropical forest biome. Thus, even
though these three cases do not conform to the rule of latitudinal spread, they obey the more
general rule of âexpansion within an ecological zone.â
All of the largest empires (with territory over 10 Mm
2
) were oriented in the east-west
direction. We have already discussed the case of the Mongol empire. The Islamic Caliphate is a
variation on the same pattern, except that the ânative biomeâ of the Arabs was the subtropical
desert, rather than the temperate grassland/desert of the Mongols. The next largest state in history
after the Mongols, the Russian empire (peak area of 22.8 Mm
2
in 1895), originated in the
transitional zone between the steppe and the forest (ecologists call such transitional zones
ecotones
). Once the Muscovite state began expanding in the sixteenth century, it spread fastest
precisely within the same ecotoneâeastward along the boundary between the Eurasian steppe
and northern taiga. Eastward expansion was extremely rapid, so that the Pacific was reached by
the mid-seventeenth century. In contrast, the southern advance into the steppes and deserts of
Central Asia took a much longer time, and they were conquered only by the late nineteenth
century.
Another example of the same dynamic is the early expansion of Rome. The territory of
the Roman Empire in the first century B.C.E. coincides almost precisely with the
woodland/shrubland biome (also known as the Mediterranean zone). Subsequent expansion took
the Romans into the forests of northern Europe. However, severe reverses, such as the battle of
Teutoburg Forest in 9 C.E., in which 20,000 legionnaires were obliterated by the tribal Germans
(Wells 2003), persuaded the Romans to abandon plans of further conquest. The general rule, thus,
seems to be that expansion is easiest and most lasting when occuring within the same ecological
zone. Expansion into other biomes is possible, but more difficult, slow, and requires greater state
resources. The case of China is probably the best illustration of this principle. The native biome
of the Chinese is the temperate seasonal forest (Endnote 3), and this was precisely the area that
was first unified by each of a long succession of Chinese empires. The strength of the Chinese
state, however, allowed it to expand into alien biomes, and at their peaks the great Chinese
empires intruded into the steppe (Inner Mongolia, Chinese Turkestan), the alpine biome (Tibet),
and the tropical rain forest (Vietnam).
Is the influence of ecology detectable in the shapes of modern states? At first glance, no.
The average latitude index for the 29 modern states whose territory exceeds 1 Mm
2
is positive,
but not significantly different from 0. However, if we exclude South American countries, where
biomes extend in the longitudinal direction, the statistical test indicates that the pattern is
detectable even today (
t
= 2.66,
P
= 0.014). The tendency to east-west orientation in modern
countries, nevertheless, is much weaker than for historical empires. Partly this could be due to the
effect of modern technology, but we suspect that it is also an artefact of the propensity of modern
states to claim territory even if it is not used by their populations. A striking example of this
tendency is Canada, whose population is squeezed into a narrow band running east-west along its
southern border with the US, but which nevertheless claims extensive territories in the Arctic.
Because of the addition of these lands, which are almost totally devoid of human occupation, the
latitudinal index of Canada is slightly negative. Algeria and Lybia provide other examples of the
same tendencyâtheir populations are largely confined to the east-west band along the
Mediterranean littoral, but their latitudinal indices are essentially zero, because they claim huge
territories to the south, in the Saharan desert.
In conclusion, our results indicate that the physical and biological environment has a
detectable effect on the shapes of historic and modern states. It appears that projection of
military/political power is easier within the same ecological zone (biome). Our results should not
be interpreted as a kind of âecological determinismââalthough ecology is important, its
influence on state expansion patterns is transmitted by entirely social mechanisms. On the other
hand, certain techniques and ideas from ecological sciences have proved to be fruitful in
suggesting novel approaches to the study of social systems (Turchin and Hall 2003). Diamondâs
original insight, which motivated our study, is one example; another is the recent demonstration
that cultural variability exhibits a latitudinal gradient (Pagel and Mace 2004). Our results also
have interesting implications for the study of historical dynamics (Turchin 2003). Researchers
working within the world-system paradigm have noted that the rise and fall of populations, cities,
and empires is characterized by a broad-scale synchronicity (Chase-Dunn and Hall 1997, Chase-
Dunn et al. 2000). For example, there is a substantial correlation between the dynamics of
Western Europe and China. On the other hand, South Asian dynamics are completely
uncorrelated with the rest of Eurasia. Our finding that the propagation of âsignalsâ within
military-political networks is facilitated in the latitudinal, but not longitudinal directions suggest
one possible explanation for this macrohistoric pattern.
Endnotes:
Note 1. Our list of large historical states was based on the compilation by Taagepera (1978, 1978,
1979, 1997), which has been systematized and posted on the web by Chase-Dunn and coworkers
(http://www.irows.ucr.edu/). We checked the Taagepera list with all major historical atlases in the
library of the University of Connecticut and found additional eight empires that fitted our criteria
(Axum, Hsi-Hsia, Kara-Khitai, Srivijaya, Maurian, Kushan, Gupta, and Maratha). We excluded
the maritime empires of the European Great Powers, because our measure of the latitudinal
tendency is not applicable to such noncontiguous, widely distributed collections of territories.
One difficulty in constructing the list was presented by the repeated rise of empires in the same
location, such as in China. We adopted the middle road of counting each major dynasty (Han,
Tang, Ming, etc) as a separate empire, but did not distinguish between cycles within a dynasty
(e.g., Early versus Late Han). Analysis of a reduced dataset, which included only the largest
empire for each geographic location, yielded qualitatively the same result. See Table 1 for the list
of empires.
Note 2. Log-transforming the ratio of distances was necessary to make the distribution of the
index symmetric. Positive values indicate east-west orientation, and negative values â north-south
orientation.
Note 3. It may seem strange to call the Chinese home biome a âforest,â because in present-day
China, of course, very few forests are left. Remember, however, that the biome names reflect the
types of ecological communities that would be present before substantial human impact; the
names are simply a short-hand reference to particular combinations of the climate and soil types.
References
Barfield, T. J. 1994. The devil's horsemen. Pages 157-182
in
S. P. Reyna and R. E. Downs, editors.
Studying war: anthropological perspectives. Gordon and Breach, Langhorn, PA.
Chase-Dunn, C., S. Manning, and T. D. Hall. 2000. Rise and fall: east-west synchronicity and Indic
exceptionalism reexamined. Social Science History
24
:727-754.
Chase-Dunn, C. K., and T. D. Hall. 1997. Rise and demise: comparing world-systems. Westview Press,
Boulder, CO.
Diamond, J. 1997. Guns, Germs, and Steel: The Fates of Human Societies. W. W. Norton, New York.
McNeill, W. H. 1964. Europe's Steppe Frontier. University of Chicago Press, Chicago, IL.
Pagel, M., and R. Mace. 2004. The cultural wealth of nations. Nature
428
:275-278.
Ricklefs, R. E. 2001. The Economy of Nature, 5th edition. W. H. Freeman, New York.
Taagepera, R. 1978. Size and duration of empires: systematics of size. Social Science Research
7
:108-127.
Taagepera, R. 1978. Size and duration of empires; growth-decline curves, 3000 to 600 BC. Social Science
Research
7
:180-196.
Taagepera, R. 1979. Size and duration of empires: growth-decline curves, 600 B.C. to 600 A.D. Social
Science History
3
:115-138.
Taagepera, R. 1997. Expansion and contraction patterns of large politices: context for Russia. International
Studies Quarterly
41
:475-504.
Turchin, P. 2003. Historical dynamics: why states rise and fall. Princeton University Press, Princeton, NJ.
Turchin, P., and T. D. Hall. 2003. Spatial synchrony among and within world-systems: insights from
theoretical ecology. Journal of World Systems Research
9
:37-64.
Wells, P. S. 2003. The Battle that Stopped Rome: Emperor Augustus, Arminius, and the Slaughter of
Legions in the Teutobutg Forest. W. W. Norton, New York.
Table 1.
The large historical states used in the analysis.
Date
(peak)
Empire name
World
Region
Area
(Mm
2
)
Latitude
Index
-1300 Egypt (New Kingdom) Africa
1.00
-1.292
350 Axum
Africa
1.25
0.241
969 Fatimid
Africa
4.10
0.782
1120 Almoravid
Africa
1.00
0.561
1200 Almohad
Africa
2.00
0.864
1380 Mali
Africa
1.10
0.512
1400 Mameluk
Africa
2.10
-0.225
1527 Inca
America
2.00
-1.139
-176 Hsiung-Nu (Hunnu)
Central Asia
9.00
0.818
405 Juan-Juan
Central
Asia
2.80
0.740
557 Turks
Central
Asia
6.00
1.026
800 Uigur
Central
Asia
3.10
0.213
800 Tufan (Tibet)
Central Asia
4.60
0.605
850 Khazar
Central
Asia
3.00
0.139
1100 Hsi-Hsia
Central
Asia
1.00
0.655
1210 Khorezm
Central
Asia
2.30
0.054
1210 Kara-Khitai
Central
Asia
1.50
0.362
1270 Mongol
Central
Asia
24.00
0.737
1310 Golden Horde
Central Asia
6.00
0.153
1350 Chagatai
Central
Asia
3.50
0.383
1405 Timurâs
Central
Asia
4.40
0.426
-1122 Shang
East
Asia
1.25
0.050
-50 China-Early
Han
East
Asia
6.00
0.661
579 Liang
East
Asia
1.30
0.137
715 China-Tang
East
Asia
5.40
0.375
947 Liao (Kitan)
East Asia
2.60
0.606
980 China-Sung
East
Asia
3.10
-0.164
1126 Jurchen (Chin)
East Asia
2.30
-0.147
1450 China-Ming
East
Asia
6.50
-0.138
1790 China-Manchu
East
Asia
14.70
0.246
117 Rome
Europe
5.00
0.204
441 Huns
(Atillaâs)
Europe
4.00
1.003
555
East
Roman Europe 2.70
0.516
814 Frankish
Europe
1.20
0.092
1000 Kiev
Europe
2.10
-0.132
1025 Byzantine
Europe
1.35
0.806
1480 Lithuania-Poland
Europe
1.10
0.079
1683 Ottoman
Europe
5.20
0.320
1895 Russia
Europe
22.80
0.303
1200 Srivijaya
Southeast
Asia
1.20
0.272
1290 Khmer
Southeast
Asia
1.00
-0.665
-250 Mauryan
South
Asia
5.00
0.191
200 Kushan
South
Asia
2.00
0.095
400 Gupta
South
Asia
3.50
-0.031
648 Harsha
(Kanyakubia) South Asia
1.00
0.668
1312 Delhi
South
Asia
3.20
-0.082
1690 Mughal
South
Asia
4.00
0.435
1760 Maratha
South
Asia
2.50
-0.280
-670 Assyria
Southwest
Asia
1.40
1.845
-585 Media
Southwest
Asia
2.80
0.141
-500 Achaemenid Persia
Southwest Asia
5.50
0.200
-323 Alexanderâs
Southwest
Asia
5.20
0.478
-301 Seleucid
Southwest
Asia
3.90
0.882
0 Parthia
Southwest
Asia
2.80
1.374
550 Sassanian Persia
Southwest Asia
3.50
0.292
750 Caliphate
Southwest
Asia
11.10
0.730
928 Samanid
Southwest
Asia
2.85
-0.194
980 Buyid (Buwahid)
Southwest Asia
1.60
0.142
1029 Ghaznavid
Southwest
Asia
3.40
0.689
1080 Seljuk
Southwest
Asia
3.90
0.409
1190 Ayyubids
Southwest
Asia
2.00
-0.300
1310 Il-Khan
Southwest
Asia
3.75
0.664
Table 2
Region Mean
SE
n
t
P
Africa 0.21
0.28
7
0.72
0.50
Central
Asia
0.48 0.08 13 5.80 <0.001
East Asia
0.18 0.10 9 1.71
0.12
Europe 0.35
0.12
9
2.92
0.02
South & SE Asia
0.07 0.13 9 0.51
0.62
Southwest
Asia
0.52 0.15 14 3.40
0.005
America
â1.14 â 1 â â
All
regions
0.31 0.06 62 4.83 <0.001
Figure 1. Distribution of world biomes (Ricklefs 2001).
Latitude Index
Fr
eq
uenc
y
1.5
1.0
0.5
0.0
-0.5
-1.0
18
16
14
12
10
8
6
4
2
0
Egypt
Inca
Khmer
Figure 2. Frequency distribution of the Latitude Index in the sample of large empires.