JOST A MON

This is the sort of interdisciplinary thing that I find fascinating. Murat Iyigun of the University of Colorado has written a paper titled Lessons from the Ottoman Harem (On Ethnicity, Religion and War), in which he brings several ostensibly disparate disciplines into one in his work. Politics, economics, history and statistics are blended with considerable panache.

The Ottomans, a Turkic people who, under the leadership of Osman, conquered much of Anatolia and established one of the great empires of the world, were also inveterate keepers of records. This treasure trove is available to researchers in all sorts of areas. But Dr Iyigun picks up some of the information he requires from the Conflict Catalog, a comprehensive dataset of violent conflicts in all regions of the world from the 1400s to date, and data on the imperial family from such sources as Turkish Wikipedia.

There appear to be two not quite mutually exclusive theories to explain Ottoman conflicts with their neighbours. The Gaza theory states that religion was a driving force for Ottoman bellicosity, leading them to initiate war against Europe, whereas eastern conflicts were started by others, their co-religionists for the main part. What fueled the energies of the early Ottoman conquerors was essentially their commitment to Gaza, and 'ideology of Holy War' in the name of Islam. Ottoman power was built on that commitment... (Paul Wittek.) Another theory is that as the Ottomans became a multiethnic empire, the Imperial Harem wielded considerable influence on the Emperor and determined to a large extent the direction of his wars. Various historians averred that members of the Harem with different ethnic or religious backgrounds often lobbied the Sultan to influence the geography of Ottoman conquests.

Dr Iyigun finds that the Gaza theory was an important but not sufficient cause for Ottoman war, but that the ethnicity of the Valide Sultan was an important and independent determinant of whether the Empire engaged in military conquests in Europe versus North Africa or the Middle East.

It appears that conversions into Islam, even among the elite in the Harem who were influential in policy matters, was insufficient to promote the cause of holy war. The ethnic roots seemed to dominate over the religious, and subverted the Sultans' ambitions toward the Middle East and North Africa.

A look at the genealogy of the Ottomans shows the varied maternal ancestry of many of them. [Source: Shaw, Peirce, Turk Vikipedi, http://turkboard.com]

The Conflict Catalog comprises violent conflicts on five continents, with Richardson ratings exceeding 1.5 (a logarithmic scale, such that a value of n implies 2ⁿ fatalities). The number and identities of the participants in each listed conflict, the common name for it, and its date and geographical location are the primary data in the catalog. The duration of the conflict, and the number of fatalities are also available for about a third of the sample.

The crux of the paper is the following linear regression, used to estimate the impact of ethnic identities on Ottoman military operations.

OTTOWAR_t = L₀ + L₁ EUROMOM_t + L₂ X_t + e_t

where OTTOWAR is one of four alternative dependent variables, EUROMOM is a dummy variable for whether the sultans had a European maternal link, and X is a control variable including trend (TIME), the lagged variable OTTOWAR_t-1, estimates of Ottoman and European populations (OTTOPOP_t, EUROPOP_t), and an indicator variable for each of the three centuries (CENTURY_t). 300 annual observations in the period 1401-1700 are used in the regression.

What are these four alternative dependent variables? 1) the number of newly initiated conflicts between the Ottoman Empire and European powers at time t, OTTOMAN_t; 2) the count at time t of the newly initiated number of Ottoman conflicts with its non-European neighbours, OTHEROTTOMAN_t; 3) the aggregate number of conflicts the Ottoman Empire had with continental Europeans at time t, both that began at t and began earlier, AGOTTO_t; 4) the aggregate number of Ottoman conflicts with non-European foes, AGOTHER_t.

The reason for including 3) and 4), of course, is that medieval warfare was a seasonal affair: hostilities would be postponed at the onset of winter. Therefore, all unresolved military confrontations were renewed every year. Also, Dr Iyigun wanted to know if matrilineal ancestry determined not only the short-running wars, but also the longer ones.

What conclusions can be derived if we assume that the Valide Sultan's ancestry mattered for the patterns of Ottoman conquest? Well, we would expect L₁ to be negative and statistically significant when 1) and 3) are the dependent variables; and L₁ should be positive and significant, or insignificant and of any sign, if 2) and 4) are dependent variables.

Dr Iyigun uses other control variables as well, e.g. the Sultan's age at ascendancy to the Sublime Porte, the length of his reign, and whether he reigned before or after the great military defeat of the Ottoman navy at Lepanto. For the most parsimonious specification, though, he finds indeed that L₁ is negative and statistically significant at the 5% level. Hence, statistically at least, the sultan's tie to Europe via his mother reduced his military ventures in Europe by more than 70%. In fact, a subsequent regression implies that having a Venetian mother resulted in almost no aggression towards Europe on the part of the Sultan concerned. Meanwhile, on the other front, there is almost no relation between the maternal ancestry of a Sultan and his conflicts with his Oriental neighbours. In fact, having a European mum didn't affect an Ottoman's fights with his coreligionists to his east. A truly remarkable finding.

Couple of minor points: the older a Sultan was on ascending the throne, the likelier he was to engage the Europeans rather than other Muslims in battle. Likewise, the longer he reigned ,the less likely he was to fight on his Eastern fronts.

Dr Iyigun addresses all sorts of subtleties in his paper. One interesting point: considering that the average age of ascension was 22 years, the reason that the younger Sultans were less likely to engage the Europeans in battle could have been because their mothers had been absorbed into the Harem two decades earlier upon a previous Ottoman victory. By a suitable lagged variable in his regression, Dr Iyigun is able to show that this effect is negligible.

A very enjoyable paper, indeed.

References

1. Shaw, S (1976). History of the Ottoman Empire and Modern Turkey, Vol 1, (Cambridge University Press).

2. Peirce, L. P. (1993). The Imperial Harem: Women and Sovereignty in the Ottoman Empire (Oxford University Press).