Byzantium: its internal history and relations with the Muslim World

Speros Vryonis Jr.


Byzantium and the Muslim World


XIV. Review Article

Basilike D. Papoulia, Ursprung und Wesen der “Knabenlese” im osmanischen Reich. Südosteuropaische Arbeiten 59 (Munich, 1963).

(In: Balkan Studies, V (1964)) 



The book of Miss Basilike Papoulia, concerned with the Turkish devshirme, is the most detailed scholarly treatment on the origins of this important Ottoman institution. Thus it is a welcome addition to the growing body of literature concerned with early Ottoman history. She has brought together from a wide variety of sources a considerable body of factual material that relates to the topic. Much of this material was previously known, but a significant portion was either little known or at any rate not specifically connected by scholars with the problem of the child tribute. The interesting results of the author’s research are presented in four chapters.


The first chapter deals with the general phenomenon of slavery in the political and military institutions of the Islamic world. Beginning with the assumption that the extensive use of slaves in the military and government was unique to the Islamic world, Miss Papoulia notes the appearance of military slavery from the time of the caliph Uthman (644-56), and comments briefly on the use of slaves by the Ummayids of Cordova, Abbasids, Samanids, Seljuks Ayyubids, Chorids, Ghaznevids, Ottomans, and Mamelukes. The rulers generally acquired these slaves as captives in military expeditions, by purchase, gift, and most interestingly as tribute from their provinces. Some attention is given to the contents of Nizam al-Mulk’s Siyasat-Nama as regards the systematization of the gulam system and the training of its members. Miss Papoulia accepts the contents of these passages in the Siyasat-Nama as being a reflection of actual conditions. In this she accepts the earlier view of Barthold which Bosworth (Der Islam, XXXVI, 4546) has rejected. This initial chapter closes with the observation that the slave institution remained closer to its original purpose in Ottoman society than it did amongst the Ayyubid and Mameluke dynasties. Under the Ottomans the system served the dynasty, but under the Ayyubids the slave system replaced the dynasty.


Chapter two describes the conditions and factors which led to the spread of the slave institution in the pre-Ottoman Islamic world, in





the Ottoman world, and also the appearance of the devshirme. In all three cases the basic factor was the need of the dynasty to obtain obedient military support against decentralizing and insubordinate forces. In the Ummayid period the spread of military slavery corresponded with the increase in tribal strife and with the thinning out of the Arab ruling class as it was spread out over an increasingly great area. Under the Ottomans, slaves were utilized against the internal foes of the dynasty, viz. the timar holders and the tribal groups. But that which differentiated the Ottoman system from the earlier Islamic practice was the devshirme. The Ottomans did not restrict themselves to obtaining slaves by capture, purchase, or gift, but also raised them in the form of a tribute taken from their Christian subjects. The author reviews the attempts of Palmer and Wittek to explain away the seeming incompatibility between the devshirme and the sharia. Palmer had attempted to do so by asserting that the Ottomans considered the taking of child tribute as a development of the pencik. Wittek’s explanation employs the Shafii principle which differentiated between those peoples already in the category of ahl al-kitab at the time of the prophet Muhammud, and those who became ahl al-kitab after the time of the prophet. According to Wittek only the former were legitimate dhimmis and therefore immune from the child tribute. In the Balkans only the Greeks and Jews belonged to this category, hence their children were not taken. But Papoulia rightly rejects Wittek’s explanation, as previously have other scholars who demonstrated that Greeks were taken in the devshirme. In addition she re-emphasizes the fact that the basic motive for the creation of the devshirme was not proselytizing zeal.


Chapter three concentrates on the rise of the slave institution and the devshirme during the fourteenth century, and it is this portion which constitutes the central core of the author’s thesis, namely that the Janissaries and the devshirme were instituted in the reign of Orhan I and the pencik was applied under Murad I. The discussion centers about the Byzantine and Ottoman sources, of which only some of the former are of the fourteenth century, the latter dating from the middle of the fifteenth century and after. Still the most important document on the devshirme is the sermon (1395) of Isidore Glabas edited by Basil Laourdas (1954) from a manuscript of the early fifteenth century. In addition reference is made by the author to Cantacuzenus and Symeon of Thessaloniki (between 1410-1429), who refer to the general captivity of Christians and their children. The author brings to our attention a new reference to the devshirme in the last quarter of the fourteenth





century (after 1387). The text is an unpublished Athonite manuscript (132, Dionysiou) dated variously to the sixteenth and seventeeth century, and implies that the system was by now taking considerable numbers of children. Thus the contemporary Greek sources bear testimony to the existence of thesdevshirme by the late fourteenth century.


The real difficulty, however, lies in the Ottoman sources, all of which were written after the mid-fifteenth century. According to Miss Papoulia, Idris al-Bitlisi (first decade of sixteenth century) sets the foundation of the Janissaries and the devshirme under Orhan, whereas Uruc (1460-70), Aşıkpaşazade (fl. 1490), and Anonymous Giese (c. 1490) present a different story. These latter authors relate that as a result of the suggestion of Kara Rustem and Kara Halil Candarlĭ, Murad ordered that onefifth of prisoners were to be given to the sultan. These youths were sent to Anatolia for a few years to labor and to learn Turkish. Then they were brought to the sultan, white bonnets were placed on their heads, they were called Janissaries, and, these chronicles continue, the Janissaries had their origin at this time (reign of Murad I)... “asĭl οἱ vakitden berü oldu.” But she prefers the testimony of Idris al-Bitlisi, and so places the beginning of the Janissaries and devshirme in the reign of Orhan, prior to 1330-31. This of course contradicts the assertions of other recent scholarship on the subject (Ménage, Giese, etc.) and reverts to a view expressed by western scholars as early as the time of von Hammer. Her choice she justifies on the basis of the following considerations:


a) Contemporary sources bear testimony to the existence of the devshirme by the last quarter of the fourtheenth century.


b) The version of Idris is a much better account for the rise of the slave institution than is the version of the other chronicles.


c) The chonicles do not mention the devshirme at all.


d) The version of Idris is the difficilior.


Miss Papoulia realizes, however, that it would be difficult to reject in toto the testimony of the other chronicles. So she accepts most of their testimony, rejecting only the final phrase “ol vakitden beri adïnï yeniçeri qodïlar” (and other variants of the phrase). This latter, she asserts, was added by the chroniclers because unlike Idris they were not really aware of the facts and simply felt that the Janissary corps was instituted at the same time as the pencik. Her conclusions then are: Creation of the Janissaries and devshirme under Orhan I, and application of the pencik under Murad I.


It has long been pointed out that the best approach to the investigation





and consideration of early Ottoman institutions is that which begins with the institutions of the Anatolian Seljuks and the beyliks. It would, perhaps, have been of some advantage had the author considered the slave institution of this period in detail. After concentrating on the provisions of the Siyasat-Nama (eleventh century) she skips to the early Ottoman period, dispensing with the Seljuks of Rum in one footnote, stating merely that they too had a slave institution (p. 17, 33). She implies that the Ottomans may have had before them the example of the Siyasat-Nama as an inspiration for the implementation of their own slave system. But it is even more likely that it was the traditions of the Seljuks of Rum which influenced them in a more direct manner. In fact the use of gulams on an extensive scale in the military and administrative system of the Seljuks in Anatolia provides the immediate ancestry of the Ottoman slave institution. The principal documentation for the existence and shape of the slave system in thirteenth century Anatolia is the Persian chronicle of Ibn Bibi, recently translated by H. Duda, Die Seltschukengeschichte des Ibn Bibi (Copenhagen, 1959). This is supplemented by the Persian chronicle of Kerimuddin Mahmud, the Arabic inscriptions, vakif documents, etc. There also exist decriptions of modern scholars. I. Uzunçarşĭlĭ, Osmanlĭ Devleti Teşkilatĭna medhal (Instanbul, 1941), a work which Miss Papoulia did not mention, gives a history of the general development of the slave institution from the time of the Abbasids and includes a chapter on the Seljuks of Anatolia. These sources, and primarily Ibn Bibi, reveal a full grown and well developed system of slave troops and administrators based on the gulams. The slave basis of the palace service, central and provincial administration is clearly manifested.Among such posts and services which are mentioned as directed and or manned by gulams are the following: royal stables, royal wardrobe and cleaning service, treasury, secret archives of the divan, corps of iğdiş, justice, sultan’s table and pantry, sultan’s bodyguard, royal kitchen, office of translations, office of the sahib tugrai, harem, naib of Konya, various provincial governments, atabey, and possibly the court musicians. Bodies of slave troops are mentioned, as for instance the 1,000 gulams sent to garrison the town of Chliat. The size of this garrison, as also the frequency with which troops of gulams appear in the pages of Ibn Bibi, indicate that the slave institution must have been of comparatively significant size. A further indication of the importance and developed character of the slave institution amongst the Seljuks of Rum are the following facts. Important officials had their own private corps of





gulams, a fact which made them powerful above and beyond their own official positions. Λ palace school, the gulamhane, is mentioned where the gulams were turned over to the babas for their education (Duda, 242, 330), after which they were often turned over to the various palace services for further training under the direction of the older members of the slave institution. There is enough material to trace the histories of a few of the more important gulams, such as Celaluddin Karatay, Aminuddin Mikail, and Hass Oguz. These three individuals, who were of Greek origin, rose to high positions in the state, and contributed significantly to the cultural and political life of the Seljuk state (Extremely important in the history of the Seljuk gulams are the vakif materials edited by O. Turan, “Semseddin Altun aba vakfiyesi ve hayatı.” Belleten, XI (1947), 197, ff; “Mübarizeddin Er-Tokuş ve vakfiyesi,” Belleten, XI (1947), 415 ff; “Celaleddin Karatay vakıfları ve vakfiyeleri,” Belleten, XII (1948), 17 ff.) These gulams came to the state through the customary methods; from the ganimat or law of one-fifth, as hostages, as ‘gifts’ from the begs of the borders, etc. Ibn Bibi relates that most of the important emirs in the thirteenth century were gulams taken from the dar ul-harb (Duda, 328). However the question has been raised whether there was not also a form of child tribute amongst the Seljuks in Anatolia which was similar to the Ottoman devshirme. The Turkish scholars O. Turan [“L’Islamisation dans la Turquie du moyen age” Studia Islamica, I (1959), 147-150; Türkiye Selçukluları hakkında resmi vesikalar (Ankara, 1958), 178] and I. Uzunçarşĭlĭ [Osmanlĭ devleti teşkilatĭna medhal (İstanbul, 1941), 115-116], have called attention to the existence of the igdiş. This was probably a military corps which was composed of Islamized youths taken from their Christian parents within the Seljuk domains. Thus, if the conjecture of these Turkish scholars is correct, as seems highly likely, something similar to the devshirme was already in existence amongst the Seljuks of Rum. There are also a few scattered references to the actual taking of children from the Christian subjects of the Turks in Anatolia during this early period which would tend to reinforce the suggestion that something similar to the devshirme existed previous to the Ottomans. Inasmuch as these cases have not, as far as the reviewer is aware, been discussed in connection with the devshirme it may not be out of place to mention them here. At the end of the eleventh century the Turks took the male children of the Greek towns between Dorylaeum and Iconium [Gesta Francorum et aliorum Hierosolimitanorum, ed. L. Brehier (Paris, 1924), 55; Tudebodus, Receuil des historiens des Croisades, Hist. Oc., III (Paris, 1866), 29,





“Adhuc quoque et Christianorum filios secum tolerabant.” Tudebodus, 26, mentions ‘Angulani’ in the armies of the Turks opposing the Crusaders in western Asia Minor. These Angulani are quite possibly the gulams]. In the same period the Turkish emir of Samosata had likewise taken children from the Christian inhabitans of the city (William of Tyre, IV, 4). At Antioch, after 1085, the Turks were short of manpower and so made use of Armenian and Greek youths whom they had, evidently, forcibly converted to Islam (Raymond of Aguilers, R.H.C., H.O., III, 250-251. “Quoniam Turei ante annos quatuordecim Antiochiae obtinuerant, atque Armenios juvenes et Graecos quasi pro penuria domesticorum turcaverant, et uxores eis dederant.” See also “Tureare” in Ducange, where he quotes again from Raymond; “Quidam de Turcatis, qui erat in eivitate, per Boimundum principibus mandavit nostris etc.” Also the following interesting passage. “Si qui per Dei gratiam contempsissent, cogebantur tradere pulchros parvulos suos ad circumcidendum vel Tureandum.” Nicephorus Gregoras, I, 58, refers to the recruitment of military corps of Greek Christians (in contrast to gulams) from those inhabiting the Seljuk state, «ἐπεὶ γὰρ ἦσαν ὑπ᾿ αὐτῷ πάλαι δεδουλωμένοι Ῥωμαίων συχνοί, τούτους εἰς μοῖραν καταλέξας στρατοῦ...»).


A consideration of the Seljuk slave institution in pre-Ottoman Anatolia seems to suggest: (a) That the Ottoman system was directly inspired by that of the Seljuks of Rum. (b) There is great probability that the levying of Christian children from amongst the subject population by the Muslim rulers was practiced in Anatolia between the eleventh ant thirteenth centuries.


The section with which the reviewer finds himself least able to agree is chapter III, in which the author analyzes the Ottoman sources. She attempts to demonstrate that the narrative of Idris in regards to early Ottoman institutions is more reliable than that of Uruc, Aşĭkpaşazade, and Anonymous Giese. Inasmuch as this is critical to the entire thesis, one would have expected a more rigorous and thorough proof of Idris’ accuracy in institutional matters. However Mis Papoulia has failed to demonstrate this in a convincing manner. Relying on the judgement of Palmer (Bulletin of the John Rylands Library, 1952-53, 473) and Sükrü (Der Islam, 1931, 149 ίΐ.) she asserts that Idris contains material not to be found in the other chroniclers and therefore we cannot reject lightly his contention that the Janissaries and the devshirme were introduced in the reign of Orhan. However, in the creation of the Janissaries we are dealing with an event which is not recorded exclusively in Idris, but one which is chronicled by the other sources as well.





And these latter contradict the account of Idris and agree on the matter of chronology with one another. Secondly, inasmuch as the Idrisian text is so important to her thesis, one would have expected a more detailed examination of Idris. Instead, the reader is referred at one time to the summary of Palmer (based on Br. Museum Add. Mss 7646-7), at another time to the edition and translation of the relevent section by V. Ménage (based on the above manuscript and on Bodleian, Ousely 358, in Bulletin of the School of Oriental and African Studies, 1956, 1813), and at still other times to a third manuscript (Westdeutsch Bibliothek, Marburg, Orient folio Nr. 3179, dated 1560-61), without any specific pagination in the last case. The summary of Palmer and the translation of Ménage contain so many differences in content that these must be explained, as for instance in the following. The Palmer ‘version’ refers to the collected Christian youths as Janissaries, whereas the Ménage ‘version’ does not call them Janissaries, nor does this latter version attribute the innovation to Kara Halil Candarlï, at least not in the section edited by Ménage.


It is interesting that both Palmer and Ménage reject the account of Idris, wherein the Janissaries and devshirme are attributed to the reign of Orhan, as anachronistic. This view is derived from the fact that Aşĭkpaşazade, Anonymous Giese, and Uruc (who wrote their accounts earlier than did Idris) agree in dating the institution of the Janissaries in the reign of Murad. This is possibly strengthened by the work of Mehmed Nešri, Kitâb-i cihan-nümâ, ed. F. Unat and M. Köymen, I (Ankara, 1949), 197-9, which likewise attributes the creation of the Janissaries to the reign of Murad I and adds a few new details.


In the realization that the unity of the chronicles (that is the chronicles exclusive of Idris) constitutes a difficulty in rejecting their account in toto, the author attempts to solve her dilemma by deleting the passage in each of these chronicles which states that ‘the inception of the Janissary corps dates from that time’ (reign of Murad). By so doing she accepts everything that they have to say in regards to the taking of one out of every five prisoners and their training for and induction into the Janissary corps at the suggestion of Kara Halil Candarlĭ. But, as the Janissary corps and devshirme existed in the reign of Orhan I, she continues, the chronicles have confused these two institutions with that of the one-fifth levy of prisoners of war implemented under Murad I.


The reviewer really sees no justification for the deletion of the phrase “asĭlda yeniçerinin οἱ vakitden berü oldu,” (or variant thereof) which occurs at the end of the accounts of the four chronicles dating





the Janissaries to the reign of Murad I. If one proceded on the premise of the author that such deletions are allowable because the texts are confusing, the Pandora’s box of deletions would be flung open to the whims of any particular point of view. This phrase could legitimately form an integral part of any text dealing with the history of institutions and does not necessarily constitute an awkward addition on the part of the chroniclers. Anonymous Giese contains the same type phrase after the institution of the yaya in the reign of Orhan, stating that this body date from the time of Orhan (Anonymous Giese, p. 14 text, p. 22 translation).


The author sees a possible support for her thesis in the fact that Ibn Batuta upon visiting Antalya in 1333 noted that the emir had a body of mamelukes. She remarks, p. 79;


“Interessant ist die Angabe insoweit, als ein anderen Kleinasiatischer Emir schon im Jahr 1333 eine derartige Garde besass. Es fragt sich also, warum wir nicht dasselbe fur die Osmanen annehmen sollten, wenn es von einer wichtigen Quelle behauptet wird?”


Here there seems to be a confusion on the part of the author between the usual slave troops and administrators (gulam, kul, bende, mameluke) and the Janissary corps raised by the devshirme or by the pencik. As the name implies, Janissaries were a new troop of slave soldiers, and it does not seem extreme to assume that the Ottomans had slave bodies of sorts before the creation of the Janissary corps, just as did the ruler of Antalya.In fact Ibn Batuta also saw gulams or mamelukes in the retinues of the Muslim rulers of Pirghi, Castamon, Sinope, and Kayseri (Ibn Batuta, ed., tr. Defremery and Sanguinetti, II, 307, 309, 310-311, 317. In his audience with the emir of Pirghi he saw in the reception hall twenty Greek pages clothed in silk and with strange coiffures). Al-Umari (ed. F. Taeschner, 25) notes their presence in the service of the Germiyanids. In fact there is evidence of sorts that such slaves existed already under both Osman and Orhan, prior to the appearance of the Janissaries. When Osman carried on his sieges of the Bithynian cities, he entrusted part of the operations against Bursa to his slave (kul) Balabancik (Anonymous Giese, p. 20 tr., p. 12 text). By the reign of Orhan the Ottoman rulers had a well defined personal troop of slaves (Uruc, ed. Babinger, p. 15, kullar; Nešri, p. 155, has bendelerü). Though the sources are distant from the events which they describe some 150 years, the testimony of Ibn Batuta as to the presence of such kul amongst other emirates tends to confirm the statement of these chronicles. The probability of the existence of a slave institution prior to the Janissaries amongst the Ottomans





has already been pointed out by others, viz. Palmer, Gibb and Bowen, etc.


Thus kuls certainly existed in Ottoman institutions before the appearance of the Janissaries and the devshirme. In addition one is not as yet convinced of the superiority of Idris over the other Ottoman chronicles in relation to the origins of the Janissaries and devshirme. The chronicles (Idris excepted) attribute the founding of the Janissaries to the reign of Murad I. There is no specific mention of the devshirme prior to 1387 and possibly not until 1395. But the accounts of Neşri, Uruc, Aşĭkpaşazade, and Anonymous Giese describe the collection of the one fifth under Murad as follows: “oğlanlar cemi’ idüb,” (Nešri, 199); “oğlanlar cem’ olĭndĭ.” ((Aşĭkpaşazade, ed. Atsĭz, 128); “oğlanlar devşirdiler,” (Anonymous Giese, 22 text); “oğlan devşirdiler,” (Uruc, 22). All four texts state that “they collected the youths,” or, ‘the youths were collected.’ In two cases the actual verb used derives from devshirmek, and in all cases the passages referred to youths. Devshirme in the sixteenth century seems to have become a definite terminus technicus referring to the levy of the children of Christian subjects. But did it mean this exclusively in the fourteenth century when the practice arose, did it also include the oğlanlar taken from the captives in war? It is quite possible, though the evidence is certainly not conclusive, that as Palmer has suggested we have a reference to the devshirme here as it evolved from the suggestion of Kara Halil Candarlĭ. In short, the reviewer finds the view that the Janissaries and devshirme were instituted in the reign of Murad I somewhat more consistant with the sources.


However, given the nature of the sources it would seem that any position one takes (as to the dating of these institutions) is fraught with difficulties. Certainly the reviewer does not wish to imply that the publication of Miss Papoulias is without merit. She has brought together, for the first time, most of the important material which bears on the subject. In addition she has pointed out that the Ottoman slave institution has its roots in a tradition going back as far as Ummayid times. It is to be hoped that the author will continue her researches on the later history of the devshirme and the Janissaries, subjects still awaiting research and clarification.


University of California

Los Angeles


[Previous] [Next]

[Back to Index]