Hostname: page-component-cd9895bd7-7cvxr Total loading time: 0 Render date: 2024-12-26T18:53:15.143Z Has data issue: false hasContentIssue false
  • English
  • Français

Some Notes on the Devshirme

Published online by Cambridge University Press:  24 December 2009

V. L. Ménage
Get access Rights & Permissions [Opens in a new window]

Extract

No Ottoman institution has aroused more bitter criticism than the devshirme the ‘tribute of blood’, especially—and naturally—among those Christian peoples whose forebears were subjected to it; and at the same time none touches on so many fundamental problems. Dr. Basilike Papoulia, bringing together the Western and oriental sources and submitting them to a close and careful analysis, has now presented the first full discussion of its origin and its character. The scope of her study appears already in the definition with which she begins and ends it, that the devshirme devshirme was ‘the forcible removal, in the form of a tribute, of children of the Christian subjects from their ethnic, religious, and cultural environment and their transplantation into the Turkish-Islamic environment with the aim of employing them in the service of the Palace, the army, and the state, whereby they were on the one hand to serve the Sulṭān as slaves and freedmen and on the other to form the ruling class of the State’.

Type
Articles
Information
Bulletin of the School of Oriental and African Studies , Volume 29 , Issue 1 , February 1966 , pp. 64 - 78
Copyright
Copyright © School of Oriental and African Studies, University of London 1966

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Papoulia, Basilike D.: Unsprung and Wesen der ‘Knabenlese’im osmanischen Reich. (Südosteuropäische Arbeiten, 59.) X, 139 pp. + errata slip, plate. München: Verlag R. Oldenbourg, 1963. DM. 15.Google Scholar

2 He speaks of a of the ruler for

3 Vryonis, S., Jr., ‘Isidore Glabas and the Turkish devshirme’, Speculum, XXXI, 1956, 433–43.CrossRefGoogle Scholar

4 Balkan Studies (Thessaloniki), V, 1964, 145–53.Google Scholar

5 Sir Gerard Clauson suggests to me that the word is a deverbal noun in ikit-, ‘to rear, bring up’, and so means primarily ‘reared, [home-]bred’. We may perhaps compare : al-'alūfa min al-ḥayawān ‘hand-reared animal’, (Codex Cumanicus) ikti: domesticus ‘tame’, Old Ottoman (TTS, s.v. ekti, definition 2) and provincial Anatolian (SDD, s.v. ekti) ikti/ekti: ‘tame’. Sir Gerard also kindly refers me to the earliest attestations of the word, in the bilig: verse 1554 (ed. R. R. Arat), ‘I am your servant brought up in your household’ (the speaker is the son of the ruler's recently-deceased adviser Ay-Toldi, and certainly not a bought or captured slave); V. 5590, taki yilkiči üklitsüni ‘and let the herdsman increase the home-bred stock’; and verses 4439 ff., speaking of the duties of the kamug yilkilarka bular ‘it is these who are in charge of all the flocks’.

From this meaning it is only a short step to the meanings noted by I. H. Uztuiçarşih (Osmanli devleti teṣkilcitina medhal, Istanbul, 1941, p. 115, n. 3) as attested by the thirteenth-century Ibn Muhannā (ed. Melioransky, 51, 10; ed. Kilisli Rif'at, 147), al-muwallad ‘child born of one brought up in the Muslim world’ (see W. Heffening, in Enc. Islam, first ed., s.v.) and by the early fourteenth-century Waṣṣīf, ‘half-Turkish’. In the thirteenth-century MS of Maḥmūd (facsimile, p. 611) one reads clearly igdish or (as interpreted by Besim Atalay) igdish with the translation banū ‘uterine brothers’; but the context, a proverb contrasting the behaviour of half-brothers having a common father with that of half-brothers having a common mother, strongly suggests that the copyist may have been misled by the meaning hybrid, ‘half-breed’ which igdish bore in his day. The eighteenth-century dictionary of Turkish, the (ed. Sir Gerard Clauson (GMS, NS, XX), 1960, fol. 108v, line 28), gives two forms, and the definition ‘horse of mixed blood’. The Ottoman meaning ‘gelding’ (Redhouse, with the variants igdish igdič—and also the doublet igdish [Persian] ‘hybrid’) seems to be modern; it developed, perhaps, from the idea of docility.

Either of the contemporary meanings attested by Ibn Muhannā and Waṣṣāf fits a corps raised from the sons of of men brought into the Islamic world where they married local girls. Such a Muslim-born second generation represented something of a problem in a Muslim state and various solutions were found in order to provide for them; with all reserve I suggest that the igdish are to be compared with the awlād al-nās in the Mamluk state (Ayalon, D., in BSOAS, XV, 3, 1953, 456 ff., and in Enc. Islam, second ed., s.v.) and the müteferriqa (M. Tayyib Gökbilgin, in IA, s.v., especially the second paragraph) under the Ottomans.Google Scholar

6 Present state, Bk. I, ch. iii.

7 Published by Turan, Ṣ. in Belleten, XXVI, 103, 1962, 539–55, at p. 547.Google Scholar

8 See, for example, in İnakik, H., Sûret-i defter-i sancak-i Arvanid, Ankara, 1954, the references to sipāhis who, through blindness or old age, kullukdan yaramaz (tīmārs 132, 333); the first of these is called but the second is one of the sürglin, presumably free-born Turks, from the term for a sipāhi who fails to report for duty is kulluk etmez (tīmār 159).Google Scholar

9 For the technical meaning of this phrase see Uzuncarph, İ. H., Kapzkulu ocaklari, I, Ankara, 1943, 115 ff.Google Scholar

10 Barkan, Ö. L., Kanunlar, Istanbul, 1943, p. 306, £ 12, and cf. p. 381, £ 38.Google Scholar

11 That all three terms kulluk, , and might, however, be practically synonymous is nicely illustrated in a report of Dāwūd of c. 1482 (Arṣiv Ktlavuzu, Istanbul, 1940, pl. XVII, lines 13–14); some members of the house having submitted, Dāwūd had proposed to send them to the Porte, but they volunteered to join his forces: …āsitāne-i devlete īṣāl etdük-idi.Ḥāliyā maḥallinde gitmezüz; elden geldükče A'azz Allāhu anṣārahu yolinda ve kulluk edelüm ’ deya bunda

12 Published by Inalcik, Halil, in Fatih devri üzerinde tetkikler ye vesikalar, I, Ankara, 1954, 204–Google Scholar15 and plates III–V; and again by Uzuncarph, I. H., in Vaktflar Dergisi, IV, 1958, 1–17.Google Scholar

13 A grant of manumission with effect from the master's death is known as tadbīr (see Enc. Islam, second ed., art. ‘'Abd’, by R. Brunschvig, col. 30a; D'Ohsson, M., Tableau général…, second ed., VI, Paris, 1824,Google Scholar 31 ff.; Juynboll, T. W., Handbuch des islamischen Gesetzes, Leiden and Leipzig, 1910, 206).Google Scholar Here, however, it is stipulated that the manumission takes effect not at the testator's death but as from 40 days before his death-sickness (maraḍ al-mawt). A slave freed at death (mudabbar) formed part of the estate, and hence complications might arise in that a testator could dispose of only one-third of his estate, the rest going to his heirs according to the Qur'anic provisions (Juynboll, Handbuch, 206–7; Schacht, J., An introduction to Islamic law, Oxford, 1964, 169–74);Google Scholar again, a gift made during death-sickness was treated as a legacy (Enc. Islam, first ed., art. ‘Wasiya’, by J. Schacht; idem, Introduction, 174; Fyzee, A. A. A., Outlines of Muhammadan law, third ed., Oxford, 1964, 363 ff.); but the provision ‘40 days before my death-sickness’ would presumably remove altogether the risk that the grant of manumission, interpreted as a bequest, should be hindered in the event that the testator had exceeded the one-third limit of his property that he was able to dispose of by will.Google Scholar

It is a point of some interest that the wording of this testament (kullu 'abdin amlakuhu al-ān …/āzādsuz kullarimdan …) shows clearly that the members of the ‘slave-institution’ were regarded as belonging not to ‘the Crown’ impersonally, but to the Sulṭān personally, and (as a corollary) were inherited by his son and successor.

14 Like kul, the word āzād must be interpreted with caution. It may bear the neutral meaning ‘released (from prison)’, as in ferāḥindan āzād eyledi, Tokat ḥiṣārinug ḥabsinden čilcartdi, and Fekete, L., Die Siyāgat-Schrift, Budapest, 1955,Google Scholar doc. 3; but in other contexts āzād et- means, more precisely, ‘to manumit’: see for example Elezovie, G., Turski spomenici, I, 2, Belgrade, 1952, p. 87,Google Scholar a letter of Hersek-zāde Aḥmed to Ragusan ambassadors who are claiming the return of an enslaved lad: biz ol [sic] āzād etdüb-dürüz; Müsülmān oldi…; cf. also Barkan, , Kanunlar, p. 102, £ 47.Google Scholar

15 Also published by H. Inalcik, Fatih devri…, 215–17 and p1. VI.

16 Three are Serbians, two Saxons (sāsī), four Bosnians, five Albanians, and one Vlach; five are described as and four as čakirji (two groups of falconers, see Enc. Islam, second ed.),

17 Juynboll, Handbuch, 206: ‘Solange der Herr lebt, ist die rechtliche Stellung eines solchen Sklaven durchaus dieselbe, wie die anderer Sklaven’.

18 kamā [so to be read, not] 'l-'abdi'l-muṭlag.

19 Bodleian Library MS Marsh 313, fol. I33r: fetḥ, u ẓafere muṭṭali' olub ol seferde olan kullarin āzād etdi.

20 See Minorsky, V. (tr.), Calligraphers and painters: a treatise by Qāḍī Aḥmctd…. (Freer Gallery of Art Occasional Papers, III, 2), Washington, 1959, 85.Google Scholar

21 This seems to be the interpretation of Münejjim, III, 391, for he mentions slave-girls as well: ne kadar kul ve jāriyeye mālik iseler jümlesini i'taq buyurdilar; and cf. Hammer, , GOR, II, 122: ‘gab er… alien seinen Sclaven and Sclavinnen die Freyheit’.Google Scholar

22 , MS Nuruosmaniye 3209 (fair-copy autograph), fol. 468V:

23 British Museum MS Add. 7869, fol. 201r: ḥattā mervīdür ki bu bir günde olub kirk bin kulin āzād eyledi; Sa'd a1-Dīn, I, 542: bir kelime-i jāmi'a ile 'aẓbqa-i nāfi ‘a mebde’ olub ‘jemī'-i 'abid ye memālikim aḥrār cdsunlar’ demekte hisāb olindukda kirk bin raqabe ribqa-i riqqdan āzād oldi.

24 ‘Breve narratione della vita… del signor Ussuncassano’, in Ramusio, , Navigationi…, II, 1559Google Scholar, 66r–78r, at 69r: ‘Hauuta questa rotta it Turco dubitò fortamente, & deliberò di ridurre it suo essercito per la piu corta nel suo paese, & per confortar li suoi soldati, oltra it soldo ordinario dette un' altra prestanza, & donò la prima, che haueua data alla sua partita: & fece anche liberi tutti li suoi schiaui, cbe si trouauano in campo, con questa conditione, che niuno fusse in libertà di abbandonarlo, ma fussero huomini del Signore, come gli altri stipendarij, che non sono schiaui, & posson fare della for robba quel che for piace: & fece molte altre prouisioni carezzando & donando alli Capitani’. (Cf. the very similar account incorporated in Lezze, Donado da, Historia turchesca, ed. Ursa, I., Bucharest, 1909, 54.) The remitted ‘earlier loan’ to which Angiolello refers was an advance of ‘one quarter's pay’ (to the kapu and a ‘subvention to the timariots’ made at the beginning of the campaign Ursu, 49). This detail also is confirmed by the Oriental sources (who say that the sum involved was 100 yüks = 10 million akčes), though they represent this as being, like the manumission, a gesture of thanksgiving after final victory (Idris, Sa'd al-Dīn, Hammer, loc. cit.).Google Scholar

25 I note only that what Angiolello calls ‘the condition that none should be at liberty to desert him but that they should be the Sulṭān's “men” like the other paid troops who are not slaves’ presumably refers to the bond of clientage (walā'). Dr. Papoulia is evidently right in seeing this as an important element in the Ottoman ‘ruling institution’, but the question is when, and to what extent, the master-slave relationship was modified to that of patron and client.

26 See Ayalon, D., L'esclavage du Mamelouk, Jerusalem, 1951, 17. If the Ottoman slave-institution was in some features consciously based on the Mamluk system, it may be that this striking difference, the absence among the Ottomans of a wholesale manumission, was a conscious ‘improvement’, aimed at eliminating the intense group-loyalties which in Egypt so easily led to faction.Google Scholar

27 p. 6, n. 15: ‘Es ist offensichtlich, dass das “eiqma” ein staatsrechtlicher Begriff war mit einer speziellen Bedeutung bezüglich der zukünftigen Stellung dieser Sklaven. Es handelte sich um einen Staatsakt, für den die Erlaubnis des Sultans erforderlich war…’. Long before the Ottomans came on the scene, the doctrines of servitude and manumission had been fully elaborated by the , so that (in principle, at least) no scope remained for the operation of 'ōrf. Certainly the permission of the Sulṭān was required when a čikma was held; but he was acting (I suggest) not as ‘Head of the State’ but as the personal owner of the slaves involved.

28 D'Ohsson, M., Tableau général…,VI, 18–23; Enc. Islam, first ed., s.v. idhn, by T. W. Juvnboll.Google Scholar

29 Any officer of the state held his position by virtue of a berāt. In the case of a slave this would presumably, in law, amount to a delegation of idhn.

30 P. Wittek, , BSOAS, XVII, 2, 1955, 271–8.

31 Islamic society and the West, I, 2, London, 1957, 223.Google Scholar

32 Quoted in V. L. Ménage, ‘Sidelights on the …’ BSOAS, XVIII, 1, 1956, 181–3.

33 Supplement to TOEM, nos. 14–15, p. 27.

34 Enc. Islam, second ed., art. , by Chafik Chehata; J. Schacht, Introduction, 121 f.

35 Tevárih-i Âl-i Osman, VII. defter, ed. S. Turan, facs. (Ankara, 1954), 10 = transcription (Ankara, 1957), 9.

36 Juynboll, Handbuch, 52 f.; idem, in Enc. Islam, first ed., s.v.; J. Schacht, Introduction, 60–2.

37 The must have been unpopular with landholders, for whom it entailed the loss of hardy peasants; thus in an early sixteenth-century document (Uzunçarşili, I, 92–4) punishment is threatened to anyone who hides away a lad ‘in his tīmār or in his house or in his village’: the first of these can be directed only at the sipāilvīs.

38 pp. 54–5 and n. 33. The document is transcribed in Uzunçarşili, I, 114.

39 ḥimāyetleri ziyāde lāzim olmakla.

40 Uzunçarşili, I, 99, last para., and 97, second para.

41 Palmer, J. A. B., ‘The origin of the Janissaries’, Bull. of the John Rylands Library, XXXV, 2, 1953, 448–81, at p. 448.CrossRefGoogle Scholar

42 Except that Dr. Papoulia sees here (pp. 76–7) two separate episodes, the Gallipoli levy being an ‘economic measure’ and the Evrenos levy a recruitment of men. This interpretation is possible, but unlikely. Gallipoli, the main crossing-point from Rūmeli into Anatolia, is the obvious place for ‘customs control’, where dues may be levied from the sipāhīs of Anatolia as they return to sell their booty, human and otherwise, in the markets of Bursa before dispersing to their fiefs; but the of Rūmeli have no occasion to pass through here, and so the powerful Evrenos is given the thankless task of collecting penjik from them. In later years, when great raids were made across the Danube, the penjik could conveniently be levied at the crossing-place on the river, but even so the control was not easy: in describing a raid 'Ali Beg, Ibn Kemāl, to emphasize the great number of prisoners taken, says (facs., 410): ‘at the crossingplace [on the Danube] the paid penjik to the treasury on 32,000 prisoners, apart from those [evidently numerous] who were hidden away in remote frontier regions’.

43 Uruj, MS Oxford (ed. Babinger, 22): aṣilda yeničerinün bunyādi budur, ol vaqitclenberü adint yenčeri kodilar; MS Cambridge (ed. Babinger, 94) lacks aṣilda and ol vagitelenberli; MS Manisa (fol. 21r) begins yeničerinün aṣli bunyādi… and continues as Oxford; MS Paris, Bibl. Nat. supp. t. 1047 (fol. 17V): aṣilda yeničerinün bunyādi budur, ol vagitdenberü adi yeničeridür. The W1, recension of the Anonymous Chronicles has the clauses transposed (ed. Giese, 22): ve hem adini yeničeri kodilar, bunlarun aṣla ol vaqitdenberü, oldi; the W3 recension has a single, telescoped, clause (MSS M2L): yeničerinün adini ol vaqitdenberü, yeničeri kodilar (cf. Annales: et inde ab illis usque temporibus nomen hoc Genizarorum adepti retinent). 'Apz. reads (£ 46) adin yeničeri kodilar, yeničeri bunun zamāninda vāqi' with the same transposition as W1.

44 Not the ‘menāqib of Faqīh’ but an anonymous work, see Ménage, V. L., in Lewis, Bernard and Holt, P. M. (ed.), Historians of the Middle East, 1962, 171Google Scholar and 175, and in BSOAS, XXVI, 1, 1963, 52.Google Scholar

45 Barring, of course, the case of contamination, which has admittedly occurred (elsewhere) in the Paris texts of Uruj (they contain some interpolations from 'Āpz.); but otherwise the Anon.-Uruj group and 'Apz. do not seem to have influenced each other.

46 See, for example, 'Ālī, in TOEM, 48, 355 ff., and Khalīl Edhem, Meskūkāt-i niyye, 2–4. There is no need to postulate a written source for this statement of Idrīs: he may well have seen one of these early coins.

47 Uzunçarşili, I. H., Osmanli devleti teṣkilâtina medhal, Istanbul, 1941, 302 f. and index, S.V. kadilkuzat.Google Scholar

48 This is the appropriate point to clear up a misunderstanding. At p. 151 of his review (cited at p. 65, n. 4 above), Professor Vryonis speaks of the discrepancies between the ‘Palmer version’ of Idrīs (i.e. Palmer's summary at p. 470 of the article cited at p. 72, n. 41 above) and the ‘Ménage version’ (the passage quoted in BSOAS, XVIII, 1, 1956, 181 f.). The ‘Palmer version’ is a summary of the whole chapter as presented by Sa'd el-Dīn (1, 37–41); the ‘Menage version’ is a quotation, from Idrīs, of a small passage in the whole chapter: it occupies 15 lines out of a total of 135 in the Nuruosmaniye MS 3209, Idrīs's autograph, of which I now have a microfilm (the equivalent in Sa'd al-Dīn is p. 40, 1. 24 to p. 41, 1. 5; and in Palmer, p. 470, lines 25–30, ‘This piyāda… service and obedience’). Idrīs certainly does refer to the collected youths as becoming Janissaries—but in contexts before and after the passage which I quoted:Google Scholar

Again, Dr. Papoulia's aside (p. 72), that perhaps not all MSS mention Kara Khalīl, is unwarranted (unless her MS, Berlin Or. fol. 3179, has very different readings from those I have seen). Idrīs begins his ‘infantry’ section by stating that consulted his brother ‘Alā’ al-Dīn pointed out that in order to take fortified towns (cf. Sa'd al-Dīn, I, 39, line 10) infantry were more necessary than cavalry (another anachronism); was put in charge of recruiting yaya. The levy of lads, however, is ascribed to the advice not of Kara Khalīl but of arkān-i dawlat (line 3 in BSOAS, XVIII, 1, 181 = MS Nuruosmaniye, fol. 93v, line 7, and cf. Sa'd al-Dīn, 40, line 26). Dr. Papoulia is therefore mistaken in saying (p. 72) that Kara Khalīl is mentioned by Idrīs ‘als Ratgeber Orhâns’ on the question of the and (p. 89) that ‘die Einführung des Devširme mit dem Namen des Qara verbunden ist’; it is Hammer who makes this assumption (GOR, 1, 91).Google Scholar

Even if there were a link between Kara it is rather hard on him to say (p. 89) that he was ‘für seine Verstösse gegen das Scheriat-Recht bekannt’. This notoriety rests solely on the criticisms of the redactor of the Anonymous Chronicles (ed. Giese, p. 30, lines 21 ff., and cf. 'Āpz. £ 63), who is saying in effect that in the good old days honest were not pestered by the central government: there was no penjik (= one of the basic prescriptions of Islam) to tax private enterprise; there were no laws compelling the surrender of an earlier sound currency in exchange for a debased new one; and there were no nasty (everyone knows how they won favour) coming out of the Palace to lord it over free-born Turks. Kara evil reputation derives in fact not from his infringement of the but from his attempts to apply it.

49 I have microfilms of MS Nuruosmaniye 3078 and of part of MS Ali Emiri 30. The former is wrongly described in 1st. küt. tarih-cografga yazmalari kataloglari, I, 2, 1944, p. 122, as containing books I–IV: it contains only I, II and IV; the latter (books I–IV) lacks some pages between fol. 81v (the incidents of 'Āpz., £ 44) and fol. 82r (defeat of the Serbs, 'Āpz., £ 49).

50 MS Nuruosmaniye 3078, fol. 45V:

51 Closely linked with this point is the question whether the devshirme was first applied in Rūmeli and then extended to Anatolia or whether (perhaps as a continuation of SelJūq practice) it had been carried from Anatolia into Rūmeli. This is crucial for Dr. Papoulia's thesis, for if the devshirme were shown to have originated in Rūmeli it could hardly have been instituted by and still less in the first years of his reign. Dr. Papoulia points out that there is no evidence in the earliest sources that it was confined to the European domains; and she argues convincingly (p. 85 and n. 19) that the letter of 1456 to the Grand Master of Rhodes (which, in my article ‘devshirme’ in Enc. Islam, second ed., I suggested might refer only to piracy) does indeed indicate that the devshirme was then practised in Anatolia. But the existence of an Anadolu which she cites in corroboration (p. 87), does not in fact help her argument. This officer's duty, apparently, was to supervise the lads from Rūmeli who were undergoing their ‘basic training’ in Anatolia, and that of his colleague, the Rūmeli , the converse (see e.g. Hezārfenn, , quoted in Tūrkiyat Mecmuasi, X, 1953, 383). Of the two, however, the Anadolu was senior in rank (Uzunçarşili, I, 44–5); and this suggests that his post had been established earlier (just as the beglerbegi of Rūmneli out-ranked his colleague of Anatolia), and hence that the devshirme had been established first in Rūmeli.Google Scholar

52 Bojanić-Lukač, Dušanka, ‘Povodom izraza čilik’, in Vesnika Vojnog Muzejajna (Belgrade), VI–VII, 1962, 237–9.Google Scholar

53 Ḳānūnnāme-i sulṭānī ber mūceb-i 'ōrf-i 'osmānī, ed. Anhegger, R. and Inalcik, H., Ankara, 1956, 76Google Scholar (= ed. Babinger, F., Munich, , 1956,272,Google Scholar and cf. Beldiceanu, N., Les actes des premiers sultans…, I, Paris, 1960, p. 149, £ 3).Google Scholar

54 Enc. Islam, second ed., art. col. 564b.