VARIEDADES SOME REMARKS ON CONVERSION TO ISLAM IN AL-ANDALUS *

The book by R. W. Bulliet, Conversion to Islam in the Medieval Period, ^ is well known for being the first and, until now, the only work dealing with conversion to Islam fi'om a quantitative point of view. In order to construct conversion curves in Iran and al-Andalus, Bulliet made use of the data provided by genealogies (nasab, pi. ansàb) contained in biographical dictionaries, assuming that the first member of the sequence bearing a non-Arabic name was the convert to Islam. At first, Bulliet's method, which I will describe fijrther on, seems acceptable, although with reservation, as it has some weak points which may call for revision. In the particular instance of al-Andalus, which is the one we are interested in here, he uses too few cases to consider the results as being completely valid, thus making it difficult to accept unreservedly the book's final conclusions. Furthermore, since Conversion to Islam has been published, later studies have revealed that some information provided by biographical dictionaries or other kind of sources —such as historical chronicles or juridical sources—, although does not invalidate Bulliet's method, does suggest that there should be some reservation in accepting the results he obtains. Moreover, this information leads one to speculate about the possibility of establishing new and more reliable methods applicable to the quantitative study of conversion to Islam in al-Andalus by increasing the number of cases used by Bulliet, thereby rendering it possible to draw conversion curves that are more precise for al-Andalus. Briefly, the assumptions these new methods would be based on are: — The convert's father was named ""Abd Allah. — The convert received the nisba al-Islâmî.

The book by R. W. Bulliet, Conversion to Islam in the Medieval Period, ^ is well known for being the first and, until now, the only work dealing with conversion to Islam fi'om a quantitative point of view.In order to construct conversion curves in Iran and al-Andalus, Bulliet made use of the data provided by genealogies (nasab, pi.ansàb) contained in biographical dictionaries, assuming that the first member of the sequence bearing a non-Arabic name was the convert to Islam.At first, Bulliet's method, which I will describe fijrther on, seems acceptable, although with reservation, as it has some weak points which may call for revision.In the particular instance of al-Andalus, which is the one we are interested in here, he uses too few cases to consider the results as being completely valid, thus making it difficult to accept unreservedly the book's final conclusions.
Furthermore, since Conversion to Islam has been published, later studies have revealed that some information provided by biographical dictionaries or other kind of sources -such as historical chronicles or juridical sources-, although does not invalidate Bulliet's method, does suggest that there should be some reservation in accepting the results he obtains.Moreover, this information leads one to speculate about the possibility of establishing new and more reliable methods applicable to the quantitative study of conversion to Islam in al-Andalus by increasing the number of cases used by Bulliet, thereby rendering it possible to draw conversion curves that are more precise for al-Andalus.Briefly, the assumptions these new methods would be based on are: -The convert's father was named ""Abd Allah.
-The convert received the nisba al-Islâmî.
Before further explanation, I will describe the methodology applied by Bulliet and the results he reaches with its application, making it clear that I do not intend to question the essence of his method but to examine some of its debatable points, which do not diminish the undeniable value of his work.Moreover, we join with Bulliet in hoping that his innovative work will give rise to new efforts to apply quantitative methods for the study of Muslim medieval history.^ For Iran, the author of Conversion to Islam drew a bell-shaped conversion curve through analysing the ansâb of three biographical dictionaries that contain individuals from Nishapur and Isfahan deceased before 525/1130 and 390/1000 respectively.^ From the almost six thousand biographies included in these dictionaries Bulliet takes 469 genealogies in which the last member of the sequence has a Persian name assuming that this person was the convert to Islam.Starting from the death date of the biographee -datum that is usually provided by biographical dictionaries-, Bulliet carries out calculations taking as figures 70 years for average lifespan, "^ 34 for average intergeneration difference, and 25 for age at time of conversion.^ Bulliet's theory seems quite reasonable: although not all converts to Islam changed their names to a Muslim one, they gave their children Muslim names.Bearing in mind that, with very rare exceptions, in genealogies only Muslim members appeared, if the member who initiates the genealogy has a non-Arabic name, it is likely that he is a convert who has decided to keep his native name.Obviously, BulUet's method is not valid "if the initial convert changed his own name to an Arabic name, which was probably the normal practice", as Bulliet recognizes ^.Nevertheless, if the instances obtained through the application of that method are numerous enough, they may al-^ See Conversion to Islam, p. 5. ^ For Nishapur, Bulliet uses the histories by Abu ''Abd Allah Muhammad al-Hakim al-Naysâburî and Abu 1-Hasan ''Abd al-Gàfir al-Fârisî (contained in The Histories of Nishapur, ed. R. N. Frye, Cambridge, Mass., 1965); for Isfahan, the Kitàb dhikr akhbàr Isbahàn by Abu Nu'^aym Ahmad b. ""Abd Allah al-Isbahânï (ed.S. Dedering, 2 vols., Leiden, 1931Leiden, -1934)).
^ Actually, the figure obtained for average lifespan was not 70 but 75, which Bulliet has considered it appropriate to adjust downward {Conversion to Islam, p. 143, note 7).
^ Bulliet carries out the following operation: he subtracts 70 from the death date of the first member of the sequence, namely the biographee, so he gets the biographee's date of birth; fi"om this figure he subtracts 34, and the resultant figure would be the father's date of birth, fi'om which figure he subtracts 34 again and he gets the grandfather's, and so on until arriving at the date of birth of the last member of the sequence, namely the convert to Islam; finally, to this figure he adds 25, which is the age around which he would reasonably convert {Conversion to Islam,[19][20][21][22].
Bulliet applies to al-Andalus the same methodology as that used to construct conversion curves in Iran, ^ utilizing five biographical dictionaries which contain biographies of ""ulamà'from the 2^^/8^^ century until the first half of the 7^V13^*^ century.^ These dictionaries altogether contain more than seven thousand biographies.Bulliet takes just 154 genealogical sequences whose last member has a non-Arabic name.After applying to these sequences his criteria to determine the approximate age of the convert at the time of conversion, the author concludes that the conversion line is rising until 300/913; from this year, the line is descending except for the period between 350/961 -which marks the midpoint of the curve ^-and 400/1010, when there is a slight increase.Without analysing the conclusions Bulliet arrives at in detail, which is not the purpose of this study, he relates these results to several historical events that happened in al-Andalus, such as the collapse of the Umayyad caliphate and the arising of the petty kings (mulük al-fawà'if), and tries to establish parallels between al-Andalus and other Eastern countries, especially Iran, based on the conversion curves drawn.
This is an outline of the method used by Bulliet to determine moments of higher and lower repercussion of conversion in al-Andalus, as well as the curve resulting fi*om its application.This method has some weak points, the most obvious being the very small number of individuals he relies on for al-Andalus: just 154 individuals for a span of more than four hundred years.Bulliet himself admits that "the result is obviously not a tidy bell-shaped curve of the type similar data produced for Iran", for which "the curve was made up from three times as many genealogies".^ '^ As Bulliet explains, this method is not applicable to Iraq, Egypt, Tunisia, and Syria because "all became Arabic-speaking regions in time," which fact "makes the occurrence of non-Arabic names in genealogies from these areas quite rare", (Conversion to Islam, p. 115; see also p. 72).For these regions Bulliet uses another method based on the curve of popularity of Muslim names.
^ There are some further aspects of Bulliet's work that should be taken into account, especially in the event that we need to rely on it for future studies.Firstly, it seems that in doing calculations, Bulliet has not taken into account the fact that genealogies may be incomplete.In order to avoid this oversight, which may badly affect the results, the reconstruction of family trees is very useful, namely the gathering of biographees who descend from the same individual, like that carried out by L. Molina.^^ Molina showed that the 1650 individuals contained in the Ta 'rîkh 'ulamâ' al-Andalus of Ibn al-Faradï could be gathered in about 1300 families.^^ The reconstruction of family trees helps to complete the information on an individual -concerning, for example, his ethnic and/or geographical origin-as well as to reconstruct the incomplete genealogical sequences, of which the biographer has omitted, by ignorance or negligence, one or more members.Obviously, this kind of fault in the genealogies may affect works based on biographical dictionaries misinforming the results, especially in quantitative studies.With regard to Bulliet's work in particular, if the genealogical sequence were not complete, the figure obtained for age at time of conversion would not be correct.^^ In order to calculate the average length of a generation Bulliet himself made use of the data obtained through reconstructing a group of families from Nishapur "in which death dates of fathers and sons were recorded".^^ As has akeady been noted, the figure obtained was 34 lunar years.Molina carries out the same operation for the period between the 2°^/8^ and the gth/j4th centuries, and concludes that the average intergeneration age is 40.1 lunar years.^^ This shows, at least, that data that, at first, seem to be valid for Ch.Veauvy, Paris, 2000,114.Within the same volume, there are two other interesting articles on "Mozarabs": Ch.Picard, "Les chrétiens mozarabes: encadrement religieux et déclin.L'exemple du Gharb al-Andalus", 97-110; J.-P.Molénat, "L'identité mozarabe dans ribérie reconquise, spécialement à Tolède", 123-131).
^^ Conversion to Islam, 21. ^"^ "El estudio de familias de ulémas", 166.Iran are not necessarily so for al-Andalus.In my opinion, the fault of Bulliet's work lies mainly here: to apply to al-Andalus the results he obtains for Iran -a region that he has studied in greater depth and knows better-, without having previously carried out a specific study for al-Andalus.
As noted above, the most debatable point of Bulliet's work is the very low number of individuals he uses to draw the conversion curve for al-Andalus.With regard to this question we wonder if it would be possible to use different methods that allow us to increase the nxmiber of 154 converts so as to construct, with a greater degree of accuracy, a curve that reflects the moments of higher and lower conversion in al-Andalus.In this regard, a fi-agment from the book by the Cordobán jurist Ibn al-''Attar (d.399/1009) Kitâb aUwathâ'iq wa-l-sijillât contains two reports that upon a first examination lead one to speculate about the possibility of increasing this number.
Ibn al-^'Attár's book is a collection of notarial documents accompanied by ^fiqh or juridical basis, which the document is based on.On very few occasions it is also accompanied by a tafsîr or comment, as happens in the case of the conversion document of a Christian.In the aforementioned article devoted to the reconstruction of family trees, L. Molina pointed out the importance of the content of this tafsîr about the convenience of changing not the name of the convert, which seems to be taken for granted, but the name of the convert's father.^^ The jurist says, "If his father's name is unknown or it is one of those unpleasant and hatefiil non-Arabic names {asma' al-''ajam), you will call him Fulán b. ''Abd Allah, and will make his father a God's servant by rights, since everybody is God's servant".^^ Further on, in the conversion document of a majüs, Ibn al-'^Attár refers to the convert as follows: "Fulán ibn Fulán, or ibn ''Abd Allah, al-Islámí (convert to Islam)...".^^ The des ulémas andalous du v/xf au vil/Mlf siècle, Geneva, 1978, 30-31); M. L. Ávila obtains 72.8 lunar years (70.7 solar years) for the period between A.H. 350 and 450 {La sociedad hispanomusulmana al final del califato, Madrid, 1985, 31); J. Zanón gets 71.5 lunar years (69.3) for individuals deceased between A.H. 540 and 650 ("Demografía y sociedad: la edad de fallecimiento de los ulémas andalusíes", in Saber religioso y poder politico en el Islam,[333][334][335][336][337][338][339][340][341][342][343][344][345][346][347][348][349][350][351].It is important to note that these ñgures cannot be applied to society as a whole, not even to the general group of''ulamâ', but only to the elite of the intellectual class, since longevity was closely connected to prestige, and individuals deceased before arriving at 40 years do not usually appear in biographical dictionaries (see Urvoy, Le monde des ulémas, 30; Avila, La sociedad hispanomusulmana, 15; Zanón, "Demografía y sociedad", 342).Zanón suggests that the figure of 60-64 solar years would be nearer to the general average age of death ("Demografía y sociedad", 343 and 345).
^^ "Famihas andalusíes", footnote 3. ^^ Formulario notarial hispano-árabe, ed.P. Chalmeta and F. Corriente, Madrid, 1983, 406.17  (d. 179/795) regarding the change of the name of the convert's father ^^ seems to reflect that this practice was at least not unusual.Malik condemned the change of the father's name to an Arabic-Muslim one because, since he was not Muslim, that would be a lie.^2 According to A. Fernández Félix and M. Fierro, remarks such as that of Ibn al-'^Attár may suggest that this prohibition faced some opposition in practice.^^ There are some known examples of converts to Islam who changed their fathers' name to that of ''Abd Allah, one of them, though quite a late instance, being the Franciscan Anselm Turmeda (d.ca.1424).On converting to Islam, Turmeda was called ''Abd Allah b. ""Abd Allah al-Tarjuman.'^^ M. de Epalza states that this name was quite usual among converts from Chris-tianity, sons of Christians, even before their conversion, given that all of them were God's servants.^^ In the same comment to the notarial document of Christians' conversion, Ibn al-''Attar goes on to say: "And the correct way is to say 'al-Islâmî', as we do, because you are giving him the nisba from Islam.Who says 'al-Aslamf is giving him the name of Aslam, which is one of the Arab tribes," ^^ comment that is echoed by Ibn Mugïth.^7 Without doubt, this would be a reliable way of identifying converted people: all those bearing the nisba 'al-Islâmî'.Unfortunately, the biographical dictionaries contain hardly any biographees bearing this nisba.As a matter of fact, in the biographical dictionaries examined ^^ I have found one person whose nasab contains the nisba al-Islâmî: the convert ancestor of Ibn Hafsün is given that nisba.^^ Maybe it would be too clear an indication of the individual's non-Muslim origin.^0 Moreover, Ibn al-'Attar's words may reflect that people converted into Islam adopted the nisba al-Aslamî instead of the appropriate al-Islâmî, so hiding their non-Muslim origin.
To conclude, I have tried to show how information provided by biographical sources may help us to establish new quantitative methods for the study of conversion to Islam in al-Andalus.It may also be meaningful, for example, the fact that a biographed individual has a short nasab, a genealogy consisting of just two or three members.Why does the interruption of the genealogical sequence happen?Is it due to mere ignorance on the part of the biographer?Or has the biographer consciously omitted the rest of the members?And, if this is the case, why?This would be another interesting line of research.Through the application of new methods we would intend to draw conversion curves using a higher number of examples than that used by Bulliet.In light of the reached results, we could study, for example, if repercussion of conversion in al-Andalus was determined by a given political and/or social situation, which seems quite logical a priori.
However, when it comes to applying these methods, we face some problems which make the quantitative study of conversion to Islam not easy to imdertake: through the application of some of them, so few instances are found that it is not possible to construct conversion curves with a modicum of reliability; in other cases, we cannot know for certain which member of the genealogical sequence was the convert to Islam unless the source provides information in this respect, which the biographical dictionaries do hardly ever.