Dating Muslim Traditions: A Survey

Author(s): Harald Motzki

Source: Arabica, T. 52, Fasc. 2 (Apr., 2005), pp. 204-253
Published by: Brill
Stable URL: http://www.jstor.org/stable/4057795
Accessed: 04-01-2016 12:34 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/4057795?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Brill is collaborating with JSTOR to digitize, preserve and extend access to Arabica.

http://www.jstor.org

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS:
A SURVEY

BY

HARALD MOTZKI

Introduction'

Throughout the centuries Muslim scholars have devoted themselves to

Hada-tstudy for various reasons.2 Legal theorists, for example, sought in
Hadit texts a source of law. Others found in them moral and religious
inspiration. Still others saw in the Hadatan important source for the
history of early Islam. The interests of scholars in the West have been
less varied. Their interest in Muslim traditions has been almost exclu-
sively historical. They seek knowledge from the Had&tprincipally to find
out what really happened ("wie es eigentlichgewesen").3 This is true not
only in the case of traditions purporting to recapitulate historical events.
It is true for hadatstouching on legal, exegetical and theological mat-
ters as well. In short, the aim of occidental scholars has concentrated
on hadatsas sources for the reconstruction of Islamic history: the his-
tory of events, the history of jurisprudence, of religious ideas and insti-
tutions, and exegesis of the Qur'an, etc.
For the history of early Islam the Hadit is certainly a source of prime
importance, if only for the reason that there are not many other sources
available. A prerequisite of historical reconstruction is source criticism,
one of the methodological achievements of modern historical studies.

A first draft of this paper was read at the conference "Hadtth:Text and History"
organized by the Center of Islamic Studies, School of Oriental and African Studies,
London, March 1998. I wish to thank Dr. Paul Hardy for the careful revision of my
English text.
2 The term Hadtt as I used it in this article means the sort of traditions found in
the pre-canonical collections such as Malik's Muwatta'. It is not limited to traditions of
the Prophet.
3 A famous expression of the German historian Leopold von Ranke (1795-1886).

? Koninklijke Brill NV, Leiden, 2005 Arabica,tome LII,2

Also available online - www.brill.nl

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 205

Source criticism sets out to evaluate the sources available by checking

the authenticity, originality and accuracy of the source's informational
content.4 Two examples may illustrate its importance. Consider a doc-
ument which, although it purports to be a Genoese title-deed of the
ninth century of the common era (C.E.), can be proven to have been
composed at Rome in the eleventh century of the same era and is,
therefore, a fabrication. Hence, the reliability of its information on
Genoa of the ninth century is uncertain. The document can be used,
however, as source for aims and practices of forging documents at
Rome in the eleventh century. Or consider a document which is trans-
mitted by writing over a longer period of time. Obviously, it can undergo
any number of changes. That is, passages can be omitted, added or
distorted, intentionally or not. Such changes must be taken into account
and documented (if possible) if we wish to extract from the document
its original intention. This is the task of source criticism.
One aim of source criticism is the dating of documents. When try-
ing to determine the degree of reliability of a source the first questions
a historian usually asks are: How far away in time and space is the
source from the event of which it informs us? Are the date and place
of origin which the source assigns to itself correct? Dating a source is,
therefore, the first step in determining what historical use can be made
of it. The methods which can be used to date a source depend on the
character of the source in question. Consequently, dating methods are
many and diverse. In fact, each historical discipline developed its own
methods. Scholars working in the field of early Islam likewise devel-
oped methods adapted to their discipline. Now whether the different
methods they are using in dating 4adits are reliable is a matter of dis-
pute. But it is a dispute in which the participants are few, given that
the number of scholars who engage in the critical study of dating meth-
ods is small. All the same, it is a dispute of the very first importance
for every scholar who works in the area of early Islam.
In order to review the existing methods used in the studies con-
cerned with Hadft I have classified them into four groups: 1) methods
which use the matn,2) dating on the basis of the collections where tra-
ditions appear, 3) dating on the basis of the isndd,and 4) methods using

fiberEnzyklopdie und Methodologie

Cf. J.G. Droysen, Historik, Vorlesungen der Geschichte,
cd. R. Huibner, Darmstadt, 1972, 98-99.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
206 HARALD MOTZKI

matnand isndd.5In each group, the approaches are not always the same
and can be further classified. Additionally, it must be said that schol-
ars often use combinations of different methods. For each method, I
shall present one or more representatives and discuss their approaches.
The main questions which I shall try to answer are: How does the
method in question function? On what premises it is based? Are method
and premises reliable? What results does the method produce?

I. Dating on the Basis of the Matn

A. First Stepsin Dating:Ignaz Goldziher

One of the most famous examples of the use of dating by means of
the matnof a had7tis Ignaz Goldziher's article "Ueber die Entwicklung
des Hadith", published in 1890 in the second volume of his Muham-
medanische Studien.6In this article - the first fundamental study on Hadiit
written by a Western scholar - Goldziher mentions that there is some-
thing called isnad but does not mention it further.7 His statements on
the origins of hadatsare solely based on their matnsand other criteria.
Two types of dating can be distinguished in Goldziher's article: first, a
general dating, i.e., a dating of the Hadztas a whole; second, a dating
of a particular hadi-tor tradition.
The principle behind Goldziher's general dating of the Had-t is well-
known: Most of the material available in the canonical collections is a
result of the religious, historical and social development of Islam in the
first two centuries, the reflection of the efforts which emerged in the
Islamic community during their more mature stages of development.8
On the basis of this principle of general dating of the HadRL, Goldziher
denies that the bulk of traditions concerning the Prophet and also most
reports on the Companions might possess any worth as historical sources
for the time about which they purport to inform us. This does not
mean that they cannot be used as sources for the time when they actu-

5 A fifth category would be "methods using other criteria". It is left for another
article.
6 J. Goldziher, Muhammedanische Studien,Halle, Max Niemeyer, 1889-90, II, 1-274.
English translation, Muslim Studies,trans. C.R. Barber and S.M. Stern, London, George
Allen & Unwin, 1971, II, 1 ff.
7 Goldziher, Muhammedanische Studien,II, 6 (I quote the English translation according
to the pagination of the German original given in the margins).
8 Ibid.,5.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 207

ally originated which is defined by Goldziher as the Umayyad and the

first century of the Abbasid caliphate.
Goldziher did not formulate his general dating of the Hada-tas a
universal statement. He did not say: "All hada-tsare the result of later
developments". He formulates it as a partial generalisation, i.e., "the
vast stock of hadats".This means that some authentic hadatsgo back to
the first half of the first Islamic century, although about these Goldziher
expresses no concern. This division between a major component of
non-authentic and a minor part of possibly authentic traditions leaves
us with an epistemological problem. If we have to do with a tradition
which is not clearly a late fabrication - which is most frequently the
case - then, into which category must we place it? If Goldziher's gen-
eral dating is correct, then, for statistical reasons, we have to conclude,
indeed, it is safer to assume that the tradition is late, or rather, not
authentic; for the probability of coming across an early and possibly
authentic hadatis not large.
But on what arguments does Goldziher base his general dating of
the Hadat?On what grounds does he rest his judgement on its authen-
ticity? His conclusions are based merely on a limited sample of tradi-
tions he collected. The following represent the indications or reasons
which may have motivated their invention or fabrication:
1) Political quarrels and religious disputes within the nascent Islamic
community. Goldziher, like a number of others, assumes that the more
secular regime of the Umayyad dynasty had driven "more pious Muslims"
to create a religious world of their own and to project it back to the
Prophet and the first four caliphs. The rulers reacted to this development
by having their political principles justified by opportunistic scholars in
the same way. Namely, they ordered them to forge hadatsand ascribe
them to earlier authorities. According to Goldziher, a large number of
traditions said to go back to the Prophet or the Companions arose pre-
cisely in this way during the second half of the first century A.H.9
2) Other hadatscame into being when the Abbasids took over the
caliphate from the Umayyads in the course of the second Islamic cen-
tury. The new religious policy of the rulers gave a strong impulse to
the development of Islamic jurisprudence. This impulse, at the same
time, prompted the study and production of hadatssince under Abbasid

9 Ibid., 73-83.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
208 HARALD MOTZKI

rule subservience to the heads of state led some scholars ("court theolo-
gians") to forge traditions in favour of whatever the regime currently
in place wished to carry through.'0
3) During the second half of the second century A.H. many tradi-
tions emerged from the dispute between the old style jurists, the ahl al-
ra'y, and scholars who advanced arguments on the basis of traditions,
the ashdbal-hadat.According to Goldziher, the latter wished to base the
law as much as possible on the example of the Prophet and his
Companions and in cases where they could not find a tradition, they
simply invented one. The scholars of the ancient schools answered the
challenge to their doctrines by looking for traditions which supported
their point of view and even invented hadatswhenever they thought it
was appropriate.ll
4) Many 4adztshave their origin in or became distorted during the
many political and religious struggles within the Muslim community or
else they derive from groups or circles dissatisfied with or opposed to
the ruling family. To give expression to their claims, the different par-
ties created hadztsfor or against rebellion, for or against the dynastic
principle of rule and for or against the claim of particular clans of the
Prophet's tribe to the caliphate. In fact, rivalry between tribes, towns
or scholarly circles must never be underestimated as source of fabri-
cated traditions.'2
Goldziher's set of causes and motives for the invention and fabrication
of hadits during the Umayyad and Abbasid period is derived from a
wide range of sources. However, the choice of the source material and
the use to which Goldziher puts it display two major points of weak-
ness: 1) Goldziher's source material consists mostly of traditions about
transmitters and kadztsand only rarely of the traditions themselves."'
When Goldziher falls back on the traditions themselves, he relies on
kadits, which are rarely considered reliable by Muslim scholars them-
selves. Traditions from the collections of al-Buhiariand Muslim, appear
but rarely amongst his pieces of evidence. 2) Goldziher seldom ques-
tions the historical reliability of the reports which he is using, although
they often have an anecdotal character.

10
Ibid., 53-73.
" Ibid., 73-83.
12
Ibid.,88-130.
13 In the light of my typology of dating methods, these cases belong to the fifth cat-

egory "information derived from other sources". See note 5.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 209

To form some idea of Goldziher's mode of reasoning an example is

helpful. From a late compilation of ahbarhe quotes the following anec-
dote: A scholar at the court of the Caliph al-Mahdi (158-169 A.H./775-
785 C.E.) added a word to a hadatin order to make pigeon racing
permitted, a sport of which the caliph was fond. However, to the "ortho-
dox scholars" pidgeon racing was repugnant. This prompts Goldziher
to conclude: "the tale nevertheless shows what a court theologian was
capable of doing in matters of the tradition. Theologians, who wished
to reconcile theory with the practices of life, had to haverecourseto such
subterfuges, and this consideration became one of the chief factors in
the history of the growth of the IHadith".14 Still, does not Goldziher
owe the reader proof that this distorted hadatfound its way into one
of the authoritative compilations of hadatsand/or was taken seriously
by other scholars?
What this example shows is how, on the one hand, Goldziher is able
to move from singularcases- where the historical character of a nar-
ration is by no means certain - to a conclusion about the entire cor-
pus of Hadit. On the other, it illustrates how he can move from the
possibilit that something could have happened to conclude that it actu-
ally did happen. That is, he can deduce a fact from a mere possibil-
ity. Even if the story about al-Mahdi's court theologian was true and
if there were some other reliable cases of forgery, it takes some degree
of audacity to conclude that manyor mosthadctsare forgeries. I do not
wish to deny that Goldziher succeeded in his study on the Hadatto
present a number of texts which possibly or probably reflect reactions
on later religious, political or juridical developments. I only question
whether it is methodologically correct to conclude on the basis of a
limited number of reports about invented or falsified hadatsand a few
traditions which can hardly derive from the time from which they pur-
port to emanate, that the vast majority of the hadTtshave been fabri-
cated at a later time and came into being as a result of the develop-
ments mentioned above. Goldziher's dating may be true for a number
of individual traditions. That this dating holds true for the majority of
the hadfts he has not demonstrated.
Apart from general dating, Goldziher sometimes tries to pin down
the time of origin of a particular tradition or some of its elements. In
these cases, he does not mention explicitly which criteria he uses to

" Goldziher, Muhammedanische

Studien,II, 70 (emphasis mine).

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
210 HARALD MOTZKI

distinguish between an earlier and a later tradition. Some of his exam-

ples will reveal his methodological principles. In fact, four follow.
1) Anachronisms indicate that a text originates from a time later than
purported.15
2) Traditions of which the content clearly displays a secondary stage
in the development of an issue are younger than those with a less de-
veloped content.'6
3) When the Prophet or early Muslims appear in a tradition in an un-
favourable light, the tradition can be accepted as authentic and early.'7
4) Reproaches among opponents against each other have probably a
historical kernel.'8
The first principle is certainly a safe one. The second principle takes
for granted that we know what the development was. However, knowl-
edge of that kind is, for the most part, based on the traditions them-
selves. So such an argument seems suspiciously circular in its reason-
ing. One adopts Goldziher's premises three and four, however, at one's
own peril. Their validity depends on a supply of background informa-
tion about the persons who circulated the traditions in question for
they may have had a bias. But information about such persons is for
the most part lacking. Even if principles three and four may be of some
use in particular cases, they cannot be considered as general rules.
In sum, Goldziher's methods of dating particular traditions on the
basis of their matnsare very rudimentary. His conclusions concerning
the origin of a tradition seem often to derive from intuition and appear
quite arbitrary. Therefore, the validity of his general dating of HadAt
based on the dating of particular traditions seems as limited as that of
the rules of thumb he employs.

B. Joseph Schacht'sDating with the Matn

Another famous advocate of dating hadftson the basis of their matnsis
Schacht. As in the case of Goldziher, we can distinguish between his
dating of the Hadit in general and his dating of individual hadats.But
unlike Goldziher's, Schacht's general dating is not based only on the

15 See ibid.,23-27, 138-40.

16 E.g., ibid.,25-26.
7 E.g., ibid., 29-30.
18 E.g., ibid.,35.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 211

matns but results from a combination of different methodological

approaches:
1) The hypothetical reconstruction of the development of the juridical
theory during the second Islamic century based on a study of al-Safi'c's
(d. 204 A.H./820 C.E.) theoretical treatises. 2) The method of dating
traditions on the basis of the collections in which they first appear. 3) A
comparison of matnsof individual traditions. 4) A comparison of their
isnads.'9
This order of methods shows that Schacht dates on the basis of the
matn only after other methods have been used. Even so, one must
uncover the premises on which Schacht works when he is dating by
using the matn. This is even more urgent because his method and
premises have been adopted by many scholars after him. The most im-
portant of them are:
1) A tradition must first be dated by placing its content (the problem
and its solution) into the legal development as he had reconstructed
it.20 2) Traditions which have the form of short legal maxims are ear-
lier than narratives.2' 3) Anonymous maxims are earlier than those
ascribed to a particular authority.224) Terse statements are earlier than
detailed ones. 5) Texts which contain the problem implicitly are ear-
lier than those which expound it explicitly.23
The first rule shows that Schacht's dating on the basis of the con-
tents of traditions is dependent on his assumptions regarding the gen-
eral development of Islamic legal thinking and his view of the particular
juridical problem in question. These assumptions derive from a study
of the material using different methodological approaches of which
the analysis of the matnsis one among others. Arguing in this manner
gives the appearance of circular reasoning. Furthermore, the other four
premises or methodological rules are secondary generalisations. They
result from a study of legal traditions in which other premises and
methodological approaches played a primary role. To give an exam-
ple: The second premise is not plausible as such, because legal max-
ims can also be formulated on the basis of reports on legal cases and

' J. Schacht, The Originsof Muhammadan

junsprudence,Oxford, Clarendon Press, 1950.
20 Ibid., 176-79.
21
Ibid., 180, 188.
22
Ibid., 180-89.
23 J. Schacht, "Modernism and Traditionalism in a History of Islamic Law" in Middle

EasternStudies,1 (1965), 393.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
212 HARALD MOTZKI

their solutions and thus be secondary.24This premise is dependent on

other premises of Schacht and cannot claim an independent and uni-
versal validity. It can be only used as working hypothesis. For it may
prove false on the basis of legal traditions which Schacht has not stud-
ied and/or when some of his other premises prove to be false.25

C. FormAnalysasand Dating:MarstonSpeight
In the seventies of the century just past a method which was originally
developed in Biblical studies entered into Islamic studies: form analy-
sis. It was applied to Islamic traditions by Marston Speight. In his arti-
cle, "The Will of Sa'd b. a. Waqqas: The Growth of a Tradition", he
attempts to reconstruct the chronological development of a Prophetic
haditby comparing its matnvariants.26Speight proceeds from the assump-
tion that all textual variants have been part of an oral tradition before
they "became frozen in a written compilation".27His method consists
of the following steps. Firstly, he compiled a corpus of nineteen tradi-
tions which he considered to be variants related by their content. In
step two, he arranged the texts according to their complexity. As a
third step, he analysed each text with respect to: its degree of devel-
opment; the internal cohesion of its elements; indications of style and
vocabulary since these may suggest an earlier or later stage of devel-
opment of the text in question. In the fourth and final step, Speight
classifies the texts from the standpoint of related content. On the basis
of all these steps, a chronology of the nineteen traditions is established.
In his analysis of the texts, Speight starts from several premises which
he seems to consider as self-evident, at least, he does not question them:
1) Concise texts are older than more detailed and descriptive ones.28

24 Some examples in H. Motzki, Die Anfangeder

islamischenJurisprudenz.
IhreEntwicklung
in Mekicabis zur Mitte des2./8. ahrhundertsin series Abhandlungenflir
die KundedesMorgenlandes
L, 2, Stuttgart, Steiner Verlag, 1991, 115-20; English translation, The OrWinsof Islamic
jurisprudence.MeccanFiqh beforethe ClassicalSchools,trans. M.H. Katz, Leiden, EJ. Brill,
2002, 126-131 and in idem, "The Murder of Ibn Abi l-Huqayq: on the Origin and
Reliability of some Magh&zr-Reports"in H. Motzki (ed.), The Biographyof Muhammad:the
Issue of the Sources,Leiden, EJ. Brill, 2000, 188 ff., 201 if.
25 See below pp. 214-215, 220-223. For a more detailed critique of Schacht's approachcf
Motzki, Anfangeder islamischen jurisprudenz;idem, 7he Originsof IslamicJurisprudence,
passim.
26 R. Marston Speight, "The Will of Sa'd b. a. Waqqas: The Growth of a Tradition"

in Der Islam, 50 (1973), 249-267.

27
Ibid.,249.
28
Ibid.,250.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 213

2) Reported speech is earlier than direct speech.29Additionally, Speight

distinguishesa vertical and horizontal development of traditions. Vertical
means the internal development of a group of texts which are contex-
tually related; horizontal denotes the development from elementary texts
which consists of only a few elements to more complex traditions.30
Speight's analysis leads to a "rough chronological pattern" which
consists of three stages: 1) The oldest text (which a hypothetical more
original version preceded); 2) A group of three texts which he dates
somewhat later but which belong to the early Umayyad period; 3) The
remaining fifteen texts which are later than those mentioned developed
during the (later) Umayyad period. In a relative chronology, based
exclusively on the matnsof the traditions, one would not expect absolute
dates like early and later Umayyad period. Speight bases them, firstly,
on the date of death of Sa'd b. Abi Waqqas, the central figure of the
traditions, who died in 55 A.H./675 C.E. and, secondly, on the obser-
vation that in some variants other names appear. According to him,
this could only have happened after the death of Sa'd; these versions
must thus be later.3'
Speight's dating of traditions on the basis of pure matn analysis is
not convincing for the following reasons. First, it is questionable whether
all the texts of his sample really belong to the same tradition.32Secondly,
the general validity of the premises on which his form analysis is based
is doubtful. The first premise, which is borrowed from Schacht, is used
by Speight as if it is a rule generally valid in the study of Muslim tradi-
tions, a conclusion which can be challenged as I have just argued when
discussing Schacht's method. The second premise cannot claim general
validity either. The same story can be reported by different transmit-
ters not only in different wordings but also with different emphasis. It
is not less plausible to assume that reports which were originally vivid
and colourful, using direct speech, became more sober in the course
of time and changed into reported speech. The weakness of the premises
undermine Speight's relative chronology.
Thirdly, Speight's absolute dating which is based on the difference
of names is not convincing either. According to him, the central figure

29 Ibid., and passim.

30 Ibid.,251-52, 265.
31Ibid., 266-267.
32
D.S. Powers already pointed to this in his article "The Will of Sa'd b. Abi Waqqas:
A Reassessment", in StudiaIslamica,58 (1983), 33-53, esp. 41.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
214 HARALD MOTZKI

of the tradition, Sa'd b. Abi Waqqas, is called in two variants Sa'd b.

'Afra'. He concludes from it that there has been another primitive "sick
visit story" originally connected to the name Ibn 'Afra' which became
later confused with the similar story of Sa'd b. Abi Waqqas.33Because
such a confusion can only be thought of after the death of the latter,
Speight dates those variants in which the name Ibn 'Afra' appears into
the early Umayyad period.34 This reasoning is erroneous. The name
(Sa'd) ibn 'Afra' does not belong to the "sick visit story" but to the
story of muhagiru-n dying in Mecca, which is in some variants combined
with the former. In the latter story the central figure is Sa'd b. Hawla,
not Sa'd b. Abil Waqqas as Speight rightly states. He did not realise,
however, that Sa'd b. 'Afra' is only an erroneous transmission of the
name Sa'd b. Hawla which may be due to bad handwriting. The
difference of names tells us nothing about the date of the versions in
question. The mistake in the names can have been made by a trans-
mitter or copyist at a much later period than Umayyad times. Indeed,
the editor of Ibn Hanbal's Musnad in which both variants are found
may be responsible for it.
Criticism of the methods used by Goldziher, Schacht and Speight
for dating of traditions on the basis of the matnsought not lead us to
the conclusion that matnsare worthless for purposes of dating. The crit-
icism conducted here merely shows that the premises and methods used
by these scholars are unsafe. There is much to be learned from the
Methode".In this sense, Speight's approach is a step in
'!formgeschichtliche
the right direction. In my experience, it is seldom possible to find
sufficient indications for the dating of traditions in the matnsalone.35
All the same, matnanalysis can and, sometimes, must contribute to the
dating of traditions. But it does this best when used in combination
with other methods of dating as we will shortly see.

of Traditionsin Collections
II. Dating on the Basis of the Occurrence

Again, Joseph Schacht was the first to use this method of dating in a
systematic fashion. He describes it as follows: "The best way of prov-
ing that a tradition did not exist at a certain time is to show that it

3 Speight, "The Will of Sa'd b. a. Waqqas: The Growth of a Tradition", 257-58, 266.
34 Ibid.,266-67.
35 For my view on John Wansbrough's approach that is also based only on the texts
see H. Motzki, "The Origins of Muslim Exegesis. A Debate" in Der Islam (forthcoming).

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 215

was not used as a legal argument in a discussion which would have

made reference to it imperative, if it had existed".36This e silentioargu-
ment has two weak points, one theoretical, the other, practical. On the
theoretical side, the fact that a tradition was not used by someone may
have several reasons; non-existence is only one of them. The simplest
explanation may be that this person did not (yet) know the hadatin
question. This, of course, is not the same as the tradition not having
existed at all. The person may also have had reasons, ones not known
to us, which prevented him from citing the tradition. On the practical
side, the weak point in Schacht's reasoning is that in most cases we
do not know whether or not the sources actually reflect a juridical dis-
pute. Whether collections of legal traditions are compiled as complete
arsenals of legal ammunition to be used in disputes or whether they
contain a personal choice of the compiler is not a matter which we
can know with complete certainty.
G.H.A. Juynboll has employed the same method in his article "The
man kadhabaTradition and the Prohibition of Lamenting the Death",
published in his book Muslim Tradition.37 In his treatment of the man
kadabatradition, he applies the method to a tradition which is not obvi-
ously legal in character. His dating of it will be discussed in the fol-
lowing section. Juynboll examines first "in what early collections avail-
able in printed editions" the hadatin question "is not found and those
in which it is found". He proceeds in two steps. First, he investigates
the collections compiled in the HIijazand Egypt, next, the Iraqi ones.
The result of his investigation of the collections compiled in the Hijaz
and Egypt is that: "The man kadhabatradition does not occur in HIijazi
or in Egyptian collections from before the 180s/800s".2 Here he is
speaking of the Muwatta' of Malik (d. 179 A.H./796 C.E.) and the
Gdmi' of Ibn Wahb (d. 197 A.H./812-13 C.E.). This hadatis found,
however, in the works of al-Safici7(d. 204 A.H./820 C.E.) and in al-
Ijumaydl's (d. 219 A.H./834 C.E.) Musnad,both HijazTscholars. This
leads Juynboll to the conclusion that the tradition in question must
have come into circulation in the Hijaz between Malik's Muwatta' and
the books of al-Safi 1 and al-HumaydL. He thinks those transmitters
mentioned in the isna-dof the tradition who died in the 180s or 190s

36 Schacht, The Origns of Muhammadan Jurisprudence,

140.
3 G.H.A. Juynboll, Muslim Tradition:Studiesin Chronology,
Provenanceand Authorshipof
Early Hadith,Cambridge, Cambridge University Press, 1983, 96-133.
38 Ibid., 109.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
216 HARALD MOTZKI

are responsible for it. The isnddsby which they trace the tradition back
to the Prophet are correspondingly fabricated.39
The investigation of the Iraqi Hadzitcompilations produced the fol-
lowing results:The mankadabatradition is not found in collections which
originate before al-Tayalisi's (d. 203 A.H./818-19 C.E.) Musnad,as, e.g.,
the Gami'of al-Rabi' b. Habib (d. second half of the second century,
perhaps 170 A.H./787 C.E.). This leadsJuynboll to conclude that: "We
are [. . .] justified in determining, with the non-occurrence of the com-
plete dictum in this collection in mind, a terminus post quemfor its emer-
gence in Iraq".40The hadit in question must then have come into cir-
culation in Iraq between the date of death of al-Rabi' b. Habib and
that of al-Tayalisi. "Responsible for the dictum are probably the various
pupils - or people using their name - of the key figures or common
links in the man kadhabaisndds, such as Shu'ba b. al-HaIjaj;(d. 160
A.H./777 C.E.), active in Basra and Ktufa,Abtu 'Awana al-Waddhahb.
cAbd Allah (d. 176 A.H./790 C.E.), active in Wasit and Basra, and
cAbd Allah b. Abi Awfa (d. 174 A.H./792 C.E.), active in Egypt, although
the majority of his masters and many of his pupils were Iraqi".4'
Compared to al-Tayalisi's Musnadin which only a handful of vari-
ants are found, the collections of the third century contain many more
versions with different isndds.Juynboll seems to think that these isnads
originated only after al-Tayalis!, although he does not state this explicitly.
The most extensive list of variants of the man kadabatradition is con-
tained in Ibn al-Gawzf's (d. 597 A.H./1200-01 C.E.) Kitabal-mawdui'dt
which has thirty-one versions more than the collections of the third
century. This leadsJuynboll to the conclusion that those thirty-one vari-
ants are fabricationswhich emerged "from the fourth century onward".42
On the basis of his investigation of Hijazi, Egyptian and Iraqi col-
lections of traditions,Juynboll finally concludes that "every piece of evi-
dence [. . ] points to Iraqi sunnite traditionist circles flourishing in the
second half of the second century as the breeding ground of the man
kadhabasaying".43The isnddswhich reach back to the Prophet must be
considered as fabrications of the transmitters living in this period, the
same holds true for the isnads appearing only in later collections.44

39Ibid., 112-14.
40 Ibid., 125.
41 Ibid., 125.
42 Ibid., 130.
43
Ibid., 132.
44
Ibid. 132-33.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 217

Additionally, he postulates a general rule concerning matnsand isnads:

"The more elaborate or composite a tradition, the later it came into
circulation. This holds also true for isndds".45
This is a very short sketch of Juynboll's detailed study. His method
is characterisedby the use of the argument e silentio.Schacht had justified
its use by arguing that the traditions were used as arguments in the
disputes of jurists and we can therefore expect that traditions which
are suited to support the position of a jurist or a school of jurispru-
dence are quoted if they existed. The weaknesses of this assumption
have already been mentioned.46Because there is no legal discussion dis-
cernible, Juynboll defends the use of the argument by silence with the
following claim: "Muslim collectors used to put all the material they
had gathered from their predecessors into their collections which have
thus to be considered as completerecords of the material available in a
certain region at a certain time". Therefore, he reasons, "the absence
of certain material in certain collections may be considered as relevant
fact with significant implications for the chronology of that material or
its provenance".,,47
This premise seems to be doubtful in view of the fact that during
the second and third centuries A.H. HadUtfor the most part, were not
collected by gathering the manuscripts of collections made by prede-
cessors, but by hearing it in the classes and recording it. Additionally,
it is to assume that the collection of a teacher of Hadatcontained his
personal choice of traditions, not necessarily all he knew, and that the
collection grew in the course of time. That means that not all trans-
mitters of a scholar necessarily received identical corpora of texts. Apart
from the general objection againstJuynboll's premise, his concrete dat-
ing of the hadatin question is not convincing. In fact, there are a num-
ber of arguments to be made against it.
First,Juynboll has discovered that the mankadabahadTtwas known to
the Higazi collectors al-4afi'c and al-Humaydi. He ascribes the spread of
the hadit to their informants. All these informants are, according to the
isndds,Medinese or Meccan scholars who mention, in their turn, Higizi
scholars as their own informants.48In addition, the Higazi informants of

45 Ibid., 128.
46 See p. 214-215.
4 Ibid.,98 (emphasis mine).
48 This is also true in the case of al-Humaydi's informant Sufyan b.
'Uyayna who
must be regarded as a Meccan scholar, not as Kuifi. He moved to Mecca already in

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
218 HARALD MOTZKI

al-Safi'i and al-Humaydi are older than the Iraqi scholars who invented
the tradition according to Juynboll. The latter are the pupils of Su'ba
and Abiu 'Awana. Among them one finds al-Tayalisf, the earliest col-
lector with whom Juynboll finds the hadit.The existence of earlier Higazi-
informants in the HigazT collections contradicts his general conclusion
"that every piece of evidence [...] points to Iraqi sunnite traditionist
circles [. . .] as the breeding ground of the man kadhabasaying".49
Second, it is remarkable that Juynboll does not hold the common
links of al-Tayalisl's variants responsible for the hadit, e.g., Su'ba (d. 160
A.H./776 C.E.). The common links suggest a dating of the tradition
in the first half of the second century, not in the second half. Obviously,
Juynboll prefers here the e silentioargument (the tradition is not found
with al-RabT'b. Habib) to the common link phenomenon. To my mind
his preference is highly questionable.50
Third, Juynboll has overlooked the fact that several versions of the
mankadabatradition are contained in Ma'mar b. Rasid's Gami'.'Ma'mar
was born and grew up in Basra. However, he left the city as a young
student around the year 120 A.H./738 C.E. to study in the Higaz. He
finally settled in San'a', where he died in 153 A.H./770 C.E. Ma'mar
knows already three different versions of the hadat.The matnsof two of
them are short, one has only the man kadabadictum. The second adds
another well-known saying of the Prophet. The third version is a longer
story which ends with the mankadabasaying. These three versions show
that short versions of a tradition can exist simultaneously beside a long
version. This contradicts Juynboll's rule that "the more elaborate or
composite a tradition, the later it came into circulation".52It seems
more likely in the case of Ma'mar's variants that the short versions
containing only the saying are the abridged ones, than that the longer
version was created secondarily by adding an invented history to the

123 A.H./741 C.E., perhaps even in 120/738 when he was sixteen or thirteen years
old (see Ibn Sa'd, Muhammad, al-Tabaqdtal-kubrd,ed. J. 'Abbas, Beirut, Dar Sadir,
1957-1960, V, 497-498), not, as Juynboll claims (according to Ibn H.agar, Ahmad b.
'Ali al-'Asqalani, Tahdfbal-tahdfb(Beirut,Dar Sadir, 1968 [repr. of the edition Haydarabad:
1325-1327 A.H.] IV, 122) in 163 AH. The latter date seems to be a printing error.
49 See above note 43.
50 Only in his recent article "Shu'ba b. al-Haiiaj (d. 160/776) and his Position among
the Traditionists of Basra", in Museaon,111 (1998), 187-226, esp. 193-196 Juyboll identifies
Su'ba as the originator of the man kadabasaying.
5'Abd al-Razzaq b. Hammam al-San'ani, al-Musannaf,Beirut, al-Maglis al-'ilmi,
1391/1972, XI, no. 20493-95. Ma'mar's Gdmi'transmitted by his pupil 'Abd al-Razzaq
is part of the edition of the latter's Musannaf:
52 See above note 45.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 219

saying of the Prophet. In short, it is improbable that Ma'mar invented

the different matnswith their defective isndds.
All this leads to the conclusion that the mankadabatradition in long
and short versions circulated already in the first half or rather in the
first third of the second century, not only in Iraq - which may indeed
have been its place of origin - but also in the HIigazand Yemen.53 This
example illustrates how dangerous it is to date with the argument
e silentio,when only a few sources are available as is the case for the
second century after the Hijra. A single source overlooked or edited
afterwards can destroy the whole argument. The method can and even
should be used to establish a terminus post quemfor a tradition, but we
should not conclude from it that it couldnot be earlier and that the lists
of informants given in the isnddsare inevitably fabricated.

III. Dating on the Basis of the Isnad

Among the third group of methods, two wholly different approaches

can be distinguished: A) Establishing the date of a particular hadft or
complex of hadits with the same content on the basis of its or their
isnddvariants; B) Establishing the origin of traditions which according
to their isndds, derive from the same informant of a collector (recon-
struction of sources).

A. Isnad Analysisof a SingleTradition

1. Schacht on Isndd Analysis
When the topic of isndd analysis is mentioned, the name of Joseph
Schacht immediately comes to mind. Although he was neither the first
nor the only one to recognise the potentialities of the isndd for dating
purposes,54 he is to be credited with popularising the method. He
describes it in the chapter of his book, The Originsof Muhammadan
jurisprudence,entitled "The Evidence of Isnads". Schacht proposes five
rules to be applied when one tries to establish the date of a tradition
on the basis of the isndd:55

5 See also M. Muranyi, "'Man halafa 'ald minbariatiman...' Bemerkungen zu einem

fruhen Traditionsgut" in Die Welt des Orients,18 (1987), 92-131.
5 Alois Sprenger did it already in the nineteenth century; see his Das Lebenund die
Lehredes Mohammad,Berlin, Nicolaische Verlagsbuchhandlung, 1861-65, III, 235-36 and
passim.
5 Schacht, The Originsof Muhammadan Jurisprudence,
163-75.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
220 HEARALD MOTZKI

1) "The most perfect and complete isnads are the latest".562) If there
are isnddsof the hadUtwhich stop at a later level of transmission, e.g., at
the Successors' level, in addition to isnadswhich reach back to a higher
authority, the latter isnddsare secondary. This is the result of what he
calls "backwards growth of isndds".573) Isnad variants which appear in
later sources with "additionalauthorities or transmitters"are fabrications.
Schacht called this the "spread of isndds".584) "The existence of a
significant common link, N.N., in all or most isnadsof a given tradition
would be a strong indication in favour of its having originated in the
time of N.N."595) Isnadvariants that by-pass the common link are later.60
Schacht considers rules one through five as general in character. But
are they as general as he supposes? His observation that the isnddsof
traditions contained in later sources are in general more complete is
undoubtedly correct. This fact was also known to Muslim Hadatschol-
ars. They also knew that defective isnadswere sometimes improved. Yet
none of this need lead us to the conclusion that all or most early tra-
ditions must have a defective isndd nor should it induce us to believe
that early hadatswith unbroken isnddscannot exist.
The "backwards growth of isndds"is a phenomenon known to every
Hadzitscholar. The Muslim scholars called it raf' literally, "raisinghigher"
in the chain of transmission. However, the fact that there are cases in
which, e.g., a Companion tradition is "raised up" to a Prophetic one
by adding the Prophet to the isnad does not justify the conclusion that
all Prophetic hadatsof which variants are known which stop at the
Companion's or the Successor's level are necessarily secondary. If we
rid ourselves of Schacht's theory that the Muslim traditions generally
came into being only by fabrication and developed from Successor to
Companion and finally to Prophetic hadAts, we become capable of imag-
ining that a certain legal opinion can have been expressed by the
Prophet and also held by a Companion or a Successor. It cannot be
excluded a priori that there are Prophetic hadAtswhich are earlier than
similar Companion or Successor traditions.
Schacht considers the asnddthe "most arbitrarypart of the traditions".6

56 Ibid., 165.
57 Ibid., 161, 171.
58
Ibid., 164-169.
59 Ibid., 172.
60 Ibid., 171.
61
Ibid., 163.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 221

He does not only assume that all isnadshave a fictitious part (i.e., the
last part which contains the persons of the first century).62He also thinks
that the remaining part of the isnadstrands (containing the transmitters
of the second and third century A.H.) "were often put together very
carelessly".63This is the reason for his sceptic attitude towards isnad
variants. He thinks that isnaddsdisplaying different transmitters on the
same level of generation are "uncertain and arbitrary".64The exam-
ples he quotes in his book show that he cannot imagine that a tradi-
tion of the first third of the second century or earlier could really have
been transmitted by two or more different persons. Schacht is con-
vinced that most isnddsof a tradition originated by "creation of addi-
tional authorities or transmitters"65or by fabrication of complete isnadds
(spread of isndds).66But these views, too, are generalisations made on
the basis of a few cases. What's more, they are assertions not proven
facts. I shall come back to the theory of spread of isndds when dis-
cussing attempts to apply it.
Schacht claims that "family isndds",i.e., isnadds in which transmitters
are related to each other (e.g., father - son - grandson, or uncle/aunt -
nephew, or patron - client) are, in general, inauthentic. Rather, they
are later fabrications which merely simulate authenticity. This, he con-
cludes from his study of the sources. He gives some examples of fam-
ily isnads which he considers to be fabricated. These examples show
that his reservations against family relations in the isnddsconcern only
what he calls "the fictitious part" of isndds,i.e., the part with the ear-
liest transmitters and the authority to which it goes back.67 It is not
clear, however, on what basis he arrives at his negative judgement with
respect to concrete examples. Possibly, it is based on his general dat-
ing and the relative chronology of the legal problem in question, as he
postulates it, plus the five rules of matnanalysis mentioned above. Yet
even if every one of his examples were examples of isnad fabrication -
which is far from certain - it is not justified to generalise them and to
brand every family isnddfictitious. It seems natural for traditions to be
transmitted to family members. Moreover, if there has been real trans-
mission at all, such cases must have been frequent.

62 Ibid., 171, 175

63
Ibid., 163.
64
Ibid., 163.
65
Ibid., 166.
66 Ibid., 166-169.
67 Ibid., 171.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
222 HARALD MOTZKI

Schacht's name is particularly connected with the phenomenon he

called "common link", a peculiarity of the isndds which was known
already to the classical Muslim Hadatscholars as Schacht rightly pointed
out.68But according to Schacht, the common link is the junction between
the fictitious and the real part of the isnads. The fictitious part which
mostly has the form of a single strand is the part which reaches from
the common link to an earlier authority, e.g., a Companion or the
Prophet.69 The real part consists of the several strands which reach
from the common link to the authors of the collections in which the
hadt in question is found. In this part we have to do with real transmission
which can be used for dating purposes. The common link is then,
according to Schacht, the first who brought the hadit into circulation.
Therefore, it must originate from the time of the common link at the
earliest.70Schacht states about the value of the common link for dating
purposes: "The existence of common transmitters enables us to assign a
firm date to many traditions and to the doctrines represented by them".71
However, Schacht obscures the clarity of what the common link phe-
nomenon is all about by assuming that a tradition cannot be brought
into circulation by the common link himself but "by a person who used
his name".72He expounds his reservation against dating with the com-
mon link with examples in which the common link belongs to the gen-
eration commonly called the tdbi'uin(Successors).73The asnddsof a tradi-
tion which allegedly go back to the same Successor were fabricated,
according to Schacht, a generationlater and the traditionis falselyascribed
to him.74 In these cases the common link can only give "a terminusa
quo".The traditions which allegedly go back to Nafi' and which were
possibly available in form of a written source around the middle of the
second century are such cases.75All transmitters who used that source
referred directly to Nafi' who became, thereby, a common link, with-
out the transmitters having had contact with him. However, Schacht's

68
Ibid., 172.
69 But can "acquire additional branches by the creation of improvements which would
take their place beside the original chain of transmitters, or by the process which we
have described as spread of isnads". Ibid., 171.
70 Ibid., 171-172.
71 Ibid., 175.
72
Ibid., 171.
73 Ibid., 176.
7' G.H.A. Juynboll labels them "seeming common
links".
75 Ibid., 176-179.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIMTRADITIONS 223

conclusions about Nafi"s traditions are, for the most part, mere alle-
gations or statements based on doubtful arguments.76
In spite of these limitations, Schacht considers the common link
phenomenon as a suitable basis for dating traditions.77Schacht already
detected that sometimes isndds bypasses the common link.78 He con-
siders such isnads as fabrications because he thinks that the common
link brought the tradition into circulation. Whether this conclusion is
acceptable or not depends on the question whether Schacht's inter-
pretation of the common link phenomenon is acceptable or not.

2. Juynboll's Method of Isndd Analysis

Schacht himself did not apply the method of dating on the basis of
common links very often in his book The Orginsof Muhammadan Jurisprudence
nor in later publications. It is the merit of Joseph van Ess and G.H.A.
Juynboll to have put the method into practice and to have developed
it further. In the following pages I focus on Juynboll's application of
the method. In his book Muslim Tradition,Juynboll gave a detailed
description of the premises on which the method is based and how it
functions.79 He illustrated for the first time in more detail the phe-
nomenon that common links are sometimes by-passed by isnads - a
peculiarity which Schacht had only mentioned in passing - and gave
an explanation of it. In Muslim Tradition,however, Juynboll was still
reserved and cautious with respect to the benefits of its use. He stated:
"[... ] it is mostly impossible to prove with incontrovertible certainty
that isnads are not invented in their entirety. Thus the common link,
if there is one, is often only a useful tool from which to distil an approx-
imate chronology and possible provenance of the hadath".80
In his article "Some Isnad-AnalyticalMethods Illustrated on the Basis
of Several Women-Demeaning Sayings from IHadathLiterature", pub-
lished six years later, he attaches much more value to the common

76 On the issue of Nafi' traditions see Motzki, "Quo vadis Hadit-Forschung?Eine kri-

tische Untersuchung von G.H.A. Juynboll: 'Nafi' the Mawla of Ibn 'Umar, and his
Position in Muslim Hadtth Literature,"' in Der Islam, 73 (1996), 40-80, 193-231 (English
translation forthcoming).
77 These limitations concern above all the period of the Successors.

78 Juynboll calls them "dives".

79 Juynboll, Muslim Tradition,206-17.

80 Ibid., 214.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
224 HARALD MOTZKI

link for dating purposes.8' Obviously, in the time between the book
and the article he had gained extensive experience with this method.
In what follows I shall focus on his method as set forth in this article.
Juynboll agrees with Schacht on the interpretation of the common
link and of the part of the isnid which goes back from the common
link to earlier authorities. The common link, according to Juynboll, is
the originator of the tradition, or put it otherwise, the tradition "is his
own, or (if somebody else's) he was the first to put it into so many
words"; "the single strand from the cl [i.e., the common link] down to
the prophet [. . .] is a path invented by the cl [... .]". He further states
that (real) common links only appear from the level of the Successors
onwards.83Juynboll seems to consider these statements as methodological
rules, not as statements about historical facts. This is obvious from his
view that the content of traditions may be older than the date arrived
at on the basis of the common link. Since this cannot be proven, it is,
according to him, not possible to go back in dating before the com-
mon link.84However, Juynboll obscures his methodological interpreta-
tion of the common link by statements which follow those of Schacht.
For example, he claims that the common link must be considered the
originator of the tradition.85So far, there is little difference between
Juynboll and Schacht.
Much of Juynboll's refinement of the common link method concerns
the part of the isnddswhich Schacht called "the real part of the isndds",
i.e., the transmitters between the common link and the compilers of
the later Hadatcollections. According to Juynboll, this part is much less
real than it looks. There is, on the one hand, a difference between
strandswhich run from the common link through "partialcommon links"
or "knots" to the collectors, and, on the other hand, "single strands"
who do not cross others. The former alone can be considered histori-
cal, the latter must be suspected of having been fabricated.86The single
strands must be considered unhistorical as long as new sources do not

81 G.H.A. Juynboll, "Some Isndd-AnalyticalMethods Illustrated on the Basis of Several

Women-Demeaning Sayings from Hadith Literature" in Al-Q,antara,10 (1989), 343-84.
82
Ibid., 353.
83 Ibid., 369.

84 Ibid., 381. Juynboll writes: "although the breeding ground of this type of saying

may be older than the proposed date, this is in any case not borne out by the tradi-
tion material discussed here". Cf 370.
85 See ibid., 359 ("more often than not they were just religious dicta which they

ascribed to older authorities and very often all the way back to the prophet") and 369.
86 Ibid., 354.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 225

reveal that the transmitters through whom the single strand goes to the
common link, have indeed passed the tradition to more people, not
only one, and thus become (historical) partial common links.87
With this distinctionJuynboll responds to the objections which Michael
Cook had raised against dating with the common link in his book Early
MuslimDogma.88Of Cook's several objections, I mention here only one:
the possibility that isnddvariants came into being by the process, which
Schacht has called "spread of isnads". Cook emphasises more sharply
than Schacht that spread of isnad can occur on every level of the
transmission process. In doing so, he describes some hypothetical pos-
sibilities of how a tradition could have been transmitted from persons
other than the common link without the isnddsshowing that.89
Juynboll also studies the phenomenon of isnadssometimes by-passing
the common link in more depth than Schacht. He thinks that these
isndds,which he calls "dives", are fabrications for which the authors of
the Hadiitcompilations or their informants are responsible.90This view
is based on two sorts of arguments. Firstly, the dating method on the
basis of the collections in which a tradition first appears. The argu-
ments are as follows: The collector Ibn Hanbal is responsible for these
single strands, "because they are only found in his Musnad'";91or: "every
non-occurrence of an Ibn 'Uyayna strand in al-Humaydil's collection
automatically throws doubt on that strand";92or: this single strand "is
most probably the handiwork of Ibn Hanbal, because the tradition is
not found in 'Abd al-Razzaq's Musannaf".93Secondly, Juynboll thinks
that his view is corroborated by an isnad-cum-matn analysis.94
In order to date the origin of the dives, Juynboll formulated the rule:
"The deeper the 'dive' under the common link, the more recent is the

87 Ibid., 358.
88
M. Cook, Earl MuslimDogma:A Source-Citical Study,Cambridge, Cambridge University
Press, 198 1.
89
Cook's view is discussed in more detail below in the present article.
90Juynboll, "Some Isndd-AnalyticalMethods", 366, 375-77.
91 Ibid., 366.
92 Ibid., 356, note 19.
93 Ibid., 376-78.

94 Juynboll writes: "on the basis of countless analyses of isndd bundles accompanied

by detailed comparisons of the matnsof common link-supported strands with those of

'diving strands', I have come to recognise that the latter almost invariably shows up a
more sophisticated wording and embellishments and short comments, lacking in what
I call the proto-version brought into circulation by the common link". Ibid., 367. For
the methodological approach of the isndd-cum-matn analysis see section IV below and
Motzki, "Quo vadisHadit-Forschung?".

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
226 HARALD MOTZKI

date of origin of that particular strand".95Analogous to Schacht's model

of the development of Muslim traditions, Juynboll postulates that "div-
ing strands" that end in an early Successor have to be considered as
the oldest (albeit "of much more recent origin than those strands that
end in the common link"), "diving strands" ending in a Companion
have to be considered as "of again more recent origin", and those
"ending up in the prophet of the most recent origin".96
Juynboll's studies have improved the method of dating traditions by
means of the isnddconsiderably, especially as far as the study of the com-
mon link phenomenon is concerned. This does not mean, however,
that question marks cannot be put after some of his premises, method-
ological rules and conclusions. Consider, for example, the following
three items.
1) The assumptions that the common link is the originator of the
tradition and that "the single strand from the common link down to
he prophet" is "a path invented by the common link" are problematic
generalisations.Juynboll admits that the common link can have the tra-
dition from a contemporary and that its content may even be older,
but he categorically excludes the possibility that the person named by
the common link as his source really is his informant. He states: "If
he [i.e., the common link] did hear it from somebody else, his isnad
down to the prophet does not contain a clue as to his source.97This
is not plausible. The argument that we cannot know for sure on the
basis of isndd analysis whether the common link really heard the tra-
dition from someone else or not, does not justify the definite exclusion
of the possibility that he did. It is of course possible that a common
link received it from someone else and named his informant when trans-
mitting it as is the normal case in Hadit transmission. A priori we can-
not rule this out and there are means to prove it in some cases. A
closer look at the examples Juynboll uses to illustrate his assumption
that the name of the common link's informant is generallyunhistorical
makes clear that it is groundless.
Juynboll analyses a haditof the Prophet in which the Basran Sulayman
b. Tarihan al-Taymi (d. 143 A.H./760-61 C.E.) is the common link.
He concludes that Sulayman must be the originator of the hadit. As
informant of Sulayman the Basran Abiu 'Utman ('Abd al-Rahman b.

95 Ibid.,368.
96
Ibid., 369-70.
97 Ibid., 359.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 227

Mull) al-Nahdi (d. ca. 100 A.H./7 18-19 C.E.) is given in the isndd.Juyn-
boll considers this claim "highly dubious",98because this Abiu 'Utman
"seems to belong to a generation of early Successors who can rightly
be called very peculiar".99Among the peculiarities are: ripe old age,
topoiin their biographies, settlement in Basra or Kufa and that they
are often monopolised by one particular common link. Surely, these
peculiarities as such do not preclude the possibility of Sulayman hav-
ing received the tradition from Abui 'Utman. In another saying of the
Prophet, analysed by Juynboll in his article, he identifies the Basran
Abtu Raga' al-'Utarid! (d. 107 A.H./725-26 C.E. or 109/727-28) as
"undeniable common link" and, by virtue of it, as originator the hadat.
Juynboll dates this tradition between the eighties and the year 107 or
109 A.H., which is the terminus post quem.'00This Abiu Raga', however,
displays the same "peculiarities"as Aba 'Utman. I wonder whyJuynboll
rejects the latter as a possible informant of Sulayman and perhaps orig-
inator of the tradition because of these peculiarities, but accepts the
former as originator of the hadatin spite of these peculiarities.
Juynboll's conclusions concerning the tradition of Abtu Raga' show
other inconsistencies as well. In the isnddbundle, some of the variants
name the older Companion 'Imran b. Husayn as informant of the com-
mon link Abtu Raga', other isnad variants have instead the younger
Companion 'Abd Allah b. 'Abbas. Both groups of isnads are charac-
terised by partial common links, which speaks, according to Juynboll,
in favour of the historicity of the transmission. The conclusion should
be, then, that Abtu Raga' has named both Companions as his source.
To some of his pupils he must have mentioned 'Imran, to others Ibn
'Abbas and to a third group both. Juynboll concludes, however, that
the strands which go via Ibn 'Abbas "hail from later times" and means
the time of the authors of the large compilations which emerged in the
course of the third century.1'OThis conclusion contradicts, on the one
hand, his premise that the common link is not only the originator of the
text of the tradition but also of the single strand which refers to earlier
authorities, on the other hand, it contradicts his view that partial com-
mon link transmissions are historical.
I propose two alternative interpretations, one of the common link
phenomenon and another of his single strand. The early common links

98
Ibid.,359.
99 Ibid., 360.
'?? Ibid.,370.- d1tethn'ma)
'?' Ibid.,364-365 (since Ibn 'Abbasdidlaethnamd)
r

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
228 HARALD MOTZKI

(the generation of the Companions excluded) were the first major col-
lectors and professional disseminators of knowledge in general, and of
traditions about individuals of the first Islamic century in particular.
This does not exclude the possibility that some traditions were revised,
combined or even invented by the common links.
The single strand of the common link reproduces first of all the
name of the informant from whom the common link received or alleged
to have the tradition, then the way by which the informant claimed to
have received it or the way by which the common link thought that
his informant has got it. Juynboll rejects the single strand as fictitious,
with the argument that if it were a real transmission path, we must
expect to find not only this one but many others. Such an expectation
is however not plausible if we assume that a common link is a major
teaching collector. In this case, his single strand reflects only the path
he has mentioned. The single strand does not signify that this has been
the only channel through which the tradition was spread. Other chan-
nels of transmission may have existed but remained unknown because
they were not quoted by one of the major collectors and professional
disseminators of traditions, or perhaps found the way only into collec-
tions we do not know (yet).'02This explanation of the common link
phenomenon seems more in harmony with our knowledge concerning
the transmission processes during the first three Islamic centuries than
that ofJuynboll and Schacht. Whether it is a workable hypothesis has
to be tested.
2) Another problem is Juynboll's division between historical and
unhistorical common links. It says that only those transmission lines
can be considered historicalwhich contain partial common links between
the common link and the authors of the collections. In these cases
Juynboll speaks of a "real common link" which has to be distinguished
from a "seeming common link". "Spidery bundles", i.e., transmissions
which consists mostly of single strands between the common link and
the collections, must be considered unhistorical. He expresses this the-
sis in the form of a general rule which he calls "a major adage": "the
more transmission lines there are, coming together in a certain trans-
mitter [... ] the more that moment of transmission [...] has a claim
to historicity".103

102 See also Motzki, "Quo vadis HadAt-Forschung?".

103
Juynboll, "Some Isnad-Analytical Methods", 352.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 229

This general rule is plausible. It is not, however, very practical, as

his own examples prove. When we look atJuynboll's diagrams illustrating
his isnddanalysis, it is striking that in many cases partial common links
appear almost exclusively on the level of those transmitterswho directly
refer to the common link. They show up more rarely in the later gen-
erations of transmitters. Most transmission lines between these partial
common links and the collections are single strands.'04If one takes seri-
ously Juynboll's general rule, then most of the traditions dealt with in
his article must be considered unhistorical. Obviously, Juynboll's con-
ditions for traditions to be accepted as historical make too great a
demand on the scanty sources available to be useful in practice.
What an usnadbundle of a tradition shows are its transmission lines
which we find in the compilations of a few (seldom more than a dozen)
later collectors. Each of these compilators reproduce one or more trans-
mission lines by which he has or claims to have received the tradition.
If one assumes that there has been transmission of traditions at all, and
that (at least) a part of the transmission lines between the compilations
and the common link are real ones (an assumption with whichJuynboll
agrees), then, they can only represent a tiny part of the many trans-
mission lines which must have actually existed. Therefore it cannot be
expected that the part reflects the structure of the whole. Partial com-
mon links or "knots" in the transmission lines are ideal cases which
our sources by chance can reveal. The demand that only those trans-
mission lines which are completely filled with partial common links
must be accepted as historical would leave us with only a few histori-
cal traditions.'05
3) The last point I would like to discuss is Juynboll's claim that the
"diving single strands" are fabricated by the authors of the HadAtcol-
lections or their informants. This claim is based mainly on e silentio
arguments of which examples have been given above. In a more gen-
eral form the argument goes as follows: Isnddswhich contain the names
of early collectors of whom HadAtcompilations are available (e.g., Malik
or 'Abd al-Razzaq) or names of individuals (e.g., Ibn 'Uyayna or Su'ba)
which are richly documented in the collections of their pupils (e.g.,

104 See ibid., 349, 351, 363, 365, 368. Diagram 7 (p. 373) displays a
"spider" with
Dawuid b. Qays as "knot". In spite of it, Juynboll considers him an "unmistakable"
c[ommon] l[ink]. Ibid., 372.
105 For a more detailed presentation of my arguments see Motzki, "Quo vadis HaAt-
Forschung?".

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
230 HARALD MOTZKI

al-Humaydi or al-Tayalisi) but which are only found in later Hadatcol-

lections (e.g., those of Ibn Hanbal or Muslim) are fabrications. For if
these isnddswere authentic ones, we could expect to find them in the
earlier compilations already. This argument is based on the assump-
tion that the earlier compilations contain completely or exhaustively the
transmitted material of the persons in question. For the second and
third century, at least, this assumption appears unlikely.
Firstly, the material which teachers passed on to their pupils during
lectures and seminars was, for practical reasons, surely only a selection
of what they had learned or possessed themselves. Secondly, it seems
likely that the content of the lectures was not always the same in the
course of decades, so that not all the students of a teacher learned the
same subjects from him. Thirdly, it is doubtful whether all early col-
lections which were compiled by a scholar and transmitted by his pupils
were transmitted completely and accurately. We must reckon with
processes of edition, selection and rearrangement.'l06

3. Michael Cook's Critique of Dating by Employing the Common Link

The most detailed and subtle critique of the dating method which makes
use of the common link phenomenon has been put forward by Michael
Cook. He has articulated his general reservations first in the chapter
"The dating of traditions" of his book Early Muslim Dogma, a critique
of Joseph van Ess' study ZwischenHada-tund 7heologie.Later, he tried to
prove the unreliability of the method with some examples.'07 Cook's
reservation about dating with the common link derives from two sorts
of arguments:first, general considerations concerning the value of knowl-
edge in early Islamic civilisation and concerning the motives of forgery,
secondly, concrete ideas about how isnddfabrication had happened or
possibly could have happened. Cook's general considerations are based
on the following assumptions:
1) "In a traditionist culture [... ] the relevant value is not originality
but authority: sharp practice consists in falsely ascribingmy view to a

106
As regards Juynboll's additional argument based on the matnsof the dives, one
can wonder whether this is not a rash generalisation. Besides, it starts from the assump-
tion that generally shorter matnsare earlier, which is far from certain. A comparative
study of the matns of common links and of diving transmissions would be most wel-
come.
107 In his article "Eschatology and the Dating of Traditions" in PrincetonPapersin Near
EasternStudies,1 (1992), 23-47.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 231

greater authority than myself".108 2) Transmission lines have to be as

short as possible in order to be considered elegant. 3) "Isolated tradi-
tions", i.e., traditions which have been only transmitted with one or a
few varying isndds,"09 are not accepted as proof. These three "values"
result in a sort of compulsion of "the system" that makes forgeries
acceptable and plausible."0
Is the argument implied here convincing? Let us begin with the first
assumption. I wonder whether the Muslim educational system during the
first and second century after the Hijra is adequately characterisedby the
label "traditionistculture" in which "the relevant value is not originality
but authority"."' This is a black and white picture. Was the educational
system in early Islam really so one-sided or is it more fairly described
as a system in which both values, originality and authority, played a
role? Just the first Islamic centuries were characterised by a conflict
between both values, a conflict in which authority eventually triumphed,
but victory was not yet won at the end of the second century.
According to Cook, the first assumption explains the phenomenon
of what Schacht has called "backwards growth of isndds", i.e., "the
process whereby [. ..] isnddsare 'raised' from oneself to one's teacher
to his teacher and ultimately to the Prophet"."2 Cook illustrates the
phenomenon with an anecdote which reportsthat 'Amr b. Dfnar ascribed
a saying to Ibn 'Abbas, but when a colleague spoke to 'Amr about it,
the latter admitted that he had received the saying from an informant
without explicit mention that it was a saying of the Companion. This
prompts Cook to formulate the rule: "Where one isnad reaches only to
A and a second goes back through him to his teacher, then given the
values of the system we are entitled to suspect that the higher isnad is
secondary, rather than the other way round"."3
Is it legitimate to draw such a general rule from the anecdote? If
we generalise the report about 'Amr, it says that it could happen that
a scholar ascribed a saying of his informant or teacher to an earlier

108 Cook, Early Muslim Dogma, 107-108.

'09 See for a more precise definition G.H.A. Juynboll under the entry "Khabar al-
wahid", in Engyclopaediaof Islam, 2nd Edition, ed. B. Lewis et al., Leiden, EJ. Brill,
1960-, IV, 896. Henceforth abbrev. El2.
"? Cook, Early Muslim Dogma, 11 1.
"' Ibid., 108.
112 Ibid.
113
Ibid.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
232 HARALD MOTZKI

authority whom the informant had not mentioned. What does this
prove? First, the anecdote does not say that it was the opinion of the
scholar himself, which he ascribed to an earlier authority, as Cook sup-
poses, calling it a "sharp practice". Secondly, if the anecdote has a
historical value at all and did not result from rivalry between different
centres of scholarship, then the anecdote merely proves that such cases
of dishonesty or inaccuracy happened. No one would deny that. Yet
Cook's conclusion reaches further; he suggests that it was "a system",
i.e., early Muslim scholars generally behaved in this manner. Based on
one anecdote (or even a few more), is such a conclusion warranted?
The second method of forgery which Cook considers to be a con-
sequence of the system of values and exerting pressure on the scholars
is "spread of isndds",i.e., - as Schacht has put it - "the creation of
additional authorities or transmitters for the same doctrine or tradi-
tion"."'4 This sort of forgery is more tricky than others because in-
stances of it are more difficult to detect. Moreover, according to Cook,
it has serious consequences for the dating on the basis of common links.
Cook distinguishes between three types of "spread of isndds".The
first type is one where a transmitter ascribes a tradition which he has
only received from a contemporary or a fellow student to the latter's
teacher thereby suppressing his real informant in the isnad.By this trick,
the latter's teacher becomes a common link which in fact he is not.
This does not fit precisely the definition which Schacht gave of "spread
of isndds",namely, "creation of additional authorities or transmitters".
In these cases transmitters or authorities are omitted from the isnadd.
Cook imagines that in the same manner even two levels of informants
could be jumped over and an even earlier virtual common link could
be created. That such things really happened, Cook demonstrates by
anecdotes with doubtful historical reliability, such as that about Sufyan
b. 'Uyayna who is reported to have suppressed in an isnad two trans-
mitters."5 This anecdote is aimed at portraying Ibn 'Uyayna as a bad
transmitter, whereas other anecdotes say just the contrary, namely that
he tried to hear personally from an old scholar whose traditions he
received from someone else.
Next to "striving for higher authority", Cook mentions in this con-
text his second "basic value of the system", namely elegance. Again,
he illustrates this with anecdotes to the effect that shorter isnddswere

114
Schacht, The Originsof Muhammadan 166.
Jurisprudence,
115 Cook, Early Muslim Dogma, 111.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 233

preferred to longer ones."6 However, several of these anecdotes only

report that there had been scholars who tried to hear a tradition which
they had heard from a colleague (in one case a much younger scholar),
from the source himself."7 It is not mentioned in the anecdotes that
they did so in order to have a more elegant isndd.It seems somewhat
doubtful that in the first half of the second century after the Hijra the
elegance of isnads was already an issue. The motive behind the be-
haviour of those scholars was the desire to transmit from an old and
famous scholar themselves instead through a younger colleague because
it provided more scholarly prestige. A similar but less anecdotal report
is transmitted for example from Yah.ya b. Yahya al-MasmtidT,the trans-
mitter of the most current version of Malik's Muwatta'. It is reported
that Yahya had studied Malik's Muwatta'in al-Andalus with his teacher
Ziyad b. 'Abd al-Rahman, but then travelled to Medina in order to
hear the same text from Malik himself."8 Did he make the voyage from
al-Andalus to Arabia only in order to have a more elegant isnad?Besides,
the anecdotes quoted by Cook prove not only dishonesty but also the
contrary, namely that there were also scholars who did not suppress
their informant in order to be able to relate from an older authority.
Cook's second variant of "spread of isndds"is the hypothetical case
that a transmitternot only conceals his direct informant but also replaces
the latter's teacher by his own teacher. In this way the transmitter cre-
ates a virtual common link two generation before himself, because his
fictitious informant (his real teacher) and the teacher of his real infor-
mant appear as pupils of the same early authority. This would be a
true "spread of isndds",according to Schacht's definition, since in this
case a fictitious transmitter is added. Cook gives no evidence for this
hypothetical case, not even an anecdote.
Speculation over possible kinds of forgery can be carried on until it
finds its natural end in the assumption that whole isnddscan be fab-
ricated and added arbitrarily to traditions. Cook gives no documenta-
tion for this third type of "spread of isndds",which has been mentioned
already by Schacht.'19

116
Ibid., 109.
17 Ibid., 202-203, note 7.
118 Cf Zurqani, Muhammad 'Abd al-BaqY,Sarh 'aid Muwatta' al-imdmMdlik, Beirut,

1990, 1, 19.
19 There are, of course, anecdotes which report such cases; see for example Goldziher,
Muhammedanische Studien,II, 160.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
234 HARALD MOTZKI

Touching on the last two types of forgery, Cook mentions his third
value of "traditionist culture": "the objection which used to be made
to 'isolated' traditions". Supposedly, this compelled Muslim scholars to
forge isndds.'20But as in the case of the other two "values of the sys-
tem", the assumption that this value played an important role as a dri-
ving force in isnad forgery is too vague and undifferentiated. Was the
objection to "isolated traditions" already an issue in the first century
and during the most part of the second century? It seems doubtful that
this motive applies to the isnadsof this period. In addition, one wonders
whether the reservationagainst al-habaral-wahid(isolatedtradition)affected
all sorts of traditions equally or only one genre, namely, legal Hadat.
Cook himself wonders whether "spread of isnads"was "a process
operative on a historically significant scale" and admits "that the evi-
dence does not lend itself to a conclusive answer".'2' He thinks, how-
ever, "that some store must be set by the fact that the process [of the
"spread of isndds"]as outlined is thoroughly in accordance with the
character and the values of the system".'22In view of the reservations
which can be brought forward against Cook's "values" - the gist of
these reservations being that Cook's ideas are, historically speaking, too
vague, undifferentiated and provable only by anecdotes - the whole the-
ory that there existed a compulsion to forgery by reason of those val-
ues is unconvincing. Besides, it is doubtful, first, that "spread of isndds"
was really practised "on a significant scale", second, that all isndds,no
matter the genre of the tradition, are affected equally, thirdly, that
"spread of isnads"was practised at all times to the same extent, and,
finally, that the assumption of forgery on a huge scale applies also to
the traditions contained in the collections of the critical Hadatscholars.
Based on his conviction, that forgery was stimulated by certain cul-
tural values, Cook concludes that scholars who are using hadatstoday
have only two methodological choices: if they deny that "spread of
isndds"happenedon a significant scale, they must also accept mutawatir
traditions'23as historically reliable; if they admit, on the contrary, that
Muslim scholars forged isnads in huge dimensions, they must give up
the idea that it is possible to date traditions on the basis of the isndds,
in general, and the common link phenomenon, in particular.

120 Cook, Early Muslim Dogma, 110.

121 Ibid., 1.
122 Ibid.
123 See for them p. 38.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 235

However, in view of the reservations against his arguments, these

are not the only positions which can be chosen. Neither Schacht nor
Cook have convincingly shown that "spread of isndds"was really prac-
tised on a significant scale. They have only shown that there were sev-
eral possible ways how isnddscould be forged and that Muslim schol-
ars could have had different motives to do so. Apart from possibilities,
Schacht and Cook produced only scarce evidence that isndd forgery
really happened.
On the basis of mere possibilities and a few instances of the real
forgery, it makes no sense to abstain completely from using the isndds
for dating purposes. The historians of the European Middle Ages would
not abstain from using diplomas as historical sources because there were
cases of forgery which are not easy to detect. The intention of the isnad
system was to assure the reliability of the transmission process. The
basic value linked with it was that I have to name the informant from
whom I had received the information. Doing otherwise intentionally
was forgery and dishonesty. Certainly, this must have been clear to
anybody familiar with that system and the whole scholarly community
as a whole must have watched to ensure that its norm was not vio-
lated. This does not exclude that forgery could happen. But it seems
unlikely that it happened at a huge scale in scholarly circles, not to
speak of the circles of muhadditun.'24 If the scholarly isnad system was
only or mainly used to feign reliability, then the whole system of val-
idating traditions by isnadswould have become absurd. Al-Safici's insis-
tence on traditions with reliable isnadswould have been pointless and
hypocritical if he had been convinced that most traditions available in
his time were equipped with fabricated isndds.Was the whole system
of Muslim Hadatcriticism only a manoeuvre of deception? Who had
to be deceived? Other Muslim scholars? They must have been aware
of the pointlessness and vanity of all the efforts to maintain high stan-
dards of transmission, if forgery of isnddswas part and parcel of the
daily scholarly practice.'25
It seems, therefore, more appropriate to keep the premise that, gen-
erally speaking, the isnadsystem served the expectations of the traditionist.
Otherwise, we would expect that they would have quickly abandoned
it. Until we have proof to the contrary, we must, therefore, presume

124Their use of isnddsmust not be equated with that of popular storytellers.

125 See also my "The Prophet and the Cat: on Dating Malik's Muwatta' and Legal
Traditions" in JerusalemStudiesin Arabicand Islam, 22 (1998), 32, note 44.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
236 HARALD MOTZKI

that isnads are, in principle, reliable, except, perhaps, around the time
when the system came into being. Still and all, we have to be on our
guard against possible cases of error, well meant improvement or forgery
in the isndds. The question of the possible motives for improvement
and forgery, an issue which occupied Cook in his chapter, "The dat-
ing of traditions", may help in this connection, as may the study of
the relationship between isnads and matnsof a particular tradition, a
method to be dealt with in the next chapter.
Before we leave the issue of the type of isnddanalysis which is focused
on a single tradition, Cook's attempt to check the reliability of Schacht's
dating methods will be reviewed. In his brilliantly argued article
"Eschatology and the Dating of Traditions",'26he examines traditions
which "can be dated on external grounds", i.e., independently of the
isnads.'27Such an external dating seems to be possible in case of tra-
ditions which predict certain historical events, though not quite cor-
rectly. This reservation is important, for a tradition which predicts what
really happened could have been created post eventumand is thus not
suitable for an external dating. Cook rightly assumes that half-true pre-
dictions must originate from a time before the predicted event had
actually happened.'28 However, a crucial point for dating on external
grounds is that there should be no doubt that the prediction in ques-
tion really intended a particular historical event.
Among the three examples Cook presents in his article, only two are
datable with some certainty on external grounds: the tradition about
"the reign of Tiberius, son of Justinian" (which is to be dated between
93 A.H./711 C.E. and 119 A.H./737 C.E.) and the tradition about
"Ibn al-Zubayr and the Mahdii" (datable into the time of the Second
Civil War, i.e., 64-73 A.H./683-692 C.E.).'29These two traditions are
preserved in several variants with different isnads and thus permit an
isndd analysis. Cook examines whether the dating based on an isndd
analysis, following the principles defined by Schacht, is corroborated
by the external dating of the two traditions. His conclusion is that "in

126 Cook, "Eschatology and the Dating of Traditions", 23-47.

127 This method belongs, actually, to my fifth category of dating methods which is
not discussed in this article (see note 5). Cook's article is, nevertheless included here
because of its importance for isndd analysis.
128
Cook, "Eschatology and the Dating of Traditions", 25.
129 In the case of the "tradition on the Andalusian invasion of
Egypt", Cook admits
"the possibility that the apparent relationship between prophecy and event is fortuitous".
Ibid., 29.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 237

none of the [...] cases examined does the obvious external dating
match the obvious Schachtian dating".'30Among Schacht's criteria of
dating the following proved to be unreliable: first, the general dating
"that fabrication of legal traditions began only around the year 100
A.H./718 C.E."; secondly, the rule "that the better the asnadin terms
of the classical norms of Hadatcriticism, the later the real date of the
tradition";'3'thirdly, the claim that traditions going back to the Prophet
are the most recent ones; and, finally, the common link method. Cook
summarises the result of his study in the sentence: "There is nothing
in my findings that could serve to encourage recruitmentto the Schachtian
school" of dating traditions.'32This discovery, based on two concrete
examples, is not surprising in view of the theoretical weaknesses of
Schacht's dating methods which I discussed above.'33Based on a much
broader textual basis and by using other methods of analysis, I came
to a similar conclusion in my study Die AnfdngederislamischenJurisprudenz.'34
However, I am convinced that some of the criteria proposed by
Schacht for the dating of traditions remain valuable, under the condi-
tion that they are freed of their Schachtian interpretation and that they
are removed from their connections with his other principles of dat-
ing. The common link method is one of the principles which deserves
more confidence than Cook is prepared to give it. In his amiable man-
ner of discussion, Cook writes: "I am sympathetic to critics of this
papers who urge that the existence of common links must mean some-
thing; but just what it means, I do not pretend to know".'35
In view of the results which the examination of Schacht's dating
methods have produced, Cook's reservation is understandable. Schacht
considered the common link as the originator or fabricator of the tra-
dition in question and denied emphatically that he had transmitted it
from the preceding generation and even less from the individual he
named as his informant. Cook, on the other hand, came to the con-
clusion that in the two cases which are datable on external grounds,

130
Ibid., 35.
131
Ibid.,24.
132 Ibid., 35.
33 See pp. 7-9, 11-12, 16-20.
134
Except for the common link phenomenon. See note 24. Cook's and my own neg-
ative judgement on Schacht's dating method affects his method as a whole, the interplay
of individual premises and rules on which his instances of dating are based, and the
interpretation which Schacht gave to certain phenomena occurring in the field of HadLt.
135 Ibid., 46, note 74.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
238 HARALD MOTZKI

the common links cannot be the originators of the traditions in ques-

tion, because these traditions must be earlier.
But Cook's conclusion that "the common link method performs
poorly" has to do with the fact that he slavishly follows Schacht's idea
of what the common link represents.'36When discussing Schacht's and
Juynboll's interpretationof the common link phenomenon, I have already
suggested that an alternative interpretation is possible and that there
are reasons in many cases not to consider the common links as the
originators or rather fabricators of the traditions in question but as the
first major collectors and professional disseminators of traditions.'37One
reason is my conviction (which I share with Schacht) that the trans-
mission lines which fan out from the common link onwards until they
reach the later collections are, for the most part, the real paths of trans-
mission (Schacht's "real part" of the isnddstructure). This is not meant
to exclude occasional instances of improvement or fabricationsof isndds.38
If most of the transmitters of the isnad strands which fan out from
the common link onwards are real transmitters,it is implausible to deny
a prioriand categorically that the common link could be a real trans-
mitter as well. Why must the common link always be a fabricator?
Why could he not have received the tradition (at least the gist of it)
from the person he gives as his informant? I have discussed the argu-
ments brought forward by those who regard the common link as the
fabricator elsewhere in some detail.'39So in the present context I shall
reject this interpretation of the common link without additional argu-
ment. That the common links which appear from the generation of
the successors onwards should be considered mostly as major collectors
is not meant to exclude that, occasionally, they can have invented a
tradition, changed - intentionally or by mistake - the matnor isnad of
a tradition which they had received, or can have not given the right
names of their informants.
If one applies this conception of the common link to the two eschato-
logical traditions which Cook served to refute the common link method,
it becomes obvious that the external dating fits very well the position of
the common link in the isnddbundles. In the case of the tradition con-

136
Ibid.,35.
137 This definition concerns only common links from the generation of the Successors
onwards, not those on the level of the Companions. The latter must not be mixed up
with the former. See below p. 38.
138 For the reason why only occasional see above pp. 31-32.

139 See Motzki,

"Quo vadis Hadft-Forschung?".

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 239

cerning "Ibn Zubayr and the Malidil", the isnaddbundle suggests Aba
al-Halil as source of the common link Qatada (1 17 A.H./735-6 C.E.).'40
The tradition about "The reign of Tiberius" is more complicated
because it represents the rare case of a tradition with two independent
common links who are contemporaries but lived in different countries.
In addition, each common link has different informants, a fact which
is also somewhat rare. In such a case it is difficult to decide what hap-
pened. In order to explain how the different informants and isnads of
each common link came into being, one could imagine that the com-
mon link received the tradition from more than one informant, possi-
bly with different matnsand isnads, and transmitted them at times as
individual traditions with their original matn,at other times in the form
of a combined tradition, that is with one matnbut varying isnadds. The
fact that a similar tradition appears both in Syria and Egypt without
showing a common source in the isnadscan, perhaps, be explained as
diffusion. The turn of the first century was certainly a period which
was susceptible to eschatological predictions which must have been cir-
culating widely and quickly as often happens at turns of centuries. There
could have been slightly different versions spread by different people.
The several names which appear as informants of the common links
can reflect this situation; they must not necessarily be fictitious. The
informants of the common links are the generation in which the tra-
dition must have its origin according to the external dating. If we con-
sider the common links not as originators/fabricators but as collectors
of the traditions the dating based on the common link fits perfectly the
external dating.141
The thesis that the common link phenomenon can best be explained
by the collection and the systematic or institutionalised, as it were,
spreading of traditions in circles of scholars, takes the fact into account,
already realised by Schacht, that most of the common links we find in
Hadatliterature belong to the first three generations active during the
second century, i.e., the time between roughly 100 and 175 A.H. 142

140
He is Salih b. Abi Maryam al-Duba'c, his date of death is unknown. Cf Ibn
Hagar, Tahdtb,IV, 402-403.
141 For the issue of dating eschatological traditions by the common link see the recent

article of A. Gorke, "Eschatology, History, and the Common Link: A Study in

Methodology" in H. Berg (ed.), Methodand Theogyin the Studyof Islamic Origins,Leiden,
EJ. Brill, 2003, 179-208.
142
Schacht, The Originsof Muhammadan Jurisprudence, 174-175.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
240 HARALD MOTZKI

This is exactly the time in which the first substantial collections of tra-
ditions were compiled and their material transmitted more systemati-
cally in institutionalised scholarly circles. It became the basic material
of the more substantial collections compiled at the end of the second
and during the third century A.H.143
When transmitting their material to their pupils, the first major col-
lectors mostly confined themselves - probably for practical reasons - to
give the name of only one informant, even if they had heard the con-
tent from more than one. This assumption explains the fact that the
part of the isnadswhich reaches from the common link back to earlier
authorities has mostly the form of a single strand. The rarer cases in
which two or more names are mentioned as informants of the com-
mon link show that occasionally scholars found it suitable to indicate
that they had received the information from different people. This pro-
cedure was particularly called for when the collector combined different
reports to a single new one. In such cases the common link transmit-
ter could give the names of all his informants in the isndd (collective
zsnad)or mention sometimes one of the informants, sometimes another
one. This explanation of the relation between the common link and
his informant(s) is not meant to exclude the possibility that difference
in the names of the informant of the common link can also be caused
by carelessness or intentional change on the part of the common link
or a later transmitter.
The idea that most common links from the tdbicungeneration onwards
were collectors, not fabricators, has consequences for the dating of their
traditions. Then, the time of the common link's activity as a scholar
is, in many cases, not the terminus post quemhis traditions have existed
(as Schacht and Juynboll claimed), but the terminusante quem.We are
entitled to assume that the common link received the tradition - at least
the gist of it - from the individual(s)he gives as his informant(s) as long

14' The generation of the earlier tabi'un (fl. in the last third of the first century A.H.

must - at least partially - be added to this category of common links, even if they
appear more rarely. The common links of this generation, such as 'Urwa b. al-Zubayr
and Sa'cd b. al-Musayyab, can be integrated in the conception of collectors in assum-
ing that they were the first minor collectors of traditions which were active at a time
in which the system of institutionalised scholarship was only just starting. Actually, it
was the generation of the teachers of the first major collectors, such as Ibn Sihdb al-
Zuhri. The scholars from among the early tabi'in appear more rarely as common links
because their teaching circles were yet small, and because only a few of their pupils
became famous scholars themselves and collected and spread traditions systematically.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 241

indications are lacking to the contrary.'44The informant of the com-

mon link is crucial to the dating of the tradition, not the common link
himself. The informant's date of death - or, more exactly, the time in
which the common link had contact with him - is the terminus post quem.
Epistemologically, the shift of the terninuspost quemfrom the common
link to his informant(s) is accompanied, by a decrease in certainty.
We can be sure that the tradition existed at the time of the com-
mon link (if he is a real one) because two or more people transmitted
it from the common link, but the isndddoes not give us a clue to decide
whether the common link really had the tradition from the person he
gives as his informant. This loss of certainty is compensated to a certain
degree. It is compensated first by the legitimate premise that in most
cases the collector will have given the real informant. For there is no
reason to assume that most collectors invented them. Some collectorsmay
have had reasons to hide their real informant in some cases or may
have boasted occasionally with a spectacular authority. But these cases
must be considered the exceptions, not the rule. Secondly, a system-
atic investigation of a common link's practice of giving his informants
can help to judge the reliability of a common link.'45
The conception of the common link as systematic collector does not
explain common links which belong to the generation of the sahabaor
the case of the Prophet himself being the common link. These cases
require other explanations. Since this issue has not yet been sufficiently
studied, I confine myself to a few remarks. We should differentiate
between traditions in which only one Companion appears as common
link and mutawatirtraditions in which the Prophet is the common link
from which several companions are alleged to have transmitted. In the
first case, we cannot exclude a priorithat such traditions are the relics
of a real transmission process. Circles of pupils developed probably
already around some Companions. Out of these circles grew the first

14' Such indications can be detected for example by studying systematically the tra-

ditions of the common link in question and by examining his practice of giving infor-
mants. For examples of such investigationssee Motzki, Die AnfdngederislamischnJurispmudenz;
idem, The Originsof IslamicJurisprzdece,idem, "Der Fiqh des-Zuhr!:die Quellenproblematik"
in Der Islam, 68 (1991), 1-44; English transl., "The Jurisprudence of Ibn Sihab az-Zuhri.
A Source-critical Study" (Nijmegen, 2001, http://webdoc.ubn.kun.nl/mono/m/motzki_h/
juriofibs.pdf), 55 pp.
145 There are only very few investigations of this kind until now. For the common

link Ibn Ourayg, see Motzki, Die Anfangeder islamischen passim and esp. 209-
Jurasprudenz
212; idem, The Originsof IslamicJurisprudence, 234-238.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
242 HARALD MOTZKI

scholarly circles of the early tdbi'ungeneration in the great administra-

tive and intellectual centres of the early Islamic world.'46 Besides, we
must reckon with family or tribal traditionswhich focus on a Companion
and may be very early.'47 Such traditions could even have been pre-
served for some time independently of the scholarly circles and could
have been covered by the systematic collection of traditions only at a
later date.'48 How traditions developed in which the Prophet appears
as a common link, I do not know. I assume that there is no general
solution of this problem; different possibilities must be accounted for.
In any case, the issue deserves further study.

B. SourceReconstruction
on the Basis of Isnads
Our knowledge of the first two centuries of Islam is based on sources
dating mostly not earlier than the third century. These sources pretend
to be collections of pieces of information which are much earlier and
which were transmitted by several generations of scholars. Yet from
later sources we also know that there were already earlier collections
compiled during the second or even the first century, of which only
the titles and the names of their putative authors are preserved. For
the historian of early Islam, who has to rely on the later sources be-
cause they are almost the only ones available, the question arises whether
these sources really contain earlier material and, if so, how early it is.
This question can hardly be answered by analysing the content of the
texts. The isnads,on the contrary, can be of much help. Many sources
provide chains of transmitters for every piece of information or give,
at least, the name of the person of whom the information is said to
derive. If these chains of transmitters are not entirely fictitious, they
can tell us something about the history of the texts before they became
taken down in the later collections.
The possible usefulness of the isnddsfor the "reconstruction of ear-
lier sources" - this expression which will be clarified below - was realised
by Western scholars already in the nineteenth century. Thanks to their

146 Cf
H. Motzki, "Die Entstehung des Rechts" in A. Noth/J. Paul (eds.), Der islamische
Orient.GrundzigeseinerGeschichte,
Wurzburg: Ergon Verlag, 1998, 167 if.
147 I
have reconstructed such Companion traditions in "The Prophet and the Cat"
and "The Murder of Ibn Abi l-Huqayq". For another example see D.S. Powers, "On
Bequests in Early Islam".
148 An example of an early tradition emerging only in later collections is the Surraq

kadft. Cf H. Motzki, "Der Prophet und die Schuldner. Eine had-t-Untersuchungauf dem
Prufstand" in Der Islam, 77 (2000), 1-83.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 243

experience in Biblical studies, early scholars working in the field of

Islamic history like Julius Welihausen, not only had a keen sense for
''source reconstruction" but they also had right at their disposal most
suitable sources for such a reconstruction, namely, al-Tabarl-'s(d. 310
A.H./923 C.E.) monumental Ta'rThal-rusulwa l-muluikand al-Baladuri's
(d. 279 A.H./892 C.E.) FutWhal-buldan.In these works every report is
equipped with an isnad.By investigating the isnadsof a compilation and
looking for common transmitters in its transmission lines, material of
earlier compilers was detected, such as Ibn Ishaq (d. 150 A.H./767
C.E.), Abiu Mihnaf (d. 157 A.H./775 C.E.), Sayf b. 'Umar (d. 180
A.H./796 C.E.), al-Waqidl (d. 207 A.H./823 C.E.) and al-Mada'ini
(d. 228 A.H./843 C.E.).'49Most scholars of early Islamic history accepted
the method in principle, although some details of its application remained
controversial.'50It was also used in other fields of Islamic studies, such
as tafsar,Hadatin the stricter sense, and adab.'5'An investigation into
the sources of al-Tabarl's voluminous commentary of the Qur'an, for
example, detected quite a number of substantial earlier collections on
which his Tafsaris based. Among them are such early ones like that
of Ibn Abi Nakils (d. 131 A.H./748-9 C.E.), Ibn Gurayg (d. 150 A.H./
767 C.E.), Ma'mar b. Rasid (d. 153 A.H./770 C.E.), Abtu Ga'far
al-Razi (d. ca. 160 A.H./777 C.E.), Asbat b. Nasr al-Hamdani (fl. first
half of the second A.H./eight century C.E.), Ibn Ishaq (d. 150 A.H./
767 C.E.) and Sufyan al-Tawri (d. 161 A.H./778 C.E.).'52Among the

149
See J. Wellhausen, "Prolegomena zur altesten Geschichte des Islams" in his Skizzen
und Vorarbeiten, Berlin, 1844-1899, VI, 1-60.
150 See, e.g., A. Noth, "Der Charakter der ersten grof3enSammlungen von Nachrichten

zur fruhen Kalifenzeit" in Der Islam, 47 (1971), 168-199; U. Sezgin, Abu Mihnaf Ein
Beitragzur Historiographie der Uma'yadischen Zeit, Leiden, EJ. Brill, 1971; G. Rotter, "Zur
Uberlieferung einiger historischer Werke Mada'inis in Tabaris Annalen" in Oriens,23-
24 (1974), 103-133; Kh. Athamina, "The sources of al-Baladhuri's Ansdb al-ashrdf', in
3erusalemStudiesin Arabicand Islam, 5 (1984), 237-262; J.A. Bellamy, "Sources of Ibn abi
'l-Dunya's Kitdb Maqtal Amaral-Mu'minan"in Journal of the AmericanOrientalSociey, 104
(1984), 3-19; S. Leder, Das Korpusal-Haitamibn 'Adi (st. 207/822). Herkunft,Uberlieferung,
Gestaltfilher Texte der Ahbdr Literatur,Frankfurt am Main, Vittorio Klostermann, 1991;
S. Gunther, Quellenuntersuchungen zu den "Maqdtilat-Taliby fn" des Abu l-Farag al-Isfahani
(gest.356/967), Hildesheim, Georg Olms Verlag, 1991.
15' For the field of adab see L. Zolondek, "The sources of the Kitdb al-Aghdni"in
Arabica,8 (1961), 294-308; F. Fleischhammer, "Hinweise auf schriftliche Quellen im KIitab
al-Agdna " in Wissenschaftliche derMartin-Luther-Universitdt
Zeitschrifl Halle- Wittenberg.
Gesellschafts-
undsprachwissenschaftliche Reihe,28 (1979), 53-62; W. Werkmeister, Quellenuntersuchungen zum
Kitdb al-'Iqd al-farid des AndalusiersIbn 'Abdrabbih(246/860-328/940): ein Beitragzur ara-
bischenLiteraturgeschichte
in series Islamkundliche
Untersuchungen70, Berlin, K. Schwarz, 1983.
152
Cf H. Horst, "Zur Uberlieferung im Korankommentar at-Tabaris" in Zeitschrift

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
244 HARALD MOTZKI

collections of Had-tin the strictersense, only al-Buhari's(d. 256 A.H./870

C.E.) al-6dmi'al-.saUhand 'Abd al-Razzaq al-San'ani's (d. 211 A.H./826
C.E.) Musannafhave been systematically investigated with the method
until now.'53
Although the use of the isndds for the "reconstruction of earlier
sources" is in principle accepted, there is much disagreement about it
in detail. First, how is it to be used? Second, what does "reconstruc-
tion of a source" mean? Third, how far back in time does the method
leads us? I shall review these topics, which are connected with each
other, focusing on the field of Had-t in the stricter sense.
According to Fuat Sezgin the method works as follows: First, all
isnads given in a compilation, such as al-Buihari's Gdmi',are arranged
according to their youngest transmitters (the informants of al-Buhari).
Second, the isnddsof those among the youngest transmitters who are
frequentlymentioned by the compiler (al-Buhari)must be further checked
for whether in their transmission lines they have names in common. If
they do not, they must be regarded themselves as authors of sources
which the compiler has used. If they do, the last common name of
their isnddswhich contains the same transmitters, must be considered
the author of an earlier source (while the other common names indi-
cate only transmitters).'54
Sezgin gives the following isndd as example which occurs frequently
in al-Buhari's Gami':'Abd Allah b. Muhammad [al-Musnadl] (d. 220
A.H./843 C.E.) - cAbd al-Razzaq [al-San'ani] (d. 211 A.H./826 C.E.) -
Ma'mar [b. Rasid] (d. 153 A.H./770 C.E.) - Hammam [b. Munabbih]
(d. 130 A.H./747 C.E.) - Aba Hurayra(d. 58 A.H./678 C.E. or 59/679).'55
One would expect that, according to his rule, Sezgin singles Aba Hurayra
out as the author of the source, because Abtu Hurayra is the last name
which all those isnddshave in common. Surprisingly, Sezgin does not.
He maintains that all individuals in the isnad are authors, except Abtu

der DeutschenMorgenldndischen Gesellschaft,103 (1953), 290-307. The list of compilations

given above is my own based on Horst's findings. Horst himself assumes that there
were even earlier ones, such as the Tafsirs of Mugahid (d. 102 A.H./720-1 C.E.
or 103/721-2), 'All b. Abi Talha (d. 120 A.H./738 C.E. or 143/760) and al-Suddi
(d. 127 A.H./744-5 C.E.).
153 Cf F. Sezgin, Bu4drP'ninkaynaklanhakkznda Istanbul, 1956; Motzki, Die
araptrMarlar,
Anfangeder islamischen
Jurisprudenz;idem, "The Musannafof 'Abd al-Razzaq al-San'ani as
a Source of Authentic ahddithof the First Century A.H". in journal of Near EasternStudies,
50 (1991), 1-21.
154 F. Sezgin, Geschichte
des arabischenSchrifttums,
Leiden, EJ. Brill, 1967, I, 82.
155 Ibid., 8 1.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 245

Hurayra.'56 This inconsistency may be caused, on the one hand, by

Sezgin's idea that the names occurring in an isnad indicate "authors
and authorised transmitters of books"'57and, on the other hand, by his
conviction that Abui Hurayra, although he possessed written notes, can-
not be the author of a book of almost one hundred and forty 4adats
from the Prophet.'58
Be that as it may, Sezgin thinks that the early source from which
al-Buihari draws, is a collection of Hammam b. Munabbih which is
Kaof Hammanm"
known from later bio-bibliographicalliteratureas "the sah4
and which is also preserved as an independent booklet transmitted
through 'Abd al-Razzaq and Ma'mar.151In what form had al-Buhari
the collection of Hammam at his disposal? In his original form trans-
mitted through Ma'mar, 'Abd al-Razzaq and al-Musnadi, or more or
less scattered and rearranged in the own books of the three transmitters?
However, this question - which Sezgin does not dare to answer- is
only marginal for him because he is convinced that al-BuWhari's tradi-
tions with the isnad mentioned above, derive from Hammam's sahafa
anyway.'60He assumes that they were transmitted carefully, irrespective
of whether they were combined with other traditions to make larger
collections or not.
According to Sezgin, the reliability of the transmission process is
guaranteed mainly by the fact that it happened by writing down the
texts and additional authorisation of them by the author or transmit-
ter from whom they were received or by using written sources with-
out authorisation. The respective type of transmission, Sezgin thinks, is
reflected in the transmissionformulas used.'6' This method which Sezgin
proposed for the reconstruction of earlier sources on the basis of later
ones - not only in the field of Hadatin the proper sense, but for Arabic
literature of the first Islamic centuries in general - provoked a lively
debate among Western scholars. The criticism was chiefly directed at
three of Sezgin's assumptions.

156 Ibid.
157 Ibid., 70. Also ibid., 77 the author writes that "the isn&ds(denote) written texts [...]"
and "the formulas in the chains of transmission [ ...] actually refer to written sources".
158 According to M. Hamidullah, the sah4fais, on the contrary, a collection of Abu

Hurayra which he dictated to Hammam. Cf M. Hamidullah, SahIfaHammamibnMunabbih,

Luton, U.K., Apex, 1979, 60.
159 Ibid., 88-97.
160 Sezgin, Geschichte des arabischenSch7ifttums,I, 81.
161
Ibid.,77-78.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
246 HARALD MOTZKI

First, this criticism challenged Sezgin's idea that the transmission

process happened generally on the basis of written texts and that the
names given in the isnddsare authors or transmitters of written texts.
This, in turn, lead to an extensive discussion about the question of
what role the oral and written transmissions have played during the
first Islamic centuries.'62The outcome of this fruitful discussion was that
Sezgin's interpretation of the transmission process as one of passing
only books on, is too undifferentiated. The transmission must rather be
understood in terms of a living teaching system in which both oral and
written forms of passing information on were instrumental. The part
which both forms played could vary with respect to time, region, scholar
and subject, and must not be generalised, not even for the transmis-
sion of Hadatin the stricter sense.
Secondly, the discussion of Sezgin's method also made clear that he
over-estimated the reliability of transmission and, for that matter, the
possibility of reconstructinglost sources on the basis of later compilations.
It is dangerous to conclude on the basis of the isnadsalone that traditions
of which the isnads contain a putative author of a book (dealing with
the same subject as the traditions in question) reproduce parts of that
book in their original form. We would be on safer ground in assum-
ing only that the texts go back to an earlier compiler, not to a par-
ticular compilation of his.'63Sezgin's claim, for example, that al-Buhiari's

162 Cf, for example, G. Stauth, Die Uberlieferung des Korankommentars Mugahidb. Gabrs.
Zur Frageder Rekonstruktion der in den Sammelwerken des 3. Jh. d. H. benutztenfihislamischen
Quellenwerke,Ph.D. thesis, Giessen, 1969; S.M. al-Samuk, Die historischen Uberlieferungen
nach
Ibn Ishaq. Eine synoptische
Untersuchung.
Ph.D. thesis, Frankfurtam Main, 1978; R. Sellheim,
"Abii 'Alf al-Qal_. Zum Problem mundlicher und schriftlicher Uberlieferung am Beispiel
von Sprichwortersammlungen" in Studienzur Geschichteund Kulturdes VorderenOrients.
BertoldSpuler,ed. H.R. Roemer/A. Noth, Leiden, EJ. Brill, 1981, 362-374;
Festschriftfuir
idem, "Muhammeds erstes Offenbarungserlebnis.Zum Problem mundlicher und schriftlicher
Uberlieferung im 1/7. und 2./8. Jahrhundert" in JerusalemStudiesin Arabic and Islam,
10 (1987), 1-16; W. Werkmeister, Quellenuntersuchungen zum Kit&bal-'Iqdal-fariddesAndalusiers
Ibn 'Abdrabbih (246/860-328/940); G. Schoeler, "Die Frage der schriftlichen oder mund-
lichen Uberlieferung der Wissenschaften im fruhen Islam" in Der Islam, 62 (1985), 201-
230; idem, "Weiteres zur Frage der schriftlichen und mundlichen Uberlieferung der
Wissenschaften im Islam" in Der Islam, 66 (1989), 38-67; idem, "Mundliche Thora und
Hadit: Uberlieferung, Schreibverbot, Redaktion" in Der Islam, 66 (1989), 213-251 (English
translation in H. Motzki (ed.), Hadith: Originsand Developments, in series The Formationof
the ClassicalIslamic World28, Aldershot, Ashgate/Variorum, 2004); idem, "Schreiben und
Veroffentlichen. Zur Verwendung und Funktion der Schrift in den ersten islamischen
Jahrhunderten" in Der Islam, 69 (1992), 1-43; Motzki, Die Anfdngederislamischen Jurisprudenz,
87-95; idem, The Originsof IslamicJurisprudence, 95-104.
163 Cf H. Motzki, "The Author and his Work in Islamic Literature of the First

Centuries. The Case of 'Abd al-Razzaq's Musannaf", in JerusalemStudiesin Arabic and

Islam, 28 (2003), 171-201.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 247

traditions with the above mentioned asnddgoing back via Hammam b.

Munabbih to Abtu Hurayra reproduce the traditions such as they were
contained in Hammam's sahafacan only be considered as one of sev-
eral possibilities.164 Even if one could prove that in this particular case,
the same cannot be supposed for other cases of Hadattransmission as
well and still less for the transmission of other sorts of knowledge.'65
Thirdly, Sezgin's idea that the transmission formulae in the isndds
reflect faithfully the types of transmission such as they have been defined
in the classical handbooks of Hadatis doubtful, too, in its general claim.
It has been shown that it is not generally true, at least, for the second
century A.H. This is just the period in which most of the early com-
pilations originated.'66
Yet the shortcomings of Sezgin's application of the method does not
justify the conclusion that the isnddsare useless for the reconstruction
of earlier sources. We must only be aware that "reconstruction" does
not necessarily lead to books and their original texts, and we must look
for additional indications in a compilation and its traditions which might
help to substantiatethe conclusions which can be drawn from the names
contained in the isndds.
In my study Die Anfdngeder islamischen Jurisprudenz,I used the asnads
found in 'Abd al-Razzaq's Musannafto recover earlier "sources",namely,
material going back to Ma'mar b. Rasid, Ibn Gurayg, Sufyan al-Tawri
and Ibn 'Uyayna, who are a generation older than 'Abd al-Razzaq.
Although, at least, three of the four scholars are mentioned in later
sources as authors of books of the same genre as 'Abd al-Razzaq's, the
conclusion that the latter derived his material from those books (as
Sezgin would conclude) cannot be substantiated because the books seem
to be lost. A safer conclusion would be that 'Abd al-Razzaq received
the texts in the lectures of the four scholars, a conclusion which is cor-
roborated by biographical traditions. The idea suggests itself that their
supposed compilations may have formed for a large part the basis of
their lectures and that 'Abd al-Razzaq wrote the lectures down, but
how he did it - word-for-word or in taking notes - we cannot ascertain
without an investigation of the texts itself and their comparison with

164 Another possibility is that they underwent

changes in the course of the transmis-
sion process.
165 A concise summary of the problems which
must be taken into account when try-
ing to reconstruct earlier sources is given in S. Leder, Das Korpusal-Haitam ibn 'Adt
(st. 207/822), 3-14.
166 Cf Motzki, Die Anfdngederislamischen ?furisprudenz,
92-95; idem, The Originsof Islamic
Jurisprudence,101-104.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
248 HARALD MOTZKI

transmissions preserved from the same scholars by other transmitters

than 'Abd al-Razzaq.
In order to corroborate the dating of the texts and to show that
'Abd al-Razzaq did not ascribe his traditions arbitrarily to his infor-
mants, invented zsnddsor equipped invented traditions with well-known
isndds, it is worth the effort to study the isnads of 'Abd al-Razzaq in
detail trying to find indications of forgery or the opposite. There are
different approaches that can be applied for this end.
1) The isnddsof the alleged major "sources" can be analysed in order
to draw up what I called individual profiles. The items of such a profile
could be: the number of major informants; the quantity of traditions
ascribed to them; the number of minor informants and the quantities
of their traditions; the number of rarely mentioned informants and the
quantity of traditions allegedly going back to them; the amount of tra-
ditions giving the personal opinion of the respective scholar; the amount
of anonymously transmitted texts etc.'67 A comparison of the profiles
of the different sources enables us to draw conclusions as to whether
the assumption is probable that the compiler forged his sources. In the
case of 'Abd al-Razzaq's Musannaf;the result was negative.'68
2) Peculiarities of the collector's (in our case 'Abd al-Razzaq's) refer-
ences to his informants can be used as indications of reliability; such
as: expression of ignorance or doubt as to who the informant or what
his precise wording was; indirect transmission from a major (direct)
informant; explanations of or critical comments on the traditions of the
informant; or giving variants of the same tradition known from the
same or another source. In the case of 'Abd al-Razzaq, such pecu-
liarities could be found, and it seems not reasonable to assume that he
inserted them sporadically in his isnddswith the intent to deceive his
colleagues and pupils and to mask the fact that his ascription of tra-
ditions to certain sources is arbitrary.
This same procedure can be applied to a collector's major sources
(in our case to Ibn Gurayg, Ma'mar, al-Tawri and Ibn 'Uyayna). The
profiles of their major sources and their references to their sources in
general can provide us not only with indications as to the reliability or
unreliability of the earlier collectors (such as Ibn Gurayg, Ma'mar, al-

167 Ibid., 56-58; idem, The Origins

of IslamicJurisprudence,
58-6 1; also idem, "The Musannaf
of IAbdal-Razz.q", 2-5.
168 Ibid.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 249

Tawri and Ibn 'Uyayna) but also with additional arguments to decide
the question of whether the later collector's transmission (such as 'Abd
al-Razzaq's) can be considered reliable or not.
I tested this hypothesis in detail with the material transmitted by
'Abd al-Razzaq from Ibn Gurayg and I found that the individual profile
of his transmission and that of his major sources can be brought out
very clearly by looking at the differences among the traditions which
are ascribed to different sources. Significant differences could be found
concerning the type of tradition (transmission of the teacher's personal
opinion or of a tradition on his authority); types of transmission (pref-
erence for family isndds, major informants, local authorities etc.); the
preference for a certain type of authority (Successors, Companions, the
Prophet); the quality of the isnaids;the terminology of transmission (i.e.,
what formulas are used); and the genre of the tradition (statement or
dialogue).'69In our case, the findings corroborated the conclusions which
could be drawn on the basis of an analysis which focused only on the
names of the transmitters in the isnads.
In this way, it was possible to reconstruct on the basis of 'Abd al-
Razzaq's Musannafnot only the material going back to Ibn Ouray', a
source to be dated into the second quarter of the second century, but
also the material of two earlier sources, 'Ata' b. Abi Rabah (d. 115
A.H./733 C.E.) and 'Amr b. Dinar (d. 126 A.H./744 C.E.). Material
of a source, such as 'Ata' means the content of his teaching such as
his pupil, in our case Ibn Gurayg, has reproduced it.
The question of whether the pupil's reproduction was good or not
so good, can be tentatively answered on the basis of formal peculiari-
ties of his transmission, like those described above. A definite judge-
ment on the quality of a pupil's transmission, however, can only be
given if the transmissions of the same material by other pupils are avail-
able and are compared with it. I showed this for the case of 'Amr b.
Dinar of whom not only Ibn Gurayg's transmission is available but
also that of Ibn 'Uyayna,l70 and in more detail for al-Zuhrl.171 The re-
construction of al-Zuhrtl's(d. 124 A.H./742 C.E.) teaching, based on

169 For more details see my Die Anfangeder islamischen

Jurisprudenzand 7he Orwinsof
Islamicjurisprudence,passim, and the summary in "The Musannafof 'Abd al-Razzaq",
2-12.
170 Cf Motzki, Die Anfange der islamischenJurisprudenz,161-167; idem, The Origins,
177-185.
1'} Cf Motzki, "Der Fiqh des-Zuhri"; idem, "The Jurisprudence of Ibn Sihab
az-Zuhrf".

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
250 HARALD MOTZKI

two independent later collections, Malik's Muwatta'and 'Abd al-Razzaq's

Musannaf;shows particularlywell both the possibilities of the isnddanaly-
sis described above and the need to complete them by a thorough com-
parative study of the matns. Only the combination of both methods
leads to convincing results. Unfortunately, however, we must often be
content with conclusions drawn from the isnadsbecause substantial vari-
ant transmissions are not available.

IV. Dating with Isnad and Matn

The conclusion drawn at the end of the preceding section that the inves-
tigation of the isniadsneeds to be completed by the analysis of the matns
applies not only to source reconstruction. It also applies to the dating
of single traditions. This insight is visible already in an article by Jan
Hendrik Kramers, published in 1953, andJoseph van Ess' book Zwischen
Hadft und Theologie,published in 1975, benefited much from this
approach.'72The method of the two studies has not been appreciated
very much until recently. Kramers' article passed unnoticed and the
effect of van Ess' contribution was cut short by Cook's criticism of its
method.'73 The recent revival of this method seems to be due to both
the insight that a combined approach can lead to more reliable results
than the investigation of isndds or matns alone, and to an uneasiness
with the actual development of isnddanalysis which tended to become
a too artificial interpretation of isnddbundles.
The combined method can be called isndd-cum-matn or matn-cum-isnad
analysis, depending on the starting point of the investigation or the
intensity with which the two items are used for conclusions. Among
the several possible approaches, the one which starts from the assump-
tion that there must be a correlation between isnddvariants and matn
variants of a tradition, if they were part of a real transmission process,
seems to be the most rewarding. The scholars who adopt this assump-
tion are convinced that such a correlation is unlikely to be the result
of systematic forgery because the phenomenon of correlation is so wide-
spread that almost every muhadditmust have participated in forgery.
The fact that there is often a correlation between the different branches

172 Jan Hendrik Kramers, "Une tradition A tendance manicheenne (La 'mangeuse de

verdure')" in Acta Orientalia,21 (1950-1953), 10-22.

173 See above pp. 27-31.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 251

and strands of the isnad bundle belonging to a tradition, on the one

hand, and, on the other, the different variants of its matnallows check-
ing of the isnddanalysis by the matnsor vice versa. The method is best
illustrated in two studies which both appeared in 1996: In Gregor
Schoeler's book Charakterund Authentieder muslimischen Uberlieferunguber
das LebenMohammeds the origin and transmission process of two sira tra-
ditions are traced; and in my study "QuovadisHadAt-Forschung?" a tra-
dition ascribed to Nafi' is investigated.174
The isndd-cum-matn analysis as it is applied in the two studies pro-
ceeds in five steps: 1) All the variants of a tradition which can be found
are compiled. 2) A bundle of all isnddvariants are composed in form
of a diagram in order to document the transmission process as it is
reflected by the transmission lines and to detect partial common links
and a common link.'75In accordance with the interpretation proposed,
when discussing the common link phenomenon in the preceding sec-
tion, the common link is, provisionally, assumed to be the collector and
professional disseminator.176 3) Whether the common link can be con-
sidered really as the collector or professional disseminator has to be
checked by a matnanalysis. It consists in compiling the texts belonging
to the different transmission lines in order to make possible a synoptic
comparison of one to the other. 4) Groups of matnvariants and groups
of isnad variants are to be compared to check whether there is a cor-
relation or not. 5) If a correlation exists, then conclusions can be drawn
regarding the original matn transmitted by the common link and the
one responsible for whatever changes have occurred in the course of
the transmission after the common link.
In this way reliability of the date established on the basis of the isndd
analysis, or the relative chronology established on the basis of the matn
analysis can be confirmed or disproved. The soundness of the conclu-
sions grows with the number and diversity of variants available. With

174 The three traditions studied in these publications had been investigated already

before by G.H.A. Juynboll in his articles "Nafi', the Mawld of Ibn 'Umar, and his
Position in Muslim Hadtth Literature" in Der Islam, 70 (1993), 207-244, and "Early
Islamic Society as Reflected in its Use of Isndds" in Le Musion, 107 (1994), 151-194,
esp. 160-166, 179-184, both articles focusing on the isndd analysis.
175 This is not meant to exclude the possibility that some transmission lines or parts
of it are fictitious.
176 Schoeler speaks of him as the one "[der] die betreffende Tradition schulmassig

verbreitet hat"; G. Schoeler, Charakterund Authentieder muslimischenUberlieferunguiberdas

LebenMohammedsin series Studienzur Sprache,Geschichteund Kulturdes islamischenOrients,
NJeueFolge 14, Berlin, Walter de Gruyter, 1996, 24.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
252 HARALD MOTZKI

this method the risk becomes considerably reduced that a common link
which is the result of isn&dforgery remains undetected.'77 I leave the
description of the usndd-cum-matn
analysis with that. Several recent stud-
ies tested this method with encouraging results.'78

Conclusion

At the end of this overview of methods which modern scholars in the

West used or are still using in order to date Muslim traditions, the fol-
lowing conclusions can be drawn: The different methods have been
considerably improved since the end of the nineteenth century.
Nevertheless, some methods seem to be more reliable than others.
Dating particular traditions on the basis of the matnalone seems to be
most inaccurate. A dating which argues by silence that a tradition has
not existed before the time of the compilation in which it appears first,
is uncertain and tends to be too late. A dating based solely on an inves-
tigation of the isnadsof a particular tradition and on the common link
phenomenon are less sound than those who check the results of the
isnad analysis by the thorough study of matn variants. This does not
mean that the isndd-cum-matn method does not raise problems. It does,
and efforts must be made to solve them in the future. Two questions
remain, nevertheless: Can this or any other method provide, first, reli-
able datings in all circumstances and, second, datings which will be
generally accepted? Two factors stand in the way of it. First of all, the
scarcity of our sources, and, secondly, the fact that all dating methods
must rely on assumptions derived from other sources. The first factor

177 See also Schoeler, Charakter

undAuthentieder muslimischen uberdas Leben
Uberlieferung
Mohammeds,24, 26.
178 I. Zaman, "The Science of Rjal as a Method in the Study of Hadiths" in Journal
of Islamic Studies,5 (1994) 1-34 where the author is not really concerned with dating
hadits but also takes the correlation between isndds and matns as a starting point;
A. Gorke, "The Historical Tradition about al-Hudaybiya. A Study of 'Urwa b. al-Zubayr's
Account" in H. Motzki (ed), The Biographyof Muhammad:the Issue of the Sources,240-275;
idem, "Eschatology,History, and the Common Link";Motzki, "Quovadis IHadct-Forschung?";
idem, "The Prophet and the Cat"; idem, "The Murder of Ibn Abi l-Huqayq"; idem,
"The Collection of the Qur'an: A Reconsideration of Western Views in Light of Recent
Methodological Developments" in Der Islam, 78 (2001), 1-34; idem, "The Origins of
Muslim Exegesis"; U. Mitter, Das fruhislamische Patronat. Eine Untersuchung zur Rolle
von fremden Elementen bei der Entwicklung des islamischen Rechts, Ph.D. thesis,
Nijmegen, 1999; R. Peters, "Murder in Khaybar: Some Thoughts on the Origins of
the QasdmaProcedure in Islamic Law" in IslamicLaw and Societ, 9,2 (2002), 132-167.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions
 DATING MUSLIM TRADITIONS 253

needs no further comment, the second does. Dating traditions is not

possible without having recourse to assumptions. They can be partly
derived from general human experience, but partly more concrete
assumptions are needed: for instance, on the dimensions of fabrication
and falsification in the field of Hadat on the ways how knowledge was
transmitted in the first two centuries of Islam; on the nature of the
common links and single strands etc. In addition, all these assumptions
must take into consideration that there may have been variation in
time and place. The concrete assumptions mentioned can be based on
different source material (e.g., reports on fabrications or on the ways
how traditions were transmitted by different persons), but these assump-
tions will always be generalisations based on a limited number of par-
ticular facts. Depending on which facts we generalise, the views on the
cultural history of early Islam can be very different. Therefore, whether
the dating of a tradition is considered reliable or not, depends not only
on the dating methods applied, but also on our preconceptions of early
Islam which we have formed.

This content downloaded from 130.240.43.43 on Mon, 04 Jan 2016 12:34:43 UTC
All use subject to JSTOR Terms and Conditions