1 The ḥadīths and Their Use of Rural Space

The fourth/tenth-century Ḥanafī jurist al-Ṭaḥāwī presents versions of all three ḥadīths, as follows:⁶

In most versions, Khubayb and his companion responds to this rejection by converting.

Before taking a closer look at the texts of the three ḥadīths, it is worth noting two historical points of obvious significance. The first is that the questions that are raised and answered in these ḥadīths – questions concerning military recruitment and communal belonging, whether tribal or monotheistic – were existential ones for the Islamic state in its first century. All three ḥadīths posit, for example, a declaration of faith and allegiance as a prerequisite for military participation on the Muslims’ behalf. In the first and third ḥadīths, this requirement is overlaid on a background of tribal affiliation. This implies that belonging to a certain tribe matters less than belonging to Islam when it comes to participation in warfare. The overlapping and sometimes conflicting claims of tribal and religious affiliation were also central to the Constitution of Medina.¹⁴ Indeed, for a movement (the so-called “jihad state”) that was intent on conquering the Late Antique Near East, where political boundaries shadowed religious ones, the question of how religious difference affected the possibility of military cooperation was of perennial urgency. It is well known that non-Muslims frequently fought alongside the Arabian conquerors.¹⁵ Such cooperation was probably even more commonplace than our sources indicate. It seems that conversion was not strictly necessary in order for an individual to become a mawlā or “client” of an Arab tribe, the primary avenue to inclusion in the early Islamic community.¹⁶ The three ḥadīths thus represent a firm normative position that would make religious alterity a barrier to military cooperation. This position became increasingly dominant in the Islamic juristic tradition. The ḥadīths are thus significant both because they intersect major religious and political trends of the first two Islamic centuries and because they lie at the head of a restrictive legal tradition that has cast a long shadow.

The second point to note – and the crucial one for this volume – is that the three ḥadīths are all set on the rural periphery of Medina. All three represent the delivery of the dictum as occurring after a physical movement from urban to rural space. Why? Whatever one thinks about the events’ historicity, they need not have been represented in this way.¹⁷ I would argue that the rural setting reflects a Medinan rhetorical appropriation of rural space for the purpose of asserting normative authority. As will be shown below, these ḥadīths represent a Medinan viewpoint as it existed in that city in the early second/eighth century. They offer us a glimpse of a strain of control that was exercised by ḥadīth transmitters by means of imagined, rhetorical rural space.

To appreciate how imagined rural space could discharge this function, one ought to begin by noting that in early Islam, authority belonged to God, as the Qurʾān implies by the phrase al-ḥukm li-lāh (“Rule is God’s,” e.g., 12:40, 67, etc.). In principle, God conferred political authority on Muḥammad and his successors the caliphs or imams, whoever they should rightfully have been. How that authority was parlayed into control – coercive or otherwise – soon came under scrutiny from another quarter: that of the scholars, among them the ḥadīth transmitters. These last sought to leverage Muḥammad’s authority for causes they cared about by claiming proprietary knowledge of his paradigmatic vita, as conveyed in the ḥadīth.

Imagined rural space played a strategic dual role in facilitating this project with respect to our three ḥadīths. In all three, Muḥammad “went out” from Medina, placing the events that follow beyond the view of all but his travelling companions. The exeunt fortified claims to proprietary knowledge of what he proceeded to do and say. Moses received the Law alone on a mountain. Jesus of Nazareth retired to lonely places to teach his disciples. Muḥammad himself received early revelations alone in a cave. Similarly, the well-armed squadron proffered its services at the appropriately obscure Thanīyat al-Wadāʿ, not at the mosque in Medina. Jesus’ own principled rejections of would-be followers took place after he and his disciples had left a village, “as they went in the way” (Luke 9:56–62). An effect of setting such events in rural space was to circumscribe debate about what had taken place there.¹⁸ Eighth-century ḥadīth transmitters were a mainly urban group, but in the Prophet’s countryside that they imagined their political authority was amplified precisely because his policy statements in rure were spoken into empty space. True, those who accompanied Muḥammad at Badr and Uḥud included his most prominent Companions. This is why it is so striking that they should be marginal and anonymous in the three ḥadīths, in which their names are overlooked in favor of such obscure toponyms as Ḥarrat al-Wabra and al-Baydāʾ, the spatial transition among which opens rhetorical space for repetition and reinforcement of the dictum.

If the transmitters’ authority waxed in this way, it did so at the expense of the Prophet’s coercive control. Rejecting the aid of warlike mushrikūn not only deprived the Prophet’s forces of their aid, but left them roaming the hinterlands of Medina. Here the limits of state control over rural space are not only highly visible, but rhetorically advantageous. Beyond those limits lies contested space where state agents face choices unimaginable within or before city walls. Representation reflects experience; indeed, it relies on accurate reflection for its normative power.¹⁹ Pre-modern urban space, like the rural battlefield, had a binarizing effect on military forces, concentrating or dissolving armies as discrete units. At cities, armies form up, attack, defend, loot, or are routed. They break or are broken. To garrison troops in a city creates instability, as residents of al-Muʿtaṣim’s Baghdad or al-Mustanṣir’s Cairo might testify. Partly for this reason, the early Muslims established the amṣār – garrison cities – away from existing population centers. In a garrison city, matters are of course different, and if these accounts had been formulated among the jabbānas of Kūfa – open urban spaces where the resident tribal fighters gathered and dispersed – rather than in Medina, the exeunt might have been superfluous.

Arid and semi-arid hinterlands, by contrast, when not playing host to the temporary conurbations of combat, were zones of military transition, where fighting men were raised and recruited, and where armies break camp for the march and disperse to requisition. Group agency and identity is diluted in the dust of the road, and the flanks exposed. Dilution, in turn, facilitates change in the vectors and composition of the army. It is no surprise, then, that Muḥammad should be depicted sealing the religious boundaries of his military forces in this anomic setting, where group boundaries were most permeable.²⁰

The setting, in fact, throws the dictum into sharp relief; if space for interreligious cooperation were to be found anywhere, then surely it would have been here, in the transitional countryside, and now, on the verges of Muḥammad’s most improbable triumph (Badr) and his most bitter defeat (Uḥud). That no such space was made lent authority to the transmitters’ controverted views about the religious modalities of legitimate state control. In a sense, they harvested rhetorical authority for themselves from the excluded mushrikūn, the undermanned army of Muḥammad, and the non-cognoscenti among their colleagues, in the fertile landscape of an imagined Ḥijāzī hinterland.

2 The Milieu of Early Circulation

It remains, however, to show that the three ḥadīths in question did, in fact, originate in Medina, and to consider the question of the dates at which they were composed. The following schemata and accompanying tables present the putative paths along which each ḥadīth was transmitted, according to the testimony of the sources, as well as the overlap between shared paths of transmission and shared contents.

View Full Size

Figure 2.1

A well-armed squadron of Jews

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Figure 2.2

The mushrik of Badr

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Figure 2.3

The rejection of Khubayb

• Download Figure
• Download figure as PowerPoint slide

(For Notes to Figures 2.1–2.3, please see the Appendix)

The “Travelling Tradition Test” (TT Test), recently formulated by Behnam Sadeghi, is a useful tool for dating ḥadīths that share a common feature. It uses geographic clustering of wording, themes, or legal positions to put bounds on the dates of traditions that purportedly moved from one city to another. If an idea, word, phrase, or some other feature is thereby uniquely linked to a city, its presence in a tradition of uncertain provenance can be used to assign the tradition to that city. The report can be dated approximately to the time interval spanned by the transmitters hailing from the tradition’s city of origin.²³

The TT Test requires that “candidate birthplaces” (“the first city in which a prototype containing the shared features of the traditions … circulated”) be identified for each of a cluster of traditions sharing a common feature. In the present case, the common feature comprises variations on the dictum – “I’ll not accept aid from a mushrik” – in close combination with the distinctive narrative sequence in which it is embedded: 1. exit from Medina into rural space; 2. encounter with warlike non-Muslim(s); 3. dictum; 4. conversion. In some versions one or more of these components is omitted, while in others there is additional material. Never, however, is one of these components replaced with an incompatible alternative. In principle, the dictum might have been deployed in a wide variety of alternative narrative sequences, but in fact it was not.

What, then, are the candidate birthplaces of each of these three ḥadīths? The candidate birthplaces are determined on the basis of the cities in which each ḥadīth was known to circulate during its early transmission history, based upon the names of the transmitters and information known about their lives. (For what next follows, see Figures 2.1–2.3.) Thus for the squadron of Jews, the candidate birthplaces are Medina and, if one chooses to treat the isnāds bypassing the later potential “common link” (al-Faḍl b. Mūsā) as Juynbollian “dives,” a choice that I will argue on textual grounds to be unwise,²⁴ Marw.²⁵ For the mushrik of Badr, Medina is the only viable candidate birthplace. For the rejection of Khubayb, Wāsiṭ and Medina are both candidate birthplaces.

The TT Test yields Medina as the only candidate birthplace common to all three ḥadīths. The isnāds thus give these ḥadīths the appearance of regionalism (“geographic clustering of contents”).²⁶ But why should we believe that isnāds carry historical information about the actual paths, human and thus geographical, along which the ḥadīths were disseminated? Sadeghi argues persuasively that “isnāds often carry valid geographical information.” Supposing that “unity of distinctive contents implies unity of origin,” if isnāds do not carry valid geographical information then there is no reason that the isnāds of ḥadīth that share features should habitually give the appearance of regionalism. Yet they often do so, suggesting that the geographical information conveyed in the isnāds is valid. Put differently, the regular concurrence of common features with common milieus is unlikely to be due to chance, and common geographical origin “usually provides the most plausible explanation” for this concurrence.²⁷

These formulations apply to the present case, and point to a Medinan origin for all three ḥadīths. They arise from a single insight about ḥadīths: shared distinctive contents tend to go along with shared geographical paths of transmission. I would propose a minor refinement to Sadeghi’s formulation, based on our case: there is reason for the appearance of regionalism to come about even if the isnāds are wholly fictitious in cases where the ḥadīths in question narrate events that happen in a certain place. This is because of the natural presumption that residents of a certain place are most likely to have knowledge of events that occurred there. We might say of a shared feature that the more geographically specific it is, the less confident we may be that the appearance of regionalism arises from a historical transmission process. This assumes that the place mentioned corresponds to the place of the observed regionalism; if the places differ, we may be relatively more confident that the appearance of regionalism reflects reality. Thus one can readily imagine a motive for spuriously attributing each of our three ḥadīths to early Medinan transmitters.²⁸

This refinement recognizes that transmitters of any ḥadīth relating events that occurred near Medina, or any given city, might want to cite the authorities of that city. It does not explain how certain clusters of ḥadīths, including the three at hand, came to have related non-region-specific contents as well (in this case, the shared narrative structure and dictum). Moreover, it is undermined by the fact that many ḥadīths about events that occurred in Medina were attributed to Arabs who settled in Syria, Iraq, and elsewhere very early in the Islamic period. Nevertheless, it will be well to test the reliability of these isnāds in another way, using not only principles that apply to any cluster of textually related ḥadīths, but also the data for these particular ḥadīths. A further reason for such an independent test is that we have only three related ḥadīths that converge on Medina – not an unusual number for that city, as Sadeghi points out, and thus not a sure indicator that that dictum was peculiar to it. However, one might here invoke Sadeghi’s own caveat: “shared distinctive features can significantly strengthen the likelihood of shared origins even for a small number of traditions.”²⁹ The dictum in combination with its associated narrative sequence is distinctive to these three ḥadīths.

We possess an independent test of the isnāds’ integrity in matn-cum-isnād analysis: comparison of the contents of the different versions of a certain ḥadīth, with attention to the putative routes of transmission found in their isnāds. Matn-cum-isnād analyses of the three ḥadīths that contain variations on the dictum “I will not accept aid from a mushrik” strongly suggest that their isnāds give historically valid information about the human and thus geographical paths along which they were disseminated.

For what follows, refer to Tables 2.1–2.3. The suggestion of the isnāds’ historical validity is strong because in numerous cases, versions whose isnāds claim for them a shared path of transmission tend to share specific features or combinations of features. For example, in Table 2.2, the six versions of the ḥadīth about the mushrik of Badr whose isnāds claim that they were transmitted from Mālik, the “common link,” by Wakīʿ b. al-Jarrāḥ (d. 197/812) contain only the dictum. The six versions allegedly via Ibn Wahb (d. 197/813) all give a much more elaborate narrative, in which the phrase by which the mushrik attempts to enlist is virtually identical. Only the two versions that pass via Bishr (d. 207/822) mention, of toponyms, only Badr, while only versions via Yaḥyā b. Saʿīd (d. 198/813) include the verbal form akūnu/akūna in the mushrik’s enlistment pitch. Other similarities may be observed in Table 2.2.

Analogous patterns are observed for the other two hadīths. In the case of the rejection of Khubayb (Table 2.3), for example, there is a perfect correlation between the four shared paths of transmission and the respective words used to designate the objective of Muḥammad’s expedition. In the case of the well-armed squadron of Jews (Table 2.1), only three versions give the number of Jews (600): the two that were allegedly transmitted from al-Faḍl b. Mūsā by Khālid b. Khidāsh, and a third version that, although it does not claim to have been transmitted by Khālid, was allegedly transmitted by al-Ṭabarānī (d. 360/971), who also transmitted a version from Khālid. More arresting is the correspondence between the two versions that are unique not in sharing but in lacking a certain transmitter; the only two versions that do not claim to have passed via al-Faḍl b. Mūsā in Marw also happen to be the only two versions ignorant of the “fact” that the Jews were of Banū Qaynuqāʿ, a detail that creates chronological difficulties that are discussed below. This coincidence suggests that the Medinan Muḥammad b. ʿAmr (d. 144–145/761), or a Medinan near contemporary of his, actually transmitted a proto-version of this ḥadīth, and that he did not (or at least did not always) mention the Jews’ tribal affiliation. A similar though less striking example of this phenomenon may be observed in the case of the rejection of Khubayb, where only one version consists of the naked dictum: that which “dives” beneath Yazīd b. Hārūn (d. 117/735) to his putative source, Mustalim b. Saʿīd (d. ?). The most persuasive account of all this evidence is that the concurrences were in fact produced by the transmission process that the isnāds indicate to have taken place, in more or less the way they depict it.

Matn-cum-isnād analysis’ strong suggestion that these isnāds convey historically valid information remains a suggestion rather than a demonstrated proposition because the correlation between shared putative paths of transmission and shared content is imperfect. There are of course minor, random differences in the contents of ḥadīths allegedly transmitted by common routes – this is expected because the common routes are usually only partial, leaving room for non-shared transmitters, including the named compilers themselves or their students, to have introduced modifications. More problematic are cases where there is a strong correlation between the contents of versions that do not claim to share routes of transmission after the common link, or where there is little correlation between the contents of versions that do claim to share routes of transmission. For an example of the first scenario, in the case of the rejection of Khubayb, one can easily explain why the version of al-Ṭabarānī that allegedly came to him via Ibn Abī Shayba (d. 235/849) should use the same distinctive, unforseeable word for Muḥammad’s objective – wajh – as the version found in the latter transmitter’s Muṣannaf. It is less easy to explain why one and only one other version – that of al-Bukhārī (d. 256/870) – should use the same word, when it allegedly came via an independent route. For an example of the second scenario, relating to the ḥadīth about the mushrik of Badr, one version cited by Ibn Ḥanbal (d. 240/855) gives the longer narrative, while the two other versions from Ibn Ḥanbal’s putative source for this version (Ibn Mahdī, d. 198/814) that are found in other compilations correspond quite closely to one another and show no knowledge of the longer narrative.

There are two plausible, overlapping explanations for this phenomenon. The first is borrowing or influence of matn material that the isnāds do not reflect. Thus al-Bukhārī’s source, ʿAbd Allāh al-Juʿfī (d. 229/844), might have heard a version that used the term wajh from someone other than Yazīd b. Hārūn, but narrated the ḥadīth as though he had heard it straight from Yazīd (he might of course really have heard a version from Yazīd, but been influenced by a version he heard elsewhere). Or, Ibn Ḥanbal’s source for the longer version of the ḥadīth, Ibn Mahdī, might have heard the ḥadīth about the mushrik of Badr from someone familiar with the elaborate version, but ascribed his transmission straight to Mālik (d. 179/796). The second explanation for unexpected concurrence of contents is variation in the wording of versions that a single person transmitted at different times. Thus Mālik might have transmitted different versions of the ḥadīth about the mushrik of Badr at different times, and Ibn Mahdī and Ibn Wahb might have heard the same version from him on the same occasion, while Wakīʿ (for example) might have heard from Mālik only the dictum. Then, Ibn Mahdī, having heard the long version from Mālik, might have at times transmitted it in full, and at times transmitted the abbreviated version that other collectors claim to have from him. The second explanation accounts not only for unexpected correlation, but also for some portion of the observed variation. Combinations of these two phenomena undoubtedly operated in the transmission history of these and most other ḥadīths. This fact erodes but does not efface the strong impression made by the matn-cum-isnād analyses: that the isnāds of these ḥadīths contain historically accurate human and thus geographical information.

The “single-strand” portion of the isnāds below (in Figures 2.1–2.3) the “common links” is immune to matn-cum-isnād analysis. Here Sadeghi’s TT Test is most helpful. The TT Test indicates that the common portion of the three ḥadīths, comprising dictum and closely associated narrative sequence, originated in Medina. This indication has implications for dating that are in line with views advanced by Harald Motzki.³⁰ Specifically, two of the ḥadīths should be assigned to a period earlier than the common-link phenomenon might lead one to believe on the Schachtian assumption that the common link of a ḥadīth was usually responsible for bringing it into circulation and attaching to it a fictitious isnād.³¹ According to the TT Test, a prototype account of the squadron of Jews predates al-Faḍl b. Mūsā (d. 192/807) and originated in Medina. Thus it may well have been narrated by the earlier common-link candidate, Muḥammad b. ʿAmr, the latest Medinan transmitter in the stemma. A prototype account of the rejection of Khubayb b. Yisāf would really have been narrated by a Medinan transmitter before Mustalim, perhaps even by Khubayb b. ʿAbd al-Raḥmān (d. 132/749), who is the latest Medinan in the stemma. It would not have been invented two generations later in Wāsiṭ, as a Schachtian reader of the stemma (Figure 2.3) might conclude.

The combination of the TT Test and matn-cum-isnād analysis yields the probable conclusion that a proto-account embedding the dictum in this distinctive narrative sequence first circulated in Medina, and is to be dated before the death of the earliest last Medinan transmitter in any of the three stemmata: Khubayb b. ʿAbd al-Raḥmān (d. 132/749). In my view, this is the only plausible account of how the three ḥadīths came to share distinctive contents. To appreciate the significance of the stemmata found in Figures 2.1–2.3, consider a counterfactual. If all three ḥadīths had the same two candidate birthplaces (say, Medina and Baṣra), it would be impossible to determine whether the shared contents indicated a Medinan or Baṣran birthplace. In fact, however, the ḥadīth about the squadron of Jews allegedly moved diffusely from Medina to Kūfa, Damascus, and Marw, that about the mushrik of Badr must have been transmitted by the Medinan Mālik, and that about the rejection of Khubayb moved only to Wāsiṭ. It is of course possible to conceive of scenarios in which parts of one or two of the ḥadīths originated somewhere other than Medina. These are of low probability, however, and will not be detailed here. Furthermore, the most plausible of these low-probability alternative scenarios require the import of information from Medina – the names of transmitters, the dictum, etc. – along paths that must have closely resembled those we already have. Even if they were somehow to prove true, these alternative explanations would have little effect on the notion that the dictum and associated narrative sequence is of Medinan origin.

3 Early Muslim View on the Recruitment of Non-Muslim Troops

There is thus no evidence to indicate that Muslims possessed widely shared, principled reasons not to recruit non-Muslims in warfare during the main period of the conquests, roughly coinciding with the pre-Abbasid period. This conclusion is the product of a syllogism. The major premise is that all of the versions of these three ḥadīths, cited from numerous compilations, whose origins reach from Egypt to Transoxania, contain reports with isnāds that converge on pre-Abbasid Medina. Given the general reliability of their isnāds postdating this convergence, as established above, these ḥadīths, with the prophetic dictum they bear, are unlikely to have circulated widely beyond Medina before the Abbasid period. The minor premise is that in numerous juristic discussions of even greater geographic, temporal, and sectarian diversity, such as those cited in note 2 above, one finds no widely shared, principled reasons against allying with non-Muslims in warfare apart from these three ḥadīth. These juristic discussions may be considered reliable weirs, set in the current of the tradition that bore with it the available, principled rationales against recruiting non-Muslims. Now, if principled Islamic rationales for this view are restricted to these three ḥadīths, and if these three ḥadīths were restricted to Medina in pre-Abbasid times, then principled Islamic rationales against recruiting non-Muslims were restricted to Medina in pre-Abbasid times. The conquering Arabs may of course have had a range of reasons to refuse aid from willing, warlike infidels, from tactical to tribal. But these reasons are unlikely to have been widely shared, or to have been overtly Islamic in character. This conclusion highlights the significance of the geographical concurrence of these isnāds; if they had travelled along independent routes in diverse locales (say, Medina, Baṣra, and Ḥimṣ) during the first century, it might be argued that they were widely known among the Arabs, perhaps because they reflected a shared memory of Muḥammad’s policy. Since this is not the case, the reverse holds true.

Of course, this conclusion is not apodeictically certain. Other ḥadīths opposing interreligious military alliance might have circulated widely among the early Muslims, but been overlooked by later jurists who wrote on the topic. Or, certain Qurʾānic verses (such as 3:28, 3:118, or 5:51) might have been widely read as forbidding the recruitment of non-Muslims, but their significance for the matter been ignored or rejected by later jurists.³² Or, versions of these three ḥadīths might have left Medina earlier than the extant isnāds reveal, and circulated widely, but memory of these early transmissions have been lost.³³ These and other such possibilities are remote. The permissive views on military alliance with non-Muslims held by such early figures as al-Zuhrī (d. 124/742), Abū Ḥanīfa (d. 150/767), al-Awzāʿī (d. 157/774), and al-Thawrī (d. 161/778) also suggest that there were no early, widely accepted Islamic reasons against such alliance.³⁴ As I suggest below, this diversity of opinion also highlights the possibility that the three ḥadīths entered circulation in the context of debates on the issue, and not prior to these debates.

4 Toward Using the ḥadīths as Historical Sources

At present it is not possible to reach conclusions about the historical accuracy of the three ḥadīths that contain the dictum. Such conclusions will require study of all available evidence of the limits set by Muḥammad on inclusion in the polity he founded, and in its military activities in particular. Nevertheless, it will be expedient to prepare the way for such study by gesturing to a few features of the evidence that are relevant to the question of historical accuracy. Preliminary indications are that Fred Donner’s vision of regular military cooperation among various monotheists in Islam’s early decades does not face a serious challenge from this quarter. The evidence is addressed under two headings: isnāds and matns.

5 Conclusion

It may eventually be possible to learn more about the circumstances under which the dictum and the narratives in which it is embedded were formulated. Even in the early Islamic heartland of Medina, it is not difficult to identify potential reasons to condemn the recruitment of non-Muslim fighters: the Christian Taghlibites in the army of Yazīd b. Muʿāwīya who marched under the banner of St. Sergius, for instance, or the two hundred Christians of Ayla who were employed by Muʿāwiya to keep order in Medina.⁴⁵ Or, the catalyst might have been not an event but an objectionable doctrine, such as the one implied in the ḥadīth that Ibn Shihāb al-Zuhrī narrated in Medina, describing Muḥammad’s military cooperation with Jews. On this model, the Medinan origin of these three ḥadīths might be explained by reference to the fact that al-Zuhrī, arguably the Marwānids’ most important learned apologist, happened to be Medinan. These are hypotheses to be tested in future studies, not the conclusions of this one.

Instead, the following arguments have been made in this essay: all versions of the three ḥadīths that embed variations on the dictum “I will not accept aid from a mushrik” in a distinctive narrative sequence arose from a proto-version that circulated in Medina no later than the first half of the second/eighth century. The principle that the dictum expresses influenced conceptions of the religious modalities of legitimate state control in early Islam. There is a variety of reasons, most notably obscurities in putative paths of transmission and striking contradictions with parallel accounts, to be wary of assigning these ḥadīths an earlier date. The common geographical origin indicated by the isnāds, which can be shown to convey accurate data about paths of transmission, offers the most plausible explanation for their shared content. I have also suggested that before the ʿAbbāsid period, during the period of their most dramatic military gains, the Arabian conquerors are unlikely to have shared any principled reasons not to recruit non-Muslims, as such.

In addition, the rural setting of the events narrated in these three ḥadīth provided rhetorical advantages to their transmitters. It reinforced the proprietary nature of their knowledge of Muḥammad’s normative dictum, raising the authority value of the reports by implicitly limiting their potential circulation. It leveraged the real implications of rural settings for pre-modern fighting forces, implying that if Muḥammad did not accept non-Muslim aid in the transitional, invisible desert, where the boundaries of fighting forces were porous, then such aid was a fortiori illicit in cities, or on the rural battlefield. Finally, movement among the arid spaces of the Ḥijāzī hinterland created rhetorical space for the repetition and consequent reinforcement of the dictum, and thus of the normative principle it encapsulates. In both this narratological analysis and in adapting the TT Test formulated by Behnam Sadeghi, I have suggested that imagined rural spaces played significant roles in the spatial extension of normative authority in the early Islamic period, and that Islamicists should appreciate the multiple roles played by extra- and inter-urban physical space in their study of the ḥadīth and akhbār, our most important sources for the history of early Islam.

Appendix

Acknowledgements

I am grateful to Marie Legendre, Petra Sijpesteijn, Michael Cook, Behnam Sadeghi, Harald Motzki, Amr Osman, and participants in the Princeton Islamic Studies Colloquium, particularly Dale Correa, for their comments on drafts of this essay.