1. Introduction
North Africa occupies a special position in the history of Islam, as the first region to be separated politically from the caliphate while maintaining the religion. By the early ʿAbbāsid period, the coast had passed out of effective ʿAbbāsid control, while the fringes of the Sahara were dominated by Ibāḍī and Ṣufrī states, which rejected not only the rule of the ʿAbbāsids but also their Sunnī orthodoxy, and were hence rather reserved in their attitude towards the works of the most influential religious scholars of the ʿAbbāsid era. This position is reflected in the religious terminology of the non-Arabic languages of the region, which preserve reflexes of variant forms attested in early Islamic traditions but are obsolete in most other parts of the Islamic world.Footnote 1
A significant number of Berber languages, along with some of their neighbours, refer to the five daily prayers of Islam using words that cannot be derived from their usual Arabic names; most of these are internal formations, briefly discussed by Naït-Zerrad (Reference Naït-Zerrad1998), while others are Arabic loans. Van den Boogert and Kossmann (Reference Boogert and Kossmann1997) give Berber-internal etymologies for these forms, and argue that their consistent use by both Sunnī and Ibāḍī groups indicates that they must have been produced by a single centre of Islamization, probably Ibāḍī, in a period before the Sunnī–Ibāḍī split became a significant factor in the region, i.e. no later than the Umayyad era. Comparable forms, which have attracted less comment, are found in a number of West African languages – notably Soninké and Songhay, the languages of the first major Sahelian states to come into direct contact with Islam – as well as their neighbours to the south, such as Mandinka and Pular. Section 6 of this article tabulates all available prayer names in languages for which data was available, along with references.
In this article I will show that some of these unexpected forms can be traced directly back to variants attested in the early ḥadīth literature, via borrowing or calquing. While modern Islamic literature consistently refers to the prayers by the names Fajr / Ṣubḥ (in the morning), Đ̣uhr Footnote 2 (at noon), ʿAṣr (in the afternoon), Maghrib (at sunset), and ʿIshā ʾ (at nightfall), mainly derived from terms referring to the time of day, authoritative earlier sources including Ṣaḥīḥ al-Bukhārī attest to the existence and wide usage within Arabic of alternative names for several of these prayers, as noted briefly in Wensinck (Reference Wensinck1987). These names have long since disappeared from Arabic usage, but reflexes of them continue to be used across much of North and West Africa. Their ubiquity throughout the Sahara constitutes compelling evidence that the first Arabs to introduce Islam to the region habitually used these alternative terms in their speech, and not the ones that have become usual. In this particular instance, seemingly non-standard traditions of Islam in the Sahara do not reflect local innovation, but rather provide information on the distribution of variation that was accepted within early Islam and only later came to be levelled out.
Other Berber forms that are almost as widespread, however, cannot be derived from Arabic through calquing or otherwise: they are rather derivations based on Berber numerals, with ʿAṣr = four, Maghrib = five, and ʿIshāʾ = six. The reasons for the choice of these numbers have not been convincingly explained. I will argue below that these forms reflect a system of time-keeping widespread among Saharan oases, and used within living memory. The wide usage of these forms suggests that this time-keeping system too dates back at least to the Umayyad period.
2. Early prayer names in Arabic
An obvious starting point for the early variation in prayer names is the ḥadīth literature: convenient sources include Mālik (d. 795), ʿAbd al-Razzāq (d. 826), al-Bukhārī (d. 870) and Muslim (d. 875). For the most part, these all use the familiar names mentioned above; however, they also faithfully record variants that do not correspond to the preferred terminology. Of several ḥadīth reflecting these variations, the following (Bukhārī #547, Reference al-Bukhārī2001: 114) is particularly revealing:
… عن سيار بن سلامة، قال دخلت أنا وأبي، على أبي برزة الأسلمي، فقال له أبي كيف كان رسول 
صلى 
عليه وسلم يصلي المكتوبة فقال كان يصلي الهجير التي تدعونها الأولى حين تدحض الشمس ويصلي العصر ثم يرجع أحدنا إلى رحله في أقصى المدينة والشمس حية ونسيت ما قال في المغرب وكان يستحب أن يؤخر العشاء التي تدعونها العتمة وكان يكره النوم قبلها والحديث بعدها وكان ينفتل من صلاة الغداة حين يعرف الرجل جليسه ويقرأ بالستين إلى المائة.
... from Sayyār b. Salama: “My father and I entered into the presence of Abū Barza al-ʿAslamī. My father asked him: ‘How did the Messenger of God, blessings and peace be upon him, pray the prescribed (prayers)?’ He replied: ‘He used to pray al-hajīr, which you (pl.) call al-ʿūlā, when the sun declines (from the meridian), and pray al-ʿaṣr such that one of us could return to his home at the far end of Medina while the sun was still lively.’ I forget what he said about al-maghrib. ‘And he used to prefer to delay al-ʿishāʾ, which you (pl.) call al-ʿatama, and he used to hate sleeping before it or speaking after it. And he used to return from the ghadāt prayer when a person could recognise the one sitting next to him, and read sixty to a hundred (ʿāyāt.)’”
The people in this conversation were most likely living in Basra in the early eighth century, but this ḥadīth presents them not only as not using the set of prayer names familiar to modern Arabic speakers, but as each using a slightly different set of names. The familiarity of Abū Barza with the set he attributes to “you (pl.)”, however, suggests that the latter rather than the former was prevalent in contemporary Basra. By implication, the terms for which he does not give an alternative were presumably common to both groups.
A similar usage combining features of both sets – possibly with reference to the times rather than to the associated prayers – is attributed to the Prophet himself in Mālik 3:3 (Reference ʿAnas1985: 78) = Bukhārī #652 (Reference al-Bukhārī2001: 132):
... عن أبي هريرة أن رسول 
صلى 
عليه وسلم قال لو يعلم الناس ما في النداء والصف الأول ثم لم يجدوا إلا أن يستهموا عليه لاستهموا ولو يعلمون ما في التهجير لاستبقوا إليه ولو يعلمون ما في العتمة والصبح لأتوهما ولو حبوا.
… from Abū Hurayra, that the Messenger of God, blessings and peace be upon him, said: “If people knew what is in the call and the first row, and found no (alternative) but to draw lots for it, they would draw lots. If they knew what is in al-tahjīr they would race each other to it. If they knew what was in al-ʿatama and al-ṣubḥ, they would come to it even crawling.”
Outside the ḥadīth literature, the book Al-Kāmil by al-Mubarrad (d. 898), from Basra, mentions all three of the relevant terms, in a discussion of genitive constructions (Mubarrad Reference Al-Mubarrad and Wright1864: 128):
وعتمة اسم للوقت، فلذلك سميت الصلاة بذلك الوقت، وكل صلاة مضافة إلى وقتها، تقول: صلاة الغداة، وصلاة الظهر، وصلاة العصر.وما قولك: الصلاة الأولى، فالأولى نعت لها إذا كانت أول ما صلي، وقيل أول ما أظهر.
And ʿatama is the name of a time, and the prayer at that time was named after it; every prayer is muḍāf (annexed) to its time, so you say: ṣalāt al-ghadāt, ṣalāt al-đ̣uhr, ṣalāt al-ʿaṣr. As for your saying: al-ṣalāt al-ʔūlā, “first” (ʔūlā) is an adjective modifying it, it being the first to have been prayed, and some say the first to be made known.
A practically isolated term nevertheless attested quite early is ṣalāt al-nawm, reported in ʿAbd al-Razzāq #2136 (Ṣanʿānī Reference Ṣanʿānī1970: 562):
عن يحيى بن العلاء ، عن الأعمش ، عن أبي وائل قال : طلبت حذيفة ، فقال : لم طلبتني ؟ قال : قلت : للحديث ، فقال : إن عمر بن الخطاب رضي 
عنه ، كان يحذر بالحديث بعد صلاة النوم .
From Yaḥyā b. al-ʿAlāʾ, from al-Aʿmash, from Abū Wāʾil who said: I asked for Ḥuḏayfa, and he said: Why have you asked for me? I said: For conversation. He said: ʿUmar b. al-Khaṭṭāb, may God be pleased with him, used to warn against conversation after ṣalāt al-nawm.
A further dimension of variation applies to the meaning of al-ʿishāʾ. Some speakers used this term to refer not to ʿIshāʾ but to Maghrib – a usage condemned as a Bedouin solecism in the following ḥadīth (Bukhārī #563, Reference al-Bukhārī2001: 117):
حدثنا أبو معمر ـ هو عبد 
بن عمرو ـ قال حدثنا عبد الوارث، عن الحسين، قال حدثنا عبد 
بن بريدة، قال حدثني عبد 
المزني، أن النبي صلى 
عليه وسلم قال "لا تغلبنكم الأعراب على اسم صلاتكم المغرب ".قال الأعراب وتقول هي العشاء.
… ʿAbd Allāh al-Māzinī told us that the Prophet, blessings and peace be upon him, said: “Do not let the Bedouin prevail over you in the name of your prayer al-maghrib.” He said the Bedouin, and they call it al-ʿishāʾ.
This confusion might be avoided by referring to ʿIshāʾ as al-ʿishāʾ al-ʿākhira, as in Muslim 465c (Reference Ḥajjāj2006: 215):
... عن جابر بن عبد 
، أن معاذ بن جبل، كان يصلي مع رسول 
صلى 
عليه وسلم العشاء الآخرة ثم يرجعُ إلى قومهِ فيصلي بهم تلك الصلاة .
… from Jābir b. ʿAbd Allāh that Muʿādh b. Jabal would pray al-ʿishāʾ al-ʿākhira with the Messenger of God, peace and blessings be upon him, then go back to his people and lead them in that prayer.
We can conclude that at least the variants shown in Table 1 were current in Arabic in the Umayyad era. A study of narrators might indicate whether these variants can be associated with particular regions or tribes, but that would take us beyond the scope of this paper. However, the same sources that indicate these variants also help explain their disappearance. The emergence of these ḥadīth compilations reflects a wider move to codify the religion in writing, which in itself gave the preferred variants cross-regional prestige; on top of this, the main variant names of Maghrib and ʿIshāʾ were explicitly condemned as improper in several ḥadīth. The example above is considered ṣaḥīḥ, as is what looks rather like a variant of it in Muslim #644a:
وحدثني زهير بن حرب، وابن أبي عمر، قال زهير حدثنا سفيان بن عيينة، عن ابن أبي لبيد، عن أبي سلمة، عن عبد 
بن عمر، قال سمعت رسول 
صلى 
عليه وسلم يقول " لاَ تغلبنكم الأعراب على اسم صلاتكم ألا إنها العشاء وهم يعتمون بالإبل ".
[…] From ʿAbd Allāh b. ʿUmar, who said: I heard the Messenger of God, blessings and peace be upon him, say: “Let the Bedouin not prevail over you in the name of your prayer, for it is al-ʿishāʾ; (they call it ʿatama because) they stay late (ʿtm) with their camels”.
But some less well-supported ḥadīth object in much more extreme terms, such as Ibn ʿAbī Shayba #36971 (Reference Ibn ʿAbī Shayba and ʿAwwāma2006: 537), narrating from the same source:
حدثنا وكيع قال ثنا شريك عن أبي فزارة العبسي عن ميمون بن مهرانَ قال قلت لعبد 
بن عمر من أول من سماها العتمة ؟ قال : الشيطان .
… from Maymūn b. Mihrān, who said: I asked ʿAbd Allāh b. ʿUmar “Who first called it al-ʿatama?” He replied: “The Devil!”
The end result, observable today, is that Muslim prayer terminology is broadly uniform, in the Maghreb as elsewhere, and the variant oral traditions to which these ḥadīth attest are known only to scholars. However, at the time of these compilations, most of North Africa was outside of ʿAbbāsid control, and much of it, particularly the northern Sahara, was non-Sunnī. Mālikī Sunnism, though already present locally, would not begin to attain its current overwhelming preponderance until centuries later. This left ample time for the prayer names based on alternative oral traditions to become fixed in the region's languages, and the evidence below suggests that this is precisely what happened.
Table 1. Variant prayer names in Arabic
3. Variants in the western Islamic world
3.1. Ṣalāt al-ʿŪlā (Đ̣uhr)
The term literally means “the first (f.)”, ʿūlā being the feminine form of ʿawwal. A rather later authority, the Andalusian scholar al-Qurṭubī (d. 1273), offers an explanation for the name in his commentary on Quran 12: 109 (Reference Qurṭubī2006: 411):
ومن قال صلاة الأولى فمعناه: عند صلاة الفريضة الأولى؛ وإنما سميت الأولى لأنها أول ما صلي حين فرضت الصلاة، وأول ما أظهر؛ فلذلك قيل لها أيضا الظهر.
Whoever says ṣalāt al-ʔūlā, its meaning is “the first among the obligatory prayers”, called “first” because it was the first prayer to be prayed when prayer was made obligatory, and the first to appear; that is why it is also called al-đ̣uhr.
At an early period, this name must have been quite widespread in the western half of the Islamic world; the only part of North Africa where it is consistently absent is the Mediterranean coast and its mountains. Even within Arabic, some Moroccan varieties still use luwli (Harrell and Sobelman Reference Harrell and Sobelman2004: 73). As a loan, it survives in scattered points stretching from southern Syria to the Niger bend: Western Neo-Aramaic ʿalūla “Mittag” (Arnold Reference Arnold1990: 333); in two eastern Berber languages – Sīwa luli “Đ̣uhr” (western Egypt), Awjila alûli “Đ̣uhr” (eastern Libya); and in Songhay aluula “Đ̣uhr” (eastern Mali, western Niger). (See section 6 for sources and tables.) The i in the Berber forms reflects final imāla, echoed in Arabic dialects of the Egyptian oases (Woidich and Behnstedt Reference Woidich and Behnstedt1982: 49).
As a calque, it survives in an area even larger than and even more distant from the Middle Eastern heartland of early Islam. Within Berber, the commonest form for “Đ̣uhr”, excluding loanwords, is one that can be reconstructed as ti-zwar-(nin), a feminine plural derived from the Berber root √zwr “be first, precede” (Tashelḥiyt zwur, Zenaga äẕ̌bər, Tamasheq ǎzzar). The post-consonantal w regularly corresponds to b in Zenaga, while many varieties replace zw with zz. Such forms are used to mean “Đ̣uhr” across an enormous area including: Tashelḥiyt and Tamazight in southern Morocco; the Zenati Berber-speaking oases stretching across the northern Algerian Sahara, including the Gūrāra and Mẓāb; north-western Libya, including the Nafūsa mountains and Zuwāra; Tuareg in the central Sahara; and Zenaga in Mauritania. Wolof (Senegal) and Korandjé (Algeria) have borrowed the Zenaga form: tisbar, təzbəṛṛən respectively.
The relevant Berber root has both the senses of “be first” and “precede”, and Naït-Zerrad (Reference Naït-Zerrad1998: 63) rendered these forms with the latter sense, as “celles qui précèdent”, suggesting tentatively that it was so named because it was followed by additional recitations, as Lanfry noted for Ghadames. However, the Arabic loans exemplified above indicate that the relevant sense is not “precede”, but rather “be first”.
Soninké, the language of the early Sahelian state of Ghāna, has calqued rather than borrowed the term, whether directly from Arabic or from Berber, yielding sállì-fànà “prayer-first”. This term has in turn been borrowed into West African languages to its south, such as Mandinka and Pular. In Bambara, it has been reinterpreted as based on fàná “meal” (Bailleul Reference Bailleul1981), but the order indicates that this is a folk etymology, since the Bambara possessive construction is head-final.
3.2. *Ṣalāt al-Fiṭr (Maghrib)
Unlike the other terms discussed here, this term, literally “prayer of fast-breaking” (fiṭr being a deverbal noun from faṭar a “break apart; break (fast)”), appears to be non-classical in the appropriate sense; the normal sense of the phrase is “(congregational) prayer on ʿĪd al-Fiṭr (the fast-breaking festival after Ramadan)”. The semantic shift is natural, given that in Ramadan one breaks one's fast each day at the time of Maghrib. No Arabic attestations of this usage have so far been found. However, their existence is implied by a series of West African terms for Maghrib which appear to be direct loans: Soninké fùtúrò, Mandinka fitíri, Timbuktu Songhay fitirow / futurow, etc. The vowels of the Soninké form suggest a possible alternative derivation from another form of the same root, fuṭūr “breakfast, fast-breaking meal”; however, the absence of a long vowel and the other West African forms make this less probable.
This form was calqued into Berber at an early period, yielding ti-n-wučči “that (f.) of eating”. It survives mainly in the westernmost Berber varieties – Tashelḥiyt, Tamazight in southern Morocco, and Zenaga in Mauritania – but also in Zuwāra in north-western Libya. It was loaned into Korandjé (Algeria) as tsyunəs; the Wolof form timis may also derive from this, assuming assimilation *nw > m.
This form is not attested in Tuareg, which consistently uses alməẓ “twilight”, presumably a not very literal translation of maghrib based on a pre-existing time period. The predominant Zenati forms will be discussed below.
3.3. Ṣalāt al-ʿIshāʾ (ʿIshāʾ/Maghrib)
In the sense of ʿIshāʾ, this form is dominant throughout modern Arabic, so its presence as a loan hardly merits comment. However, it is also present in one case as a calque within Berber: Nafusi ti-n-mənsi “that (f.) of supper”. Since the homonymy of reflexes of ʿishāʾ (“evening”, from the root ʕashiy a “be night-blind”) and ʿashāʾ is historically secondary, resulting from the weakening or loss of short vowels, this is probably a late development; comparison to Medieval Eastern Berber, which used ti-n-iḍas like most other Saharan Berber varieties (see below), would appear to confirm this.
A potentially more interesting calque based on this word is provided by Awjila mnishiw “p. del tramonto”, to be derived by metathesis from *n-mishiw, cp. n “of”, amishiw “cena [supper]”. Available data on this form is ambiguous; Mauri (Reference Mauri2011, ms) indicates that this refers to ʿIshāʾ, but Paradisi's gloss rather suggests Maghrib. If the latter is correct, this usage would reflect the confusing dialectal practice, mentioned and specifically condemned in Bukhārī, of calling Maghrib al-ʿishāʾ. However, since Paradisi (Reference Paradisi1960) gives no gloss for ʿIshāʾ, it is difficult to be certain of the intended meaning.
3.4. Ṣalāt al-ʿAtama (ʿIshāʾ)
Several ḥadīth suggest that, in principle, the difference in name here corresponds with a difference in timing: al-ʿatama (“first third of the night”, from the same root as ʿatam a “be late, become dark”) refers to the late end of the time in which ʿIshāʾ should be prayed, whereas al-ʿishāʾ refers to the early end. This form is rather less widely attested in North Africa than those seen so far, and no obvious calques are attested, but it occurs as a loan at opposite ends of the region. Alongside luli < al-ʿūlā for Đ̣uhr, it is still used in Siwi for ʿIshāʾ: lʿətmət. This is likewise the source for Morisco Spanish alatamo. However, reflexes of both al-ʿishāʾ and al-ʿatama were adopted by the Church for translating canonical hours, confirming that the original semantic distinction between the two times was preserved in at least some registers of Andalusi Arabic: Alcalá and Lagarde (Reference de Alcalá and de Lagarde1883: 329) renders Complines as al âyxĭ, and Matins as al âáteme.
While not a calque, the widespread Tuareg name of ʿIshāʾ, aẓuẓəg “milking”, may be indirectly associated with this name. Some of the ḥadīth cited as condemning the term ʿatama explain its usage as the result of a Bedouin practice of praying it after milking the camels after nightfall, cf. Musnad Aḥmad #4688 (Ibn Ḥanbal Reference Ibn Ḥanbal, al-Arna'ūṭ and Murshid1995: 315) (apparently a variant of the ḥadīth from Muslim cited above):
(حديث مرفوع) حدثنا يحيى ، عن سفيان ، حدثني عبد 
بن أبي لبيد ، عن أبي سلمة ، عن ابن عمر ، عن النبي صلى 
عليه وسلم قال : " لا يغلبنكم الْأَعراب على اسم صلاتكم ، فإنها العشاء ، إنما يدعونها العتمة ، لِإعتامهم بالإبل لحلابها " .
[…] From Ibn ʿUmar, from the Prophet, blessings and peace be upon him, who said: “Let the Bedouins not prevail over you in the name of your prayer, for it is al-ʿishāʾ; they (the Bedouins) call it al-ʿatama because they keep the camels out late to milk them.”
3.5. Ṣalāt al-Nawm (ʿIshāʾ)
This form, literally “prayer of sleep”, is very marginal in the ḥadīth literature, but its rarity in Arabic contrasts with the overwhelming frequency in Berber of its calque, ti-n-iḍəs “that (f.) of sleep”. This form is used in Tashelḥiyt and Tamazight in southern Morocco; the Zenati Berber-speaking oases stretching across the northern Algerian Sahara, including the Gūrāra and Mẓāb; western Libya, including Zuwāra, Ghadames, and Al-Fuqahāʾ; and Zenaga in Mauritania. It has been borrowed into Korandjé (Algeria), as tsyạḍə̣s.
The same calque was made in Soninké, whether directly from Arabic or from Berber: sákhú-fó (“sleep-thing”). From Soninké this form was borrowed, with regular loss of kh and compensatory vowel lengthening, into many other West African languages: Mandinka sáafu, Songhay saafoo, Pular safʿi, etc. The Songhay form is explained by Prost (Reference Prost1977: 637), following a suggestion by Marty (Reference Marty1920: 228) for Peul, as deriving from Arabic shafaq “twilight”, but if that were the case we would expect a short vowel: *safoo rather than saafoo. Nor would this explain the vowels and the position of the kh in the Soninké form. To make matters worse, ʿIshāʾ cannot be prayed until after shafaq has passed (Monnot Reference Monnot1995: 960); shafaq is in fact the time of the Maghrib prayer. The Peul form cited by Marty, safako, clearly is based on shafaq, but given its meaning, it is better viewed as the result of a folk-etymological reinterpretation (among Arabic-literate scholars) of a term originally based on sákhú-fó. Marty (Reference Marty1920: 70) links the Songhay form itself to shafʿ, the name of the optional prayer immediately following ʿIshāʾ; this fits the Pular form, but otherwise again fails to account for the long aa and the other regional forms.
3.6. Ṣalāt al-Ghadāt (Ṣubḥ)
This form has already been exemplified above. Medieval Eastern Berber used what I take to be a calque of ghadāt (properly “early morning”), folk-etymologized as based on ghad “tomorrow”: ti-n-uzəčča “that (f.) of tomorrow”. This term, however, has not survived into any reported modern variety. Some of the Tuareg forms are ambiguous between ṣubḥ and ghadāt: tufat means both “morning” and “tomorrow”, like Spanish mañana. Other Berber terms for the morning prayer show far more variation than the other four prayers; where they are not Arabic loans, they may be calques of ṣubḥ “(early) morning” (Tamasheq, Zenaga, Sīwa), or just terms meaning “early” (Zuwāra, Nafūsa).
3.7. Ṣalāt al-Ghabasha (Ṣubḥ)
Within North Africa, borrowed reflexes of this form are reported only in the Berber languages of the Mẓāb and Ouargla, two of the last strongholds of Ibāḍism, and Awjila (Paradisi glosses it only as “morning”, but Mauri (Reference Mauri2011, ms) confirms the prayer time sense). No early authority for this form as a prayer name has yet been found, but it is in current use in colloquial Arabic in the other main region where Ibāḍism has been maintained, eastern Arabia. Examples from Oman in a highly colloquial register can readily be found online, e.g.:
جاينكم بقوة بس تو اذن معنا..صلاة الغبشة قصدي الفجر
Coming in force but now the aḏān is with us ... the Ghubsha, I mean Fajr, prayer.Footnote 3
سحور حال إيش !! نص الشعب من بعد صلاة الغبشة ، […]
Suḥūr like what!! Half the people after Ghubsha prayer, […]Footnote 4
While the name itself seems absent from the ḥadīth literature, its motivation can readily be found there; cf. Mālik #1.9 (Reference ʿAnas1985: 8):
[...] أنه سأل أبا هريرة عن وقت الصلاة فقال أبو هريرة أنا أخبرك صل الظهر إذا كان ظلك مثلك والعصر إذا كان ظلك مثليك والمغرب إذا غربت الشمس والعشاء ما بينك وبين ثلث الليل وصل الصبح بغبش يعني الغلس
[…] he asked Abū Hurayra about the time of prayer. Abū Hurayra replied: “I will tell you. Pray Đ̣uhr when your shadow is your size, and ʿAṣr when it is twice your size, and Maghrib when the sun sets, and ʿIshāʾ within the (first) third of the night, and pray Ṣubḥ in ghabash, meaning ghalas (just before daybreak).”
4. Berber-internal innovations
4.1. Prayer names based on calling
In several varieties spoken in or around south-western Libya, Đ̣uhr is called iməghri, a verbal noun from ghər “to call”. The same term is found in a Tuareg variety further from Libya, Tamasheq, but with reference to Ṣubḥ. A probably independent development of the same kind is Ayt Hdiddou tighuṛiwin, a feminine plural based on the same root, for Maghrib.
4.2. Prayer names based on numerals
While most Berber languages have borrowed most of their numerals from Arabic, several preserve the original system, as exemplified in Table 2.
Table 2. Some numerals in Berber
As noted by Van den Boogert and Kossmann (Reference Boogert and Kossmann1997), these seem to provide the sources for some Berber prayer names. Throughout a wide area, roughly coinciding with the distribution of ti-zwar(-nin) above (including Tashelḥiyt and Tamazight in southern Morocco; the Zenati Berber-speaking oases stretching across the northern Algerian Sahara, including the Gūrāra and Mẓāb; north-western Libya, including the Nafūsa mountains and Zuwāra; Tuareg in the central Sahara; and Zenaga in Mauritania), ʿaṣr is referred to as t-akkʷẓ-in, a feminine plural based on kkuẓ “four”. Beyond Berber, this form has been borrowed from Zenaga into Wolof, as tàkkusaan, and Korandjé, as tsakʷẓẓən.
In a rather smaller area, including most of the Zenati Berber-speaking oases of the northern Algerian Sahara, along with Nafūsa and Al-Fuqahāʾ in western Libya, the maghrib prayer is referred to as ti-səmms-in, a feminine plural based on səmmus “five”. The term is also attested with a vaguer meaning “afternoon” in Tamezret (southern Tunisia).
In many Tuareg varieties, the ʿishāʾ prayer is ti-suḍs-en, an etymologically ambiguous feminine plural. Naït-Zerrad (Reference Naït-Zerrad1998) derives it from the causative verb suḍəs “to make sleep” (Foucauld Reference Foucauld1951), thus fitting it with the tin-iḍəs forms seen above. However, Van den Boogert and Kossmann (Reference Boogert and Kossmann1997) note that a derivation from səḍis “six” cannot be excluded. In light of the other two numerical forms, this derivation seems plausible.
Naït-Zerrad (Reference Naït-Zerrad1998: 63) explained the forms for ʿAṣr and Maghrib on the basis that the (24-hour) day starts after sunset, making Maghrib the fifth and final prayer within it. This view is reasonable in itself; such a definition of the day was commonplace in the Islamic world, and is supported by the Morisco Spanish form alajere “Maghrib” < al-ʿākhira (cp. Alcalá’s (Reference de Alcalá and de Lagarde1883: 329) al aḳirĭ, given as “Nones” but probably transposed from “Vespers”). However, this does not sit well with the fact that, in all the languages which use these numerical forms, Đ̣uhr is consistently named as “the first”. If Maghrib is the fifth prayer, then Đ̣uhr should be third; or, conversely, if speakers who view Đ̣uhr as “first” are naming prayers ordinally, we expect ʿAṣr to be “second”, and Maghrib “third”. Moreover, we would have to dismiss the otherwise promising similarity between the Tuareg terms for ʿIshāʾ and “six” as a coincidence: there is no sixth obligatory prayer! Nor are there three prayers between Đ̣uhr and ʿAṣr. For an explanation to fit the numerical forms for Đ̣uhr, ʿAṣr, and Maghrib all at once, let alone ʿIshāʾ, we must abandon the idea of ordinally named prayers.
Happily, an alternative solution is available. Let us take the numerals to represent not an ordering of prayers, but rather the times of the prayers, as measured by some numerical time-keeping system. The average time between Đ̣uhr and Maghrib is six hours; in fact, if we use seasonal rather than equinoctial hours, as medieval sources commonly do, the time between them is always six seasonal hours by definition. Thus, if the interval between Đ̣uhr and Maghrib is (5–1) = four units, the unit in question must equal (6/4) hours, i.e. 1 hour 30 minutes. The day would be made up of eight such units, and the night of another eight such units.
Such a unit is not a mere artefact of imaginative reconstruction; it is well attested in Saharan Berber culture. In the desert, irrigation water is frequently too scarce to allow everyone unlimited use, and measurement by volume is rarely practical. Most Saharan oases instead handle water rights through time measurement: for any particular spring or canal, each owner has the right to irrigate from it for a specific time-span each day, determined contractually. In order to measure this time-span, they divide the day into a fixed number of units, just as modern society does. In Ouargla, Ghadames, and Sīwa, the day was divided into eight units of 1 hour 30 minutes each, called ttmən at Sīwa and Ouargla (< Arabic ṯumn “one-eighth”; Souag (Reference Souag2013: 119) for Sīwa, Delheure Reference Delheure1988: 196 for Ouargla), and anattam at Ghadames (< Berber, cp. tam “eight”; Lanfry Reference Lanfry1973: 273).
This would yield, on average, the following times for prayers:
1st anattam Đ̣uhr 12:00 (normalized; add or subtract depending on time zone)
2nd anattam – 13:30
3rd anattam – 15:00
4th anattam ʿAṣr 16:30
5th anattam Maghrib 18:00
6th anattam ʿIshāʾ 19:30
The time given here for ʿAṣr will strike most modern North Africans as a little too late. Contrast the actual spring equinox times for Ghadames as calculated by IslamicFinder.org,Footnote 5 subtracting thirty minutes to normalize for the time zone:
Đ̣uhr 12:00
ʿAṣr 15:27
Maghrib 18:05
ʿIshāʾ 19:22
According to this calculation, Maghrib and ʿIshāʾ are correct to within less than ten minutes. However, ʿAṣr is incorrect by an hour. This, however, can easily be remedied. ʿAṣr has not one recognized time, but two: while most schools prefer to pray it when the shadow of an object is longer than at noon by the same length as its height, the Ḥanafī school instead prays it when it is longer by twice its height, and both alternatives are supported by ḥadīth (King Reference King1990). If we change the calculation method to Ḥanafī on the same site, and again subtract thirty minutes, then we get the following:
Đ̣uhr 12:00
ʿAṣr 16:22
Maghrib 18:05
ʿIshāʾ 19:22
On the simple assumption that the twice-height method was followed, the anomaly thus disappears, and the implied timing is almost exactly correct.
Archaeological investigations of Garamantian settlements have made it clear that irrigation-based oasis agriculture was already practised by the Berbers of central Libya centuries before Islam (Wilson Reference Wilson2005). Such irrigation is hardly practicable without some means of dividing up time in order to divide water appropriately. It is therefore probable that, at the arrival of Islam, the settled Berbers of the Sahara and its fringes already had a time-telling system corresponding to the anattam of Ghadames and the ttmən of Sīwa and Ouargla. They adopted noon as “the first” based on the prayer's alternative Arabic name, and named the following prayers based on the number of anattams elapsed counting from noon.
5. Conclusion
Saharan and Moroccan Berber, Soninké and Songhay have all preserved, and transmitted to their neighbours, traces of Islamic religious terminology found in the ḥadīth literature but largely obliterated elsewhere by pressures towards standardization which can already be seen in that very literature. Indeed, Siwi Berber preserves the variant system labelled “Basran” above almost in its entirety. The terminology al-ʿūlā for Đ̣uhr and al-nawm/al-ʿatama for ʿIshāʾ must have been dominant among the Muslims who reached the region in the Umayyad period. These variants' distribution correlates well with the fact that, at the time when what would come to be accepted as the key ḥadīth collections and Sunnī madhhabs were emerging, the trans-Saharan trade was largely in the hands of Muslims who did not accept the political authority of the ʿAbbāsid empire or the religious authority of its scholars. In addition to calquing Arabic terminology, Berber also coined new terms based on pre-existing Berber time-keeping systems, whose relative antiquity this fact confirms.
6. Data: Names of prayer times in languages of the western Islamic world
Tables 3 and 4 summarize the data used in this paper. A few Berber languages also have attested native terms for tarawīḥ, the long supererogatory prayers of Ramadan: Ouargla tizgrarin, Zenaga təẕ̌əgrärən. The etymology is transparent in both cases, being the feminine plural of the Berber adjective “long” in Ouargla (though Zenaga uses a different root for the adjective).
Table 3. Variant prayer names in Berber
Table 4. Variant prayer names in non-Berber languages
