al-Ṭalāq 65:2

Fa-idhā balaghna ajalahunna fa-amsikūhunna bi-maʿrūfin aw fāriqūhunna bi-maʿrūf, wa ashhidū dhaway ʿadlin minkum wa aqīmu'l-shahādata li'llāh, dhālikum yūʿaẓu bihi man kāna yuʾminu bi'llāhi wa'l-yawmi'l-ākhir, wa man yattaqi'llāha yajʿal lahu makhrajā

"When they reach the end of their term, keep them in the recognised good way or part from them in the recognised good way. Call to witness two just persons from among you, and hold the testimony upright for God. With this, admonition is given to whoever believes in God and the Last Day. And whoever guards himself against God, He opens for him a way out."

أَجَل — root أ ج ل (ʾ-j-l). The sense of the word is not "period" but something more exact: the end of a period, its due date, its terminal point.

This is a nicety in Arabic that has to be attended to. Ajal is used both for the period itself and for the moment at which the period ends; but its central sense is the end. The word is commonly met in only one of its uses, the appointed time of death; but in Arabic it serves for the maturity of a debt as well — daynun muʾajjal, a deferred debt.

And precisely this commercial use is meaningful here: the waiting period is a term of maturity. It falls due on a determined date, and when it falls due a decision must be made. The commercial-legal language of the sūra frames the process of divorce like a relation of debt: a counted period, a maturity, two witnesses, and a record.

بَلَغْنَ — root ب ل غ (b-l-gh): to arrive at a place, to reach, to attain. From the same root balāgh (delivery, conveying a message), bāligh (one who reaches; one come of age), balāgha (a word's reaching its target). The root's second occurrence in this sūra is in verse 3: inna'llāha bālighu amrih — there what reaches is God's command, here what is reached is a term of maturity.

The verb is feminine plural: balaghna — "they (the women) reached". That is, it is the women who reach the end of the period; and this prepares the party addressed in the next sentence.

Architecturally the most beautiful place in the verse is here.

أَمْسَكَ — root م س ك (m-s-k). The concrete sense: to hold firmly by hand, to grip, not to let go. From the same root misk (musk, a scent that holds and stays), imsāk (holding; abstaining from food in the fast), mustamsik (one who holds on).

فَارَقَ — root ف ر ق (f-r-q). To separate, to divide. This root was taken up in al-Bayyina in connection with tafarraqa (to be divided); what was recorded there was this: at the centre of the root there is a whole's being parted into pieces. Farq, firqa, furqān, tafriqa are from the same root.

The difference of pattern matters. Fāriqa is of the مُفَاعَلَة (mufāʿala) form — like tafāʿul this form too carries reciprocity, but in a different manner: mufāʿala announces that the act issues from one party and is directed at the other and meets a response. Qātala (he fought with someone), sāfara (he travelled — he parted from a place).

That is, the verse does not say "separate them" (farriqūhunna); it says "part from them", "part mutually". The separation is set up not as a one-sided operation but as a movement that covers both parties. This is an observation about the grammar; I draw no legal conclusion from it.

And now the main matter: بِمَعْرُوفٍ, twice.

Look at the structure of the sentence:

fa-amsikūhunna bi-maʿrūf — aw — fāriqūhunna bi-maʿrūf

Two options are offered and the same qualification is attached to both of them. This is a symmetry that must not be passed over.

What one would expect is that the qualification be attached only to the second option: "keep them; or, if you are going to separate, separate well." For the risk of behaving badly is higher in the separation. The verse does not do this. Keeping too has a measure.

And the reason for this is worth thinking about. A man can do harm by keeping his wife too: not divorcing her and leaving her in suspense, holding her in a position neither married nor divorced, neither letting her go nor taking her in. This was a known form of wrong in that period, and the Qurʾān forbids it explicitly elsewhere:

وَلَا تُمْسِكُوهُنَّ ضِرَارًا لِّتَعْتَدُوا۟ "Do not keep them in order to do harm, so as to transgress against them." (al-Baqara 2/231)

There too the same verb (tumsikūhunna) is used and harmful keeping is named. Al-Ṭalāq 65/2 expresses the same matter from its positive side: if you are keeping, keep bi-maʿrūf.

That is, the verse oversees both doors. Neither staying nor going is a free field. The measure of both is the same and comes from outside.

Root ع ر ف (ʿ-r-f). To know, to recognise — but a particular kind of knowing: to recognise what has been met with before. Maʿrifa is not learning for the first time but recognising. From the same root ʿirfān, taʿrīf (introducing), ʿurf (the conduct known and accepted in a society), and the day of ʿArafa (the meeting/being known at ʿArafāt).

مَعْرُوف is a passive participle: known, recognised. That is, it does not mean "good"; it means "what everyone knows to be good".

This distinction carries the whole force of the word. The verse does not leave the measure of goodness to the conscience of the parties; it ties it to the common knowledge of the community.

Why? Because at the moment of divorce the parties' own measures are not trustworthy. Everyone finds his own conduct reasonable. The qualification maʿrūf carries the measure outside: what you do is not what you find reasonable; it is the good that those around you recognise, know, and have a name for.

And this has a concrete consequence: maʿrūf is a measure that can be demonstrated. "I behaved as I felt like behaving" is not defensible; "I behaved in the way everyone knows" is either true or not, and someone looking on from outside can decide.

The second face of the word. The opposite of maʿrūf in the Qurʾān is مُنْكَر (munkar): from the root n-k-r — unrecognised, strange, found odd. That is, in the moral vocabulary of the Qurʾān good is named "familiar" and evil "unfamiliar". Evil is what does not fit the setting, what stands out, what makes people say "what is this now?"

This does not amount to reducing morality to social relativity — the Qurʾān does not define munkar merely as "outside custom", it also forbids it. But the choice of the word itself asks that the moral measure be visible and shared.

The third mechanism of the regulation: record.

شَهِدَ — root ش ه د (sh-h-d). Its sense has two layers: to be present and to report what one has seen. The witness is the one who was there at the moment of the event and afterwards conveys it. From the same root shahāda, mashhad (the place where an event occurred), shāhid, shahīd, mushāhada.

That the root's original sense is "to be present" is important: witnessing is not a transmission of information but a presence. The witness becomes part of the event.

The verb is in the إِفْعَال form (ashhidū) — causative: "call to witness", that is, make some people witnesses. The command is directed not to those who witness but to the party: have the record set up.

ذَوَىْ عَدْلٍ — "two possessors of justice". Dhawā is the dual of dhū (possessor). The number is two and it is given in the dual form.

عَدْل — root ع د ل (ʿ-d-l). The concrete sense: equivalence, the equality of two sides. ʿAdl is the levelling of the two pans of a balance; from the same root muʿādala (equation), taʿdīl (equalising), iʿtidāl (balance), ʿadīl (equal, of like value).

That is, "a possessor of justice" is the balanced person: the one who does not lean to one side.

مِّنكُمْ — "from among you". The witnesses are not from outside but from within the community. This shows that the record is not institutional but belongs to the congregation: there is no court, no register; but there is a community.

To see the function of the ruling about witnessing, it is enough to think what would happen in an order without witnesses.

In an order without witnesses divorce is an entirely private event. When it was done, with what words it was done, how many times it was done — all of it is left to the memory and the assertion of the parties. And when a dispute arises, which of the assertions is to count as valid is determined not by evidence but by power.

The witness closes this gap:

• It fixes the time. When the divorce occurred, and therefore when the waiting period ended, becomes beyond dispute.
• It fixes the content. The defence "that is not what I said" is cut off.
• And it carries the event from the private into the public. This is perhaps the most important: an operation a community knows about runs differently from one it does not. The man's conduct now becomes observable — and only by this means does the oversight of the bi-maʿrūf qualification become possible.

Two rulings complete one another: bi-maʿrūf sets a measure from outside, ashhidū makes the application of that measure visible from outside. Had measure and observation not come together, neither would have been of any use.

At which moment the witnessing belongs — to the divorce itself, or to the decision to keep or to part at the end of the waiting period — has been debated in the sources; a divergence of view is also transmitted as to whether witnessing in this context is a condition or a recommendation. I do not go into the detail of these disagreements or adopt a preference among them.

And now the verse turns to the witnesses.

أَقِيمُوا۟ — root ق و م (q-w-m). This root was treated in detail in al-Fātiḥa (mustaqīm), in al-Baqara at 2/3 (iqāmatu'l-ṣalāt) and in al-Bayyina (qayyima, dīnu'l-qayyima). I do not repeat those analyses. A reminder: the centre of the root is to stand and to be upright — both to be without crookedness and not to collapse. إِقَامَة (Form IV) is causative: to hold upright.

And the choice of the verb here is striking. The verse does not say "bear witness". It does not say ishhadū. It says أَقِيمُوا۟ ٱلشَّهَٰدَةَ — "hold the testimony upright".

This is the very verb the Qurʾān uses for prayer: aqīmu'l-ṣalāt. In al-Baqara the reason for this verb's being chosen for prayer was treated: prayer is not something "performed" but something held upright; because it tends to fall down it needs continual support.

The same logic is set up for testimony. Testimony by its nature tends to lean:

• The witness may know one of the parties.
• Telling the truth may have a price.
• Keeping silent looks easier and less risky than lying.

That is, testimony left to itself does not stay upright; it leans. The verb says this.

And the Qurʾān describes this risk of leaning explicitly elsewhere:

يَٰٓأَيُّهَا ٱلَّذِينَ ءَامَنُوا۟ كُونُوا۟ قَوَّٰمِينَ بِٱلْقِسْطِ شُهَدَآءَ لِلَّهِ وَلَوْ عَلَىٰٓ أَنفُسِكُمْ أَوِ ٱلْوَٰلِدَيْنِ وَٱلْأَقْرَبِينَ "O you who believe! Be upholders of justice, witnesses for God, even against yourselves or your parents and kinsfolk." (al-Nisāʾ 4/135)

There too the same root (qawwāmīn), the same qualification (li'llāh), and the address of the leaning written out plainly: yourself, your parents, your near kin. In the section al-Maʿārij this verse was taken up in connection with the single public item on that list.

لِلَّهِ — "for God". The function of the qualification: the address of the testimony is neither of the parties. The witness is not answerable to the man who called him; nor to the community that summoned him. The address is raised upward — and this is the only thing that frees the witness from the pressure of the parties.

"With this, admonition is given to whoever believes in God and the Last Day."

وَعَظَ — root و ع ظ (w-ʿ-ẓ). Waʿẓ: to counsel someone by reminding him of the outcome. The lexicographers record that within the root there are elements of softening and of drawing attention to the outcome.

And the word order of the sentence sets a qualification: the admonition is given to the believer.

This is not an exclusion but a finding — and it is the same as the structure of hudan li'l-muttaqīn (a guidance for those who guard themselves) at 2/2 in al-Baqara. The solution recorded there holds here too: the call of the text is to everyone; its being of actual benefit depends on the person's own posture.

But there is something additional here, and it is proper to the subject: the effect of the admonition is tied to belief in the hereafter.

The reason is concrete. Most of the rulings in this sūra have no worldly sanction, or only a weak one. There is no police force to make the man count the period rightly, keep the woman in the house, pay the maintenance in full. The rule works only on the person who takes into account a reckoning he does not see.

This is another appearance of the finding made in al-Māʿūn: there it was recorded that the only support for not repelling the orphan is belief in the reckoning — because repelling the orphan has no worldly cost at all. There is a similar structure here: counting the waiting period short, treating the woman narrowly, has no cost. Where the cost is invisible, the rule stands only by belief.

One of the most quoted sentences of the Qurʾān. And in being quoted it is almost always torn from its context.

Let us look first at the word.

Root خ ر ج (kh-r-j). To go out, to come out. From the same root khurūj (exit), ikhrāj (putting out), kharāj (what comes out of the soil; and tax), makhraj (the place of going out; and the point at which a letter issues from the mouth — makhārij al-ḥurūf).

Pattern: مَفْعَل. In Arabic this pattern is either a noun of place (where the act is done), a noun of time (when it is done), or a mīmic verbal noun (the act itself).

• On the noun-of-place reading: makhraj = the place one will go out by. A concrete door.
• On the mīmic-verbal-noun reading: makhraj = going out, exit. The act itself.

Both readings are possible and both are present in the classical sources. The difference is fine but real: on the first a place is promised, on the second a possibility.

I think the first reading suits the language of the sūra better, and my ground is the weave of the root we shall see under the next heading. This is a preference; it is not binding.

Its coming indefinite. Not al-makhraj but makhrajan. That is, not "the way out" but "a way out". What it is is not said, where it opens onto is not said. The indeterminacy can be read here both as magnification ("such a way out as…") and as real indeterminacy — and the two can stand together.

And now the main matter. Almost everyone who quotes this verse takes it on its own. But the verse is the place where the same root occurs for the third time — and the two previous occurrences change what the sentence says.

The root occurs in the sūra in these three places:

Read the table. With the same root the sūra first forbids two goings out, then promises one going out.

And what it says is this:

Do not set about opening yourself a way out. You cannot open the door — and when you try to, the door you open turns out to be putting the other person into the street. The one who will open the door is another.

This is my own reading, and I do not present it as a transmitted report. But that the root occurs three times, with three different subjects and two opposite rulings, is the text's own datum; and a repetition of this density occurs within too narrow a space (two verses) to be counted an accident.

And there is a fourth occurrence — at the end of the sūra:

لِّيُخْرِجَ ٱلَّذِينَ ءَامَنُوا۟ وَعَمِلُوا۟ ٱلصَّٰلِحَٰتِ مِنَ ٱلظُّلُمَٰتِ إِلَى ٱلنُّورِ (65/11) "In order to bring out those who believe and do righteous deeds from the darknesses into the light."

The same root, the same causative pattern (yukhrija), and again a bringing out of which God is the subject. The act forbidden to the man in the first verse of the sūra (tukhrijū), God performs in the last verses — but in an entirely different direction.

That is, the sūra takes a single root and works it from beginning to end:

Forbidden putting out (from the house) → promised going out (from straitness) → accomplished bringing out (from darkness).

I record this as the backbone of the sūra.

Now something must be said about the widespread use of this verse.

The sentence "whoever guards himself against God, He opens for him a way out" is generally used as a general consolation for every kind of hardship: work, debt, illness, trouble. This use is not wrong — the wording of the sentence is general, it is built with man (whoever) and it is not restricted to a subject. The principle of the tradition of legal theory, al-ʿibratu bi-ʿumūmi'l-lafẓ, operates here.

But something is lost, and what is lost is not trivial.

The place the sentence comes in is this: two people are separating. The woman cannot be put out of the house. The man is obliged to pay. A period is being waited out. The relationship has broken down, the parties cannot bear one another, and the law keeps them both under the same roof.

That is, the sentence is said at the moment when the way out looks most closed. And what it says is tied to what is asked of one at that moment: man yattaqi'llāh — whoever obeys the rule.

This connection is what gets lost. Used as a general consolation the sentence turns into: "fear God and your affairs will improve." In its context it says something else:

At the moment when the way out looks closed, obey the rule. The way out will open from there.

The difference is practical. The first is an advice about feeling, the second an advice about conduct. And the second is far harder — because it asks the person to obey the rule at exactly the moment when he will want to break it.

And one further detail, often passed over: the promise in the verse does not say that the hardship will be lifted. It does not say "He will deliver him from the straitness". It says "He opens for him a way out". The straitness may remain where it is; what is opened is a door. The opening of the door and the emptying of the room are not the same thing.

This is the beginning of verse 3, but it is the continuation of the sentence of verse 2; for that reason I take it up here.

حَيْثُ — an adverb of place: "place, side". Min ḥaythu — "from the place where…".

يَحْتَسِبُ — root ح س ب (ḥ-s-b). And this root was treated in al-Humaza; I do not repeat that analysis. A reminder — and it falls exactly into place here: the root carries both the verb "to count" (ḥasaba) and the verb "to suppose" (ḥasiba), and the lexicographers record that beneath the root lies the idea of "assessing a thing according to a determinate reckoning", from which both numerical reckoning and opinion are derived.

Pattern: إِفْتِعَال (iftiʿāl). Iḥtasaba — to take into account, to reckon up, to expect. What the form adds is that the act is done by the subject for himself: the person is making his own reckoning.

That is, min ḥaythu lā yaḥtasib is rendered "from where he does not expect"; but what it says literally is sharper: "from where he has not put into his reckoning."

And this, combined with the context of the verse, takes on a very definite meaning.

The man in the process of divorce is making a reckoning. He will pay maintenance, provide a dwelling, perhaps a wage for suckling too. These are entries and their sum is known. The rule loads a cost upon him and the resource to meet that cost does not appear in his reckoning.

What the verse says settles exactly here: the resource will not come out of the reckoning.

And the word-weave of the sūra keeps this up across two verses — this is a striking density:

In the sūra the word for reckoning is used for four different tasks: what a human being must count (the period), what a human being cannot count (the source of provision), the authority that settles a human being's account, and the final reckoning.

I record this weave as an observation. The sūra sets up a language of reckoning and within that language draws how far a human being's reckoning extends: you can count the period; you cannot count the source.