How do scholars determine whether a hadith is authentic?

The definition, in one sentence

Ibn al-Salah (d. 643 AH) states it as a single sentence:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

He then says what those descriptions exclude, and that he will treat each kind later:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

Two things follow. A defect only counts if it damages the report's authenticity. Not every oddity is a defect. And the definition names its conditions without explaining them; the explanations below come from Ibn al-Salah's own later sections.

The same definition, set out as conditions

Ibn Hajar (d. 852 AH) sets the definition out as a list:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

Ahad here is a technical term, and it is easy to mistranslate. It does not mean “transmitted by one person”. Ibn Hajar defines it in the same book as everything that falls short of mass transmission — «وفي الاصطلاح: ما لم يَجْمَعْ شروطَ التواتُر», “in technical usage: what does not gather the conditions of tawatur” — and he says plainly that the “one person” sense is the linguistic one: «وخبرُ الواحدِ في اللُّغة: ما يرويه شخصٌ واحدٌ», “a single person's report, in the language: what one person relates”. So the category covers reports with several transmitters as well as reports with one.

The five conditions, and the different question each one asks:

1. An upright transmitter ('adl): is this person's integrity established?
2. Complete accuracy (tamm al-dabt): did he transmit precisely?
3. A connected chain (muttasil al-sanad): is every link present, with no gap?
4. Not anomalous (shadhdh): see below.
5. No hidden defect (mu'allal): see below.

The enumeration into five is Ibn Hajar's. Ibn al-Salah gives the definition; he does not number it.

What “anomalous” means

Ibn al-Salah opens his section on the anomalous report with a definition he reports from al-Shafi'i (d. 204 AH):

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

He reports two other accounts as well, and then gives his own view. Al-Shafi'i's cases he accepts without hesitation — «قُلْتُ: أَمَّا مَا حَكَمَ الشَّافِعِيُّ عَلَيْهِ بِالشُّذُوذِ فَلَا إِشْكَالَ فِي أَنَّهُ شَاذٌّ غَيْرُ مَقْبُولٍ», “I say: as for what al-Shafi'i judged anomalous, there is no difficulty that it is anomalous and not accepted.” The other accounts, he says, «فَيُشْكِلُ بِمَا يَنْفَرِدُ بِهِ الْعَدْلُ الْحَافِظُ الضَّابِطُ» — “run into difficulty with what an upright, retentive and precise transmitter reports alone” — and he points to reports in the two Sahih collections that have exactly that profile: «فَكُلُّ هَذِهِ مُخَرَّجَةٌ فِي الصَّحِيحَيْنِ مَعَ أَنَّهُ لَيْسَ لَهَا إِلَّا إِسْنَادٌ وَاحِدٌ تَفَرَّدَ بِهِ ثِقَةٌ», “all of these are included in the two Sahihs, although each has only one chain, transmitted alone by a reliable transmitter.” So, he says, the matter calls for distinctions, and he sets them out:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

And he concludes:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

So, for Ibn al-Salah, being alone is not by itself the deciding question. On al-Shafi'i's definition, the anomaly is the contradiction. On Ibn al-Salah's own account, a fully reliable transmitter's solitary report is accepted, while a solitary report is rejected when it contradicts better-preserved transmission, or when its transmitter is not reliable enough to carry it alone.

What “hidden defect” means

Ibn al-Salah describes such defects as «أَسْبَابٍ خَفِيَّةٍ غَامِضَةٍ قَادِحَةٍ فِيهِ» — hidden, obscure causes, damaging to the report — and defines the defective report this way:

Translation: ConveyVerse, from the Arabic of the 'Itr edition.

He says that recognising such a defect is assisted «بِتَفَرُّدِ الرَّاوِي وَبِمُخَالَفَةِ غَيْرِهِ لَهُ، مَعَ قَرَائِنَ تَنْضَمُّ إِلَى ذَلِكَ» — “by a transmitter's standing alone, by others contradicting him, together with indications that attach to that.” And he describes how the examination can end: «بِحَيْثُ يَغْلِبُ عَلَى ظَنِّهِ ذَلِكَ، فَيَحْكُمُ بِهِ، أَوْ يَتَرَدَّدُ فَيَتَوَقَّفُ فِيهِ» — “so that this prevails in his judgment and he rules accordingly, or he hesitates and suspends judgment on it.” He adds at once: «وَكُلُّ ذَلِكَ مَانِعٌ مِنَ الْحُكْمِ بِصِحَّةِ مَا وُجِدَ ذَلِكَ فِيهِ» — “and all of that bars ruling authentic any report in which it is found.”

Translations: ConveyVerse, from the Arabic of the 'Itr edition.

So this condition is not a lookup. A careful examiner may end by suspending judgment, and by Ibn al-Salah's own words, a suspended judgment is still enough to withhold the verdict sahih.

Where a defect can lie

It is tempting to think the chain conditions are about the chain and the last two are about the wording. Ibn al-Salah says otherwise:

Translations: ConveyVerse, from the Arabic of the 'Itr edition. Ibn al-Salah explains the mursal and the mawquf in their own chapters of the same book.

So the last two conditions are not a separate examination of the wording; they are an examination of the report, chain included.

Key takeaways

• Sahih is a technical verdict reached by applying fixed conditions, not a description of how convincing a report sounds.
• The classical definition is Ibn al-Salah's, in one sentence. The familiar list of five conditions is Ibn Hajar's way of setting that sentence out.
• A defect counts only if it damages authenticity, and it may lie in the chain as well as in the wording, more often in the chain, by Ibn al-Salah's own statement.
• On al-Shafi'i's definition, the anomaly is contradiction, not being alone. Ibn al-Salah's own conclusion adds a second kind: a solitary report whose transmitter is not reliable enough to carry it alone.
• An examiner of a hidden defect may suspend judgment, and Ibn al-Salah counts that as enough to withhold the verdict sahih.

Sources

1. Work: Ibn al-Salah al-Shahrazuri, Ma'rifat Anwa' 'Ulum al-Hadith. Reference: ed. Nur al-Din 'Itr, Dar al-Fikr, Damascus / Dar al-Fikr al-Mu'asir, Beirut, 1406/1986, al-Naw' al-Awwal, pp. 11–12 (the definition and what it excludes). Wording used: as printed above. Translation: ConveyVerse, from the Arabic of this edition. Grade / grader: not applicable: a definition, not a report. Context: the definition opens al-Naw' al-Awwal; Ibn al-Salah says the excluded kinds are treated later. Verify: p. 11 · p. 12, al-Maktaba al-Shamila's reproduction of this edition, with its page markers shown. Does not settle: who applied the definition to any report.
2. Work: Ibn Hajar al-'Asqalani, Nuzhat al-Nazar fi Tawdih Nukhbat al-Fikar. Reference: ed. Nur al-Din 'Itr, Matba'at al-Sabah, Damascus, 3rd edn 1421/2000, p. 58 (the definition), pp. 50–51 (al-ahad). Wording used: as printed above. Translation: ConveyVerse, from the Arabic of this edition. Grade / grader: not applicable. Context: pp. 50–51 define al-ahad before p. 58 uses it. Verify: p. 58 · p. 50 · p. 51, al-Maktaba al-Shamila's reproduction of this edition. Does not settle: the other grades.
3. Work: Ibn al-Salah, same edition, al-Naw' al-Thalith 'Ashar, pp. 76–79 (the anomalous) and al-Naw' al-Thamin 'Ashar, pp. 90–91 (the defective, and where a defect lies). Wording used: as printed above. Translation: ConveyVerse, from the Arabic of this edition. Grade / grader: not applicable. Context: the definition on p. 76 is al-Shafi'i's, as Ibn al-Salah reports it; pp. 77–79 give Ibn al-Salah's own comment, evidence, rule and conclusion. Verify: p. 76 · p. 77 · p. 78 · p. 79 · p. 90 · p. 91. Does not settle: any particular report's status.