The longest part of the book, and all of it is about the chain of narrators. Seven chapters ask the same kind of question over and over: what standard did each of the five great compilers set, and what did he actually do with the chains in front of him?
The famous question: what was each compiler's bar for letting a hadith into his book? Baqa'i works through the five books one at a time, and keeps separating two different things: the general soundness bar, and the specific bar for which narrators are allowed in.
Al-Bukhari has three standards. One general one he stated himself, that everything in the book is sound. Two he alone insisted on: a very narrow bar for narrators, and his rule for the 'an chain - he wants proof the two men actually met at least once, not merely that they lived at the same time.
Muslim's three standards. Al-Hazimi puts his narrator bar at the second tier; Muslim's own account is that he leads with the most exact narrators and brings the rank below them only in supporting reports. On the 'an chain he differs openly: same period plus the chance of meeting is enough, unless it is clear they never met.
A phrase people misuse. It is not enough that each narrator's name appears somewhere in the two Sahihs. Those two had to rely on that exact pairing, this student hearing from this teacher. They use Sufyan ibn Husayn, and they use al-Zuhri, but never Sufyan from al-Zuhri, so that link fails the test. The hadith must also have no hidden flaw.
Abu Dawud's standard, taken from his own letter to the people of Mecca. Three parts: he gathers the soundest legal hadiths he knows, usually; he excludes narrators the critics unanimously abandoned; and he wants well-known, widely used hadiths rather than odd ones, even from a great narrator.
The famous 'silence' problem. Abu Dawud said he flags anything with severe weakness, and that what he says nothing about is salih (usable). Ibn Hajar warns hard against reading that as blanket approval: he does stay silent on weak narrators, so check for a supporting route yourself instead of leaning on silence.
Al-Tirmidhi's standard has two parts. First, the whole book is built on hadiths some scholar actually acted on - he names only two exceptions. Second, he does not argue from a severely weak narrator, and when he does bring such a hadith he says so, which is exactly what makes his book easy to use.
Al-Nasa'i, and a myth to retire. Some early scholars called his Sunan 'sahih', but Baqa'i says that is loose talk; al-Nasa'i himself denied it. His narrator bar is about the same as Abu Dawud's, though he would rather take a clean chain with more links than a shorter chain with a doubtful man in it.
Switches from who gets in to how the material is laid out: the order of hadiths inside a chapter, and how the chapters themselves are sequenced.
Inside one chapter al-Bukhari has no fixed rule; the order serves whatever point the heading is making, and he usually gives only one or two hadiths. The ORDER OF THE BOOKS is deliberate though: each leads into the next. Faith, then knowledge, then action; prayer needs purity, so purification comes first. It opens at how revelation began and closes at Tawhid.
Muslim states his own rule: put the cleanest reports first, then follow them with reports whose narrators are a step below. Baqa'i notes it is a general tendency, not absolute. Also worth knowing: Muslim arranged by chapters but wrote no chapter titles - the ones in printed copies were supplied by al-Nawawi.
A neat contrast. Al-Tirmidhi and al-Nasa'i usually open a chapter with the flawed or odd version and then give the correct one, because they are teaching you to spot defects. Abu Dawud leads with the sound chain instead, his eye being on the meaning. Baqa'i then warns that these are tendencies, not rules, and shows each man doing the opposite.
The puzzle this chapter solves: if the two Sahihs require a connected chain and a saying of the Prophet, why do broken chains and non-Prophetic reports appear in them at all? Answer: never as the proof. They sit in the supporting notes, and in al-Bukhari's chapter headings.
Covers the two kinds of incomplete chain that do turn up in al-Bukhari and Muslim: the mu'allaq, cut off at the compiler's own end, and the mursal, where a Follower quotes the Prophet directly.
Mu'allaq (معلّق, 'suspended') means one or more narrators are missing from the START of the chain, the compiler's own end. Note carefully: this describes the SHAPE of the chain, not its strength. A suspended report can still be sound, and many are given with a full chain elsewhere in the same book.
Muslim has very few: seventeen, and sixteen of those he also gives with a full chain somewhere in the book. Only one is left suspended on its own, in the chapter on tayammum. Al-Nawawi answers al-Maziri here: they are all sound, and for the odd one Muslim leaned on its being well known to the hadith specialists.
Al-Bukhari has many, and Ibn Hajar sorted them out. Two things decide the verdict: whether the hadith also appears with a full chain elsewhere, and the wording. Firm wording ('he said') usually signals confidence, though Ibn Hajar shows many firm-worded ones still fall short of al-Bukhari's own standard, some being weak through a break in the chain.
Mursal (مرسل) here means a Follower reports straight from the Prophet, skipping the Companion he heard it from - so the chain is not connected. These are rare in the two Sahihs and are never the proof. They appear alongside a connected version, to show a disagreement exists and that it does not damage the hadith.
The other three books are more relaxed. Abu Dawud says openly that his Sunan holds non-connected material, one whole part of it gap-chained. Al-Tirmidhi has many, and uses the word mursal loosely for any break in a chain. Al-Nasa'i spells the breaks out, and will sometimes prefer the broken version to the connected one.
Reports that stop short of the Prophet: mawquf, stopping at a Companion, and maqtu', stopping at a Follower. Al-Bukhari has more of these than Muslim, and keeps them out of the main body - they sit in the chapter headings and suspended notes, used to support a view, not to prove the ruling.
Abu Dawud keeps such reports out of his Sunan by policy, not from dislike; he even tells students to record the jurists' opinions alongside it. Al-Tirmidhi names them plainly, and can use one to lift a weak hadith to hasan (fair). Al-Nasa'i records who traced a report to the Prophet and who stopped it at a Companion, then rules which is right.
Why the same hadith shows up more than once, and why that is not padding. Baqa'i's line throughout: the repetition is on the surface only. Something is different each time, in the chain or the wording or the ruling being drawn.
Al-Bukhari and Muslim both repeat on purpose, but in different places: al-Bukhari across different chapters, Muslim mostly within the same chapter.
Ibn Hajar lists about eleven reasons al-Bukhari brings a hadith back: a new teacher in the chain, a text carrying more than one ruling, a fuller or shorter wording, a dispute over whether the chain is connected. Some commentators report him saying he did not want a plain repeat in the book at all.
Muslim states his own rule in his introduction: no repetition, EXCEPT where a second version adds meaning, or a second chain is needed because of a defect. He says that where he cannot cleanly separate the extra bit out, bringing the whole hadith back is the easier course.
The other three, and here the three genuinely differ from each other rather than just differing in degree.
Abu Dawud says he repeats only when a version carries extra wording. In practice, if the second version has a different ruling he gives it in full; if the difference is just one word, he quotes only that word.
Al-Tirmidhi repeats least of all - so little that you have to hunt to notice it, usually no more than three places. And unlike al-Bukhari he does not always add something new: a few hadiths recur with the same text and the same chain.
Al-Nasa'i repeats a great deal, and deliberately. Al-Sakhawi says he rivals al-Bukhari at drawing fine rulings out of a text, and repeats the hadith precisely so each new chapter heading can make its own point - sometimes changing nothing but a detail in his teacher's name.
Space-saving techniques. The default would be to write every chain out separately, since a change of one narrator technically makes a different hadith - which would turn one hadith into twenty. This chapter is how the compilers avoided that without hiding anything.
Two shortcuts. Join teachers with 'and' when several gave you the same hadith. And use the tahwil mark, the letter ha (ح), at the point where two chains meet, so you write the shared part once. Muslim uses it most; Baqa'i works out where exactly Muslim places the mark relative to the shared narrator.
The other shortcut: give one route, or part of a text, and point to the rest rather than writing it out - with phrases like 'so-and-so also narrated it', 'it is narrated by more than one route', 'the like of it', or 'with this chain'.
Where the compilers step out from behind the material and talk about the narrators themselves - rating them, and identifying who they were.
Praising and criticising narrators. This is a habit of the three Sunan authors, not of the two Sahihs - in al-Bukhari and Muslim it is rare, apart from Muslim's introduction. Al-Tirmidhi does it most, then al-Nasa'i, then Abu Dawud.
Identifying narrators rather than rating them: that this man was a Companion or a Follower, that he was from Kufa or Basra, when he was born or died, whether he actually met the man he is quoting. Small notes, but they are what settles whether a chain is connected. Again al-Tirmidhi does it most.
A chapter of nine short sections, each one a kind of remark the compilers drop next to a hadith. Baqa'i opens by saying who does what: al-Tirmidhi comments most and openly; al-Bukhari usually only hints, and most of his points are about the meaning of the text, while most of Muslim's are about chains.
Flagging an 'illa (علة), a hidden flaw in a chain that looks fine on the surface. The compilers say things like 'there is instability in its chain', or simply 'this is a mistake, and the correct version is such-and-such'. Common in al-Tirmidhi and al-Nasa'i.
Settling a dispute about the chain itself. Two questions keep recurring: is this traced to the Prophet or does it stop at a Companion, and is the chain connected or does it skip someone? The compiler either picks the stronger version or simply records that it is narrated both ways.
Noting that people disagreed over a narrator's name or family line - the same man called by different names in different chains. Getting this wrong can make one man look like two, or two men look like one.
Introducing a narrator, especially where his name was disputed: saying who he really was, whose freed slave he was, which grandfather a name refers to. Practical notes that let the reader pin the man down.
Siyagh al-ada' (صيغ الأداء), the wording that shows HOW a narrator received the hadith. Where two teachers gave the same hadith with different wording, the compiler records which said 'he narrated to us' and which said 'he informed us'. Muslim does this a lot. For why the choice of phrase matters, see Book One, Chapter 4.
A rawin muhmal (راوٍ مهمل) is a narrator named too vaguely to identify - just 'Muhammad', or 'Sa'id'. The compiler adds the missing piece, usually flagged so you can see the addition is his and not part of the chain he received: 'Muhammad, that is, Ibn Salam'.
When two teachers give the same hadith with slightly different words, whose exact wording is being quoted? The compiler says so: 'and the wording is his'. Small, but it matters, because rulings get drawn from precise wording.
Noting that a hadith is narrated only by the scholars of one town - the people of Medina, of Basra, of Yamama, of Egypt. Abu Dawud does this often. It is a piece of evidence about how a hadith travelled, not by itself a verdict on it.
Stating the grade outright: sound, fair, or weak. Baqa'i says this is frequent in al-Tirmidhi. Al-Nasa'i also weighs one report against others, saying things like 'there is nothing better in this chapter than this hadith'.