Online Serials Copy Dickens. They Should Copy a 900-Year-Old Con Man.

An old man in worn clothes walks into the mosque of the Banu Haram in Basra. He says he comes from Saruj and that his name is Abu Zayd. He describes his city's destruction, his daughter's captivity and his own exile, and he does it so well that the room is moved. That is how Thomas Chenery's 1867 introduction to the Assemblies retells the story told by al-Hariri's son: the father went home and wrote the piece that now stands forty-eighth in the book [1]. The last part to take its place was one of the first to be written. A form that can do that is telling you how it works.

My thesis: the maqama, as al-Hariri built it in fifty episodes around one narrator and one trickster [2], answers the central problem of serial fiction better than the European serial novel. The problem is how to make each part complete and still make the parts add up. The novel's answer is the cliffhanger, which makes each part a debt the next part has to pay. The maqama's answer is a frame that recurs: the same witness, the same rogue and the same unmasking each time, so that each part pays for itself and the series builds interest like savings. Online serials, where most readers arrive in the middle, should borrow the frame and drop the debt. I hold this at 0.5 and will end by saying what would move it.

The problem every serial has to solve

A serial installment has two jobs that pull against each other. It has to satisfy the reader who is holding it, and it has to make that reader come back. Lean too far toward the first and you get an anthology with nothing to return for. Lean too far toward the second and you get a chapter that ends mid-gasp and a reader who feels cheated if the next one never comes.

There are two old answers to this. I will call them the debt model and the frame model.

In the debt model, each part is a promissory note. The "make 'em laugh, make 'em cry, make 'em wait" line belongs here. It is usually credited to Wilkie Collins, though the attribution also goes to Charles Reade, so I treat it as a slogan of the trade and not a quotation [3]. It works. When The Woman in White ran weekly in All the Year Round from 26 November 1859, it lifted the magazine's circulation to record levels, and crowds gathered at the office on publication days [3].

In the frame model, each part ends and nothing is owed. What carries over is a set of fixed figures and one fixed event the reader comes back to watch happen again. Al-Hariri did not invent the frame. The genre began with Badi' al-Zaman al-Hamadhani (969 to 1008) [4]. But al-Hariri's version made it into the most copied book in its tradition. Wikipedia's summary cites reports of seven hundred copies authorized by al-Hariri in his lifetime, and more than a hundred surviving manuscripts [2].

What al-Hariri built

Every assembly opens with the same move, a chain of transmission one link long. In Steingass's 1898 English, the twenty-seventh begins: "Al Harith, son of Hammam, related: In the prime of my life that has fleeted..." [5]. The Arabic of the first assembly opens with the same formula, haddatha al-Harith ibn Hammam qala ("al-Harith ibn Hammam told us, saying") [6]. Before anything happens, the reader knows who is speaking, in what posture and to whom.

Then a stranger appears: a preacher, a blind beggar, a man in a law court. He talks. The talk is the event, in rhymed prose packed with verse, riddles and palindromes. Then al-Harith sees through the disguise, and it is Abu Zayd again. In the first assembly the narrator follows an ascetic preacher home and finds him eating white bread and a roast kid [7]. One summary puts the design in a sentence: each assembly takes "a separate topic, the whole being unified by the persons of the narrator and the traveler" [7].

Look at what that does for a serial. Each recognition scene shuts its own episode completely. The con is exposed, the money is spent and the narrator walks off. No question is left hanging from the last line. But the reader already knows the stranger is Abu Zayd, from the third episode on if not sooner. So the suspense moves from whether to how: how this disguise will slip, in which city and in which register. That kind of suspense does not run out, because the answer is new each time while the question stays the same. That is the core of the argument. A cliffhanger is spent the moment it is resolved. A recurring recognition renews itself.

The frame also lets a series grow from the middle outward. Chenery tells a second story from the tradition, and he presents it as a story, not a record. Al-Hariri brought forty assemblies to Baghdad. Literary men there doubted he had written them and said they came from the papers of a dead rhetorician from the Maghrib. The vizier asked him to compose something on the spot, and "God inspired him with nothing." He went back to Basra, wrote ten more and sent them to Baghdad [1]. Whether or not the anecdote is true, it describes a property the form really has. You can add ten episodes to a frame without rewriting the forty already there. Try adding ten chapters to the middle of The Woman in White.

The ending that changes the frame without cancelling it

The usual objection to anthology seriality is that it cannot end. It just stops. The maqama answers this in its fiftieth episode. Set in Basra, it shows Abu Zayd repent. He goes home to Saruj, leaves the house only to pray, and sends al-Harith off with one instruction: keep death before your eyes [8].

The ending accumulates without retracting anything. The forty-nine earlier cons are not revealed as a dream, a lie or a test. They stay exactly as they were. What changes is their weight, because now they are the record of a life that stopped. I can't stand twist endings that cancel a story, and the maqama shows the alternative: the last part turns the frame without breaking it. A reviewer in Public Books sees how unstable even that turn is. The picaresque rogue "repents, or does he?" [9]. Fifty episodes of watching this man lie will make the reader doubt his sincerity, and that doubt is the cumulative payoff. A single installment could not produce it.

Variation inside the frame

If the frame is fixed, what varies? In al-Hariri, the language varies. That is why Michael Cooperson's English translation, Impostures, is the best modern argument for my case, though it was not written as one. Cooperson rendered the fifty episodes in fifty registers of English, "from the distinctive literary styles of authors such as Geoffrey Chaucer, Mark Twain, and Virginia Woolf, to global varieties of English including Cockney rhyming slang, Nigerian English, and Singaporean English" [10]. Public Books lists Dickensian, Woolfian, Spanglish, Singlish, psychotherapy jargon and 1940s Harlem jive among them [9].

A translation in fifty styles only holds together because the frame does the holding. Al-Harith still relates and Abu Zayd is still unmasked, so the reader never loses the thread even when the language changes completely. Cooperson's translation shows how much variation the frame can carry. It also admits a loss: the original's single high Arabic register is gone, swapped for fifty borrowed ones. I love a translation that admits what it lost.

This is the design I want for my own twelve-part serial, where each part uses a different constraint. In the cities catalogue, each entry withheld one letter and the arc lived only in the order of the entries. A reader who arrived at entry 30 without the first 29 got a city and nothing else. That was a debt structure dressed up as a catalogue. A twelve-part serial with a recurring witness and a recurring trickster would let part 7, written as a lipogram, stand on its own, while part 12 still turns all eleven before it.

The strongest objection: the cliffhanger sold, and the novel won

The objection runs like this. The European serial novel is not a straw man. It was the most successful form of popular fiction in the nineteenth century, and the numbers are on its side. The Woman in White drew crowds to a magazine office [3]. Pickwick began with a print run of 1,000 for its first monthly part and ended at 40,000 [11]. Worse for me, Pickwick started as close to an anthology frame as Victorian publishing came. Robert Seymour proposed it as a run of sporting engravings with text attached, "a still somewhat loose collection of the adventures of Samuel Pickwick and his friends." Dickens dropped the sporting club and pulled the sketches "into a more cohesive novel" [11]. On this reading, the market tried the frame and chose the plot.

I take this seriously, and I think it gets the evidence half right. Two points answer it.

First, look at what the same source says changed Pickwick's fortunes. It was not the cliffhanger. It was the arrival of Sam Weller in chapter 10, after which sales rose dramatically [11]. Weller is a second fixed figure, the street-wise servant beside the innocent master. The sales jump came with the frame getting its second pole. That fits my thesis at least as well as it fits the objection, though one book cannot separate the two explanations, and I will not claim that it does.

Second, the conditions that made the debt model pay are gone. A Victorian weekly had a captive, synchronized audience: one magazine, one day and one queue at the office. The cliffhanger works on such an audience because everyone is at the same place in the story. Arthur Conan Doyle saw the weak point in 1891, from the publisher's side. He wrote that "the ordinary serial might be an impediment rather than a help to a magazine, since, sooner or later, one missed one number and afterwards it had lost all interest" [12]. His fix: "Clearly the ideal compromise was a character which carried through, and yet instalments which were each complete in themselves" [12].

Online, every reader is the one who missed a number. Readers arrive from a link, a search or a share, and land on part 9 without having read part 1. The forums of serial platforms are full of authors puzzling over the drop between their first chapter and their second. Threads like Royal Road's "Large viewer drop from chapter 1?" are a genre of their own [13]. I cite them as forum anecdote, not data, and they show nothing about the rate. My point is structural. When entry is random, a debt structure charges every new reader for parts they never saw. The frame charges nothing.

The narrower claim the objection forces

Here is where Doyle forces me to give ground. His "ideal compromise" is the frame model, reinvented: a narrator, Watson, and a recurring figure of brilliance, Holmes, with a complete case in every part. Doyle believed he was the first to see it [12]. Al-Hariri died in 1122 and the first Holmes story ran in The Strand in 1891, so Doyle was about 769 years late. The joke is on priority, not on Doyle. He arrived at the right answer from the magazine's side, independently, and that independent arrival is evidence the solution is a real one.

So the honest version of my thesis is narrower than the working title. Against the serial novel, I think the maqama wins for online publication. Against the detective series, the maqama is the ancestor, not the rival. Then what does the maqama offer that Holmes does not? I see three things, and I hold them as opinion.

• The recognition is known in advance. Holmes's reader does not know the culprit. Al-Hariri's reader always knows the stranger is Abu Zayd. That moves all the suspense onto language and manner, and that suits a writer working in constraints, where the pleasure is in how the sentence gets made.
• The style can change with every part. A Holmes story in Spanglish would break the series. A maqama in Spanglish is how Cooperson's translation works [10].
• The ending is a turn, not a solution. Reichenbach killed Holmes and readers made Doyle bring him back. Abu Zayd's repentance cannot be undone, because it does not contradict anything earlier. It only reframes it.

The form decides what a serial can notice. A debt structure notices what happens next. A frame notices what is the same and what has changed since last time, and that is a better question to put to someone arriving cold.

What follows if I am right, and what would change my mind

If I am right, an online serial should spend its formal budget on three fixed things: an opening formula, a pair of recurring figures (a witness and a performer), and a recognition that shuts each part. It should then let everything else vary, down to the alphabet. Its last part should change the frame without contradicting the parts before it. And any single part, handed to a stranger, should read as complete.

My twelve-part serial is the test, so the falsifier has to be one I can fail. I will draft two parts both ways: once in the maqama frame, and once as chapters of a continuous plot with a cliffhanger, each under the same constraint. Readers who have not seen the rest of the serial will get one version at random, blind, and answer two questions: "Did this feel complete?" and "Would you read another part?" If the cliffhanger versions win the second question by a clear margin and the frame versions win only the first, then the frame buys completeness at the cost of return, and my thesis fails where it counts. A test this small will only detect a large difference, and I will report the numbers whichever way they fall.

The archive moves me a little, mostly because of Doyle's independent arrival, so I go from 0.5 to 0.55. I am not going higher. The case for the frame against the serial novel rests on a structural argument about random entry plus one Victorian editor's judgment, and so far it has no reader data. What would move me back down is a serial platform's published retention figures showing that self-contained-episode series lose readers between parts as fast as cliffhanger series do. If someone has those figures, I would rather read them than keep rhyming.