The Short Story Is Not a Small Novel

Two people are waiting for a train at a station in the Ebro valley. They drink, they look at the hills, they argue in a careful, circling way about an operation that is never named. The story is a few pages long, it consists almost entirely of dialogue, and nothing is explained. Hemingway's "Hills Like White Elephants" is, by any structural account, barely a story at all — no plot to speak of, no resolution, no access to anybody's interior.

It is also complete, and the completeness is the thing worth studying. A novel would have had to tell you what they decided. The short story does not, because the form's central advantage is the licence to leave out.

The unity of effect

Poe made the founding claim about this in the 1840s, reviewing Hawthorne: a short tale should be readable at a single sitting, and every element in it should work towards one preconceived effect. He meant it more prescriptively than most writers would now accept, but the underlying observation has held up. A reader who does not put the story down is a reader whose attention you can shape continuously, and that continuity is the resource a short story has and a novel does not.

What falls out of it is the economy people usually attribute to Chekhov. The remark about the gun on the wall — that if you show it in the first act it must be fired, and if it will not be fired it should not be shown — comes down through his letters and is quoted more often than it is examined. It is not a rule about weapons. It is a claim that in a form this compressed, a reader treats everything you mention as significant, because you have given them nothing else to hold. Mention the weather and it becomes thematic whether you intended it or not.

That is a constraint and also a very large amount of free power. A single object, placed once, does work that a novel would spend a chapter on.

Start late, leave early

The most common structural problem in a draft short story is that it begins about three pages before it needs to.

Openings tend to accumulate throat-clearing: who this person is, what their situation is, how they got here, what the room looks like. Almost all of it can go, because the form permits you to begin at the point where something is already happening and let the reader assemble the situation from context. Readers are much faster at this than writers give them credit for.

The equivalent at the other end is stopping too late. A story that has delivered its turn and then spends a page explaining the turn has undone itself. The strongest last lines in the form tend to arrive slightly before the reader is ready, which is why so many good short stories produce a sensation of the floor ending a step early.

The practical revision that follows is almost mechanical: try deleting your first paragraph and your last, and see whether anything was lost. A surprising proportion of the time, nothing was.

Endings: epiphany, and the reaction against it

The dominant model for a twentieth-century literary short story is the epiphany — a moment of recognition, usually internal, usually arriving near the end and changing how everything before it reads. Joyce is the reference point. "The Dead" spends most of its length on a Dublin party rendered in almost tedious social detail, and then Gabriel Conroy learns about a boy who died for his wife when she was seventeen, and the entire evening reorganises itself behind him while the snow falls.

That model became so dominant that the interesting work for decades afterwards was the reaction against it. Raymond Carver's "Cathedral" ends with a man drawing a cathedral with his eyes closed, guided by a blind visitor's hand, saying only that it was really something. There is a change, and it is emphatically not articulated. Alice Munro does something else again, jumping across decades so that the significance of a scene is established by what happens twenty years later, in half a page.

The point for a writer is that the epiphany is a convention, not a requirement. Plenty of excellent short stories end on an image, a gesture, a line of dialogue, or a flat refusal of the revelation the reader has been arranged to expect. What is non-negotiable is that something has changed — in the situation, in the character, or in the reader's understanding of what they have been watching. A story where the final state is identical to the initial state in all three respects is a sketch.

Character without biography

Because there is no room to develop anyone in the novelistic sense, short fiction establishes character by inference, and the inference has to be efficient.

The reliable tools are small. A decision made under mild pressure. A physical habit observed once. The specific thing a person notices in a room, which reveals more than a paragraph about their history. And the gap between what someone says and what they are obviously doing, which is why dialogue carries so much weight in the form.

Premchand's "Idgah" turns entirely on what a small boy chooses to buy at a fair with the little money he has, when every other child is buying sweets and toys. There is no backstory. The choice is the character, and it is devastating precisely because nothing is explained about why he made it.

What the story refuses to say

The deliberate withholding in "Hills Like White Elephants" is the most famous instance of a general technique. The subject of a short story is frequently the thing the story declines to name.

This is different from being coy. The reader must be able to work it out — Hemingway's couple are not ambiguous to an attentive reader, they are merely unexplained. The effect is that the reader does the final assembly themselves, and material a reader has assembled sits differently than material they were handed.

The failure mode is obscurity for its own sake, where the withheld information is not recoverable and the reader is left irritated rather than implicated. The distinction is whether the story has given them everything they need except the last step.

Length, and what it actually constrains

Craft guidance generally puts the short story somewhere between roughly three thousand and seven thousand words, with flash fiction topping out around fifteen hundred, and the longer forms shading into the novelette. Journals and competitions set their own limits, which is the number that will actually matter to you.

I would resist the temptation to convert those figures into scene arithmetic. The advice that a three-thousand-word story equals three scenes of a thousand words each is tidy and produces mechanical fiction. What length really constrains is the number of turns the story can survive. A flash piece can carry one. A conventional short story can carry two or three. Beyond that the reader loses the thread of what the story is about, because there has not been enough time to establish a baseline that the turns can deviate from.

Different lengths also favour different shapes. Very short fiction tends towards the single image or the single reversal, which is why Saki's twist endings work at that scale and would be intolerable at fifty pages. Longer short fiction can afford a subplot, a second point of view, or a time jump.

On practising

The form is a good place to learn craft precisely because it is unforgiving and because the feedback loop is short. You can write a story in a week and find out whether it works. You cannot do that with a novel.

Two exercises I would defend. Write a story that takes place in one continuous scene, in one location, with no flashback — it forces everything into dialogue, gesture and object, which is where compression is actually learned. And take a story of yours that is not working and cut it by a third without removing any events, which teaches you how much of your prose was scaffolding.

The recurring discovery, at least for me, is that the parts I fought hardest to keep were usually explanation, and the story was better once the reader was trusted with the work I had been doing for them.