The Schillerkopf - courtesy of outdooractive.com

Here is a fragment from a ten-tercet ascent poem set in the Austrian Alps, drafted in March and cut from What Storms Disclose during the spring selection. Two tercets at the structural summit, the moment the poem arrives where it has been climbing to. The cadence I’ve reached for is generic mountain-spiritual — the register that runs from the Cold Mountain translations through Snyder’s Riprap to the Bergpoesie I’d half-absorbed reading Heidegger on Hölderlin. The register is older than any one exemplar; it is the inherited shorthand for mountain poem. I haven’t earned it.

The fragment

Narrow summit and cross

leaned against the wind,

clean air, crisp and clear,

silence working hard.

Sweat on our back cools.

We come fully alive.

Two tercets from the middle of a ten-tercet poem — lines 13 to 18 of 30. The speaker reaches a limestone ridge in the Austrian Alps; these are the lines where the summit is named and inhabited. The village below, the descent, and the return stay in the drawer.

The autopsy

— #8 The borrowed cadence

clean air, crisp and clear, // silence working hard. / Sweat on our back cools. / We come fully alive.

I’ve reached for the register before earning it. The articles drop out — clean air, not the clean air; Sweat on our back cools, not the sweat. The adjectives line up in plain Anglo-Saxon monosyllables — clean, crisp, clear. The abstraction is asked to do physical work — silence working hard. And the closing line of the second tercet announces the summit epiphany in three monosyllables and an adverb — We come fully alive — the mountain-poem arrival at its most generic.

What’s borrowed is the cadence of mountain-spiritual poetry as a tradition. Snyder’s Riprap voice. Han Shan in Snyder’s translation. R.S. Thomas on a Welsh hillside. The Bergpoesie I’d half-absorbed reading Heidegger on Hölderlin. No single poet is being copied; what’s being copied is the register — the inherited “this is how altitude sounds in English-language poetry”. The register is doing the work the body and the rock should have done.

The diagnostic test for borrowed cadence: ask what specific necessity made the original poet need the register. Snyder’s compression was earned on logging crews and trail repair — his sentences were short because his life with words was. Han Shan’s gnomic declarative was earned by twenty years on a mountain. Thomas’s plain register was earned by parish ministry on hard ground. This draft was a hiking week from a summer gasthaus. The specific necessity that earned the register for the original poets isn’t in this poem. So the register sits on top of the lines as decoration — and decoration always reads as decoration.

The repair. Lose the register entirely, then find the voice this particular climb, this particular morning, would have spoken in. Not the harder work of a craft repair. The harder work of asking what these two bodies, on that ridge, at that hour, would have said in a voice that didn’t already know what mountain poems sound like.

The general lesson

Two notes I’m taking from this autopsy, both about how cadence is inherited.

First: a borrowed cadence often arrives alongside the subject. Climb a mountain in a poem and the voice of every other mountain poem comes with it. The register is so available, so well-formed by a century of English-language practice, that it speaks itself before the writer has noticed. The diagnostic check is to ask whether the cadence of one’s draft would survive being relocated. If the same lines could sit in a Snyder volume, in a Cold Mountain version, in a Bergpoesie anthology, the cadence isn’t yet one’s own.

Second: the borrowed register often hides where the encounter should have been. We come fully alive is the line where the cadence speaks loudest, and it is also the line where the poem skips the encounter. What did being on that ridge actually do? The register makes the question feel answered. It isn’t. The line below the borrowed phrase is usually the one the draft was reaching past — the same diagnostic shape as last week’s Prayers, with the register doing the work the feeling-noun did there.

— Adam

Subscribe now

The poem opens by reaching for the feeling — for grief, for tears, for the weight of it — when the thing itself, the cancer, is sitting one line below, ready to do all of that work on its own if I had only let it.

The fragment

They vanish further than any night could fall,

to a borderline of tears. No one could stay

the cancer.

The opening three lines of a fourteen-line poem in 5–2–5–2. The poem is about a death from cancer; the rest of it stays closer to the dying — to particular images of the late summer, the doctors’ failed turns, the resignation of the survivors. Those lines stay in the drawer for further revision. Only the opening is on the slab — the two lines that reach, and the line they collapse onto.

The autopsy

— #1 The abstraction reaching (sub-case: the performed feeling)

They vanish further than any night could fall — the they without antecedent, the vanishing measured against a hypothetical night, the verb already at the metaphorical altitude where the sentence has nowhere left to go. To a borderline of tears — the feeling-noun standing in for the dying, the grieving, the room, the months. I’ve used a noun for a feeling where a body and a disease were waiting one line later.

Photo by Hush Naidoo Jade Photography on Unsplash

What’s concealed is the person. The poem has them: No one could stay / the cancer is the line where they appear, three lines in. That should have been the opening. The two lines above it are me flinching from the appearance — reaching past the cancer with a hypothetical night and a borderline made of tears, because the cancer was the line I didn’t want to start on.

The fragment also shows the cost of where the failure sits. Openings are load-bearing in a way middle lines aren’t; the first frame the reader receives is the frame the rest of the poem is read against. By starting on abstraction, I’ve licensed the whole draft to keep reaching. Tears recurs through the next stanza; prayers arrives twice more after that. The opening has paid for them all in advance.

The repair. Cut the first two lines. Begin on No one could stay / the cancer. The 5–2–5–2 becomes 3–2–5–2 — which may or may not be the shape the poem wants, but it is the shape that doesn’t lie about where the poem starts.

The general lesson

Two notes I’m taking from this autopsy, both about openings.

First: where an opening reaches for abstraction, the thing it should have entered through is often waiting one or two lines below. In Prayers it is line three: the cancer. I’ve gone back through a handful of my own drafts since seeing this, and the same shape keeps recurring — the first lines perform the mood, then a later line, often unforced, names what the mood was about. The repair is almost always the same. Cut the performance. Begin where the seeing begins.

Second: a single feeling-noun — tears, fear, grief, longing, loss — can carry an opening if the surrounding image earns it. Three in the first stanza is the diagnostic threshold. The poem has stopped showing and started naming. In Prayers, tears recurs three times across the first seven lines, and prayers arrives twice more after that. The recurrence isn’t motif. It is the same reach repeating, in lieu of the thing the poem is trying not to look at.

— Adam

Subscribe now

Photo by Veronica White on Unsplash

Two lines in, there is a noun I should never have written down — the word that names the thing the poem was about to show me — and because it is there, the other lines have stayed in the drawer.

The fragment

her flock long gone, funnelled through a gap / in the stone wall, stooped in a graze of prayer.

The closing two lines of the first stanza of a four-stanza poem about a knackerman — the man who collects dead and dying livestock from surrounding farms. The opening stanza arrives at a downed ewe in a wind-scuffed field; these two lines describe the rest of her flock, gone through a gap in the wall, grazing somewhere just out of sight. The poem itself stays in the drawer for further revision; the cut is on these two lines only.

The autopsy

— #7 The over-naming (sub-case: the pre-cooked symbol)

I’ve named the grazing as prayer in the same breath as showing it. The bowed-head posture of a flock against a stone wall already carries its own reverence — a flock with heads down, en masse, is already a religious figure for anyone who has stood near sheep at the edge of a field. A graze of prayer closes the door before the reader has walked through it. The symbol arrives pre-decided, and the image becomes the carrier of a meaning I’ve already settled.

The diagnostic is sharpened by what the same fragment does well. Funnelled through a gap / in the stone wall keeps its own shape — the sheep have a literal body and a literal route, and the line earns the reader’s attention by withholding any commentary. Then the closing four words turn that resulting grazing posture into a labelled image, and the labour the line has just done is collapsed into a religious noun.

The Verborgenheit — the concealment that lets a posture mean something the reader can find on her own — is exactly what’s been pre-empted. Once the meaning is named, there’s nothing left for the reader to bring to the image. The image stops disclosing.

The repair. Cut of prayer. The line ends stooped in a graze, and the bowed-head posture, against the wall, carries the reverence the noun was trying to assert. If the resonance survives the cut, the noun was redundant. If the resonance disappears, the image wasn’t earning it and the repair is somewhere else entirely — but in this case it survives, sharper for being unnamed.

Why fragments

Two reasons. The pragmatic one: Substack posts count as publication for almost every serious poetry journal, and at fifty-two whole drafts a year I’d empty the cellar within two or three. Some of the drafts I’d autopsy on this Substack are poems I might revive and rework — and posting the whole draft now closes that door. A fragment is a specimen of the failure, not the publication of the poem. The four stanzas around the two lines above stay in the drawer.

The diagnostic one: most of the failure modes operate at the unit level, not the whole-poem level. A single line of over-naming, an opening that performs feeling, a stanza closer that reaches for the universal — these are diagnosable in two to eight lines. The whole-poem autopsies of the launch wave were the right move for establishing the taxonomy through confessional case-work; once the taxonomy is in place, the smaller specimen does the same diagnostic work, sometimes more cleanly because the failure isn’t obscured by the surviving strengths around it.

Whole-poem autopsies will still appear when a draft genuinely justifies the whole-draft treatment — interrelated failure modes, or a structural shape that’s itself part of the diagnosis. But the steady-state register from now is fragment: one mode, the smallest legible specimen, the repair, and the general lesson.

The general lesson

Two notes I’m taking from this autopsy, both about the moment when a noun steps in front of the image it’s labelling.

First: the religious-noun test. When a religious word arrives in a poem — prayer, grace, blessing, communion, vigil, psalm — pause and ask whether it is being earned by what the surrounding image actually shows, or asserted in lieu of the image showing it. In nine cases out of ten the religious noun is the giveaway: the poet has seen something that resonates with reverence and named the resonance instead of trusting the image to produce it. The repair is almost always the same. Cut the religious noun. Leave the image. See whether the reverence survives. If it doesn’t, the image wasn’t doing what the noun was claiming.

Second: the fragment as diagnostic. A failure that survives across a fifty-line poem is rarely diagnosable from the whole poem — too much else is going on, and the failure can hide behind its neighbours. A failure that’s visible in two lines is visible in two lines because the failure is in the two lines. The discipline of selecting the smallest specimen that demonstrates the mode is itself a sharpening of the editorial eye: if the cut doesn’t work in two lines, the diagnosis is wrong, not the fragment.

— Adam

Every Monday, a new draft goes on the slab. If you find this diagnostic approach useful for your own work, consider subscribing to receive the weekly autopsy in your inbox. If you know another writer currently wrestling with a difficult draft, please share this post with them.

Subscribe now

Share

Lang Bua cave - Flores, Indonesia

By the second stanza the poem is already consoling me, and the loss it means to console has not yet landed — I have offered the reader comfort for a grief the draft never made them feel. I mourned a species before the poem had buried it.

This is a sibling poem to Baiji (I’m holding this one back for now) and Thylacine — the third extinction draft in the drawer, the one that didn’t make the What Storms Disclose (a draft pamphlet) selection. Worth putting on the slab partly to see what kept it off the boat.

The poem

Little lady of Flores

Here so long inside the gloom,

laid to earth before that oval door,

its constant sky. Behind perspex,

your caved-in eyes gaze back at me.

Ebu Gogo sets her small feet

in soft mud beside the same creek

her cousins knew. A trail of tiny

craters, daubs on a dark moon.

Childless, she left us her bone.

Stories the locals still tell

in the language of grandparents.

Eleven lines in 4–4–3. I’m standing at a museum case looking at the Homo floresiensis skeleton — Ebu Gogo, the small wild people of Indonesian folklore the locals on Flores claim to have seen in the forest. I read the case label, meet the skeleton’s gaze through perspex, imagine her alive at a creek with her cousins, register her small footprints in mud as craters on a dark moon, and arrive at the close: childless, she left us her bone, and stories, and the language those stories live in.

The structure is honest about its plan — encounter, empathy, residue — and the individual moves are strong. Laid to earth before that oval door, / its constant sky does double duty (the cave mouth, the case opening, the eye socket itself). Your caved-in eyes gaze back at me is the meeting-of-gazes line the museum encounter wants. A trail of tiny / craters, daubs on a dark moon is the best image in the poem — the smallness of her footprints redoubled as cosmic geometry.

What it doesn’t quite do — and the reason it didn’t make the cut — is sit with the species’ absence before it offers what survives.

It’s also relevant to note this poem has been around a long time and it has endured many separate attempts at resuscitation. At this stage, it may in truth have been cut too hard.

The autopsy

Failure identified:

— #4 The easy resolution (sub-case: the unearned consolation)

The closing tercet:

Childless, she left us her bone.

Stories the locals still tell

in the language of grandparents.

Three nouns across three lines: bone, stories, language. Each line broadens. Bone is material, particular, singular — one bone left from a species that disappeared. Stories is plural, collective, cultural — the locals’ folklore. Language is the broadest layer yet — not even the stories themselves but the medium they survive in, the grandparents’ language carrying them forward.

Compare the shape of this close to the another extinction poem in the drawer. Thylacine narrows: not even the scrape of claws → not a whimper → not even tiger stripes → the wide gape of the jaw. The poem closes inward, down to the one specific physical detail.

Little lady of Flores closes the other way. It broadens outward: physical residue → cultural memory → the language those memories live in. The widening is the consolation. By the time I arrive at the language of grandparents, I’ve moved the reader two steps away from the perspex case and the caved-in eyes. The species is gone, the line says, but its trace persists across generations of speech. The reader is reassured.

The trouble is that I haven’t earned the reassurance. The encounter with the case takes four lines; the imagined-alive Ebu Gogo takes four more; the close offers two layers of cultural survival in three lines. The ratio is wrong. The poem reaches past the loss before the loss has had its full moment.

What I conceal is what the other two extinction poems insist on disclosing: that the species is gone. Thylacine names blankness (and, last week, I diagnosed the line that names it too declaratively — different problem, same family of restraint). Little lady of Flores moves past the absence to the consolations it can offer the living.

The repair. End on bone.

The cleanest cut: drop lines 10 and 11. The poem becomes 4–4–1, with the closing single-line stanza doing the work that the current three-line tercet diffuses. Childless, she left us her bone. Then the white space. Then nothing. The reader sits with the singular bone.

If the asymmetric 4–4–1 feels too clipped, the alternative repair is to leave bone as the close and find the cultural-memory material a different home — perhaps as the opening of the poem rather than the closing. Stories the locals still tell / in the language of grandparents could be the introduction to Ebu Gogo: the folkloric Ebu Gogo before the museum-case Ebu Gogo, the imagined alive before the dead. The poem would move from the local stories → the museum case → what is actually left (bone). That inverts the consolation problem: the cultural-memory frame becomes a setup the encounter then disturbs, rather than a reassurance the encounter then receives.

Suggested order of revision. I’d try the simplest cut first — drop the last two lines, sit with the 4–4–1 version for an hour. If the close feels too steep, try the structural inversion (stories at the top, bone at the bottom). If neither works, the third option is to keep the current structure but make the language of grandparents darker — find one specific local detail that complicates rather than consoles. A particular story about Ebu Gogo that’s not benign. A grandparent’s word the speaker doesn’t understand.

What’s working. Laid to earth before that oval door, / its constant sky is the line I’d protect at all costs — three things at once (cave mouth, case opening, eye socket) without ever naming the doubling. Your caved-in eyes gaze back at me is the meeting-of-gazes line the museum elegy genre is constantly reaching for; most fail because they let the speaker do the gazing, where this version makes the dead do it back. A trail of tiny / craters, daubs on a dark moon is the kind of image I write and have to be careful not to lose. Childless, she left us her bone — the line the repair builds on — is the strongest declarative in the poem, and the singular bone is doing all the work I want the poem to land on. None of these are at risk from the proposed cut. The opposite: the cut gives them the space to land.

The general lesson

The easy resolution lives most often at the close — which is why every editorial pass should read the last stanza twice, once for what it says and once for the shape of how it says it. Two questions worth asking of any closing tercet, quatrain, or coda:

First: does the close narrow or broaden? The strongest closes I know narrow — they move from the general toward a single particular that has accumulated weight across the poem and now has nowhere else to go. Closes that broaden, that move from particular to general or from material to abstract, almost always lift the reader away from the specific occasion before the occasion has fully delivered. There are exceptions (e.g. Larkin’s High Windows); the rule is worth knowing as a default.

Second: is the consolation in the poem, or in the form? A closing image that names cultural persistence, or seasonal return, or shared humanity, is often the form of consolation performing itself — the closing tercet doing what closing tercets are conventionally for. The honest question is whether the speaker, in this specific encounter, has earned the consolation the form is offering. If the form is doing the work the speaker hasn’t, the close belongs to the genre rather than to this poem.

— Adam

Subscribe now

The poem has its hand on cold rock — wet, particular, present — and then, a line later, the speaker climbs down off the rock face to report a thing he says he has always known; the looking stops, the contemplation begins, and the poem leaves the weather it was made of.

Again came back from the drawer this spring and has just been published, which means the draft underneath it can go on the slab whole. Thirty-two lines, four stanzas — the early version, a man at a dry seep checking a spring that hasn’t come.

The poem knows its ground. The geology is real, observed, never reached-for — water finds the fractures rather than making them, the springs run on the quiet side where fewer come, the ferns are green and that could mean anything. And the subject is genuinely there: a man checking a seep for water that hasn’t come, again. The title carries it; the first two lines — Again the rock is wet. Again no spring — carry it. The trouble is everything between those two lines and the close keeps the man’s stake out of sight, under the rock he is describing so well.

The autopsy

Failures identified:

— #5 The buried subject

Again the rock is wet. Again no spring. … I can’t know how full the hill is.

The opening fourteen-line stanza runs from the first line to the springs without once naming what the man wants. The nearest the draft comes is the oblique I can’t know how full the hill is — and that is a fact about the hill, not a want about the man. I’ve built the ground in exact detail and left the reason for standing on it off the page. The subject — a person checking a seep for something that hasn’t come back, and wanting it to — is in the poem, but buried under the geology that was meant to serve it. Ask the draft what it is about and the honest answer, read off the page, is fractures.

What’s concealed is the destination: what the man wants, and why the return matters. The reader has the infrastructure — line, image, the worn ewe-path — and nowhere it delivers them.

The repair. The published version surfaces the spine twice, and adds almost nothing to do it. After the springs it adds a single line — But I know what I want to find — naming the want without naming the thing. And it changes the flat they are without water on them to they are when the spring fails, letting the closing image carry the failure the whole poem is about. One line added, one phrase sharpened, and the subject comes up out of the ground.

— #6 The mechanical failure

numberless downhill to the Coneygree Brook … the mud always glues … the must of aged water … I keep coming here.

Well-made particulars that don’t convert. numberless inflates, and the Coneygree Brook names a watercourse the poem has no use for naming; the mud always glues spends a second always the line above it has already spent; I keep coming here states the return — the one fact the whole poem is built to enact — as flat report. Each is defensible on its own. Together they are length without weight: the pennies fall and the hill stays silent.

What’s concealed is the reckoning. The particulars accumulate without gripping onto the man’s stake, because the stake — see above — hasn’t surfaced yet for them to grip.

The repair. The published version trims each. downhill to the brook loses the name and the inflation; the mud always glues compresses to a bare Mud.; aged becomes old; and I keep coming here becomes And this path knows my boots — the return enacted by the body and the place rather than announced.

Suggested order of revision. I’d surface the spine before touching anything else. Add the want — one line that says what the man is there for — and most of the over-accumulated geology becomes cuttable on sight, because the particulars that don’t serve the now-visible subject fall away on their own. The buried subject is the root; the mechanical clutter is half a consequence of it. If you cut first, you spend an afternoon deciding which good details to lose and never find the one detail that would have told you. Surface, then cut.

What’s working. Two lines went to print unchanged, and they are the ones a quick diagnosis would have reached for first. You come back when you come back and I know that now. I’ve always known it — a proverb and a flat admission of long knowing, the kind of pair that usually forecloses a poem’s discovery before it can happen. They survive because the body holds around them: the published version keeps them pinned between the present-tense hand — a drop of water runs down my wrist / from the rock or from the air, / I can’t tell — and the path that does the recognising. Held that way, the proverb is earned rather than announced. I’m not touching those. And the bead at the lip of the fracture — It doesn’t move. It doesn’t go away — survives whole, the poem’s cleanest disclosure, the line I’d protect at all costs.

The general lesson

Two notes I’m taking from this autopsy.

First: a buried subject is diagnosed by absence, not by a bad line. You find it by asking what the poem is about and failing to answer from the page — every line is well made, and not one of them is the culprit. The repair is rarely a cut; it is a surfacing. One added line that names the want can do more than ten cuts, because once the subject is visible every other line can be measured against it, and the clutter names itself.

Second — and this one is about method. The slab only holds what the finished poem moved. A line that survives from draft to print is, by definition, not the failure — diagnose it and you are autopsying a living patient. The discipline is to read the draft against the published version and let the cuts tell you where the body actually was. The lines I marked here are the ones the poem itself decided to lose; the lines I left are the ones it kept, and I have no standing to put those on the slab.

If you would like to read the poem (post repair) as it was published you can do so here. Thanks to and Zakia Carpenter-Hall for publishing it all the great work they do over at Ink Sweat and Tears.

— Adam

Subscribe now

Share

Last week I cut one line that carried three failures; this week the poem fails in only one way — but I committed it twice, and set both instances at the two seams of the form where a reader looks hardest: the close of the first stanza and the open of the second.

The poem

Twelve long lines in two six-line stanzas, gentle rhymes. I’m arriving at a place I used to walk to with someone, switching off the engine, listening to the dusk, hearing a vixen, remembering rooms, hearing the vixen again, and registering the absence of the moon and of any trace of either of us. The poem mostly knows what to do. The fauna are in the right tense — nightjar / still hawked, owls ... were common — and the ecological vanishing runs as a faint counter-melody beneath the personal vanishing. The vixen’s shriek-into-wail does honest work; the moonlit shore arrives as an unexpected inset of intimacy, then the close strips it back. Hooting who not why is the kind of small bright joke I’ll forgive a poem for, even my own.

What it doesn’t quite do — and the reason I’ve twice cut it from draft collections that didn’t quite need it, then twice put it back — is trust its own apparatus all the way through. The structural placement of where it stops trusting is the part the autopsy adds.

The autopsy

Failure identified:

— #3 The invisible argument (sub-case: the reach for the universal)

The closing couplet of stanza one:

Dusk is falling over the trees

as so many other dusks have fallen over the same trees.

And the opening couplet of stanza two:

There’s an hour toward dusk when air sometimes stills.

You hear more in the quiet.

Same failure — stating philosophical content rather than enacting it — appearing twice. I’ve put both instances at the two positions where stanzaic verse most rewards specificity: the closer and the opener. A stanza closer wants to land on a particular image or action that earns the white space about to follow. A stanza opener wants to enter the new movement with energy and locatedness. Neither does. The closer hedges (so many other dusks); I can hear it reaching for recurrence-as-content rather than producing it. The opener slips the pronoun — from I switch off the car to a generalising you — the reach for the universal sub-case — and I notice myself, as the writer, briefly turning into a narrator explaining a property of dusks in general.

The two thesis lines bracket the stanza break. I’m using the poem’s structural pivot — the place where the form most clearly performs its movement — to carry declaration rather than particular noticing. I’m telling the reader what the dusk is, twice, in the two positions where I could most powerfully have shown it.

What I conceal, in those two couplets, is what the rest of the poem otherwise discloses well: that the quietness, the air stilling, the recurrence of dusks over a place — all of these belong to this particular dusk, this car, this absent companion. Generalising them shrinks them. The hedge in so many other dusks is the giveaway; the you in You hear more in the quiet is the second.

The repair. Cut both couplets. What I’d be left with is two shorter and less balanced stanzas — which may or may not be what I want.

If formal symmetry matters to me, the cuts ask for replacement lines, not pure excision. Stanza one could close on the speaker’s body still in the act of arriving: a hand on the wheel, the dashboard going dark, the way the windscreen cools. Stanza two could open on the same body, the same attention, particular: something settling in the engine. The seat-belt warming the back of my neck. The poem’s apparatus is sensory and accurate when it trusts itself; the repair is to extend that trust through the structural pivot.

If formal symmetry doesn’t matter, the unequal stanzas are themselves the formal choice. A four-line stanza closing on attend to its gentle ticking — the ticking engine as the last sound before silence — then a stanza break, then As Sunday fades into dark / a vixen shrieks in the distance as the new movement’s first noise. That’s a stronger poem than the current draft, by some distance. The asymmetry would do work the symmetry currently isn’t doing.

Suggested order of revision. The two cuts are independent and reversible. I’d cut the stanza-one closer first; sit with the four-line stanza for an hour. If the lopsided shape feels right, cut the stanza-two opener too and I have my eight-line poem. If the shape feels wrong, write the two replacement couplets — particular, sensory, in the speaker’s body — and I have my twelve-line poem.

What’s working. Hooting who not why is the joke I keep coming back to; it earns its place. The fauna held in past-perfect (still hawked, were common) keep the ecological note running under the elegy without ever stating itself as ecology — I want all of that protected. The structural pivot at There were rooms where / together in the dark we listened to each other breathing / shushing a moonlit shore is the only fully present sensual image in the collection’s wake, and the shushing is doing more work than its three syllables suggest — sea-on-shore, breath, hush all in one. I’m not touching that one. And the close (no moon, no sign either of us were ever there) is the chiastic arrival the poem has been building. The cuts don’t reach it.

The general lesson

The reach for the universal is a small, common, deceptively gentle failure. It almost always looks like the poet being generous to the reader — here, let me show you what the dusk is. The cost is that the gift is the wrong gift. What the reader wanted was the dusk; what they got was the dusk explained.

Two notes I’m taking from this autopsy, and may keep coming back to. First: where a thesis line sits in the form matters as much as that it’s there. A generalisation hidden in the middle of a stanza does less damage than the same generalisation at a closer or an opener, where the form is amplifying it. I’m going to read drafts at the structural seams from now on — the line before the white space, the line after — which catches more of this kind of failure than a left-to-right read does. Second: the repair often isn’t deletion but particularisation. The lines don’t need to go because they’re long; they need to go because they’re general. Replacing them with shorter, less literary, more located lines tends to be the move.

— Adam

Subscribe now

Tasmanian Tiger (Thylacine), Beaumaris Zoo, Hobart, Tasmania, 1933, David Howells Fleay

Most craft writing shows you a finished poem and asks you to admire it. The Cutting Room does the opposite — and the first draft on the slab is one of mine, a poem about an extinct animal that I undid with a single line of explanation.

The publication's working assumption — the spine I worked out in my draft pamphlet What Storms Disclose, after Heidegger's Aletheia — bad poetry conceals; good poetry discloses. Every failure mode I’ll name in these posts is a specific kind of non-disclosure — a way the poem produces the appearance of depth without the thing itself.

So. Here is the first draft on the slab.

The poem

Thylacine

There’s a famous photograph of one in Hobart Zoo,

the animal inside a sunny cage stares out of frame.

People say they still see them. Sometimes I hear my nose

crack on the edge of the chair. Falling, the day I died.

A sound of breaking bone I can’t have heard. I had, after all

gone dark. Coming round, what I noticed was cornflakes

crunching under foot, paramedics strapping me to a gurney.

For that hour I was gone, what was it like? There is nothing

to report. If extinction is anything, it is blankness. Not even

the scrape of claws scratching concrete. Not a whimper

if that’s what it was, an unjointed tail, the dogginess.

Not even tiger stripes, that peculiar, wide gape of the jaw.

Twelve lines in six couplets. Two extinctions spliced together — the thylacine in the Beaumaris Zoo photograph, and the speaker’s hour of unconsciousness in his own kitchen — and the poem arriving at the proposition that both, finally, are the same thing: blankness.

It’s a poem I want to love. It has the cornflakes — bathos doing the work of disclosure, the return from death announced by breakfast cereal. It has a sound of breaking bone I can’t have heard, which is philosophically exact in five words: a memory that can’t be a memory. It has gone dark, which carries species extinction inside the smallest available phrase for personal blackout. And it has the structural pivot at Falling, the day I died — delivered without explanation, the swerve that the whole poem rests on.

But it doesn’t quite work. When I read the draft now, the reason it doesn’t quite work is sitting in a single line near the end — the line where I briefly stop trusting the poem’s own apparatus and start trusting the reader to take a piece of philosophy off me instead.

Failure Mode — #3 The invisible argument

If extinction is anything, it is blankness.

I read this line now and it’s the thesis stated as proposition. I’ve decided what extinction is before the negation list arrives. The line teaches rather than finds out. The hedged if hedges certainty, not enactment — the proposition still lands as a precept. The reader is being told the answer in the same breath as being asked the question.

The repair. Cut the line. I tested the poem without it: There is nothing / to report moves straight into Not even / the scrape of claws scratching concrete, and extinction is enacted by the negations rather than declared. If the negation list still lands, the thesis was redundant; if it collapses, the negations weren’t carrying their weight and the repair is somewhere else entirely. In my read it still lands. Better, in fact — the negations get more air.

Failure mode — #7 The over-naming

Same line, viewed from the side. By naming the meaning of the missing-thylacine catalogue before delivering it, I’ve closed the door I’ve just opened. The reader, told what blankness is, no longer has anywhere to find it. The Verborgenheit1 — the concealment that lets the negations build into their own conclusion — has been pre-empted.

The repair. The same cut. One line, two failure modes, single reversible action. The failure mode family map I’ve drawn predicts the convergence: failures of the reaching family and the disclosure mis-calibration family commonly cluster on a single line of premature naming.

Failure mode — #2 The speaker contemplating instead of acting

Look at the verbs across the whole poem. I hear. I had, after all / gone dark. What I noticed was. For that hour I was gone. There is nothing / to report. I’m delivering the verdict from after the event: contemplating the photograph from a distance, narrating the fall in retrospect, announcing the equation. I’ve already settled the identity between my own blackout and species extinction; I’m no longer at risk inside the poem.

The repair. Push myself, as the speaker, into the looking. The photograph in my hand or on the screen now. The somatic memory rising involuntarily now. The question what was it like? arriving as it actually does — without a settled answer waiting on the other side. The draft already half-contains this in Sometimes I hear my nose / crack — the present-tense involuntary memory the rest of the poem could be tuned to. The current frame, photograph → recall → thesis, has chosen to live on the consciousness side of the gap the poem is about.

Suggested order of revision. I’d address #3 and #7 first — same line, one reversible cut. If the poem strengthens without If extinction is anything, it is blankness, the family-root is fixed. Then test #2 experimentally: rewrite the opening couplet so the speaker is doing something with the photograph (looking at it now, turning it over, finding it somewhere) and see whether the rest of the poem redistributes its weight. The cornflakes and the sound of breaking bone I can’t have heard will hold under either treatment.

What’s working. The cornflakes under the paramedics’ boots is among the most converting particulars I’ve ever managed, and I’m not letting them go — domestic bathos as the return-route from extinction. A sound of breaking bone I can’t have heard is the philosophically exact phrase I’d want the rest of the poem to rest on, if the thesis line weren’t standing in front of it. Gone dark carries the species inside the person in two syllables — the kind of phrase I write and then can’t quite believe I wrote, which is usually the sign the poem found it rather than me. The pivot at Falling, the day I died is the structural move I’d build any version of this poem around. And the closing negation list (Not even ... not a whimper ... not even tiger stripes) is the apparatus I want, now, to let do its work — no longer guarded by the line standing in front of it. The cut frees them.

What this publication will do

That’s one autopsy. I’ll publish one a week. Same shape: a draft I’ve cut from a manuscript, or a poem of mine that almost works. The poem in full or a fragment. The failure named. The repair specified to this draft, not paraphrased from the entry.

The taxonomy is finite. There are roughly nine canonical failure modes (with eight finer-grained variants), grouped into five families: reaching, closure, disclosure mis-calibration, voice, and conversion. Each post diagnoses one cut draft by named mode. Over time the archive becomes a working clinic: same failures recur, repair patterns become recognisable, the editorial eye sharpens — mine and yours.

The first wave of posts, provisionally:

1. The abstraction reaching — the foundational failure: abstract language bypassing the concrete particulars that would have done the work.

2. The borrowed cadence — another poet’s register adopted without the necessity that earned it.

3. The easy resolution — the comforting ending the preceding lines haven’t earned.

4. The over-naming — the labelling that pre-empts the encounter.

5. The mechanical failure — length without weight; the pennies fall without a sound.

Today’s post triggered three modes at once, which is unusual. Most weeks I’ll be naming one.

Three things this is, and three it isn’t

It’s British, and of the Welsh Marches specifically. My diagnostic language carries the landscape and labour vocabulary of where I live and work; that grain will be visible. It’s diagnostic rather than aspirational — the autopsy register is the point. And it’s transparent about my own failures: the cuts come from my own drawer, by default, before they come from anyone else’s.

It isn’t a workshop. It isn’t a takedown of other poets — when a published poem is diagnosed here, it’ll be because the failure is instructive and the poet survives the cut. And it isn’t, finally, a list of stylistic preferences dressed as principles. The taxonomy is a tool. If it stops being useful, I’ll rewrite it.

The offer

Free for everything in the first three months — every weekly post, the full archive, the working taxonomy as it sharpens.

After that, a paid tier opens: possibly quarterly live workshops on a single failure mode, occasional manuscript mini-feedback (six lines, ruthlessly specific), and the full archive. Founding subscribers will get two single-poem deep edits at any point during their first year.

Free or paid, every Monday morning a draft will be on the slab. If the autopsy register suits you, subscribe.

— Adam

Subscribe now

Most writing about poetry is about how poems succeed. There are good teachers of that, and I owe them much of what I have — but the Cutting Room is for the other thing: what goes wrong in a draft, named precisely, in this line, and what the repair would be.

This Substack is for the autopsy.

The Philosophy of Failure

I have written, by now, several thousand drafts. Most have been wrong in ways I have come to recognise. There are perhaps a dozen named failures that, between them, account for almost everything I have had to cut. Each post here will diagnose one.

Beyond Conventional Poetics: The Failure Mode Analysis

While conventional poetics focuses on how a poem succeeds — cataloguing its devices and celebrating its finished form — there is another way to understand the craft: through the “autopsy” of its failures. Most writing about poetry forgets the failures a draft has “eaten” to reach its final state.

By shifting focus to failure mode analysis, we can diagnose specific reasons why a draft might falter:

• Reaching for abstraction when the work should stay with a thing.

• Naming a feeling instead of letting it disclose.

• Arriving at consolation the work hasn’t earned.

• Borrowing a cadence that belongs to someone else.

• Contemplating when they should act and so on.

Unlike traditional theory, this approach treats the poem as a site of labour and specific technical breakdowns, suggesting that we define poetry not just by what it is, but by the specific “wrong turns” it manages to avoid or repair.

How It Works

I pick a draft that didn’t make it, print it — either in full or as an extract — and walk through what stopped it.

• The failures get named.

• The repair, where there is one, gets sketched.

• The taxonomy grows post by post.

By six months in, the publication will have a working vocabulary that I hope will be useful to anyone trying to write poems with care.

The Slab

The drafts on the slab will all be mine. Other people’s cuts are theirs to autopsy if they wish. Mine I can be specific about — specific enough that you can see what didn’t work without needing to take my word for it.

Landscape and Language

I write from Herefordshire, in the country between the Malverns and the Wye. That matters because the language for landscape and labour here is its own, and the craft thinking in these posts grows out of it. You’ll find more about hedges and orchards and rivers than about workshops or theory.

Subscription Tiers

• Free: Available for the first three months.

• Paid Tier: Quarterly workshops on a specific failure mode, occasional manuscript mini-feedback, and the full archive.

• Founding Subscribers: Includes a single-poem deep edit at any point in the year.

About Me

I have an MA in Writing Poetry from the Poetry School and Newcastle University and am a regular attender of poetry workshops and writing groups. My poems are published in journals and anthologies and I’m working toward a first collection.

I had some great teachers: Glyn Maxwell and Tamar Yoseloff among them. Hat tip also to Jonathan Edwards, John Glenday and Hannah Lowe, each of whom have provided feedback on my poems.

Welcome to The Cutting Room. The first autopsy goes up Monday.

— Adam

Subscribe now