Three research groups on three continents went looking for what predicts a trafficking outcome and came back holding the same variable. Childhood sexual abuse. Not poverty, not migration status, not how long the exploitation ran. What was done to her before. Then I held up a character who did not have one, and the strongest finding in the literature had nothing to grip.
Her name is Marigold, and before that it was Ixchel Anay Tepeu. K’iche’ Maya, out of a highland village in Guatemala where her grandmother taught her which herbs to burn and how to mean what she said. She was taken at seventeen because a man looked at her face. She went north through the Sinaloa pipeline, got three letters inked on the inside of her left wrist so any buyer would know whose she was, and went into the ground somewhere short of the border.
She came back out of it a vampire on her eighteenth birthday. She has no novel. What she has is fifteen first-person notes and a canon record, both quotable line by line, and a household in Tucson where she is the quiet one who lights candles for the dead every night.
This is the fourth pass of the same exercise. Claude and I built Psychological Markers in Trafficking Victims, 2000-2025, forty-one graded sources with an explicit confidence level on each key judgment, and I have been using it to grade my own fiction. Rocket came back accurate everywhere she has no choice and invented everywhere she acts. Blondie lost that round to my own character database rather than to the prose. Babydoll agreed with the record everywhere except one survey item.
Here is the finding. The record scores Marigold correctly and cannot reach a single cause. Her only sexual abuse in childhood is the trafficking itself, so the literature’s strongest predictor goes circular the moment anyone tries to code her. The most consequential fact about her body was unknown to her while it was true, for reasons that are physically ordinary and thoroughly documented. And the clinical cluster the instruments do detect in her is produced directly by that unknowable fact.
She would answer every question honestly. Every honest answer would point at the wrong cause. The unmeasurable causes the measurable, and that is the whole article.
Babydoll’s piece was about a word: a respondent who holds every fact and refuses one noun, and an item that cannot tell her apart from a woman who never had the category. This is not that a second time. Babydoll is a semantics failure. Marigold is an ontology failure, which is a different layer. One fact that cannot be coded without eating itself, and one fact that was not in the respondent to retrieve.
She is also a vampire, which is the first objection anybody sensible raises. Rocket’s piece answered it and I am standing on that answer rather than re-mounting it, and it applies even less here. Nothing in either failure below needs the supernatural. A seventeen-year-old walking a trafficking route without knowing she is pregnant is the most mundane proposition in this article.
The Girl Who Was Never a Risk Factor
Abas and colleagues assessed 120 women returned to Moldova, with DSM-IV diagnoses assigned by structured clinical interview conducted blind to exposure history, and childhood sexual abuse came back at an adjusted odds ratio of 4.68 for any disorder, the largest term in their entire risk model. Reid and colleagues compared 913 Florida youth with accepted trafficking abuse reports against a matched sample and found sexual abuse history the strongest predictor of trafficking itself, at 2.52 times the odds for girls. Laird and colleagues at Deakin pooled 37 studies covering 67,453 participants and returned a pooled odds ratio of 3.80.
Moldova, Florida, and a pool of sixty-seven thousand. One variable on top in all three.
Now her file. Intact family, poor and tight-knit, a mother and a father in the house. A grandmother teaching her to cleanse with herbs and listen for spirits. Faith practiced out loud, in a village, among people who used her actual name. The first note in her series is eight hundred words of a childhood nothing had happened to yet, and it is the only one in the run that stays warm the whole way through.
She was taken because she was beautiful. The canon record puts it in six words: when the cartels came, “they saw faces, not names.” That is the entire risk model. A face, and a village close enough to a corridor.
So the strongest predictor in three studies scores her at zero, correctly. There is no antecedent to find because I never wrote one, and I never wrote one because what I wanted was a girl whose ruin has a start date.
The record does work on her, and it works where it always works. The man who conditioned her runs a method, and his method goes after the one thing the rest of the pipeline leaves standing. He calls her florecita. She refuses to say his name at all and calls him utza’m si’, which is a taboo about coyotes and about not feeding a thing by naming it. He has her pray, waits respectfully until she is finished, and then asks her, without heat, whether anyone came.
Baldwin, Fehrenbacher and Eisenman interviewed twelve women trafficked into Los Angeles County and reported the full range of the non-physical coercive tactics Biderman catalogued for the Air Force in 1957, and that prayer is monopolisation of perception arriving as pastoral care. I scored the chart item by item in Rocket’s piece. She gets one tactic here, not eight.
Seventeen for All of It, Eighteen for One Day
The record says she was born on 1 November 2005 and Embraced on 1 November 2023. She was a minor for every hour of the exploitation, and she turned eighteen on the day it killed her. That is not a technicality. It is the hinge the coding problem turns on.
Ask the screening question. Prior childhood sexual abuse, yes or no.
One answer is true, and it is yes. She was sexually abused as a child. She was seventeen, in a room with a window the colour of dishwater, for months. Code her yes and look at what you have done: you have entered her trafficking as the variable that predicts her trafficking. The 4.68 is now measuring itself.
Code her no and you are lying, and you are also producing the cleaner-looking result, because now she is a null case, a girl with no risk history who ended up in the pipeline anyway. Both answers are defensible. Both are wrong in different directions. There is no third box.
The field says this itself, and it says it in a National Academies report rather than in a complaint from outside. The Institute of Medicine, in Confronting Commercial Sexual Exploitation and Sex Trafficking of Minors in the United States, records that disentangling commercial sexual exploitation from sex trafficking is impossible in many instances, and that the two terms get used interchangeably in the literature. That is a committee writing down, in its own report, that its categories cannot reliably be told apart in the cases where it matters most.
The definitional wiring is what makes that unavoidable rather than sloppy. For a minor in commercial sex, US law requires no force, fraud or coercion element at all. Any person under 18 exchanging sex for anything of value is a trafficking victim, full stop.
Which is why a homeless sixteen-year-old trading sex for a couch is counted in one study and not in the next, and why the brief names the minor-victim rule as the single largest driver of divergent statistics in this entire field. Marigold sits directly on that seam. Everything done to her is the exploitation and the abuse history at once, because the statute defines it that way.
The obvious objection is that I have rediscovered epidemiology. Group findings do not determine individual cases, everybody knows it, and hanging a fictional girl on it adds nothing.
That objection is aimed at a claim I am not making. I am not saying a pooled odds ratio fails to settle one person’s case. I am saying the coding rule breaks before the arithmetic starts. The predictor and the outcome are the same event, in the same months, done to the same seventeen-year-old, and the field says in print that it cannot separate them. That is not the individual-versus-group problem. That is a variable eating itself, reported by the people who built it.
That is question one. She would have answered it honestly, and the honest answer would have been useless.
What the Record Gets Right
Now the part where the instrument works. It does, and none of the rest lands unless you believe that first.
Kiss and colleagues interviewed 387 children and adolescents in post-trafficking services in Cambodia, Thailand and Vietnam and found depression at 56.3 percent, anxiety at 32.6 percent, and PTSD at 25.5 percent. Depression outranks PTSD by better than two to one. That is the one finding in the brief scoped specifically to minors, and Marigold is the only minor victim in the cast. Five trafficked girls live in that house. The minor-scoped finding applies to exactly one of them.
Perry and colleagues analysed baseline data from 110 young people aged 11 to 19 with substantiated commercial sexual exploitation entering trauma-focused therapy and found 51.8 percent meeting clinical criteria for post-traumatic cognitions. Not symptoms. What the person has concluded about herself. What predicted it was age and the number of trauma categories, not any feature of the trafficking.
She fires on that, and her file says so in the flattest possible terms. The trauma profile field reads: negative self-concept, guilt, the most hidden of the six, avoidance and shame doing one job. That is the disturbance-in-self-organisation domain Brewin records travelling with prolonged inescapable abuse in ICD-11. What she matches is that profile, not a trafficking-specific one, because the brief’s central judgment on whether trafficking carries a signature of its own sits at Low confidence and Rocket’s piece has already worked through why. She is a clean positive on a well-evidenced marker.
So the instrument is doing its job. A girl comes through a pipeline at seventeen, and what the literature says to look for in trafficked minors is depression and post-traumatic cognition rather than the startle and the flashback most readers expect. That is what she has. A clinician who met her would see a composed eighteen-year-old keeping a nightly ritual for the dead, and would tick the right boxes. Flag raised. Correct flag.
Hold onto that. Section five is going to tell you why the flag fired, and the reason is not in her file.
The Signal She Never Had
The second question is about her body, and answering it properly took four separate literatures, not one of which is about trafficking.
The question is whether she was pregnant. She was. Roughly seventeen weeks, at the point they put her in the ground, and she did not know.
Not withheld. Not dissociated, not repressed, not walled off behind the door she keeps the rest of it behind. The fact was not in her to retrieve, and that distinction carries the weight of everything below it. No item on any instrument retrieves a fact the respondent does not have.
Here is why that is ordinary rather than convenient. Three pieces, in this order, because the order is the argument.
One, and it predates the traffickers by a decade. She did not lose a reliable cycle. She never had one.
Khan and colleagues, reporting in 1996 on 497 women from the INCAP nutrition cohort in rural Guatemala, where 78 percent were moderately or severely stunted by age three, put mean age at menarche at 13.7 years with a standard deviation of 1.3, rising to 14.1 for the severely stunted against 13.5 for the not stunted. Stunting remained an independent predictor of later menarche after adjustment.
That is not a general population and I am not presenting it as one. It is her population: poor, indigenous, chronically undernourished highland Guatemala, which is the exact place the canon record puts her, in a family it calls tight-knit and poor. A Western general-population menarche figure would be the wrong comparator here, not the safe one.
Then the second half, and it turns on the same variable. Apter and Vihko followed 200 schoolgirls aged 7 to 17 across three measurements at eighteen-month intervals, classifying cycles by serum progesterone, and reported that the time from menarche until half of cycles are ovulatory runs from about one year for girls whose menarche came before twelve to roughly 4.5 years for girls whose menarche came at thirteen or later.
Those are well-fed, wealthy, Finnish girls in 1983, and their absolute ages do not transfer to a K’iche’ village. What transfers is the relationship they measured. Later menarche means a longer wait for ovulatory regularity. That relationship is what her case runs on, and I am saying so out loud rather than letting the numbers do it quietly.
Run it. Menarche around fourteen, taken at seventeen, so roughly three gynecological years, in the slow group by construction. She was taken before she reached the point where half her cycles ovulated. A late or missing period was not a signal in her life. It was Tuesday.
The convergence is the thing. Childhood undernutrition delays menarche, and later menarche lengthens the wait for regularity. The same poverty pushes the same lever twice, and it is the same poverty that put her in front of the men who took her.
Two, during the pipeline. Gordon and colleagues, writing the Endocrine Society clinical practice guideline in 2017, define functional hypothalamic amenorrhea as chronic anovulation with no identifiable organic cause, associated with stress, weight loss, excessive exercise, or a combination. She has the first two at maximum, for months, with no floor under either.
Elias and colleagues, publishing in 2007, used the Utrecht DOM screening cohort to study roughly 12,500 women who lived through the Dutch famine of 1944 and 1945, when rations fell from about 1,500 kcal to under 700. Severely exposed women returned an odds ratio of 8.85, with a 95 percent confidence interval of 7.31 to 10.70, for concurrent irregular menses against the unexposed. Irregular menses. Not amenorrhea. I am saying what it measures because overstating it is the cheapest available way to lose an argument I can otherwise win.
The Dutch famine cohort is a wealthy, previously well-fed population forced down under 700 calories by an occupying army. That is the closest available analogue to enforced starvation in a body that used to be fed, which is roughly her situation and not exactly it.
Three, the symptoms. The eighth note has the morning sickness, and it has her explaining it away in real time. She wakes with her stomach climbing her throat and retches on an empty stomach, bringing up nothing, until her eyes stream. She thought it was the water. She thought it was the fear, which had got into her stomach the way it had got into everything. In her own words: “There was always a reason standing ready to take the blame, and I handed it to all of them, one after another, and never once to the reason it was.”
Every symptom she had came with a true competing explanation she had no reason to doubt. She was starving. She was terrified. Both of those cause nausea and both of them disrupt menstruation. The differential she was running came back correct on every item except the one nobody asked.
The women in that room knew. The note says so, and it says so from inside her ignorance. They looked at her in a way she did not understand then and has understood completely ever since, and not one of them said the word, “because in that place I was not a girl something was happening to. I was stock, and stock does not get told.”
One moment comes close, and it belongs here rather than in a footnote, because it is the hardest fact for my own argument and it still lands on my side. In the last weeks on the road she feels the quickening, low and deep, a flutter and then a turn, and her account says the part of her that had refused to know for months finally could not refuse. Then watch what she does with it. “I still would not say the word... I was not going to start asking now, this close to the end, about a thing I could not have borne the answer to.”
So the honest version is narrower than “she never knew” and worse. For almost all of it the fact was not there. At the end she had a movement, no word for it, nobody on that road she could have got the word from, and days left to live. “I walked north carrying a child I would not name even to myself.” There is no window in that sequence where an interviewer with a form gets a different answer out of her.
Now the pregnancy denial literature, which is a ceiling and never a match.
Wessel, Endrikat and Büscher ran a prospective study across Berlin from 1 July 1995 to 30 June 1996 covering 29,462 deliveries and found 62 cases of denied pregnancy. That is 1 in 475 women unaware past twenty weeks, and 1 in 2,455 unaware until a surprise delivery at term. Their own conclusion is that the common view of denied pregnancy as an exotic and rare event is no longer valid.
Delong and colleagues ran a French multicentric prospective case-control study in 2022, 71 mother-infant dyads against 71 controls, assessed in the week after delivery with the MINI and the QSSP. Not being in a stable relationship returned an odds ratio of 17.18. No high school diploma returned 1.11, a psychiatric history 6.33, and being older was protective at 0.86. Marigold is seventeen and unattached, and the pipeline ended any schooling she was going to finish. That puts her on three of the four. Two independent European datasets, twenty years and one border apart, which is corroboration rather than a single witness.
Here is the part that runs the other way from how it looks. Marigold is at roughly seventeen weeks, hand sized, which is the detail section five turns on. Wessel measures denial beyond twenty. She never reaches the bracket those studies describe, so I am not claiming she is a documented case of anything. I am using them as an upper bound.
If 1 in 475 fed, housed, insured Berlin women get past twenty weeks without knowing, and 1 in 2,455 get all the way to the delivery room, then a starved captive not knowing at seventeen weeks needs no special pleading at all. Her case is the ordinary one. The figures I just quoted are the extreme one.
The objection I expect here is borrowed authority: that I am leaning on a 1-in-475 phenomenon studied in women with a pharmacy on the corner to carry a thesis about a girl in a pipeline. The arithmetic is the answer. Seventeen is less than twenty, and every figure in that paragraph is a bound she sits comfortably inside rather than a match she has to be squeezed into.
Bick, Howard, Oram and Zimmerman, publishing in 2017 out of King’s College London and the London School of Hygiene and Tropical Medicine, found that more than one in four trafficked women became pregnant while trafficked. The pregnancy subgroup is 28 women, of whom 42.8 percent had at least one termination during the trafficking and 89.3 percent met criteria for a mental health disorder, and the authors note that access to maternity care was frequently controlled by the traffickers. Twenty-eight is a small number and I am reporting it as twenty-eight rather than standing behind the percentage.
A more convenient number was available and it did not clear. Lederer and Wetzel put at least one pregnancy during trafficking at 71.2 percent, which is precisely the figure this section looks like it wants, and it stays out on three counts: n = 66, advocacy-linked, and the same paper I already demoted once inside the brief itself over a different number.
Four different populations are being combined here and applied to one fictional girl. Every link is sourced and the chain is mine. Not one of these cohorts is a normal-population baseline, so name the class of each.
- Khan, rural Guatemala, 78 percent stunted by age three. Hers. Cited precisely because it is hers, and the general-population figure would be the error.
- Apter and Vihko, Finnish schoolgirls, 1983. Well-nourished and wealthy. The relationship transfers. The ages do not.
- Elias, the Dutch famine. A fed population forced to starve. The closest analogue to her conditions that exists in the literature, and still not her conditions.
- Wessel in Berlin and Delong in France. General obstetric populations with nutrition, autonomy, contraception and healthcare. The furthest from her of anything cited, which is the entire reason they work as a ceiling.
The gaps all run the same direction, and that deserves one plain sentence. Every rate borrowed from a fed, housed, insured population is a floor for a starved captive, never an estimate of her.
Delong makes the point from inside its own data: absence of a high school diploma is one of its risk factors, so the further down the ladder a woman sits, the higher the rate. Every general-population figure in this article understates her case instead of flattering it. That is the opposite of the usual failure, where a writer lifts a comfortable number off a comfortable cohort and quietly applies it to someone who has nothing.
One gap stays open and is worth naming once. Nobody can study pregnancy recognition under captivity, because the population is unreachable by design. There is no cohort of currently trafficked pregnant seventeen-year-olds to follow, and there never will be. Everything above is assembled out of adjacent populations for exactly that reason.
That is the problem in one line. The instrument is exquisitely sensitive to the thing it is looking at and has no opinion whatsoever about the thing in the ground eighteen inches under it.
That is question two. She would have answered no, honestly, and been wrong, and no follow-up item on any form in the field would have caught it.
Why That Cluster And Not Another
Section three said the instrument flags her correctly. Section four said the most consequential fact about her is unreachable. Those two are not a matched pair of failures. They are a chain, and this is the link.
Her cluster is negative self-concept and guilt. Not re-experiencing. Not hyperarousal. Not the startle response and the intrusive flashback a screening tool built around PTSD reaches for first. Shame and self-condemnation, the domain she hides best, which her own file calls the most hidden of the six.
Ask why that cluster and not another. The answer is in her thirteenth note and nowhere else, and the eleventh note walks her to the edge of it: “I was walking toward it with something alive inside me and no idea that either of us was about to die.”
She woke under the ground on her eighteenth birthday, dead and already thirsty, and found something in the grave with her. Her words: “Cold, and small, the size of my hand, and I knew what it was. I had carried it the whole road and never once let myself know, and the grave was where I finally knew, in the same moment I lost it, the only moment I was ever given.”
She clawed up out of the dirt carrying it. She screamed until there were no words left in it, and she wept blood until there was no blood left to weep with.
Then the thirst, which had been running underneath the grief the entire time, got louder than the grief, and she caught herself thinking about where she would feed before she had finished mourning her own child. She put it back in the ground, covered it over with her own hands, gave it no name because she had none to give, laid no flower because there was none, and walked away to feed.
Read the mechanism rather than the scene. She learned and lost inside the same second. The mourning was minutes old when the hunger took it. There is no version of that where the shame comes out proportionate, because no length of grief would have been long enough, and she knows it, and the knowing is exactly the shape of negative self-concept and guilt.
That is why the cluster is what it is. Perry’s 51.8 percent fires correctly on her, and the cause of the firing is the fact section four established that nothing in the field can reach.
Be precise about what that claims and what it does not. The instrument did not fail. It fired, on the correct marker, and a clinician reading her chart would have written down something true. What is wrong is the causal column, and it would stay wrong through any number of honest follow-up questions, because nothing is being withheld. She would tell you everything she had. The one thing that produced the finding is the one thing she was never given to know until the second it stopped being true.
The form has a box for what was done to her, a box for what she believes about herself, and no box at all for the only fact that outlived her.
She lights candles every night. She tells the household they are for the dead she could not save, and she tells herself the same thing, and both of those are true. Her own line about the rest of it: “I have never told a living soul and I never will.”
The One I Am Conceding
Every one of these articles has one and this is hers. Depression converted into saintliness.
Kiss puts depression at 56.3 percent in trafficked minors against PTSD at 25.5, and she is the cast’s only minor. Now look at what she presents as. Her archetype field says The Quiet Center. Composed, serene, endlessly accommodating, the one who knocks on an open archway before she walks through it. Grief rendered as gentleness. Depression rendered as devotion.
Anhedonia is the other thing that could have grown out of that grave, and it is at least as well evidenced as what I wrote.
Rocket converted despair into rage, which I conceded in her article, on the grounds that rage generates scenes. This is that filter one resolution finer. Devotion photographs. Anhedonia does not. A girl who lights candles for the dead every night hands you a ritual and a reason for other people to come to her. A girl who has stopped being able to want anything hands you a chair with somebody in it.
I am not re-deriving the filter, because that argument is already published and sitting over there. I am reporting its fourth appearance in four characters, which is the part that should bother me and does.
One thing I will not take as a concession, because it is not one. The Embrace ends the pregnancy, which looks like plot economy until you check the date on the mechanism: Vampires: Victorian Necrophilia established in July that a corpse funds no circulation and no metabolism, and gestation needs both. The rule was published months before the audit and it forces the outcome. This piece can afford to be right about one thing.
The Woman Who Also Did Not Know
Somewhere in the readership of this piece is a woman who found out she was pregnant and lost it in the same appointment. Somewhere else is a woman who carried one for months without knowing, got told, and has spent every year since being asked how she could possibly not have known.
I am not hedging my way past either of you. The second question in this article is the one people ask in a tone, and the answer is in the paragraphs above and it is boring. A cycle that was never regular. A body under stress and weight loss. Symptoms with true competing explanations standing ready. Nobody in the room with a reason to say the word.
1 in 475 women in Berlin, with a pharmacy on the corner and a doctor who takes their insurance, get past twenty weeks without knowing. That rate is not a measure of how closely a woman was paying attention.
And every number I quoted is about women who reached somebody: a shelter, a court, a clinic. Rocket’s close established that and I am not establishing it again. The direction of that bias is unknown. In the version of her life where Marigold stays human, she reaches nobody at all.
The case for fiction going anywhere near this material is one I made elsewhere and am not making twice. This piece does the narrower job. One character, one literature, and the place where the two stop being able to talk to each other.
The record is built out of women who answered questions. Marigold would have answered all of them, honestly, in order, in the wrecked whisper the road left her with, and the answers would have been no and no.
The instrument found the shame and filed it under the wrong childhood. The right one is in the ground outside Nogales with no name on it.
Sourcing appendix: Psychological Markers in Trafficking Victims, 2000-2025. Forty-one graded sources, an explicit confidence level on each of its eight key judgments, and a verification pass that revised four of the eight.
You may also like: - Rocket Against the Record, or the Two Places a Protagonist Has to Lie - Babydoll Against the Record, or the Noun She Will Not Answer To - Blondie Against the Record, or the Survival Skill That Got Her Sold