Forty-Eight Hours

The notes are kept in a spiral-bound pad, the narrow kind that fits in a jacket pocket. Later they will be typed into the case management system, and the typed version is the one most people will ever read. The pad is the original. It has a coffee ring on the cover and a phone number written sideways in the margin of page four. Every entry begins with a time.

This is the third document in the file we have been reading. It is still a composite, built from the patterns of real investigations and labeled as such, because the point is to learn the shape of the work and not to borrow one family's grief. The detective who keeps these notes was assigned at 3:40 in the morning. By the time she writes her first line, the child has been missing or dead for longer than she has been awake.

Her first entry reads: 0412. On scene. Perimeter est. by patrol, tape at N. edge of field + both ends of footpath. Sgt. says 2 neighbors already walked through before tape.

Her second: 0419. Request canvass, Elm + Carver, all doors, log no-answers.

Her third: 0431. Gas station corner Carver/Route 9 — cameras? Ask.

Read those three lines slowly. A tired person wrote them standing in the dark, and each one is a race. The perimeter is a race against feet, since every person who crosses a scene leaves something and carries something away. The canvass is a race against memory, and the cameras are a race against a machine that has no idea anything has happened.

The machine is the easiest of the three to understand. Most commercial surveillance systems record onto a drive of fixed size, and when the drive fills, the system writes over its oldest footage. How long that takes depends on the owner's settings, the number of cameras, and the resolution. Some systems keep a month. Some keep a few days. The clerk working the overnight shift at the corner station almost never knows which kind he has. Neither does the detective, until someone finds the box in a back room and reads the menu.

Memory erodes on a similar schedule, though nothing about it is mechanical. Decades of research on eyewitness memory, much of it associated with the psychologist Elizabeth Loftus, have shown that recollection loses its detail and absorbs whatever comes after it. A neighbor who saw a car at dusk will, by the second day, have read the local news site, talked to the woman next door, and heard a rumor at the gas pump. None of this makes him a liar. His memory of the car has simply been rewritten by everything he has heard since.

So the canvass goes out before sunrise, and that has consequences. Officers knock on doors at an hour when people answer in bathrobes, frightened and angry, and some do not answer at all. The detective has asked for the no-answers to be logged, and that small instruction is easy to overlook. A door that did not open is a door someone has to come back to. If nobody writes it down, nobody comes back.

By 0600 the pad has eleven entries. By 0900 it has twenty-six. The handwriting gets smaller as the page fills.

Now we have to stop and do what this book will ask of you at every document from here to the last. You have been doing it already, perhaps without a name for it. Hold the page and ask three questions out loud.

What does it say? Report only what is on the page, in the words that are there. The perimeter was set by patrol. Two neighbors walked through the field before the tape went up. A canvass was requested for two streets. Someone wondered about cameras at the gas station.

What does it miss? Look for what a reader needs and the page does not give. The two neighbors have no names, and we do not know which part of the field they crossed, or why. "All doors" on Elm and Carver leaves out the streets one block over. The camera question is only a question so far, with no answer attached.

What question does it force? This is the question the document makes unavoidable, the one you cannot set down once you have seen it. Here it is plain enough: Who are the two neighbors, where exactly did they walk, and has anyone asked them yet?

Call the three questions Says, Misses, and Forces. You met them in the first chapter, applied to a page signed at 6:52 in the morning, and you used them again on your own grid without the labels. From now on they have names, because you will need them quickly.

The pad does not stay neutral for long. At 0947, after the first conversation with the family at their kitchen table, the detective writes a longer entry than any before it:

0947. Parents interviewed separately, approx. 25 min each. Mother: last saw child 3:15 leaving for friend's house. Father at work until 6, coworker can confirm (get name). Both say child always cuts through field to Carver, has done so since spring. Child likely took usual route.

Run the lens on it. The entry says that both parents described a habit and that the father has a coworker who can confirm his hours. It misses the coworker's name, the friend's name, and whether the friend's household expected her at all. The last line deserves the most attention, because it is a different kind of sentence from the ones above it. The parents reported what their daughter usually did. Child likely took usual route is the detective's conclusion about what she did that day. The words likely and usual sit together in that line, and the probable has quietly become the actual.

It is a reasonable sentence, and nearly every investigator would have written something like it. It is also the first assumption in the file written in an investigator's own hand. Mark it, and leave it where it is for now.

To understand why the detective writes as she does, you need the numbers she carries into the kitchen.

The most cited federal analysis of child homicide in the United States is a 2001 bulletin from the Office of Juvenile Justice and Delinquency Prevention, Homicides of Children and Youth, by David Finkelhor and Richard Ormrod. Working from the FBI's Supplementary Homicide Reports and related data, they found that who kills a child depends sharply on the child's age. For the youngest victims, those under about five, the large majority of known killers were family members, most often parents. For teenagers the pattern resembled adult homicide, with acquaintances and, less often, strangers making up most offenders. Children in the middle years, roughly six through eleven, were killed less often than either group, and their killers were more mixed: family members still made up a large share, alongside acquaintances and a smaller proportion of strangers.

Strangers do kill children. When they do, especially in abductions that end in murder, the timeline is brutal. A study for the Washington State Attorney General's Office by Katherine Hanfland, Robert Keppel, and Joseph Weis, published in 1997 and later updated, examined several hundred child abduction murders. In most of those cases the child was killed within the first few hours after being taken, roughly three-quarters of them within three hours. The same research found that delay in reporting a child missing was common, often because parents first searched on their own or assumed the child was with a friend.

Hold both findings at once. The first tells the detective where most answers have historically been found, close to home. The second tells her that if this is the rarer kind of case, the hours she spends close to home are hours the field is not being walked and the road out of town is not being watched.

That is the knife edge the numbers put her on. The research protects investigators from chasing phantom strangers while the truth sits at the kitchen table, and it is the reason the father's hours have to be confirmed and the mother's afternoon has to be laid out minute by minute. The same research can make a grieving parent feel like a suspect for the rest of his life. The data describes populations. It cannot say what happened to one girl on one afternoon. The detective knows this, and she still has to decide where to send the next two officers.

At 1015 she sends them both to the gas station.

The pad records what happens next in fragments. 1032. Mgr. not on site, clerk says system "in office," no key. 1040. Mgr. reached by phone, en route, 40 min. 1124. Mgr. on site. DVR located. 4 cameras. Retention setting 72 hrs. Below that, underlined twice: Yesterday 1500–1900 exported, 2 copies, 1 to evidence.

Seventy-two hours. If no one had thought of the corner at 4:31 in the morning, and no one had made the call at 10:15, the footage would have survived until roughly the following afternoon, and then it would have been gone. Nobody would have noticed the loss. The file would simply never contain it.

The rest of the first day goes the same way: requests and their answers, the gap between them measured in minutes and phone calls. A second canvass team is assigned to the streets one block over, and the pad does not say why. A search team walks the field in a line at arm's length. The two neighbors who crossed it before the tape are identified at 1350 and interviewed at 1430; one walked the footpath, the other walked straight across the grass toward the place where the patrol officer was standing. Both give their shoes. The detective writes their names in full. On page eleven, beside a list of the child's classmates, someone has copied the mother's question from the kitchen: Mother asks: "Did anyone check whether she got to the corner?" There is no reply written beside it.

Day two arrives without a break between it and day one. The pad simply continues.

By the second morning the shape of the work has changed. The questions that could be answered in the first hours have mostly been answered or abandoned, and the remaining ones are slower. The coworker's name has come in and the father's hours have been confirmed, so that line is crossed through. The friend's mother says the girl was not expected that afternoon. The detective writes the sentence and does not comment on it. Eighteen doors on the canvass sheet are still marked no answer. Six of them get a second knock. Four open.

The footage has been watched once, quickly, by a tired officer looking for a girl in a red jacket. He did not find her. The pad records his finding without qualification: Video reviewed, no sighting of victim. Ask what this misses. He watched one camera angle, not four, and he was looking for a jacket that the mother described from memory at a kitchen table at 9:30 in the morning. Nobody has yet asked what the footage shows apart from the girl: what passed the corner, and when.

There is also a page the pad cannot contain. By the second evening the local news has a photograph of the field, and a community forum online has a thread with three hundred replies. Someone has posted the name of a man who lives on Carver. He is not in the notes. He has not been interviewed, because there is no reason in the file to interview him. By the time the detective learns of the thread, the man has taken his family to a relative's house. He will never be charged with anything, and he will never entirely come home.

The things that were lost in forty-eight hours are worth counting. Some memories have been rewritten. Four cameras at other businesses were never asked about, and their footage will expire before anyone does. A family on Carver Street has been driven from its house. In exchange the file has a perimeter, a canvass sheet with the no-answers logged, two sets of shoes, three hours of video in an evidence locker, a confirmed alibi, and one sentence in a detective's hand about which way a little girl usually walked.

Time in this work is always spent. The only choice is what it buys.

Now return to the grid you drew at the end of the last chapter. Find the stretch written in pencil, the hours your child was accounted for only by assumption. For each pencil cell, write in the margin the one thing that would have turned it to pen if you had needed proof tomorrow morning: a camera, a teacher, a neighbor's porch, a receipt, a phone's location history. Then write beside it how long that thing would last. Be honest with the numbers. Some will be weeks. Some will be days.

The next document in the file arrives on the third day. It comes from a doctor who never knew the girl's name before it was typed onto a form, and it will be read against everything the detective has written so far, including the line about the usual route. Bring your margin with you. You will want to see which of your own pencil hours a body can speak to, and which it cannot.