Kestrel: IELTS Academic Reading practice test

The peregrine falcon hunts by climbing high above its prey and then folding its wings and falling. Measurements of that dive, which falconers call a stoop, have exceeded three hundred kilometres an hour, and no animal has ever been recorded moving faster. For most of its history the bird has been a creature of wild places, nesting on sea cliffs and mountain ledges across six continents and taking other birds in the air. Nothing about it suggests an animal that would adapt easily to human company, which makes the last fifty years of its history all the more surprising.

In the middle of the twentieth century the peregrine came close to vanishing from large parts of its range. The cause was not shooting or trapping but a group of agricultural pesticides whose residues accumulated in the tissues of the small birds peregrines eat. The chemicals did not kill the adults. They interfered with the deposition of calcium, so that the shells of the eggs came out thin, and the eggs were crushed beneath the weight of the parent sitting on them. Clutch after clutch failed. By the mid-1960s the species had gone from the whole of eastern North America, and in Britain the breeding population had fallen by more than half in twenty years.

Two things turned the decline around. The pesticides responsible were restricted and then banned across most of the developed world, and wildlife biologists began breeding peregrines in captivity and releasing the young by a technique borrowed from falconry: the chicks are placed in a box on a tower, fed by keepers they never see, and left to fly when they are ready. Thousands of birds were released in this way in North America between the 1970s and the 1990s. Which of the two measures mattered more is not a question the recovery allows anyone to answer, since they happened at the same time.

What nobody had predicted was where the recovering birds would choose to live. A city, seen through the eyes of a peregrine, is a range of artificial cliffs: the ledges of cathedrals, the parapets of office towers, the cooling towers of power stations and the steelwork of suspension bridges all offer the same thing a rock face offers, which is a flat surface high above the ground with a clear view and nothing that can climb to it. The streets below are full of feral pigeons, a bird of roughly the size peregrines prefer, and the prey does not migrate away in winter. The birds also exploit something no cliff provides. Where a city throws enough light into the sky, hunting does not have to stop at dusk, and urban peregrines have been filmed taking migrating songbirds out of the illuminated air long after dark.

The relationship is now actively managed. Conservation groups and building owners fix nest boxes to towers, chimneys and bridges, usually with a camera inside, and in a number of cities the resulting footage has a following that a television series might envy. The publicity has a practical value: a pair of falcons that several thousand people watch every spring is a pair whose ledge is unlikely to be disturbed by maintenance work without anybody noticing.

Peregrines are not the only raptors to have come in from the countryside. Red kites, which had been reduced in Britain to a few dozen pairs in the Welsh hills, were reintroduced from the late 1980s onwards using young birds brought from Spain and Sweden, and they are now an ordinary sight over the towns of the Chilterns, where a good many residents feed them in the garden. The kestrel that hovers above a motorway embankment is there because the verge is rough grassland that nobody mows, and rough grassland holds voles. Sparrowhawks, which hunt by surprise along hedges, found that suburban gardens are hedges with bird tables in them, and moved in accordingly.

City life brings its own hazards. Young peregrines leaving the nest for the first time have a short flight and a hard landing, and in a city the landing may be a road or a canal rather than a slope of scree. Glass kills birds of prey as it kills smaller species, since a reflected sky is indistinguishable from a real one. The most persistent danger is chemical again: poison laid down for rats does not stop working when the rat is eaten, and surveys of dead raptors routinely find rodenticide residues in a majority of the birds examined. None of these is a reason to think the cities are traps. They are a reminder that a habitat which happens to suit an animal is not the same as a habitat designed for one.

The wider lesson is about the category of wilderness itself. Conservation has often been organised around the idea that nature is what is left where people are not, and the peregrine on the cathedral is an awkward fact for that idea. The bird has not been tamed, and it has not been domesticated; it has simply found that a particular arrangement of concrete and glass meets its requirements better than the coast it came from. If species are going to survive in a world with more buildings in it every year, the useful question may not be how to keep them away from us, but which of our structures they can be persuaded to use.

There is no moment in a criminal trial more persuasive than the one in which a witness points across the room and says, without hesitation, that the person in the dock is the person they saw. Jurors find it almost impossible to discount, and it is not obvious that they are wrong to be impressed: the witness was there, the witness has no reason to lie, and the witness is certain. Yet when convictions in the United States began to be overturned by genetic evidence that had not existed at the time of trial, mistaken identification turned out to be the single most common element in the cases that collapsed, present in around seven out of ten of them. Something about the most convincing evidence a court hears is badly understood.

The first part of the explanation is that memory does not work in the way the courtroom assumes. The assumption is one of recording: an event is stored, more or less completely, and later played back with some loss of detail. The alternative view goes back to Frederic Bartlett, who in the 1930s asked English volunteers to read an unfamiliar folk tale and retell it repeatedly over weeks and months. The retellings did not simply shorten. They changed, and they changed in a consistent direction, with unfamiliar elements quietly replaced by familiar ones until the story had become the kind of story the teller already knew. Remembering, Bartlett argued, is not retrieval but reconstruction, and reconstruction borrows from whatever is to hand.

What is to hand includes anything that happens after the event. In a much-copied experiment of the 1970s, Elizabeth Loftus showed volunteers a film of a collision and then asked how fast the cars were going when they hit one another, varying only the verb: contacted, hit, bumped, collided, smashed. Estimates of speed rose with the violence of the word. A week later, those who had been asked the smashed question were more than twice as likely as the others to report having seen broken glass in the film. There was no broken glass. A single word in a single question, asked after the fact, had been absorbed into what the witnesses believed they had seen, and they had no way of telling which part of the memory came from the film and which from the questioner.

This is the point at which the literature is usually summarised as showing that confidence and accuracy are unrelated, and that summary is wrong. What the courtroom studies established was that a witness's confidence at trial predicts very little, which is unsurprising, since by the time of trial the witness has been interviewed repeatedly, has been told the suspect was charged, and may have seen the defendant's photograph in a newspaper. Confidence expressed at the moment of a first identification, before any of that has happened, is a different measurement. Where the line-up has been properly conducted, witnesses who say at once that they are certain are right far more often than witnesses who hesitate. The relationship is not absent. It is destroyed, by the process that follows.

That finding has shaped the reforms of the past two decades, which are less about warning juries and more about collecting the evidence before it decays. A line-up should be administered by an officer who does not know which member is the suspect, since an investigator who knows cannot help leaking it. The witness should be told in advance that the person may not be present at all, which removes the assumption that the task is to pick the closest match. The others in the line-up should resemble the description the witness originally gave, rather than resembling the suspect. And whatever the witness says about their own certainty should be written down there and then, in the witness's own words, because it will not survive the months that follow. The psychologist Dorothea Marsh, who has advised several police forces on these procedures, describes the aim as treating an identification like a blood sample: collected once, under controlled conditions, and useless if contaminated.

Not everyone is content with the direction of travel. Idris Kwan, a former prosecutor who now teaches evidence, argues that the reform literature has been absorbed by courts in a crude form, and that juries are now routinely invited to treat all identification evidence as unreliable, including the kind collected properly and expressed with immediate certainty. The effect, he suggests, is to transfer the cost of past errors onto present victims, whose accurate identifications are discounted along with the inaccurate ones. Petra Halloran, who studies how juries reason, replies that the fear is largely theoretical: in the studies she has run, jurors who hear expert evidence about memory do not reject identifications wholesale but become more responsive to how the identification was obtained, which is exactly what is wanted.

My own reading of the evidence is that Kwan has identified a real risk and drawn the wrong conclusion from it. The remedy for a crude message is a better message, not a return to silence. What the research supports is neither faith nor scepticism but a question, and the question is procedural rather than psychological: not how good this witness's memory is, which nobody can establish, but what was done to it between the event and the courtroom. An identification obtained blind, from a fair line-up, with the witness's own words about certainty recorded on the day, is strong evidence. The same witness, equally sincere and now equally certain after six months of interviews, may be repeating the answer they were given rather than the one they had.