IELTS Academic Reading Practice Test 156

Ultrasound imaging is now so ordinary a part of medicine that few patients stop to wonder how it works. A pregnant woman is shown a moving picture of her unborn child; a surgeon uses sound to guide a needle into a tumour. None of these pictures is made with light. They are built from sound at frequencies far above the range of human hearing, and the principles behind them were established long before anyone imagined using them to look inside a living body.

The foundation was laid in 1880, when two French physicists, Jacques and Pierre Curie, discovered the piezoelectric effect. Certain crystals, quartz among them, produce an electric charge when they are squeezed; the same materials also do the reverse, changing shape when a voltage is applied across them. This second property is what makes medical ultrasound possible. A crystal driven by a rapidly alternating current can be made to vibrate millions of times a second, launching a beam of high-frequency sound into whatever lies in front of it.

The first serious use of such sound was military. During the First World War the French physicist Paul Langevin worked on a method of detecting submarines by sending pulses of high-frequency sound through water and listening for the echo returned by a submerged hull; quartz plates generated and received the vibrations. The war ended before the technique was perfected, but the principle was sound, and by the 1930s echo sounding was being used to measure the depth of the sea. The technique became known as sonar, from sound navigation and ranging.

Applying the idea to the human body took several more years. In the early 1940s an Austrian neurologist, Karl Dussik, tried to use ultrasound to examine the brain, hoping that tumours would show up in the reflected sound. In the late 1940s George Ludwig, working in the United States, used pulses of sound to search for gallstones. Then, in 1953, two researchers in Lund, Sweden, Inge Edler and Hellmuth Hertz, recorded echoes from a beating heart and so founded echocardiography. These early machines produced little more than a line traced across a screen.

The person usually credited with turning the technique into a practical tool for the abdomen and for pregnancy is Ian Donald, an obstetrician in Glasgow. Working with Tom Brown, an engineer at a local firm that made ultrasonic instruments for industry, he built a contact scanner and used it, from the late 1950s, to examine abdominal masses and pregnant women. His landmark paper appeared in The Lancet in 1958. Donald's achievement depended on borrowed expertise: the components, and much of the understanding of how sound behaves in materials, came from shipbuilding and engineering rather than from medicine.

A modern scanner works on the same principle as Langevin's apparatus, refined almost beyond recognition. A transducer, usually a ceramic material such as lead zirconate titanate, emits pulses of sound lasting a fraction of a microsecond. The same device then waits for the echoes. Because the speed of sound in soft tissue is roughly 1,540 metres a second, the delay before an echo returns reveals how deep the reflecting surface lies, while the strength of the echo determines how bright the corresponding point appears. A beam swept rapidly across the body builds up a two-dimensional image.

Two refinements transformed the usefulness of the technique. The first was the Doppler effect, described by the Austrian physicist Christian Doppler in 1842: sound reflected from a moving object returns at a slightly different frequency, which allows the speed and direction of blood to be measured. The second was the shift from still pictures to moving ones. Real-time scanners, developed during the 1960s and 1970s, produced images fast enough to show a heart beating or a fetus moving, and grey-scale displays, introduced in the same period, replaced harsh black-and-white echoes with a range of tones that trained eyes could interpret.

Ultrasound's great advantage is that it uses no ionising radiation, the kind associated with X-rays and computed tomography, which can damage DNA. This is why it is the imaging method of choice in pregnancy, and increasingly in children. It is not, however, entirely without risk: at high intensities sound can heat tissue and, where tiny bubbles of gas are present, cause them to collapse violently, a phenomenon known as cavitation. Guidelines therefore urge operators to use the lowest power that will produce an adequate image, and to keep examinations brief.

Sound does not travel well through bone or through gas, which is why the lungs and the interior of the skull are difficult to examine and why a gel must be spread on the skin to exclude air. The quality of an image also depends heavily on the skill of the operator, who must choose the angle of the probe and adjust the settings. Nor can a single frequency do everything: higher frequencies give finer detail but are absorbed more quickly, so deep structures must be imaged with longer waves and poorer resolution.

Recent years have brought smaller and cheaper machines. Probes that plug into a mobile phone now cost a fraction of the price of a hospital scanner, and have taken ultrasound into ambulances, rural clinics and emergency departments. Three-dimensional images, reconstructed from many thin slices, have become routine in some specialties. Even so, the technology has not replaced older methods; it is used alongside them, and its future may lie as much in better interpretation, including by machine learning, as in better hardware.

A single advanced robot is an expensive and fragile thing. It carries cameras, processors and motors, and it must be programmed to cope with every situation it might meet — an impossible demand in a collapsed building or on a field whose shape nobody knows in advance. Swarm robotics takes the opposite approach. Instead of one capable machine, it uses dozens or thousands of simple ones, each following a small set of rules and communicating only with its close neighbours. The result, engineers hope, will be groups that are cheaper, more adaptable and far more robust than any single device could be.

The idea is borrowed from biology. Ants, termites, honeybees and flocks of birds achieve remarkable feats of coordination although no individual understands the whole task. A termite does not know the plan of the mound it is helping to build; it responds to the shape of the material around it and to the scent of its nestmates. From thousands of such local responses a structure emerges that none of the builders designed. Biologists call this self-organisation, and the crucial point is that the intelligence of the colony belongs to the group rather than to any of its members. A colony can therefore survive the loss of most of its workers.

Many insect societies coordinate through stigmergy, a term coined by the French biologist Pierre-Paul Grassé in the 1950s to describe how termites build. The word means working through the environment: an insect changes its surroundings, and the change prompts the next insect to act. Ants, for example, leave trails of pheromone as they search for food. A shorter route is walked more quickly and therefore marked more often, so the scent becomes stronger there; later foragers are drawn to the stronger trail, reinforcing it further. No ant measures distance or compares routes, yet the colony converges on the shortest path.

Engineers have translated these observations into design principles. Each robot should sense only its immediate surroundings; control should be distributed, with no central computer issuing orders; and communication should be local and limited. Such systems are said to be scalable, because adding another hundred units changes nothing in the software. They are also robust: if a tenth of the robots fail or are destroyed, the group usually continues to function, whereas the failure of a single central controller stops everything at once. These properties matter most in places where repair is impossible or where a human being could not safely go.

Building such a swarm is not simply a matter of making many copies of one robot. In 2014 a team at Harvard University reported one of the largest demonstrations so far, in which 1,024 small robots, each about the size of a coin and able to communicate only with neighbours a short distance away, arranged themselves into a star and other shapes. The robots were deliberately simple and inexpensive; much of the research effort went into the algorithms that allowed patterns to form from thousands of local decisions. The experiment was a proof of principle rather than a practical tool.

Potential applications are numerous. A swarm of small flying robots might map a disaster zone, entering gaps too narrow for a person and too dangerous for a rescue dog. Groups of aquatic robots could track pollution plumes or the temperature of a lake more thoroughly than a few expensive buoys. In agriculture, teams of ground robots might weed fields or monitor individual plants. In medicine, researchers are exploring microscopic particles that could carry drugs to a tumour or help to clear a blocked artery, although such work is at an early stage. None of these systems is yet in routine use, and most exist only as laboratory prototypes.

Reality, however, remains some way behind the advertisements. Most demonstrations take place indoors, with careful positioning systems standing in for the satellite navigation that would be unavailable inside a building or under water. Communication is a persistent constraint: radio links share a limited bandwidth, and the more units there are, the more they interfere with one another. Robots also collide, run out of power and physically obstruct each other, problems that are hard to model in advance. And a swarm is difficult to verify, since its behaviour depends on interactions that no single line of code describes.

Researchers are now working on mixed teams, in which a flying robot acts as a parent that carries and releases small ground robots, or in which a few capable machines guide many simple ones. Others are studying how a human operator should give instructions to a group of hundreds, since issuing orders to each robot is impossible and a vague command may produce an unpredictable result. Cheap manufacturing matters too: a swarm is only useful if the loss of individual units is economically trivial, which means the units must be simple enough to be produced in large numbers.

The appeal of swarm robotics is therefore as much philosophical as technical. It asks engineers to give up the ideal of complete control and to accept that useful behaviour can emerge from simple rules and local contact — the strategy that insects have been refining for tens of millions of years, and one that may prove more durable than any single machine we could build.

In less than fifteen years the electric car has gone from a curiosity to a common sight on the roads of many countries. Sales have risen steeply, and almost every large manufacturer now offers at least one battery-powered model. That transformation owes something to engineering and something to policy. Governments have offered buyers grants, tax exemptions, free parking and access to bus lanes, and have spent public money on charging points. The question I want to examine here is whether this was money well spent — and, more urgently, whether it should still be spent now.

When the modern subsidy programmes began, in the early 2010s, the case for them was strong. A battery pack then cost around a thousand dollars for every kilowatt-hour of capacity; by the early 2020s the figure was roughly a tenth of that. Nobody knew how quickly such a fall would come, and a grant that looks wasteful in hindsight may have been necessary to start the process. Economists call the mechanism learning by doing: the more cells a factory makes, the better it becomes at making them. Infant industries may need temporary help of this kind. There was also a chicken-and-egg problem, since few people would buy an electric car without chargers and few firms would build chargers before there were cars to use them.

Norway shows what determined policy can achieve. Norwegian governments exempted battery-electric cars from the purchase taxes and value-added tax that make ordinary cars so expensive there, and allowed them into bus lanes and through tolls without charge. In recent years such cars have accounted for the great majority of new sales, and the average emissions of the country's new cars have fallen sharply. Yet even Norway has begun to withdraw the most generous perks, as the number of cars has grown and the cost to the public purse with it.

The objection most often raised is distributional. A new car is a luxury purchase, and the households that buy one are richer than average. Purchase grants therefore go disproportionately to people who would have bought a new car anyway, and who need the money least. A scheme that pays several thousand pounds towards a vehicle costing forty thousand is, in effect, a transfer from the general taxpayer to the affluent. Analyses of grant programmes in several countries have suggested that a substantial share of the money went to buyers who would have purchased the car regardless.

There is a second, related criticism. If the purpose of a subsidy is to reduce carbon emissions, it should be judged by the cost of each tonne avoided. Some evaluations of purchase incentives have put that cost at several hundred dollars a tonne, many times the price of a permit in the European Union's emissions trading scheme. Cheaper abatement is available elsewhere: insulating lofts, replacing coal-fired power stations with gas, or simply raising the duty on petrol and diesel.

Supporters reply that the comparison is unfair, because electric cars deliver benefits that a carbon price cannot capture: cleaner air in cities, less noise, and a fuel that is produced at home rather than imported. They also point out that the emissions of an electric car depend heavily on how the electricity is generated. Building a battery is energy-intensive, and the factories that make cells consume a great deal of power, much of it generated from fossil fuels. In countries where coal still supplies most electricity, the lifetime advantage over a petrol car is smaller, though most studies still find that there is one.

Whatever the merits, subsidies are hard to remove. Once buyers have arranged their finances around a grant, its withdrawal is treated as a tax increase, and the government that removes it takes the blame. Manufacturers, whose business plans have been built on incentives, argue that ending support will cost jobs. And a scheme that is closed at short notice can cause a rush of purchases before the deadline and a slump afterwards, which makes ministers reluctant to close one at all.

What, then, should governments do? My own view is that the case for subsidy was strongest at the beginning and has weakened as the market has matured. Help is now best directed where it changes behaviour, not where it merely rewards a purchase that would have happened anyway. That means support for cheaper models and for second-hand electric cars, rather than for expensive new ones. It also means public money for charging points in places where no company would build them unaided. Charging infrastructure is closer to a true public good than a discount on a luxury purchase, because the benefit of a charger is shared by everyone who drives past it. Money spent on other measures, such as raising the duty on petrol and diesel, would cut emissions faster for the same outlay. And any scheme that survives should carry a clear end date and be reviewed each year, rather than becoming permanent by default.

None of this means that the policy failed. The electric car exists as a mass-market product partly because governments paid, in effect, to bring it down the learning curve, and the fall in battery prices has been extraordinary. But a policy that was justified as a temporary intervention should be allowed to end. The most persuasive argument for phasing out grants is not that they have failed; it is that they have succeeded.