IELTS Academic Reading Practice Test 60

In the spring of 1974, several farmers were digging a well near the city of Xi'an, in the northwest Chinese province of Shaanxi, when their tools struck a hard underground deposit. The deposit turned out to be fragments of pottery and bronze weapons, and within months Chinese archaeologists had uncovered part of a vast subterranean army. The discovery was entirely accidental, yet it revealed one of the most remarkable monuments of the ancient world: thousands of life-sized clay soldiers, horses and chariots arranged in battle order to guard the tomb of China's first emperor.

That emperor was Qin Shi Huang, who united the warring Chinese states in 221 BC and died in 210 BC. Work on his mausoleum complex near Mount Li, about thirty kilometres east of modern Xi'an, had begun many years earlier, when he became king of the state of Qin at the age of thirteen, and it continued on a grand scale throughout his reign. The historian Sima Qian, writing more than a century later, described a tomb filled with models of palaces, rivers of mercury and numerous treasures, but his account says nothing about a terracotta army, which suggests that all memory of the clay soldiers had vanished after the emperor's death.

Ancient sources claim that around 700,000 workers were involved in building the mausoleum. The soldiers were not modelled as unique sculptures but manufactured in workshops by an early assembly-line system. Clay was pressed into moulds to produce standard legs, arms, torsos and heads, and these parts were joined while the clay was still wet. The joined bodies were then fired in large kilns, after which a layer of lacquer was applied as a smooth base for painting. The process allowed thousands of near-identical figures to be produced with speed.

Despite this standardised production, the army does not appear monotonous. The warriors stand roughly 1.8 metres tall, but they vary in rank, dress and expression. Most have their hair tied in a topknot, while some wear caps or helmets, and officers are distinguished by ribbons and more elaborate armour. Soldiers are shown as archers, cavalrymen, charioteers and armoured infantry. The faces, although formed from a limited number of mould types, were finished by hand, with different eyebrows, beards and facial shapes, giving each figure a slightly individual character.

The largest excavation, Pit 1, is a rectangular trench about 230 metres long that contains an estimated 6,000 soldiers and horses, together with wooden chariots that have long since decayed. Pit 2 holds a smaller but more varied force of cavalry, archers and infantry, while Pit 3 is thought to represent a command post, with high-ranking officers and a war chariot. A fourth pit was opened but left empty, perhaps because the emperor died before the entire complex was completed.

When new, the terracotta soldiers were not the grey figures seen by tourists today. They were covered in lacquer and painted in bright colours, including pink faces, black hair and colourful tunics. Unfortunately, the paint has proved extraordinarily fragile. When a painted figure is removed from the damp soil, the lacquer underneath the pigment dries and curls, and the paint flakes away, sometimes within minutes. Early excavations were unable to prevent this, and many figures therefore lost their original decoration. Some areas of the pits have been deliberately left unexcavated until better preservation techniques can be developed.

The soldiers were equipped with real bronze weapons rather than ceremonial copies. Tens of thousands of arrowheads, swords, spears and halberds have been found in the pits, and many blades remain sharp after more than two thousand years. This preservation was long attributed to a deliberate coating of chromium on the weapons, but recent analysis suggests that the protective layer may have formed naturally through the chemistry of the burial environment. Whatever the explanation, the weapons reveal advanced metalworking skills and a high degree of standardisation.

In 1980, a separate pit to the west of the main mound yielded two painted bronze chariots, each drawn by four bronze horses and about half life-size. These vehicles, designed for the emperor's journeys in the afterlife, are among the finest examples of ancient Chinese metalwork ever found. They also show that the terracotta soldiers were only one part of a much larger underground complex, most of which remains unexplored.

The Terracotta Army has transformed historians' understanding of the Qin state. The standardised components, the stamped or incised names found on some parts, and the careful organisation of the ranks all reflect the centralised control that allowed Qin to unify China and lay the foundations of its bureaucratic empire. The site was inscribed on the UNESCO World Heritage List in 1987 and now draws millions of visitors every year from around the world.

The emperor's burial chamber itself has never been opened, and no one knows for certain what it contains. Modern soil tests have found unusually high levels of mercury in the mound, which appears to support Sima Qian's description of rivers of mercury, but excavation of the tomb would be extremely hazardous and could destroy fragile materials. For the time being, therefore, the chamber remains sealed, while archaeologists continue to investigate the surrounding pits with ever more careful methods.

The idea that light could be turned directly into electricity was first demonstrated in 1839 by the French physicist Edmond Becquerel, who observed that certain materials produced a small current when exposed to light. For more than a century the photovoltaic effect remained a laboratory curiosity, because early selenium cells converted less than one per cent of the light they received. It was not until the middle of the twentieth century that scientists learned how to make a practical device.

The breakthrough came in 1954 at Bell Laboratories in the United States, where researchers produced a silicon solar cell with an efficiency of about six per cent. Although this figure seems modest by today's standards, it was far higher than anything achieved before and showed that solar power could be used to run electrical equipment. The space industry quickly adopted the cells: in 1958 the Vanguard 1 satellite carried a small panel that powered its radio transmitter for years, helping to convince engineers that solar energy was reliable enough for long missions.

A silicon solar cell works because of a junction between two layers of silicon that have been treated, or doped, with different impurities. One layer has extra electrons, while the other has missing electrons, and the meeting point between them creates an electric field. When a photon of sunlight strikes the cell, it can knock an electron loose; the internal field then pushes this electron out of the cell and through an external circuit, producing a direct current. The cell has no moving parts, requires no fuel and produces no emissions while it operates.

Silicon now dominates the solar industry, accounting for more than 90 per cent of panels sold worldwide. It is abundant, non-toxic and well understood from decades of use in electronics. To make a panel, extremely pure silicon is melted and formed into ingots, which are then sliced into thin wafers, doped and connected with metal contacts. Monocrystalline cells, cut from a single crystal, are generally the most efficient; polycrystalline cells, cast from many crystals, are slightly less efficient but cheaper to produce.

Nevertheless, silicon cells have practical limits. They are rigid, heavy and must be protected by glass, which restricts where they can be used. Thin-film solar cells, made from materials such as cadmium telluride or copper indium gallium selenide, avoid some of these problems. They can be deposited onto glass, plastic or metal in layers only a few micrometres thick, making panels lighter, flexible and less dependent on bulky frames. Their conversion efficiencies are lower than those of silicon, but they remain valuable for large ground-mounted installations and for applications where weight is critical.

A newer class of materials known as perovskites has attracted intense attention since about 2009. Perovskite cells can be manufactured from inexpensive chemicals using low-temperature solution processes, in which the light-absorbing film is essentially printed or coated onto a surface. Their efficiency in laboratory tests has raced upward faster than that of any previous photovoltaic technology, and they can already match or exceed thin-film cells. However, they degrade quickly when exposed to moisture and heat, and many of the best compositions contain lead, which raises concerns about safety and disposal.

Because perovskite cells absorb different parts of the solar spectrum from silicon, researchers can stack the two materials in a tandem cell. The upper perovskite layer captures high-energy photons while the lower silicon layer captures lower-energy light, allowing the combined device to convert more of the sun's energy than either material could on its own. Laboratory tandems have now passed the 33 per cent mark, well above the practical limits of single-junction silicon, and commercial production is widely expected within the next few years.

The falling cost of panels has transformed solar deployment. Since the late 1970s the price per watt of photovoltaic modules has dropped by more than 99 per cent, driven by larger factories, learning curves and economies of scale. Solar farms can now generate electricity more cheaply than new coal or gas plants in many sunny regions. However, solar output varies with time and weather, so grids also require energy storage, interconnection and flexible backup, and the manufacturing process consumes large amounts of energy and water.

The cell itself is only one part of a photovoltaic system. An inverter is needed to convert the direct current into alternating current for the grid or for household appliances, and mounting structures, cabling and transformers add what engineers call the balance of system. These components account for roughly half of the total cost of a solar installation. The performance of a panel also depends on temperature, shading and dust, so real-world yields are lower than laboratory results.

Researchers continue to pursue higher efficiencies, longer lifetimes and easier recycling. Silicon cells now reach about 26 per cent in the laboratory, while commercial modules typically deliver between 20 and 23 per cent. If tandem and thin-film devices become more stable, they may replace silicon in many markets; if not, improved silicon will probably remain the workhorse of the industry for decades. In either case, the fundamental challenge is no longer whether solar cells can work, but how quickly and cleanly they can be manufactured and deployed.

Facial recognition systems have moved from science fiction into everyday life with remarkable speed. They unlock smartphones, verify passengers at airport gates and help social-media companies sort photographs. Police around the world use them to identify suspects, while retailers use them to watch shoppers. Facial recognition promises efficiency and security, but it also raises questions about privacy, bias and power. These questions have become urgent because such systems are no longer a niche tool: they can operate continuously, over large areas and often without the knowledge of the people being watched.

Modern facial recognition works by measuring dozens or hundreds of geometric features of a face, such as the distance between the eyes and the shape of the jaw, and turning them into a mathematical template. When a camera captures a face, the resulting template is compared with millions of stored templates in a database. Advances in deep neural networks, combined with the enormous supply of photographs available online, have made facial recognition systems dramatically more accurate over the past decade. In ideal conditions, the best systems now match faces with error rates of a fraction of one per cent.

That accuracy, however, is not evenly distributed. Studies by the US National Institute of Standards and Technology and independent researchers have repeatedly found that many facial recognition systems are less reliable when identifying women, younger people, and people with darker skin tones. The reasons are complex: training datasets often contain fewer images of minority groups, and the features the software relies on may reflect the people on whom it was trained. A false match, in which the software wrongly links an innocent person to a crime scene, is therefore more likely for some groups than for others.

Privacy advocates argue that the face is among the most sensitive types of personal data the human body offers. Unlike a password, a face cannot be changed; unlike a fingerprint, a face can be captured at a distance without a person's consent or awareness. A face template can be stored indefinitely, shared between agencies and linked to other data, such as travel records, purchases and social media activity. Opponents fear that once such systems exist, they will be used for far more than their original purpose.

Another less discussed risk is function creep. A system introduced to catch terrorists can later be used to enforce minor city rules, watch employee attendance or score customer behaviour. Once the cameras and databases are in place, each new use seems only a small extension of the original purpose, and the total amount of surveillance can soon exceed what the public ever debated or approved.

The most intense concern is mass surveillance. In public spaces, live facial recognition makes it possible to track the movements of ordinary people, note who attends a peaceful protest and reconstruct a person's daily route. Even with no misuse, such surveillance can have a chilling effect: people may avoid certain neighbourhoods or hold back political views if they believe they are being identified. In some countries, facial recognition systems are already linked to identity cards and billions of surveillance cameras, creating the technical basis for a mass surveillance system that would have been impossible a generation ago.

Misidentifications are not hypothetical. In the United States, several people have been arrested after facial recognition systems produced false matches. In one widely reported case, a man in Detroit was detained after the software wrongly matched him to a shoplifting suspect; the charges were later dropped. Similar errors have occurred in other places. Such cases illustrate a practical danger: police may treat an automated match as strong evidence, even though a match score does not prove identity on its own.

Governments have responded in different ways. In 2019, San Francisco became the first American city to bar city agencies from using facial recognition, and several other cities followed. The European Union's Artificial Intelligence Act, which came into force in 2024, takes a more graduated approach. The Act prohibits police from using live facial recognition in publicly accessible spaces, subject to narrow exceptions for terrorism, serious crime and searches for missing persons. The Act also sets transparency and accuracy rules for many private-sector uses of facial recognition.

Supporters point out that the technology can be a powerful police tool. It has helped identify victims of disasters, find missing children and catch people accused of serious crimes. In these situations, supporters argue, a complete ban would be disproportionate and would slow legitimate police work. Many commercial uses, such as automatic identity checks at borders, also bring clear convenience and can reduce queues and fraud.

Nevertheless, the international picture remains inconsistent. Some countries have strong national data-protection laws; others have none. A few countries use the systems extensively for social control, while some liberal democracies are only beginning to debate it. This patchwork creates risks of its own, because data collected in one country under one set of rules can sometimes be processed in another country with far weaker protections.

In my view, the correct response is not to ban facial recognition outright, nor to allow these systems to spread without limits. A sensible framework would require a warrant or similar judicial approval before live facial recognition is used for public surveillance, set minimum accuracy standards, and insist on independent audits and human review before any automated match is used to detain or charge someone. It would also give people the right to know when their faces have been scanned and to challenge what happens to their face data. The technology will almost certainly improve, but so too must the rules that govern it.