AFor more than a century, scientists have been visiting mountain glaciers to measure how far they have retreated. The earliest records were simple: a painted rock marked the point the ice had reached, and a returning visitor noted how far it had moved. By the 1950s, however, researchers had begun to realise that glaciers held a far richer archive within the ice itself.

BEach winter’s snow is gradually compressed into a new layer of ice, trapping tiny bubbles of air. According to the Swiss glaciologist Anna Keller, these bubbles are “time capsules”: by analysing them, her team has reconstructed the composition of the atmosphere over the past eight hundred years. Keller stresses that the method only works where the ice has never melted, which rules out many lower glaciers.

COther researchers have focused on what the ice is losing. Since 2005, the Chilean geographer Tomás Rivas has been photographing the same twelve glaciers from fixed points every summer. His images show that several of them had already lost more than half their length before his project began. Rivas argues that photographs persuade the public in a way that statistics rarely do.

DSatellites have transformed the field. Dr Mei Lin, who has been working with satellite data since the early 2000s, estimates that the world’s glaciers lost around 270 billion tonnes of ice per year during the last two decades. However, she warns that satellite measurements must still be checked against observations on the ground.

ELooking ahead, the picture is sobering. If current trends continue, Lin predicts, many glaciers in the Alps will have disappeared by 2100. Keller, by contrast, believes the most urgent task is scientific rather than political: researchers must keep drilling the remaining high-altitude ice for as long as it survives, because once it has melted, the record it contains will be lost for ever.

AThe university lecture has survived for almost nine hundred years. Long before printed books had become affordable, a lecturer read aloud from a precious manuscript while students copied every word. Critics now argue that, in an age when every text is available online, this ritual has outlived its purpose. In my view, they are only half right.

BIt is true that the traditional lecture has rarely been an efficient way to transmit information. Studies have repeatedly shown that students who have been listening passively for an hour remember surprisingly little. Universities that have replaced some lectures with active seminars have generally seen exam results improve, and I find this evidence hard to dismiss.

CYet a good lecture was never simply about information. At its best, it shows students how an expert thinks: how she weighs evidence, admits uncertainty and changes her mind. A recorded video can capture this too, of course, but I suspect that few students who watch lectures at double speed on their laptops experience it in the same way.

DThe real problem, I would argue, is not the lecture itself but its length and frequency. Many students are still sitting through twenty hours of lectures a week, which leaves little time for the reading and discussion that deep learning requires. Some universities have already reduced this to eight or ten hours, and I believe others should follow.

EBy the end of this decade, many institutions will have redesigned their courses around online materials. That is no bad thing. But if they abolish the lecture altogether, they will have lost something that no technology has yet replaced: the experience of thinking alongside a skilled mind in real time.