102. Read the following passage, then answer the question below.

Many of the structures built by the Romans have lasted nearly two thousand years, while modern concrete commonly begins to crumble after fifty to a hundred. The harbor works of Roman cities, constantly battered by seawater, are especially striking. Engineers long assumed that Roman builders simply used more material or better stone. Closer analysis has revealed something more interesting about the recipe.

Roman concrete combined lime with volcanic ash from the region around Pozzuoli, near Naples. The ash reacts chemically with lime and water to form a hard binder, and in seawater the reaction does not stop. Researchers who examined samples from ancient breakwaters found crystals of a rare mineral, called aluminous tobermorite, that grew slowly inside the concrete over centuries as seawater seeped through tiny cracks. Rather than weakening the structure, the seawater helped to knit it together. Modern concrete, by contrast, is made with Portland cement, which sets quickly and is strong, but seawater and the chlorides in it attack the steel bars embedded inside, causing rust that swells and cracks the material from within.

A second discovery involved small white chunks, called lime clasts, visible in Roman concrete. For years they were dismissed as signs of careless mixing. A recent study proposed instead that the clasts are the secret of self-repair: when a crack opens and water enters, the exposed lime dissolves and recrystallizes, sealing the gap before it can grow. This would explain why some ancient structures appear to have healed themselves.

None of this means modern builders should return to Roman recipes. Volcanic ash is not available everywhere, and Roman concrete hardened slowly, a drawback for projects with deadlines. But engineers are studying ways to adapt its principles, because concrete production is a major source of carbon dioxide emissions, and a longer-lasting material would reduce the need to rebuild.

What did researchers find in samples from ancient breakwaters?