Antoniadi dissolves the canals at Meudon
Four nights before opposition, the 83-centimetre Meudon refractor showed Antoniadi a Mars crowded with irregular detail and not one straight line.
On 20 September 1909, four nights before a favourable opposition, Eugène Antoniadi got his first night on the Grande Lunette at Meudon — an 83-centimetre refractor, the largest in Europe. The air was steadier than anything he would ever see again through that instrument. Mars was covered in detail, and none of it was geometric. There were no canals at all.
His conclusion was blunt. The planet, he wrote, "appeared covered with a vast and incredible amount of detail held steadily, all natural and logical, irregular and chequered, from which geometry was conspicuous by its complete absence." Where Lowell's maps ran a clean line, Antoniadi saw a broken chain of unrelated dark patches and shadings that the eye, at lower resolution, would happily connect. A month later, on 19 October, he drew Mars again in ordinary seeing — and several canals came back, one pair of them neatly doubled. He kept that drawing as evidence too. The canal illusion, he said, was a product of poor seeing.
The reason this counted is that Antoniadi was not an outsider taking a shot. He was Greek-born, had worked as Camille Flammarion's assistant at Juvisy, directed the British Astronomical Association's Mars Section, and had drawn canals himself in the 1890s. He was a convert testifying against his own earlier work, with the best aperture available and the best seeing of his life. Six years earlier, Joseph Evans and Edward Maunder had shown in Monthly Notices that schoolboys shown discs of scattered dots would reliably draw lines between them. Meudon supplied the observation; Greenwich had already supplied the psychology.
It should have ended there. It did not. Percival Lowell went on publishing until his death in 1916, and the canals survived in popular atlases, encyclopaedia entries and a good deal of schoolroom astronomy for another half-century. Observers were still reporting them in the 1950s. The idea was cheap to hold and expensive to disprove: you could always argue that the seeing had been poor, or the observer's eye untrained, or the canals seasonal.
What finally killed them was not argument but hardware. Mariner 4's twenty-one frames in July 1965 showed craters and no network; Mariner 9 mapped the whole planet in 1971–72 and found volcanoes, a canyon system 4,000 kilometres long, and dry channels carved by water that had stopped flowing billions of years earlier. Real channels, as it turned out — just not where Schiaparelli drew them, and not made by anyone. Antoniadi's own summary, La Planète Mars, published in 1930, remained the standard reference until spacecraft replaced it.