“The botany in the Louvre version is perfect, showing plants that would have thrived in a moist, dark grotto,” says Ann Pizzorusso, a geologist and Renaissance art historian. “But the plants in the London version are inaccurate. Some don’t exist in nature, and others portray flowers with the wrong number of petals.”

She concludes: “It seems unlikely the same person could have portrayed rock formations so accurately in the Louvre work and so incongruously in the National Gallery one – especially considering Leonardo’s faithfulness to nature. There is absolutely nothing in his body of work that is not true to nature.

Her conclusions are supported by John Grimshaw, a leading horticulturalist, who is struck by the realism of the Louvre painting, unlike the National Gallery version. In the French painting, he can easily identify iris, polemonium and aquilegia. He says: “There’s a very recognisable iris, a Jacob’s Ladder, a nice little palm tree, all sorts of well-observed bits of vegetation there – and proper plants.”

This argument against the NG painting sounds quite plausible.

It might be condition, but the Louvre Christ child's profile is better too - the hint that the face is turned away a fraction, exactly the sort of thing that rarely translates into copies.