Matisse's Dance (I) Looks Spontaneous. The Bronzes in Front of It Say Otherwise.
The big blue canvas reads as a burst of energy. Five figures, a green mound, a flat blue sky, almost no detail, the whole thing apparently thrown down in an afternoon. Visitors treat it that way. They stand back, take the picture, and move on with the impression that Matisse was a man who worked fast and got lucky.

Then you turn around and there are the bronze heads on the plinth, the same subject remade over and over, each one further from the last. That row is the correction to the impression.
The painting is a working document
Dance (I) dates from 1909 and was made as a full-scale compositional study, part of Matisse’s response to a commission from the Russian collector Sergei Shchukin for the staircase of his Moscow house. The version that eventually went to Moscow is a different, hotter painting, with the flesh pushed to a hard red and the figures more strained. What hangs in New York is the pale, cooler first pass, the one where he was still deciding how many figures, where the break in the circle goes, how much ground to give the sky.
Look at the ring of dancers and the break is the whole composition. Two hands that do not quite meet at the left. Everything else in the picture is calm, and that one gap is where the tension sits.
The heads are the same argument in bronze
The sculptures are the Jeannette series, made across roughly 1910 to 1913, all of them heads of the same young woman. The first is close to a conventional portrait. By the last, the nose has become a ridge, the hair has turned into a mass balanced improbably off the back of the skull, and one eye has swollen out of any relation to the other.
They are not five attempts to get it right. They are five states of the same object, each made by working over a cast of the one before, and Matisse kept all of them. That is the point. He wanted the sequence visible rather than the final answer alone.
Why the pairing matters
Put the study and the series in one room and the flatness of the painting stops looking like carelessness. The pale figures are pale because Matisse was testing how little he could put down and still get movement. The bronzes are lumpy and violent because he was testing how far a face could be pulled before it stopped being a face.
The two experiments run on the same schedule and ask the same question, which is how much can go before the thing collapses.
Fast-looking work, slowly made.