‘Matisse: In Search of True Painting” is a smallish but superb show at the Metropolitan Museum of Art. It focuses on pairs and series of related paintings, and the sheer loveliness of its best pieces resounds through the huge building and out onto Fifth Avenue. But it is sad that this small-scale, dazzling-masterpieces-only approach wasn’t extended to Henri Matisse’s late cutouts—which also occur in pairs and in series. Unfortunately, this exhibit, for all its beauty, shows off one of the deepest, darkest prejudices in the arts today: that an artist’s earlier works are bound to be his best.

It’s true that they are usually his most unexpected and “disruptive.” And most art historians have long since decided that an artist’s main goal is to kick people in the teeth. This strange idea is only to be expected, though, given that art historians are mostly intellectuals who admire theories and ideas rather than truth or beauty per se. It is natural that intellectuals should reduce art to political or intellectual terms, just as it is natural for a horse to evaluate a lawn in terms of taste and hoof--comfort, not aesthetic effect. Intellectuals do not see the world in aesthetic terms, and so they are always bound to have trouble with art.

But their tendency to exalt early works over late ones has damaged our ability to appreciate Matisse. His cutouts, all late, are among his greatest pieces. And they are far too rarely seen in public. The Met itself owns an especially beautiful one (Fleurs de Neige, 1951) that it rarely deigns to display. It’s time to renew our understanding of great artists’ tendency to do their best work late. Towards the end of his career, a major artist has so mastered his technique, and so dominates his materials, that he speaks to us directly in color or music or line or language. Instead of looking at him through the window (however brilliantly polished) of technique, we see him face to face. It might almost be said that, the greater the artist, the more likely that his greatest work will come towards the end of his career. Picasso and Matisse, rivals and friends, make the point perfectly: Picasso’s best work came towards the start of his work as a painter, Matisse’s towards the end of his.

The Met’s show is wonderful, even though, strictly on its own terms, it misses the point. It shows us several of Matisse’s greatest paintings, surrounded, in some cases, by groups of studies, and by black and white photos of the painting in others—photos that record the many different states of the canvas along its way to completion. The photos reminded Matisse of where he had been so he could retrace his steps if he wanted to. They also reminded the public that to paint a Matisse was harder than it looked. His paintings look casual; their profoundly original color harmonies and the balanced grace of the drawing seem as unstudied as a rainbow. But Matisse worked hard. Often he reworked and repainted a canvas dozens of times before he was satisfied.

The show and the catalogue neatly explain why the intermediate--state photos exist. But the deeper question is, obviously, why do the intermediate states themselves exist? Why didn’t Matisse work out the problems first, in sketches and drawings? Sometimes he did; but why not always? The answer is that Matisse is a color artist, one of the very greatest. He ranks alongside the 12th-century stained-glass masters of St. Denis and Chartres—with the 8th- and 9th-century creators of the gospel books of Lindisfarne and of Kells, respectively, and with Fra Angelico and Titian. He struggled with drawing and eventually achieved a lovely, lyrical, wholly original way of doing it; but color was his native language.

Now, a painter who dreams up a new color chord naturally needs to test it out and adjust it. But there is no way to make a “color study.” The colors in a painting make their effect depending on the way they are put together, and the texture and relative opacity and sheer area of each separate color region. Which means that the only way to make a study of the colors for a painting is to make the painting. This means, in turn, that Matisse often worked his way through dozens of versions of one painting. Each version is, in effect, a study for the next. Of course, he could have made a flock of separate paintings instead of reworking one, but it is much harder to take up where you left off—when you must start by reproducing (most of) an earlier painting—than if you are merely drawing.

The catalog and curators don’t tell us much about Matisse’s unique color genius, but the paintings tell us everything. He is famous for his ability to use black as a hue, but his greatest virtuosity is reserved for green. No color has a greater range of separable nuances; and the hardest of all colors to use isn’t black but plain, bright, kindergarten green. Which is an oddly shocking color: You can buy large assortments of pastels from many leading makers and discover not one plain, bright green in any of them. Matisse is one of the few painters in history who uses this color effectively. Like van Gogh, Degas, Stuart Davis, and de Kooning, he prefers opaque color. In his greatest paintings (nearly all of them late), the colors tend to be laid on with hardly any variation within a color zone. These habits of opacity and uniform color-zones made it natural for him to work in cutouts, cut sheets of uniformly colored paper pasted together and painted in gouache (opaque watercolor).

Often his best paintings have separate zones of internally harmonious color pulled together by a louder, more dissonant color triad of black plus two dissonant tones. Interior with an Egyptian Curtain (1948) has a pink/peach/amber and a lemon/soft green/soft blue zone set off by a trumpeting curtain in black, crimson, and bright green—a color nightmare in itself, but gorgeous when it is braced up by beds of softer color. The famous Large Blue Dress (1937) has slate-blue, white, soft green, and pink set off by a brilliant chord of black, lemon, and warmish red. Bowl of Apples on a Table (1916) has a harmonious group of ambers, oranges, pinks, and white-yellows lit up by a brilliant blare of verdigris, pink, and black. In the exhibit, it is paired with a similar painting called Apples (1916). But Apples lacks the pink/verdigris dissonance and seems blandly halfhearted as a result.

Interior, Dress, and Bowl of Apples are all masterpieces. All three have the characteristics of Matisse’s late period. They fill their canvases right to the edges and press against them. The borders between color zones are crisp and clear. Each separate zone of color is relatively uniform and unmodulated. The earlier two hark forward to the last, and all three culminate in the radiance of the late cutouts and the extraordinary stained glass of the Dominican chapel at Vence, his culminating achievement.

Ludwig van Beethoven is the canonical example of the artist who grows better every moment of his life. Recall that Beethoven in mid-career (roughly 1802-12) composed the Third, Fifth, and Eighth Symphonies, the Kreutzer Sonata, and the two most perfect works in music—the G Major Piano Concerto and the Violin Concerto—along with dozens of other masterpieces. After all, he combined, in midcareer, perfect structural mastery of sublimely lovely thematic mat-er-ial with a Shakespearean sense of humor. Listen to the coda of the C Minor Piano Concerto’s last movement, where the tempo switches to 6/8 and the soloist tiptoes over a new version of the rondo theme trying not to burst out laughing and wake the orchestra in his sheer exuberance. But he fails, and the last pages are the loveliest laughter in music.

Yet, in his later years, we find Beethoven inventing a new kind of theme, in which the mat-erial is so powerfully compressed (carbon squeezed into diamond) that a single terrific statement emerges with no rhythmically separate harmonic lines: Each musical line speaks in the same rhythm, and often there is no harmony at all. The most cele-brated example is the first theme of the Ninth Symphony: It emerges out of a musical mist, atomized sound (which Beethoven also invented), coalescing like a tornado out of chaos, it is stated by the whole orchestra fortissimo, in unison. No such theme had ever existed before. There is also the powerful unison opening of the Opus 95 String Quartet, or the sudden surge (a huge wave crashing over the decks) of the first four bars of the Hammerklavier Sonata, Opus 106, or the maestoso (“majestic”) start of the Opus 127 String Quartet—a passage so powerful that no serious performer can keep his seat while he plays it.

Alongside these forged-steel themes are the fugue themes, which occur unaccompanied at the start of the fugue; and late Beethoven is obsessed by the fugue as completely as Bach had been. In these late fugues, the theme is laid out plain as a diamond bracelet against the black velvet of the cosmos. The greatest of all concludes the Credo in the Missa Solemnis, in which Beethoven struggles to make the world understand, at a time when the steady rusting-away of faith has already begun, that he is devoutly Christian. But his deepest discovery of all goes beyond thematic material and its development. Beethoven, and Beethoven alone, discovered how to make an end. In the middle period, he had already invented the colossal coda, in which the opening theme returns finally to loom over the whole movement like a Saturn rocket trembling with power: Thus, the close of the first movements of the Eroica and the Fifth Symphonies, or the incomparable Opus 74 Quartet.

But then, as the end approaches, something new happens. In the Opus 95 and Opus 127 Quartets, the first movements reach no conclusions. The music simply withdraws into a silence that is tangible. And at the close of the Grosse Fugue (for string quartet), the fugue breaks off at the end into a wild sprint, faster and faster, until it takes off in lunatic ecstasy. The most moving discoveries are in the last piano sonatas: The scherzo of Opus 110, the second to last sonata and Beethoven’s most miraculous feat of sheer compression, trails off into a boundless gentle mist; in the scant 15 bars of the coda we hear the whole mystery of the universe. And in the last piano sonata, Opus 111, the first of the two movements ends in floral garlands (memorials to the dead) floating on a gently heaving sea. The last, a theme and variations, closes on what can only be described as a still, small voice: The music bears the tragedy of mankind on its shoulders as it walks upright through the last open door, and is gone.

I return to Matisse by way of two of the greatest of modern novelists: Henry James, born a generation before Matisse, and Marcel Proust, James’s soulmate and Matisse’s countryman and near-contemporary. James and Proust resemble each other in that each produced his masterpieces at the end of his career. James’s three greatest novels, the work of his “major phase,” are The Wings of the Dove (1902), The Ambassadors (1903), and The Golden Bowl (1904). Except for the less successful The Outcry (1911), they are his last completed novels, exploring the subtlest, most allusive of psychological shadings in prose tuned down to an intimate whisper.

In the last of the three, a golden bowl (only it is gilt over glass, not gold) has a flaw and a secret, and James—slowly turning over this small thing in his vast imagination, as if it were crystal flashing colored gleams in a thousand directions—finds his main symbols, and several of his key moments, in the enigmatic bowl. It is a symbol of beauty combined with frailty and phoniness; it causes an initial revelation, a crucial shift in power from deceiver to deceived, makes plain the saving power of truth and grace, and reveals, even in the manner of its being gathered up after having been dropped to the floor and broken, important aspects of the heroine.

James is often said to have prepared the ground for Proust, and Proust’s fellow great modernist, James Joyce. But Proust’s prose is far more James than Joyce’s. It, too, is hushed and candlelit. Proust, like James in his later work, moves in long sinuous sentences, the trailing tails of lazy mermaids. One hears long, nuanced, whispered revelation, then a deep breath—and the cycle repeats. Proust only lived to be 51, but he waited until he was in his forties to begin his great novel, which appeared as a whole (its last sections not quite finished) after his death. À la recherche du temps perdu, “in search of lost time,” is an immense achievement made of seven subnovels. Proust called the second À l’ombre des jeunes filles en fleurs, “In the shadow of young ladies in bloom.” Listen to the whisperous l’ombre, the softly mellifluous jeunes filles en fleurs. It is the greatest title ever invented.

An artist’s job is to fashion beautiful things—with a master goldsmith’s delicate touch—out of the raw, red shrieking rage and pain and joy of human life. Beethoven’s almost-last great masterpiece was the A Minor String Quartet, Opus 132, of 1825. He had been too sick to work, and he wrote the A Minor Quartet while recuperating. Its middle movement of five is the famous “Sacred thanks-giving--song from a healed man to God, in the Lydian mode.” He was feeling better, but he could see death on its way. And so one of his last, greatest creations was this thanksgiving song. As his life ended, what he wanted to say was thank you.

Matisse finished his career with the spectacular cutouts of the short book called Jazz, and his other master-pieces in this medium. Among his last works was one he called “the crowning achievement of my life”—the decoration, specifically the stained glass, for the chapel of the Dominican nuns at Vence. He stripped his palette to translucent lemon, cool brilliant green, and vivid green-blue. This artist, capable of unmatched subtlety in the invention of colors in combination, retreated, in the end, to one sublimely basic, powerful triad and to simple forms of matchless grace. He always said he was an atheist, but in the presence of his own creation at Vence, he was seen to hesitate.

David Gelernter, whose show "Sh'ma/Listen: The Art of David Gelernter" just closed at Yeshiva University Museum, is chief scientist at Lifestreams.com and a professor of computer science at Yale.

Next Page