EXHIBITIONS / Expect a bumpy ride: A new show in Edinburgh challenges visitors to look at French landscape painting in a fresh light. But some of the work fails to make an impression

THERE ARE some pretty mediocre paintings - along with a dozen or so masterpieces - in Monet to Matisse: Landscape Painting in France at the National Gallery of Scotland, and that's because Richard Thomson, the exhibition's organiser, is a serious academic who wants to make a number of art-historical points. I don't mind second-rate pictures, especially unfamiliar ones, but looking at this show is rather a bumpy ride.

You keep wondering why on earth that one was selected, or that one, and the only way to find out is to settle down with Thomson's catalogue. His book, really. For the paintings on display aren't given individual entries, as in most catalogues of important exhibitions, but are discussed as part of Thomson's general argument.

What he proposes is briefly this. In the 40 years between the first Impressionist exhibition of 1874 and the start of the First World War, landscape painting - with French life as a whole - underwent rapid and convulsive changes. So we should not consider paintings in purely aesthetic terms, but look for the values that they represent or, quite often, subvert. To look at the history of the period in terms of changes of style, or to think of 'movements' successively replacing the art of a previous generation, is naive. Instead, Thomson wants to consider 'associations of space, nature, visual culture, national or regional identity, agricultural labour and social class'.

Here is an ambitious agenda, but on the whole Thomson convinces us his approach is valid. I am not completely impressed, however. His text is too short for the scope of his arguments. He is not tender enough towards the painters' personal visions and is shy of great art that transcends landscape. Manet is not considered and is not in the exhibition. Thomson excludes drawing, both in his discussion and in the show. Cubism makes him uneasy. Yet Thomson's is the most interesting book about French painting I've read for years. It's bound to be a standard work for students, but that's not much fun. My hope is that he will expand his present writing and produce a classic. For classics, if not necessarily fun, are permanently inspiring.

Meanwhile, the lack of inspiration in so many of the Edinburgh pictures still bothers me. I know many people don't like that word, but it's useful. It means raising art above a certain level. You can find inspiration in your comrades, your circumstances, your mistress, the clouds, Michelangelo, or many other sources. Wherever it comes from, inspiration is a reality. Van Gogh drove himself mad in its pursuit. Some of Matisse's greatness is in the way he knew how to nurture and cherish the things that inspired him, as is demonstrated in this show by his Acanthus. Whether fervent or wise, the votaries of inspiration belong to the early years of modern art. But some people didn't have it at all, even if they are more than salon artists. Raffaelli, Cazin, Lacombe, Tauzin, Sege, Brouillet, all illustrate Thomson's arguments but are still not much more than journeymen.

The exhibition galleries at the back of the Scottish National Gallery are rather awkward. There are ramps and turns, there is no natural light and the rooms aren't quite big enough. But this favours an installation that leads the spectator by the hand. Thomson does this by grouping his pictures by theme. There are 13 of these groups. Three are Parisian, then there are sections for the Channel coast, the Mediterranean coast, the classical landscape, the nocturne and so on. Somehow you're not encouraged to wander about. You are in fact going through an exercise in understanding. Festival-goers might think this severe, but there are pleasurable pictures en route.

Compare, for instance, the below- par Raffaelli with Monet's similarly toned picture of the Seine. We're at a different level of achievement and accordingly find more satisfaction in the better picture. I like Emile Bernard's Quai de Clichy and especially Albert Marquet's Notre-Dame under Snow. Marquet is a painters' painter and his excellence is hard to define. Thomson writes that he occupies 'a middle ground between the modern, the painterly and the mysterious' and this is just, even if not precisely enlightening. Chagall's disorganised The Pont de Passy and the Eiffel Tower was obviously painted in high excitement during his first visit to the artistic capital, but note that its palette is Russian rather than French. The big Corot, Pastorale: Souvenir d'Italie, is from 1873, therefore strictly outside the exhibition's date limits, and it seems out of place for other reasons. But it's a lovely painting, rather changeable, as the best Corot landscapes tend to be, elusive as the nymphs it portrays. I bet Thomson put it in his show because he likes it so much, and not because it has any historical lessons we ought to take on board.

What a gulf there is between Corot and Van Gogh. In Edinburgh the other day Thomson told me something interesting about the Dutch painter that I'd like to pass on. Apparently his few personal effects are in the basement of the Van Gogh Museum in Amsterdam. Among them is a Japanese box. For some reason - incurious people, the Dutch - it had never been opened before the big retrospective in 1990. Inside were a lot of differently coloured wools, some of them twined together. One curator twigged what they were for. She ran upstairs with a skein and found a perfect match in one of Van Gogh's late paintings.

So this is how he pre-planned his palette. The discovery changes our view of the painter. Although he wrote so much about colour, the impression in front of Van Gogh's canvases is that he was (in early days) a habitual or (later on) an unthinking colourist. Now we find that his palette was more calculating. In Edinburgh there are Van Gogh paintings interesting for their colour. Winter, a development of a Millet painting of 30 years earlier, has pink in the soil, blue in the vegetation and yellow in the air. Here are challenging effects, but as a picture I prefer the 1888 canvas Thomson calls Wheatfield with Reaper, Auvers. Surely this title is incorrect? The fields look as if they are outside Arles and Van Gogh did not go to Auvers, north of Paris, until 1890, and shot himself in July, before harvest-time.

Anyway, the connection with Arles is crucial because it was there, and with the Provencal June harvest, that Van Gogh at last came to his mastery of yellow. I guess that the wools account for the curious variegation of colour in Winter, which was painted inside, indeed inside the asylum, while the harvest picture responds to daylight and nature. Yellow is the palest of the primary colours, but that doesn't make it easier to manage, especially when spread around in quantity. In this picture Van Gogh does so, perfectly, and look how his brush-strokes race round but are never wrong. The painting must have been completed in one sitting. In its own way it's as perfect as an enamelled miniature, but perfected by the impetus of inspiration, which craftsmanlike and finished miniatures seldom possess.

Even beside the two awesome Cezanne seascapes, Matisse's Acanthus of 1912 is the best painting in the exhibition, and here again the question of finish is important. The history of the picture is that it was begun, and concluded, in a month and a half of work during the artist's first visit to Tangier. But Matisse did not quite know that it was finished. He took the canvas to Paris, then brought it back to Tangier the following year for reworking. Putting it in front of his original motif, however, he realised that there was nothing to be added or changed. The delicate and complex layering of Acanthus is remarkable. It is true that the picture is final. A bit more pigment and it would cease to breathe. It's as though one had become immersed in Matisse's purely painterly thoughts. How odd that he should have been reading Dickens for relaxation. The English writer doesn't seem proper, not holy enough.

National Gallery of Scotland, Edinburgh (031-556 8921) to 23 Oct.