La musique française pour clavecin des XVIIe et XVIIIe siècles est-elle jouée de nos jours telle qu’elle devrait l’être ? De quelles sources disposent musicologues et musiciens pour approcher au plus près l’authentique sonorité du clavecin, la technique de son jeu, et la compréhension de son répertoire ? Cette approche nous a conduit à discerner neuf points déterminants, essentiels à la compréhension de la facture du clavecin. La mesure d’octave des claviers français de l’époque était inférieure à celle des instruments des pays voisins, et notamment inférieure à celle appliquée dans la facture de nos jours. Pour mémoire, cette mesure conditionne l’écartement entre le pouce et l’auriculaire, lequel écartement influence le jeu. Si l’écart est de moindre taille, les doigts sont plus rapprochés, et de ce fait, la main gagne en décontraction. Le jeu à la française se distingue aussi depuis ses origines par l’extrême souplesse de l’harmonisation, le peu de tension des cordes et le diapason bas (le La3 à 392 – 406 Hz). Par ailleurs, nous avons pu révéler l’existence de clavecins à trois claviers, la présence du seize pieds et du jeu nasal dans certains instruments (alsaciens notamment), et la paternité française de l’éclisse courbe en forme de « S ». Concernant les claviers, l’évolution de l’ambitus de l’instrument depuis le début du XVIIe jusqu’à la fin du XVIIIe siècle est bien connue des musicologues et des musiciens. Toutefois, aucune étude de cette évolution n’a été réalisée pour démontrer l’élargissement des claviers depuis le premier livre imprimé en 1670, les pièces de clavecin de Chambonnières, jusqu’à l’ambitus définitif des cinq octaves imposées dans les pièces de clavecin en concerts de Rameau en 1741. Notre étude s’est également étendue aux cordes, par la vérification de leur épaisseur et des matières dont elles étaient faites. Il s’avère que les facteurs de l’époque utilisaient des diamètres inférieurs à ceux des cordes actuelles, et qu’elles n’ont jamais été en cuivre dans les basses. Seul le laiton à forte teneur en cuivre était considéré comme satisfaisant pour la sonorité des cordes les plus graves. Quant aux cordes des trois cinquièmes supérieurs du clavier, elles étaient faites en fer mou qui n’avait qu’une faible tension. Il va sans dire que l’acier, employé de nos jours, était inconnu à cette époque. Enfin, il est admis aujourd’hui que le clavecin, une fois le tempérament posé, s’accorde en octaves pures, alors qu’il en était tout autrement au XVIIIe siècle, ainsi que nous l’avons établi à travers un texte de Corrette. / Is French harpsichord music of the 17th and 18th centuries played today as it should be ? What sources can help musicologists and musicians to reproduce the authentic harpsichord sound and playing techniques of that epoch, and understand its repertoire, as faithfully and fully as possible ? The mere fact that this music went unplayed for so long prompts that question. In fact, the harpsichord was forgotten overnight. The favoured instrument of court and fashionable society under the ancien régime, it had aristocratic associations which doomed it when the Revolution came. A century later, in June 1889, the noble, silvery sound of its plucked strings made a first, hesitant comeback, thanks to Louis Diémer. But it was only in the 20th century, between the two world wars, that Wanda Landowska’s tireless enthusiasm gave this baroque keyboard instrument a new lease of life. Interest in building “old-style” harpsichords, using traditional techniques, first developed in the late 1950s, and their popularity has grown steadily ever since. Today’s enthusiasts want to go back to the origins, and revive old ideas and techniques, but they still have a long way to go. At an earlier stage, techniques used in making pianos were extended to harpsichords - and some of these “alien” elements and additions are still present. We felt the time had come to clarify the picture by consulting certain contemporary texts, which had been unduly neglected. We found indeed that these were at odds with twentieth- century improvements, had been mistranslated or misunderstood, or were, quite simply, hard to find.Anyone wishing to form an idea of the original harpsichord sound must start with organology, and the various instruments used by French musicians in the 17th and 18th centuries offer valuable clues. X-ray examination reveals their design and shows how they were regulated (keys, jacks, plectra).Thanks to this approach, we have identified nine essential factors which illuminate the design and construction of these instruments. French manuals of the time had a narrower octave span than those of instruments made in neighbouring countries - or today. Span, of course, determines the distance between thumb and little finger, which itself affects playing. The smaller the gap, the closer the fingers, and the more relaxed the hand. From the beginning, the French sound was also distinguished by its highly flexible harmonies,low-tension strings and low pitch (A3 at 392-406 Hz.). We also found that some harpsichords had three manuals, that some (particularly Alsatian instruments) had 16 foot stops and a lute stop, and that the S-shaped bentside was a French innovation. Musicologists and musicians already know in general terms how manuals evolved from the early 17th to the late 18th century, but no specific research has been done on the process by which they became wider, between 1670, when the first book, Chambonnière’s Pièces de clavecin, was published, and 1741, when Rameau’s Pièces de clavecin en concerts made five octaves the norm.We have also studied strings, their thickness and the materials of which they were made. We have found that string diameter was smaller than it is now, and that bass strings were never made of copper. Only brass with high copper content was thought to give the deeper strings a satisfactory sound. Strings on the upper three-fifths of the manual were made of soft iron, which had little tension. Steel, which is used today, was obviously unknown.Finally, harpsichords, once their temperament is established, are today tuned in pure octaves –which, as a text by Corrette has shown us, was far from being the case in the 18th century.