In July 1720 Johann Sebastian Bach returned to Köthen from Carlsbad, where he had accompanied his prince, Leopold of Anhalt-Köthen, to the spa. He had left his wife, Maria Barbara, in good health. He found her dead and already buried: the register of St James’s church records the burial on 7 July. The obituary that his son Carl Philipp Emanuel and Johann Friedrich Agricola wrote after Bach’s death tells of the return in a few dry lines, without saying what she had died of. In that same year Bach made a fair copy of six pieces for violin and wrote on the title page, in a chancery Italian: Sei Solo à Violino senza Basso accompagnato. Beneath it, the date: 1720.
Without bass. The formula is literal, and it contains the riddle of the most famous piece in the collection, the Chaconne that closes the Second Partita in D minor, BWV 1004. A chaconne, by definition, is a dance built on a ground bass: a short formula in the low register that returns unchanged, or nearly, while everything above it changes. From Monteverdi to Purcell that bass is heard; it is played by the cello, the viola da gamba, the harpsichordist’s left hand. Bach removes precisely that. What is left is four strings and a bow, and nobody underneath.
The document, that is the score, has an almost notarial regularity. Sixty-four variations of four bars each, plus a closing bar: two hundred and fifty-seven bars in all. Each variation starts from the tonic, D, and arrives on the dominant, A, which the next variation resolves. The variations fall into three blocks: thirty-three in D minor, up to bar 132; nineteen in D major, up to bar 208; twelve in minor again, to the end. Beneath everything runs the oldest gesture of the Baroque lament, the four-note descent D, C, B flat, A. It is the tetrachord on which Monteverdi’s nymph and Purcell’s Dido weep. Bach writes it out in full only rarely. Most of the time he hints at it with a note snatched in passing on the low string, or leaves it wholly implied.
Here is the way to listen to it, and five minutes are enough to learn it. The first eight bars set out the theme in full chords, in the grave rhythm of the sarabande, which leans its accent on the second beat. Listen to them the way you memorize a telephone number: four notes descending, an arrival on A every four bars. From there the violin runs in semiquavers, then in demisemiquavers, leaps across registers, feigns two voices and then three, and the novice ear clings to the high notes and gets lost. The trick is the opposite: keep the descent in mind and wait, every fourth bar, for the landing on the dominant, the way you count waves. The bass the violin does not play is supplied by the listener. Around the middle of the piece comes D major, at bar 133: the writing turns almost bare, then begins to thicken again until it reaches the sonorities of an organ. In the last block there is the passage every violinist dreads, the bariolage, in which an open string is struck again and again while another moves: the ground bass becomes, for a few bars, a fixed note vibrating beneath the melody. Then the opening theme returns, unchanged, and the piece closes on D.
The nineteenth century did not trust that emptiness. On 8 February 1840, at the Gewandhaus in Leipzig, Ferdinand David played the Chaconne in public and Felix Mendelssohn improvised a piano accompaniment beneath him; years later a written version appeared in print, and circulated for decades. Robert Schumann wrote accompaniments for all six pieces. The missing bass looked like a gap to be filled. Johannes Brahms understood it the other way round, as the substance of the piece. In 1877 he made a transcription for the left hand alone and sent it to Clara Schumann, explaining that on a single staff, for a small instrument, that man had written a whole world of the deepest thoughts and the most powerful feelings. Only by playing it with the left hand, he added, did the difficulty and the kind of technique make him feel like a violinist.
Then there is Maria Barbara. The German musicologist Helga Thoene has read the Chaconne as a tombeau for the dead wife, with coded quotations of Lutheran funeral chorales hidden among the notes. Other scholars, among them Peter Williams, have objected that such calculations can be applied to almost any dense page of Bach. The dates coincide; the proof is missing. The autograph manuscript, now in the Staatsbibliothek in Berlin under the shelfmark P 967, bears the title, the composer’s name and the year 1720. No dedication.