The search
for a subject

The word

What the title actually claims

Ricercare is the ordinary Italian verb for searching — the same word you'd use for looking something up. Applied to music around 1500 it meant something close to trying out: a lutenist plucking through the mode to check the tuning and settle the ear before the real piece began. The earliest printed ricercars, in Petrucci's lute books (Spinacino 1507, Dalza 1508), are exactly that — short, unbarred, scalar, harmonically vague, over before they commit to anything.

Within forty years the same title had been attached to something almost opposite: severe, learned, imitative counterpoint written down for ensembles and organists, and by the 1540s a ricercar could run several hundred bars. The word never changed. What "searching" meant did. That drift — from noodling to the most disciplined writing in the repertory — is the whole story, and it's why the label sits on such wildly different music.

Type one · c. 1500–1530

The preludial ricercar

Idiomatic, single-line-ish, improvisatory. Runs, repeated cadence figures, an occasional held bass note to prove the mode. No subject, no imitation, nothing to develop — it exists to place you in D Dorian and get the fingers moving. Listen for how the melody keeps circling the final (D) and the fifth (A), touching C sharp only at the cadence as musica ficta.

Type two · c. 1540–1580

The imitative ricercar — a motet without words

Adrian Willaert and his circle in Venice took the vocal motet's technique and stripped the text out. What remains is a chain of points of imitation: a short subject stated by one voice, answered by the others in turn, driven to a cadence — and then a fresh subject starts up, usually overlapping the cadence so the seam is never audible as a stop. Jacques Buus published ricercars long enough to need two part-books.

The example below is one point, three voices, entries three bars apart. Watch the entry map: the alto leads on G, the soprano answers on D, the bass restates on G an octave down.

The answer is not a transposition. The subject rises G–A–B–C (tone, tone, semitone). The soprano's answer rises D–E–F–G (tone, semitone, tone). Transposing exactly would have demanded F sharp and dragged the piece out of its mode, so the interval pattern bends instead. This is the ancestor of the Baroque tonal answer — and here the motivation is purely modal, a century before anyone speaks of dominants.

Type three · c. 1570–1640

The monothematic ricercar

Late in the century the multi-subject chain contracts to a single soggetto worked for the entire piece — Andrea Gabrieli, then Frescobaldi. The subject slows down: whole notes and halves, small range, mostly steps, one distinctive leap. That shape isn't decorative. A subject in long notes can be stacked against itself at close intervals, doubled in length, or turned upside down and still be recognisable, which is precisely what the rest of the piece will do to it.

Three entries, four bars apart: D in the alto, a tonal answer on A in the soprano (up a fourth, not a fifth — same modal logic as before), then D again in the bass. Note how little happens harmonically. There is no drive toward a key; the piece simply keeps searching, and cadences are places to breathe rather than goals.

The workshop

What you do to a subject

A ricercar's later sections are where the artifici come out. Frescobaldi's Fiori musicali ricercars carry explicit obblighi — self-imposed rules, like an unchanging five-note countersubject, or a phrase that must be sung by an extra voice at fixed intervals. Below, put any subject through the machine and hear the result. These are mechanical transformations: some are usable as they stand, some collide. Sorting that out is the composition.

The extreme case

Chromatic ricercars and durezze

Frescobaldi's Recercar cromaticho and the related durezze e ligature style push the search into the semitones between the modal degrees. The subject here is a rising chromatic tetrachord answered at the fifth; the counterpoint shadows it in parallel thirds so the chromaticism arrives twice over. It closes with the standard two-voice clausula — a minor sixth opening out to the octave.

Architecture

Three ways to organise the search

The same subject material, three structural attitudes. Entries in green, free counterpoint faint, cadences marked.

The fugue's innovation isn't imitation — it's episode: passages built from fragments of the subject that modulate, so that returns of the theme land in new keys and the piece acquires a tonal shape. A ricercar has no episodes worth the name. It restates, and the interest is contrapuntal rather than harmonic. Once you hear that difference you can date almost any imitative keyboard piece by ear.

Neighbours, and how to tell them apart

1747 · the last word

Bach's Ricercar a 3 and a 6

By Bach's lifetime the ricercar was archaic — deliberately so, which is why he reached for it. Given a theme by Frederick the Great in Potsdam, he improvised on it, then sent back the Musical Offering, whose two large fugues are titled ricercar rather than fuga. The word is also an acrostic printed in the source: Regis Iussu Cantio Et Reliqua Canonica Arte Resoluta — "the theme given by the king, with additions, resolved in canonic style." The six-voice ricercar was engraved in open score, one staff per voice: the old learned convention, and a piece meant to be studied as much as played.

Here is the thema regium itself. A C minor triad, a rise to the flat sixth, a climbing leap to the leading tone — then a long chromatic descent through eight semitones that makes every conventional answer awkward. That's the trap in the gift.

Feed it into the workshop above and its difficulty becomes obvious: diatonic inversion can't cope with the chromatic tail, strict inversion sends it out of the mode, and the descent makes close stretto nearly impossible. Bach's six-voice solution keeps the theme almost entirely in long notes and builds the surrounding counterpoint from everything else.

After 1900

Why the title keeps coming back

The ricercar should have died with the fugue's arrival, and as a living form it did. As a title it has been revived steadily ever since, because it says something no other word says: this piece is about process rather than tune. A composer who calls a movement ricercare is telling you in advance that the material will be worked rather than presented.

Stravinsky's Cantata (1952) has two of them. In Rome, the tradition ran through Goffredo Petrassi to his pupils — among them Ennio Morricone, whose Invenzione, Canone e Ricercare for piano sits among his chamber works of the mid-1950s, next to a Variazione su un tema di Frescobaldi from 1956. That is a direct line back to Example 4 on this page. Much later he returned to the word in Ostinato ricercare per un'immagine (2010), built from sketches for a film he declined and reworked as a personal piece for his wife: a slow–fast–slow arch over a recurring piano motif. Francisco Llácer Pla's Ricercare Concertante (1987) for two pianos and orchestra folds the form into a concerto, mirrored around a central cello solo.

None of those can be printed here. Everything else notated on this page predates 1750 and is out of copyright; these are not. So the section below is mine — an original ricercar built the twentieth-century way, to show what survives when the mode goes.

Nothing in it is free counterpoint. Every note is the subject: the alto answers at the tritone rather than the fifth, since a tritone is the one transposition that partitions the octave evenly and so has no modal implication; the bass carries the strict inversion in augmentation for eight bars; and the last four bars stack the subject at the tritone over its own retrograde. The machinery is Frescobaldi's, unaltered. What has gone is the mode — which turns out to have been carrying the harmony, not the counterpoint.

Load Intervallic cell in the workshop above and the difference shows up in the controls: diatonic inversion has nothing to be diatonic about, so it collapses into strict inversion, and stretto transposes by a literal fifth instead of by scale degree.

Practical

If you're writing one

• Build the subject for stacking, not for beauty. Range under a sixth, mostly steps, one leap, three to five bars in long values. Test it against itself at one, two, and three bars' distance before you write anything else.
• Decide the answer's shape first. Fifth above or fourth below, adjusted so the subject's mode survives. If a note wants an accidental to make the transposition exact, bend the interval instead.
• Order the entries by register, not by rule. Alto–soprano–tenor–bass is more characteristic than top-down, and it keeps the texture from thinning out too early.
• Cadence, but overlap it. Let the next entry begin on the cadence chord so the piece never comes to a halt. The seam is the craft.
• Accelerate by note value, not by tempo. A ricercar that starts in whole notes and arrives in eighths has shape even without a single modulation.
• Save one device for the end. Augmentation in the bass under a stretto of the original is the traditional last card, and it still works.

Further

Where to hear the real thing