By the numbers
How music begins: the first notes of 2.0 million real incipits
Short answer
RISM, the international catalogue of music manuscripts and prints, holds 2,029,719 decoded incipits (the opening notes of pieces, movements and parts). In major keys, 43.7% of them open on the tonic, 35.2% on the dominant (degree 5) and 16.5% on the third. The commonest first move is staying put: 30.3% repeat the first note, and the single commonest four-note opening is one pitch four times (7.5%). The commonest time-signature sign is C (36%), then 3/4 (18%); 20% open on a pickup.
- 43.7%of major-key incipits start on the tonic (1,321,933 incipits)
- 30.3%start by repeating the first note (614,936 of 2,026,262)
- 20%begin with a pickup rather than on the first beat
Dataset: 2,029,719 incipits decoded from RISM's export of 1 October 2026 (95.9% of the 2,115,606 with notation; 1,233,134 source records have an incipit). Mostly 17th- to 19th-century manuscripts held in European libraries — a catalogue of surviving sources, not of what people listen to today, and not pop.
The first note
In major keys the first note is the tonic in 43.7% of incipits, the dominant (degree 5) in 35.2% and the third in 16.5%; together they make up 95.4%. In minor keys the order is the same (45.4%, 37.4% and 10.6%). Vocal parts lean the other way: among the 593,624 major-key incipits of voice parts the dominant narrowly leads (38.7% against 37.8% for the tonic).
Left and middle columns: the key stated by the cataloguer (1,321,933 major-key and 285,999 minor-key incipits). Right column: every decoded incipit (2,029,719), counted from the major key named by its key signature — a convention that cannot tell major from minor, so in a minor-key piece the tonic shows up as degree 6.
| First note (scale degree) | Major key: stated key | Minor key: stated key | All incipits: signature's major key |
|---|---|---|---|
| 1 (tonic) | 43.7% | 45.4% | 35.0% |
| 2 (supertonic) | 1.1% | 1.5% | 4.2% |
| 3 (mediant) | 16.5% | 10.6% | 19.0% |
| 4 (subdominant) | 0.8% | 1.2% | 2.1% |
| 5 (dominant) | 35.2% | 37.4% | 28.1% |
| 6 (submediant) | 1.3% | 1.3% | 10.0% |
| 7 (leading note) | 0.8% | 1.9% | 1.7% |
| Not in the scale (chromatic) / accidental on the note | 0.6% | 0.6% | 0.9% |
In minor keys the degrees are letter distances above the minor tonic, so 3 is the minor third.
The first move
The most likely first interval is no interval at all: 30.3% of incipits start by repeating the first note. A step (a second) follows in 27.5%, a leap of a third or more in 42.0%. When the melody does move, it goes up more often than down (42.2% against 27.4%).
The most common openings
The ten commonest four-note shapes cover only 25% of incipits, so there is no single formula; but the top three are plain: one note repeated four times (7.5%), a rising scale of four notes (3.9%) and a falling one (2.7%).
A shape is the three moves between notes 1–2, 2–3 and 3–4, counted in scale steps (a step is 1, a third is 2, a fifth is 4, and so on; − is down). Which key or degree the shape starts on is ignored. The ten shapes together cover 25.2% of all incipits.
| Opening (first four notes) | Steps from note to note | Share of incipits | Incipits |
|---|---|---|---|
| One note repeated four times | 0, 0, 0 | 7.47% | 150,199 |
| Rising scale: three steps up | +1, +1, +1 | 3.93% | 78,962 |
| Falling scale: three steps down | −1, −1, −1 | 2.66% | 53,556 |
| Same note → same note → step up | 0, 0, +1 | 1.83% | 36,848 |
| Same note → same note → step down | 0, 0, −1 | 1.76% | 35,405 |
| Same note → step up → step up | 0, +1, +1 | 1.76% | 35,287 |
| Step up → step down → step down | +1, −1, −1 | 1.53% | 30,850 |
| Step down → step up → step up | −1, +1, +1 | 1.47% | 29,530 |
| Step up → step up → step down | +1, +1, −1 | 1.45% | 29,188 |
| Third up → step down → step down | +2, −1, −1 | 1.37% | 27,475 |
Steps are counted by note letter, so a major and a minor second both count as 1.
Chosen by a fixed rule: the lowest RISM number among incipits of the 150 best-represented named composers (and not anonymous) with that shape. Only the first four notes are shown, as a check on the counting; the scores belong to the libraries that hold them.
| Shape | Source (link to RISM) | First four notes | RISM ID |
|---|---|---|---|
| One note repeated four times | Fioravanti, Valentino, Adelaide maritata | G4 G4 G4 G4 | RISM sources/51652 |
| Rising scale: three steps up | Bach, Johann Sebastian, Allein Gott in der Höh sei Ehr | G4 A4 B4 C#5 | RISM sources/51784 |
| Falling scale: three steps down | Bach, Johann Sebastian, Einige canonische Veränderungen | C5 B4 A4 G4 | RISM sources/51795 |
| Same note → same note → step up | Cimarosa, Domenico, Gli Orazi ed i Curiazi | G4 G4 G4 Ab4 | RISM sources/100065 |
| Same note → same note → step down | Bach, Johann Sebastian, Durch Adams Fall ist ganz verderbt | A4 A4 A4 G4 | RISM sources/51779 |
Clefs, key signatures and time signatures
The treble clef opens 52% of incipits and the soprano C clef another 26%; the bass clef opens 10%. 24% have no sharps or flats; flats outnumber sharps (40% against 36%). The most frequent single time-signature sign is C, in 36% of incipits, followed by 3/4 (18%) and cut-C (17%).
C counts as four-beat (4/4) and C-slash as two-beat (2/2), the usual reading; a bare 2 or 3 is grouped with two- and three-beat. Mensural signs (such as O or C3) are only counted under 'other'.
| Family | What is in it | Incipits | Share |
|---|---|---|---|
| Four-beat | C, 4/4, 4/2, 4/8 | 810,244 | 38.3% |
| Two-beat | cut C, 2/4, 2/2, a bare 2 | 618,001 | 29.2% |
| Three-beat | 3/4, 3/8, 3/2, a bare 3 | 512,091 | 24.2% |
| Compound | 6/8, 12/8, 9/8, 6/4 | 137,061 | 6.5% |
| No time signature | field empty | 28,419 | 1.3% |
| Other | rare or mensural signs | 9,790 | 0.5% |
On the beat or before it?
54% of incipits begin on the first beat of a bar and 20% begin with a pickup (an incomplete first bar). A further 20% begin after a rest inside the first bar (a delayed or off-beat entry); for 5% the incipit is too short or too unclear to tell. A part that rests for whole bars and then enters on a downbeat counts as starting on the first beat.
Voices, keyboards and strings
Voices open on the dominant more often than on the tonic (39% against 38% of major-key incipits) and repeat their first note in 36% of cases; keyboards and strings open on the tonic (47% and 51%) and repeat the first note less (25% and 25%). Bass lines start on the tonic 86% of the time.
Tonic, dominant and third: share of the major-key incipits of that kind (stated key). Repeated note, pickup and clef: share of all incipits of that kind. The part is read from the incipit's voice/instrument field (S, A, T, B, V = voices; vl = violin; org, pf, cemb = keyboards; b, bc = bass line).
| Part named in the incipit | Incipits | Tonic | Dominant | Third | Repeated note first | Pickup | Treble clef |
|---|---|---|---|---|---|---|---|
| Voices (S, A, T, B, choir ...) | 974,237 | 38% | 39% | 19% | 36% | 14% | 28% |
| Keyboard (organ, piano, harpsichord) | 387,375 | 47% | 32% | 16% | 25% | 28% | 61% |
| Violin and other strings | 429,806 | 51% | 32% | 13% | 25% | 25% | 94% |
| Bass line (b, bc) | 26,410 | 86% | 8% | 3% | 30% | 11% | 1% |
| Wind instruments | 90,470 | 40% | 37% | 17% | 27% | 29% | 90% |
| Lute, guitar, harp | 13,437 | 39% | 35% | 21% | 27% | 37% | 75% |
| Other or mixed | 71,430 | 43% | 34% | 17% | 28% | 28% | 72% |
| No part named | 36,554 | 44% | 36% | 15% | 28% | 20% | 70% |
Centuries, manuscripts and countries
The pattern holds everywhere: tonic, dominant and third open at least 91% of incipits in every group. The tonic leads in the 18th century (49%) and falls to 40% in the 19th, when pickups become commoner and the treble clef takes over (61% of incipits, against 45% in the 18th). Across the eight best-covered countries the tonic share runs from 35% to 48% and the dominant share from 31% to 39%.
Tonic and dominant: share of major-key incipits in the group (stated key). Other columns: share of the group's incipits. Dates come from the date field (the century code or the years given); 241,681 incipits with notation have no usable date and appear only in the other rows. The country is that of the library holding the source.
| Group | Decoded incipits | Tonic | Dominant | Repeated note first | Pickup | Treble clef | C or cut-C | 3/4 |
|---|---|---|---|---|---|---|---|---|
| 17th century | 100,837 | 47% | 31% | 36% | 11% | 27% | 70% | 5% |
| 18th century | 815,432 | 49% | 34% | 28% | 19% | 45% | 53% | 18% |
| 19th century | 768,138 | 40% | 36% | 32% | 23% | 61% | 50% | 21% |
| 20th century | 91,356 | 32% | 38% | 27% | 20% | 85% | 38% | 21% |
| Manuscripts | 1,844,316 | 45% | 35% | 30% | 19% | 51% | 53% | 18% |
| Prints | 185,305 | 35% | 38% | 30% | 26% | 58% | 51% | 18% |
| Germany (D) | 872,959 | 44% | 36% | 30% | 21% | 50% | 56% | 17% |
| Italy (I) | 185,426 | 45% | 35% | 31% | 15% | 42% | 56% | 15% |
| Czechia (CZ) | 157,866 | 43% | 35% | 33% | 19% | 61% | 46% | 27% |
| Poland (PL) | 153,887 | 39% | 33% | 32% | 17% | 59% | 47% | 23% |
| Austria (A) | 146,214 | 48% | 31% | 33% | 17% | 41% | 48% | 21% |
| Switzerland (CH) | 119,172 | 35% | 39% | 30% | 25% | 69% | 44% | 22% |
| Great Britain (GB) | 97,244 | 47% | 33% | 29% | 19% | 50% | 64% | 9% |
| United States (US) | 86,384 | 43% | 39% | 26% | 30% | 65% | 50% | 16% |
A source can be counted in a century row and a country row; the rows overlap and are not parts of one total.
Compared with art songs
The two datasets agree that the dominant is the commonest or near-commonest first note of a sung melody (39% of voice-part incipits here; 42% of art songs). They disagree about the rest: only 18% of art songs open on the tonic, against 35% of 19th-century voice incipits in RISM, and the art songs open on the third far more often (32% against 21%). The data are not the same kind. A RISM incipit is a cataloguer's entry for the opening of a movement, aria or part (30% of voice-part incipits start after a rest inside the bar), the repertoire is mixed (the 18th and 19th centuries each hold about two incipits in five), and the key is the cataloguer's; the art-song figure is the first note of the voice in a finished song, with the mode estimated by software.
The art-song column is the first note of the vocal line in 880 major-key songs of the OpenScore Lieder corpus (see how melodies start and end). The RISM columns use the stated key.
| First note, major keys | RISM: voice parts | RISM: voice parts, 19th century | RISM: all parts | Art songs (OpenScore Lieder) |
|---|---|---|---|---|
| Dominant (5) | 38.7% | 39.4% | 35.2% | 41.7% |
| Tonic (1) | 37.8% | 34.9% | 43.7% | 18.2% |
| Third (3) | 18.6% | 20.9% | 16.5% | 32.5% |
| Incipits / songs | 593,624 | 262,935 | 1,321,933 | 880 |
The two sources do not measure the same thing, so the gap is not an error in either: see the reasons below.
What an incipit is
Three short excerpts from Wikipedia: what an incipit is, what a pickup is, and the code in which RISM stores the notes.
From Wikipedia
Incipit
The incipit of a text is the first few words of the text, employed as an identifying label. In a musical composition, an incipit is an initial sequence of notes, having the same purpose. The word incipit comes from Latin and means "it begins". Its counterpart taken from the ending of the text is the explicit (explicitum est). The unfolding refers to a papyrus scroll. The end is also referred to as desinit, 'it is finished'.
Source: Wikipedia, “Incipit” — CC BY-SA 4.0 — text excerpted; see the article for the full text and authors. Revision of 2026-06-14.
Anacrusis
In poetic and musical meter, and by analogy in publishing, an anacrusis (from ἀνάκρουσις, anákrousis, literally: 'pushing up', plural anacruses) is a brief introduction. In music, it is also known as a pickup beat, or fractional pick-up, i.e. a note or sequence of notes, a motif, which precedes the first downbeat in a bar in a musical phrase.
Source: Wikipedia, “Anacrusis” — CC BY-SA 4.0 — text excerpted; see the article for the full text and authors. Revision of 2025-09-28.
Plaine & Easie Code
The Plaine & Easie Code is an open international library standard that enables entering music incipits.
The code format standard uses ordinary ASCII symbols and is maintained by the International Association of Music Libraries, Archives and Documentation Centres and the Répertoire International des Sources Musicales. It supports modern and mensural notation and aims to be an exchange format in the library environment. It is used within the MARC standards to encode musical incipits.
Source: Wikipedia, “Plaine & Easie Code” — CC BY-SA 4.0 — text excerpted; see the article for the full text and authors. Revision of 2026-06-26.
How this was counted
- Source. The RISM source-records export of 1 October 2026 (
source-latest.xml.gz, MARCXML, 567.9 MB compressed; SHA-1 equal to the one in RISM's own file list, SHA-256 beginning 9beee56cfb96). It holds 1,626,011 source records; 1,233,134 have at least one incipit (MARC field 031) and 473,745 of those have more than one, in 2,547,335 incipit fields in all. 2,115,606 fields carry Plaine & Easie notation; the rest have a text incipit or no notes. - Unit. One incipit field is one count. A work that survives in many copies, or a source with several incipits (one per part or movement), counts several times. Counting every distinct incipit text once (1,891,419 of 2,029,719) gives the same rankings, as the checks below show.
- Decoding. The notes are in Plaine & Easie Code (IAML specification). A decoder written for this site reads octave marks, durations, accidentals (kept to the next bar line), rests, bar lines, ties, grace notes, chords (top note), tuplets and multi-measure rests. Anything it cannot read reliably is skipped, not guessed: 2,029,719 of the 2,115,606 incipits with notation decoded (95.9%); the 85,887 skipped are mostly mensural notation (80,454) and groups it cannot resolve. A further 56,955 were decoded up to a symbol it does not handle (such as a repeat shortcut) and counted from the notes before it.
- Checked. Against the signature in the same record, 95.7% of decoded incipits keep at least 80% of their notes in the major scale of the signature or the harmonic minor of its relative minor; against a deliberately wrong signature only 10.2% do. Notes sit in the usual range of their clef in 198 of 198 sampled incipits, and the bars add up to the time signature in 89.3% of the 1,803,538 incipits where that can be tested (the rest are mostly errors or old conventions in the encoded source). In a random sample of 200 incipits, 188 passed the key test, 198 of 198 the clef test and 158 of 181 the bar test; The first 20 of the sample were also read by hand against the specification: all 20 pitch sequences agree.
- First note and degree. The first melody note (for a chord its top note; grace notes are skipped). Its degree is the letter distance from a tonic. The default tonic is the one named by the cataloguer in the key field ($r: a capital letter is major, a lower-case one minor); the comparison column uses the major key named by the key signature, which is only a convention. 1,607,932 incipits have a plain key in $r; it agrees with the signature in 97% (major) and 87% (minor) of them, and using only those whose key matches the signature moves the tonic and dominant shares by less than a point.
- First interval and shapes. The interval between the first two melody notes, by letter distance (size) and semitones (quality); the four-note shape is the three signed letter distances between notes 1–4. Incipits with fewer notes drop out of the denominators (2,026,262 with two or more, 2,009,834 with four or more).
- Bar position. Compared with the second bar when there are two or more closed bars, otherwise with the time signature. If the first bar is shorter the incipit starts with a pickup; if the first note is after silence within a complete bar it starts after a rest.
- Strata. The century is read from the date field, manuscript or print from the source type (Print against all manuscript types), the country from the holding library's siglum, and the part from the voice or instrument field.
- Checks. Every distribution adds up to its total, and each ranking claimed in the text is tested when the page is built.
Data used
- RISM data export (source records, MARCXML) — 1,626,011 source records of music manuscripts and prints with 2,547,335 musical incipits; only aggregate counts are published here. CC BY 3.0 (Unported), credit RISM
- OpenScore Lieder Corpus — 1,462 art songs (mostly 19th-century European) transcribed to MusicXML by volunteers; key signatures, time signatures, tempo words, metronome marks, vocal lines and melodies are read directly from the files. CC0 1.0
What these numbers cannot tell you
- RISM is a catalogue of what survives in libraries, not a list of what people play or hear. It is mostly handwritten 17th- to 19th-century European art, church and chamber music, with no pop, rock, jazz or film music. 43% of the incipits with notation come from sources held in Germany.
- An incipit is the opening of a movement, an aria or a part, chosen by a cataloguer (the guidance asks for the violin or highest part, or for vocal music the highest voice plus the leading instrument), so it is not always the first note of the whole piece; and incipits of bass or inner parts are rare.
- 'Major' and 'minor' come from the cataloguer's key field where it is a plain key (1,321,933 major-key and 285,999 minor-key incipits), never from the notes; where there is no key field the incipit is left out of the degree tables. The key signature alone cannot tell major from minor.
- Plaine & Easie decoding is a reading of a text code with 85,887 skipped and a few thousand cut short; it was checked for consistency (key, clef, bar sums, a sample read by hand), not against the source images. Accidentals are assumed to last to the next bar line, as in standard notation.
- The rules ask for incipits of transposing instruments to be notated at sounding pitch; if a cataloguer did not follow that, the key of a horn or clarinet incipit looks odd against its signature, which the key check would show as a failure, not hide.
- Counts are per incipit, so widely copied works and prolific composers weigh more; the checks show the rankings do not change when every distinct incipit text is counted once.
Checks run each time this page is rebuilt (59)
- 2,547,335 incipit fields = 2,115,606 with Plaine & Easie notation + 431,729 without
- 2,115,606 with notation = 2,029,719 decoded + 85,887 skipped
- the skip reasons add up to the 85,887 skipped incipits
- decoded incipits = 2,029,719 in the aggregated table
- clef, time-signature and meter-family counts each add up to the 2,115,606 incipits with notation
- key-signature, start-position and first-degree counts each add up to the 2,029,719 decoded incipits
- first-interval and direction counts add up to the 2,026,262 incipits with two or more notes
- the listed four-note shapes (819,459) are at most the 2,009,834 incipits with four or more notes
- record counts are consistent (1,626,011 records, 1,233,134 with an incipit, 473,745 with two or more)
- the histogram of incipits per record covers all 1,626,011 records
- at least 90% of the incipits with notation decode (95.9%)
- first-note ranking against the stated key: tonic, dominant, third - in both modes
- degrees 1, 3 and 5 together hold 95.4% (major) and 93.4% (minor) of first notes
- counting every distinct incipit text once (1,891,419 incipits) gives the same ranking and a tonic share within one point (43.7% vs 43.7%)
- against the key signature's major key (the convention), degrees 1, 5, 3 also lead: 35.0%, 28.1%, 19.0%
- signature-based degrees (with and without an accidental on the note) add up to 2,029,719
- the 'relative minor' effect is visible: against the signature's major key, degree 6 is far commoner (10.0%) than against the stated key (1.3%)
- the stated key ($r) agrees with the key signature in 97.3% of major-key and 87.3% of minor-key incipits
- using only incipits whose stated key agrees with the signature (1,286,080) changes the tonic and dominant shares by less than a point
- in voice-part incipits the dominant leads the tonic (38.7% vs 37.8%)
- the 'repeat' direction count equals the repeated-note interval count (614,936)
- a repeated note is the commonest first interval; the next kind (Up a major 2nd) is under 15%
- same note + step + leap + same-letter-altered shares add up to 100% (30.3% + 27.5% + 42.0% + 0.17%)
- the ten shapes are in descending order of count
- the three commonest openings are one note repeated, a rising scale and a falling scale
- counting each distinct incipit once gives the same ten shapes in the same order
- an example exists for the shape 0,0,0
- an example exists for the shape 1,1,1
- an example exists for the shape -1,-1,-1
- an example exists for the shape 0,0,1
- an example exists for the shape 0,0,-1
- the five examples show four notes each, no more
- clef rows add up to 2,115,606
- the three commonest clefs are G2, C1 (soprano) and F4
- key-signature rows (including 0 non-standard signatures) add up to 2,029,719
- flat signatures are commoner than sharp ones (40.4% vs 35.6%), and 'none' is the commonest single signature (24.0%)
- meter families add up to 2,115,606
- the seven commonest time-signature labels, in order: C, 3/4, C-slash, 2/4, 6/8, 3/8, 4/4
- four-beat leads, then two-beat and three-beat, then compound
- starting on the first beat is the commonest start position
- the eight voice/instrument classes add up to 2,029,719 decoded incipits
- voices lean to the dominant, keyboards and strings to the tonic
- voice parts repeat the first note more often than keyboards and strings do (35.7% vs 24.8% and 24.6%)
- strings and keyboards start with a pickup more often than voices (24.5%, 28.0% vs 13.5%)
- every row of the comparison table has at least 1,000 major-key incipits
- the tonic opening peaks in the 18th century (49% vs 40% in the 19th and 47% in the 17th)
- manuscripts open on the tonic more often than prints, and use pickups less (45% vs 35%; 19% vs 26%)
- tonic + dominant + third exceed 90% in every row (lowest 91.3%)
- undated incipits are reported (241,681)
- treble clef is commoner in the 19th century than in the 18th (61.1% vs 44.9%)
- the 18th and 19th centuries each hold about two in five incipits (39% and 36%)
- the art-song figure 'starts on 5' (41.7%) is recomputed from the corpus file
- the comparison finds what the text says: dominant starts within about 5 points, tonic starts far apart (18.2% vs 34.9%)
- mensural-notation skips (80,454) are part of the 85,887 skipped
- whole-population key check: 95.7% of incipits fit their own signature, 10.2% fit a signature three fifths away
- bars add up to the time signature in 89.3% of the 1,803,538 incipits where that can be tested
- the 200-incipit sample: key test passes 188 of 200, the control (wrong signature) only 15 of 200
- 20 incipits were read by hand against the code: all 20 pitch sequences agree
- leaving out the 173,138 incipits that mark no octave changes the repeated-first-note share by less than a point (30.3% vs 30.3%)
Source and licence
The numbers on this page are computed from the data of RISM – Répertoire International des Sources Musicales (rism.online), source records as exported on 1 October 2026 (RISM data exports), licensed under Creative Commons Attribution 3.0 Unported (CC BY 3.0). Changes made by this site: the Plaine & Easie notation of each incipit was decoded by a program written for this site, and only aggregate counts and shares are published here (no incipit is reproduced beyond the four-note examples credited by RISM ID). RISM does not endorse this analysis. The licence covers RISM's data, not the images or scores held by the libraries.
Data: RISM data export, source records (1 October 2026) (CC BY 3.0); OpenScore Lieder Corpus (CC0 1.0); Wikipedia (excerpts) (CC BY-SA 4.0). Computed for this site on 2026-10-07; the numbers are counts or shares from the datasets named above, not estimates for music in general. Licences and credits for everything used on this site are on the credits page.