I'm having some trouble understanding my edition, particularly measures 23-30. For reference, use First Edition, Paris: Durand & Fils, 1905 [1]. I'll also refer to Leipzig: Edition Peters, 1970 [2].

The three problems:

  1. Bar between measures 23 and 24 has been omitted. This is likely in error as edition [2] separates these measures. However, it could affect problem #2.

  2. Rhythm of measure 24. The notation is quite vague so there are two interpretations, three counting recordings I've found on youtube, and a final, improbable, suggestion.

    1. There are six groupings of notes, 48/32 (48 notes where there would normal be 32 sixty-fourths). This is equivalent to 3/2 (triplets); therefore there are 12 notes per quarter-measure. The last group - of six notes - is therefore played slower. I find this to be most elegant, on paper.
    2. There are 11 groupings of notes, 44/32. This is equivalent to 11/8; therefore there are 11 notes per quarter-measure. The last group - of six notes - is therefore played faster. This is very messy.
    3. There are 46 notes in a measure to be played in equal tempo; therefore a 46/32 tuplet. Aesthetically, I prefer recordings in this style.
    4. Though quite unlikely, it is possible the missing bar of [1] was intended to split measure 24. This would explain why, unlike in subsequent measures, there is no tuple notation - the notes are in fact not tuples.

    All recordings I've found either fall into [2.3] - equal tempo or [2.1] - slowing for the last six notes. I should also point out that I'm not sure what meaning, if any, the particular style of beaming - groups of four beamed into groups of 8 - in measure 24 is intended to convey.

  3. Notation "Mesuré" (meaning even or equal) in measure 25 is ambiguous:

    1. "Mesuré" refers to the melody and thus the tempo; the arpeggios in the upper staff are of varying velocity.
    2. "Mesuré" refers to the arpeggios in the upper staff; the melody and tempo therefore vary to match.

    Once again, recordings fall into both categories, though predominantly biased towards [3.2].

I'm not sure how to practice this piece with these uncertainties. While I prefer an even arpeggiation (2.3 and 3.2), I don't want a musically incorrect (or inaccurate) performance. I do not know if Debussy required a strict interpretation of his music but I doubt it, given the ambiguous notation and the "rubato" tempo indication. Nonetheless I'm asking here: what should I do?

  • Well, it says "Quasi cadenza" (and also stringendo) a copule bars before that, so it's probably intended to be quite free anyway. Then "Mesuré" would indicate returning back to stricter tempo. I would end the cadenza with a small slow-down, it's also almost required to prepare for ppp.
    – nonpop
    Commented Oct 18, 2014 at 14:47
  • Not sure about other users but your link to the Peters edition is timing out for me.
    – Richard
    Commented Oct 29, 2014 at 2:02
  • The edition you point out does not notate tuplets, which is kind of a no-no and indicates that it is pretty old / out of date (1905 as it is...) with standard notation practice. Metrically, each of the six groupings corresponds to a single triplet-eighth in 4/8 time. Metrically, since there are 6 notes in the last grouping instead of 8, it will of course be slower. The beaming also further supports the notion of 6 groups. #2 doesn't make sense as it would be beamed differently. #3 doesn't work because there would need to be a time signature change to indicate the tuplet - effectively 1/8. Commented Oct 29, 2014 at 4:17
  • Fixed the link, for posterity I guess.
    – user19087
    Commented Jan 1, 2015 at 5:32

1 Answer 1


"Impressionist composers should be banned from using water as inspiration. Just way too many notes." - don't shoot the pianist

In these circumstances, the composer's manuscript is always a valuable tool to use to determine why a publisher made the decision to do something the way they did. In this case, the manuscript is fortunately available to us freely because of the generosity and collaboration of the international musicians' community.

  1. It is highly improbable that not putting a bar line in to divide Measure 23 was a mistake. Judging from the same measure in the manuscript, "measures 23 and 24" are actually measure 23, as Durand and Sons reflect. Debussy Reflets dans l'eau, m.23, manuscript Moreover, it is unlikely that Debussy made an error and simply forgot to add the bar line: he is otherwise consistent and accurate in his writing.

  2. Given what we know from the manuscript, this section should be played very freely, out of time, without a sense of strict tempo, yet evenly. This is the meaning of "like a cadenza". I am not sure what you mean by "aesthetically". Paying attention to the important notes in the line, not the meter, is important here, because there is no meter until "Mesuré". (Think of water.)

  3. "Mesuré" here means back to the original meter, and also implies that we have come from a place where there is not as strict a meter (or a different one). The tempo should not vary, outside of a typical rubato of impressionistic music. Going with your 3.1 option is correct, and the groups of 64th notes should be assembled into half-measures for each according to how they are beamed. Debussy Reflets dans l'eau, m.26 & 27 snippet of RH arpeggios, Durand and Sons first edition Durand knows this and reflects it in their edition. The evidence: a prominent melody (rhythmically "superior" to the arpeggios) and the fact that it would be awkward to jump from one time signature to the next.

Congratulations on learning this beautiful piece.

  • I ended up transcribing the measures into Denemo and using the midi-playback (at a reduced tempo) to help integrate both hands. The result was 2.1, though it eventually relaxed into a more even rhythm. That said, I'm not sure I agree "quasi cadenza" implies an evenness, rather than just free time. The different note durations might be suggestive of relative speeds outside the meter. In either case - played evenly (2.3) or varied (2.1) - the result sounds like a cadenza.
    – user19087
    Commented Jan 1, 2015 at 5:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.