Rhythmic organization in much of Schoenberg's music extends and even transcends common metrical practice. Innovation is evident especially in his middle-period works where the texture is highly polyphonic and the often canonic voices are saturated with a few distinct motives. Although regular pulses are fleetingly evident in these works, the surface rhythms change rapidly and irregularly. Constantly changing meter signatures apparently preclude the regular hierarchy of pulses necessary for hypermeter. For example, the opening of the Little Piano Piece, Op. 19 No. 1, reproduced in Example 1, presents a polyphony of temporally overlapping but distinguishable rhythmic groups, each consisting of diverse arrangements of accents. The irregular accents do not conform to the notated meter signatures, nor to any single regular pulse for that matter. Even in pieces that parody conventional tonal forms and phrase structures, such as the texturally and metrically more regular twelve-tone works, rhythmic gestures are accentually shaped contrary to the notated meter. 1 The fragmentary texture in this music suggests analyzing it in terms of pitch-class sets-that is, essentially without re-1 See, for instance, the composer's diacritics in m. 22 of the Prelude to the Suite, Op. 25, or in mm. 3-4 of the first movement of the Fourth String Quartet, Op. 37. gard to accents arising from pitch order, contour, or duration within segments. However research by Harald Krebs and Paul Johnson has revealed that Schoenberg conceived some of his pieces primarily rhythmically, insofar as he sketched several versions using the same rhythms but different pitches. 2 Moreover in some of his theoretical writings Schoenberg cites some complex polyphonies by Mozart and Brahms as instances of a "progressive" rhythmic practice that influenced his music. 3 Manifestly an analytical method is 2Harald Krebs, "Three Versions of Schoenberg's Op. 15, No. 14: Obvious Differences and Hidden Similarities," Journal of the Arnold Schoenberg Institute 8, no. 2 (1984): 131-40. His Example 1 demonstrates a "great similarity" of the vocal rhythm in the three sketched versions of Op. 15, No. 14; he also describes a similarity in motives, and in the use of hemiola. Paul Johnson, in "Rhythm and Set Choice in Schoenberg's Piano Concerto," Journal of the Arnold Schoenberg Institute 11, no. 1 (1988): 38-51, displays sketches of three versions of the concerto's opening. The sketches use different twelve-tone "sets" but the rhythm is the same in every case. Johnson comments: "The rhythm of the opening 35 bars took precedence over his pitch ideas.... Rhythm was the primary feature of all the other sketches of the opening.... It is the pitches in the final version that are adjusted, not the rhythm" (48) .