Mälzel's identification of primary tempi with basic note value related by powers-of-two (eighths, quarters, halves, wholes) provides a nice segué to irama in Javanese gamelan music (or karawitan). In the irama system, tempo and density smootly interact to create an equilibrium, a prevailing steady-state over longer forms. The process is somewhat akin to shifting gears in a car, which is also a process which most drivers internalize and something much easier to do than to explain. So I'll try to explain with a radically simplified version of the idea. Imagine a melody in a steady stream of eighth notes which smoothly slows down to around half the orginal speed. At this point, the melody is now re-notated as a stream of quarter notes, and it is now ornamented by eighth note figurations, which articulate the initial melodic tempo. The melody has been, in effect, stretched, but the overall apparent speed of the music, in terms of the density of attacks, returns to that heard at the beginning of the piece. (The analogy with gears is quite precise: whatever the ratio of size and teeth between two gears, the circumferences of the gears move with the same linear motion). In Javanese music, this process may be repeated at three to five different levels (related at approximate 2:1's), with the move from one level to the next connected by decreases and increases in the tempo of the melody, each decrease in tempo matched by an increase in the density of figuration and vice-versa. There is, of course, a tremendous amount of specific stylistic detail which I am omitting from this description, especially regarding the particular shapes of the increases and decreases in tempo as well as sustained periods at a given level, the internalization of which are quite critical to the ability of an ensemble to play with such precision that durations of performances of works sharing the same form with durations of a half-hour or more may vary by only a few seconds. I believe that the irama principle, connecting tempo and density, offers some rich possibilities for extension, for example to basic tempi proportions other than 2:1, or to discover new or alternative formal distributions of accelerandi, decellerandi, and plateaus.
See also this post about Monteverdi's genere concitato.