We first met the idea of Hypermeter briefly in the Other Rhythmic Essentials chapter. As discussed there, the strong-weak style of patterns we have seen within measures do not necessarily only operate within the measure and stop at the measure-length pulse. Especially when we have short measures and fast music (like a presto in 3/4), there can be a strong sense of such metrical relations at “higher”, multi-measure levels. In Western tonal music, just as the most common division/grouping within measures is by 2s (and thus 4s, 8s etc), so it is also with hypermeter. That said, division/grouping by 3s is also eminently possible, as are 5s and 7s, and when you look more closely, you notice that interesting composers frequently rove between these options fluidly in a way that would be extremely unusual at the counting, tactus, beat level.

Hypermeter in 3s: “ritmo de tre battute”

One of the most famous, explicit notations of hypermeter is the “ritmo de tre battute” section of Beethoven’s 9th symphony. This marking explicitly points out that the measures group in 3s. In doing so, the instruction also makes explicit the assumption that the grouping had been in 2s and 4s up to that point. Example 1 reproduces the 4-measure grouping seen in the Other Rhythmic Essentials chapter:

Example 1. 4-measure hypermeter at the start of Beethoven 9/iv from the Other Rhythmic Essentials chapter.

Now Example 2 shows the later “ritmo de tre battute” section (from m.152) with comparable “hypermeter counts” as well as (editorial) double barlines to show each group of 3.

Symphony No.9, Op.125 by openmusictheory

Example 2. 3-measure hypermeter from measure 152: “ritmo de tre battute”.

Note that we’re talking about the “period” of the grouping here: that is, grouping in 3s or 4s. There’s plenty more to say about the phase: that is, which measure in the cycle should be counted as “1”. There’s a lot to argue for the phase shown in this movement, although the dominant-to-tonic movement in the first to second measure of the figure sure might make you wonder about starting the phase a measure later.

Additional levels of 3s at the Fast End?

Like the common practice usage of 9-unit time signatures (“9/4”, “9/8”, … ),^[1] “ritmo de tre battute” explicitly sets out two levels of 3-grouping. So, can we have more than 2? Well, “27/X” time signatures are not seen (at least in this style), and neither is anything like “ritmo de 9 battute”, but Beethoven does occasionally glance onto 3 sets of 3 levels in the late style with triplets: i.e., adding another layer of 3-time at the “short” or “fast” end of the spectrum:

• In movement 4 of the string quartet no.14 (Op.131), there is an “Adagio ma non troppo e semplice” section in 9/4 time (from m.186), so we already have two levels of 3s right away (3×3 quarters). Later, it briefly uses triplet eight notes long enough to perhaps be called a metrical level (mm.223–225, or 5–7 measures before the final Allegretto reprise).
• the second movement of last piano sonata (No.32, Op.111) is an “Arietta” in compound time signatures: 9/16 at first but also 6/16, and 12/16 for some sections. Additionally, some of those sections use triplet sixteenth notes to give another 3-level. When this happens in 9/16, we have 3 levels of 3 grouping that could have been written as 27/32. That said, the hypermeter is regular, so 3x3x3 seems to be the limit for now. And the triplets in 6/16 and 12/16 add another 3-grouping, but with 2-grouping in the metrical levels.

Sensible Limits

(Where) do we stop looking for metrical levels above and below? Some accounts of hypermeter have extended the idea to entry movements, describing them, for example, as a huge upbeat-downbeat pair. The psychology literature suggests a more modest scope, with meter giving way to form not necessarily at the limit of the notated measure, but also not far beyond. The idea is that we struggle to perceive metrical cycles and patterns when the pulse is too short (about 1/10th of a second) or too long (a few seconds, depending on the internal content). So in practice, it makes sense to consider hypermeter when the measure is very short (like the 3/4 presto discussed above) but not for too many more levels.

Not limited, and perhaps not so sensible either

Composers, theorists, and especially composer-theorists reading this may be reading this and wondering about the extreme cases. Notwithstanding the psychological limits to meter as distinct from form, what’s the structural limit, the ne plus ultra? I.e., how many levels of 3-grouping can we stack on top of one another? I certainly wondered this and I couldn’t resist writing a piece that groups in threes and every structural level from the individual notes right up to the highest formal units. The title “Sierpiński’s Triangle” refers to a triangle-fractal pattern that shares something of the same structure. It’s the first movement of a set of “Tessellations”: click here for the full score on IMSLP.

1. Coming soon!