Pluck a guitar string and its overtones land at exactly 2x, 3x, 4x the fundamental frequency. That is not a coincidence or a property of the string material. It falls out of the one-dimensional wave equation: a string constrained at both ends can only support wavelengths that divide evenly into its length, and the resulting frequencies form a perfect integer series.
Hit a drum, and the physics is completely different. A drumhead is a two-dimensional membrane, and its resonant modes are governed by an entirely different set of rules. The frequencies are not 2x, 3x, 4x anything. They are irrational multiples of the lowest mode, determined by the zeros of a class of functions that a German astronomer originally developed to describe planetary orbits.
That is why a snare sounds like a snare, and not like a detuned piano string.
The One-Dimensional Case
For a string of length L under tension T with linear mass density mu, the frequency of the nth mode is:
f_n = n * (1 / 2L) * sqrt(T / mu)Every mode frequency is an integer multiple of the fundamental. The second mode is exactly 2x the first. The third is 3x. The ear hears this integer stack as a single fused pitch with timbre determined by the amplitude envelope of the partials. This is why stringed and blown instruments can carry definite musical pitch.
Moving to Two Dimensions
A drumhead is a circular membrane under radial tension. The two-dimensional wave equation in polar coordinates is:
d²z/dt² = c² * (d²z/dr² + (1/r) * dz/dr + (1/r²) * d²z/dθ²)where c = sqrt(T/sigma), with T as the surface tension and sigma as the surface mass density.
Applying separation of variables and imposing the boundary condition that displacement is zero at the rim, the radial component satisfies Bessel's differential equation. The solutions are Bessel functions of the first kind, J_n(kr). At the rim of radius a, the displacement must be zero, so J_n(k * a) = 0. The only permissible values of k are those where the Bessel function is exactly zero.
These zeros are not equally spaced. They bear no simple ratio to each other. And that is the entire problem.
The Mode Table
Each vibrational mode is labelled (n, m), where n is the number of nodal diameters crossing the center and m is the nodal circle count (including the outer edge). The frequency of mode (n, m) is proportional to the mth zero of J_n, written j_{n,m}. Normalized to the lowest mode (0,1):
| Mode (n,m) | Bessel zero j_{n,m} | Frequency ratio |
|---|---|---|
| (0,1) | 2.4048 | 1.000 |
| (1,1) | 3.8317 | 1.593 |
| (2,1) | 5.1356 | 2.136 |
| (0,2) | 5.5201 | 2.296 |
| (3,1) | 6.3802 | 2.653 |
| (1,2) | 7.0156 | 2.917 |
Compare that to a vibrating string: 1.000, 2.000, 3.000, 4.000. Every drum mode lands at an irrational multiple of the fundamental. The (1,1) mode is not an octave (2.000), not a perfect fifth above the octave (3.000). It sits at 1.593x, a ratio that corresponds to no named musical interval. The (2,1) mode at 2.136x overshoots the octave by roughly a semitone.
The ear, which has evolved to hear near-integer partial stacks as a single pitch, cannot fuse these inharmonic modes into a clear tone. What it hears instead is an attack transient with an ambiguous pitch character, which is precisely the acoustic signature of a drum.
What the Modes Look Like
Ernst Chladni (1756-1827) made these mode shapes visible in 1787 by sprinkling sand on a vibrating metal plate. The sand is thrown off the vibrating regions and settles along the nodal lines, where displacement is zero.
For a circular membrane, the (0,1) mode has no internal nodal lines: the entire surface moves up and down in unison. The (1,1) mode has a single nodal diameter, so one half of the membrane moves upward while the other moves downward. The (2,1) mode has two crossing nodal diameters, dividing the membrane into four alternating sectors. The (0,2) mode has one concentric nodal circle inside the rim, so the center and the outer ring move in opposite phases.
These mode shapes are geometrically ordered and regular. The frequencies they produce are not.
The Timpani Exception
Orchestral timpani can be tuned to definite musical pitches. This seems to violate everything above, but it does not.
In a timpani, the membrane stretches over a hemispherical copper kettle. The enclosed air volume acts as a spring that loads each mode differently. This air loading shifts the mode frequencies, but the shift is not uniform. Under normal playing and tuning conditions, the air loading in a typical kettle coaxes the (1,1), (2,1), (3,1), and (4,1) modes into frequency ratios close to 2:3:4:5 relative to a reference pitch. Research by Thomas Rossing and colleagues confirmed that kettle volume is the dominant variable, more than the geometric details of the bowl shape.
A timpanist striking the head at roughly one-quarter of the way in from the rim also suppresses the (0,1) mode. That lowest mode contributes almost nothing to the perceived pitch because it is barely excited at that strike position. What the ear hears is the near-harmonic cluster of the four preferred modes, which fuses into a perceived pitch much as a string's harmonics do.
The kettle turns an inherently inharmonic membrane into an approximately harmonic resonator through a purely mechanical coupling with air.
The History of the Functions
Friedrich Wilhelm Bessel (1784-1846) was a German astronomer working on perturbation theory for planetary orbits. His 1817 analysis of the problem produced the differential equation and the family of solutions that now bear his name. The connection to acoustics came later.
Leonhard Euler had encountered the same class of equations in 1764 while analyzing a vibrating membrane, but Bessel worked out the systematic theory. Lord Rayleigh's two-volume treatise "The Theory of Sound" (1877) formalized the application of Bessel functions to drum membranes and established the mode frequency ratios as a mathematical fact about circular boundary conditions. The table above has been experimentally confirmed many times since.
What This Means for Drum Synthesis
A synthesizer that sums harmonic sine waves cannot produce a convincing drum sound regardless of how many partials are used, because the intervals between partials are wrong. Getting the Bessel-function ratios right is necessary.
Sample playback captures the mode distribution automatically. The limitation is that pitch-shifting moves all partials in parallel, which breaks the Bessel ratios. A tom pitched down two semitones by a time-stretching algorithm will have the same inharmonic structure but transposed, which is musically acceptable. A pitch shift applied in the frequency domain that moves each partial by a fixed Hz offset will distort the ratios and make the drum sound wrong.
Physical modeling can simulate the membrane directly by solving the partial differential equation numerically, but real-time solutions of the 2D wave equation at audio rates are computationally expensive. Modal synthesis offers a practical alternative: model each mode as a damped second-order oscillator at the Bessel-function frequency, apply an initial impulse, and sum the decaying outputs. Each mode can have a different Q and decay time, which maps directly to how real drumheads lose energy unevenly across modes.
FM synthesis handles inharmonic spectra efficiently using non-integer carrier-to-modulator ratios. A modulator frequency at roughly 1.593x the carrier places sidebands near the (1,1) mode. John Chowning's 1973 paper in the Journal of the Audio Engineering Society, which introduced FM synthesis, specifically noted that non-integer C:M ratios produce the inharmonic spectra characteristic of bell and percussion timbres.
The 808 kick uses a different approach: a resonant bandpass filter or bridged-T oscillator excited by an impulse. The filter rings at one dominant frequency with a decaying pitch envelope. This is not physically accurate to membrane acoustics, but it is musically functional because the resonant decay approximates the behavior of the lowest drum mode. A detailed analysis of that circuit appears in a separate article on this site.
Why Drum Tuning Works the Way It Does
When you adjust the tension rods on a tom to "tune it to a B-flat," you are raising the absolute frequency of all modes proportionally, since the mode frequency scales with sqrt(T). The inharmonic ratios between modes stay fixed at the Bessel zeros. The drum does not become more harmonic at higher tension. It becomes a higher-pitched inharmonic drum.
What the ear does is latch onto the dominant partial in the attack transient, which depends on where the stick strikes and how hard. Tightening the head moves that dominant partial upward. Mixing engineers who layer a tuned sine below a kick drum during production are supplying the harmonic anchor that the membrane cannot provide on its own.
The Bessel function zeros are not a design choice. They are what emerges when you solve the wave equation on a disk with a fixed boundary. Every circular membrane in existence, from a studio snare to the eardrum in a listener's head, vibrates according to the same irrational ratios. The harmonic series is a special case of one-dimensional physics. Two dimensions are less cooperative.