Alison has drawn my attention to this video. It demonstrates vibrational modes of a square plate by using sand. At certain frequencies, there are well defined modes of oscillation, in which parts of the plate 'nodal lines' are stationary. The sand will find its way to these parts and trace out some lovely pictures.
Vibrational modes are often illustrated through waves on a guitar string. Here, the string is held stationary at both ends, but is free to vibrate elsewhere. There is a fundamental frequency of oscillation, where the distance between the ends of the string is half of a wavelength (this ensures the displacement of both ends of the string is zero since they are clamped). Since wavelength is related to frequency (frequency = speed/wavelength) that means if the wavelength is 2 L where L is the distance between the ends of the string, we have frequency = speed/2L.
But that's not the only possible mode. Another one would have L equal to a whole wavelength (equals two half wavelengths). Or one-and-a-half wavelengths (equals three half-wavelengths.) This gives us, rather neatly, frequency = n speed/2L, where n is an integer. We see that our 'harmonics' are just integer multiples of the fundamental frequency. Rather neat.
However, if you look at the frequencies given in the video, they appear to be all over the place. I challenge you to pull out the relationships between these (I've tried). There are a few reasons why the case shown on the video is considerably more complicated than the waves on the string.
1. The boundary conditions. The edges of the plate aren't clamped in place. This makes it less straightforward to define the modes geometrically.
2. The plate is square, giving rise to 'degeneracy' in the modes. This term refers to two or more distinct modes having the same frequency. You can see it rather well with the 4129 Hz mode. Basically, there are horizontal stripes shown. But equally, with the same frequency, you could get vertical stripes. Why don't the two occur together? They do. You can see the effect of having a little bit of vertical stripe most clearly at the far end of the plate, where the pattern becomes more square-like. Also, with a square, you can get two completely different types of mode with the same frequency. This occurs because what matters are the sums squares of pairs of integers. Broadly speaking (at least for a square clamped on the edges, which I must point out this ISN"T), our modes follow the relationship:
f = C sqrt(n^2 + m^2)
where C is a constant, 'sqrt' means square-root, and n^2 is n-squared. So, for example, not only is 50 equal to 5-squared plus 5-squared, it is also equal to 1-squared plus 7-squared (or 7-squared plus 1-squared). This gives us three modes all competing to appear at exactly the same time. What happens then isn't easy to tell.
3. Non-linear effects. This a physicist's code-word for 'it's all too difficult'. That's not quite true – arguably most of the interesting physics research happening in the world is looking at non-linear effects. What this really means is that, if A and B are both solutions of a problem, then some combination of A and B is NOT a solution. A lot of physics IS linear – Maxwell's equations in a vacuum is a good example – but a whole lot isn't. With waves, the speed of the wave usually depends on frequency (i.e. is not constant) which means we lose the nice, integer-multiple relationship of our waves-on-a-string mode.
So, enjoy the video for what it is, and don't try to analyze it TOO closely.