Reading the stillness.

A printed atlas invites comparison. An interactive atlas can invite a prediction: if the waves stop balancing, will those perfect crossings survive?

Design study · November 18, 2026 · Inspired by interactive optical and wind studies, interface-design research, and a Chladni print.

Make an experiment visible

The engraving is the primary object. Wave controls sit beside it, while the nine original figures form a recognizable visual index. Changing the balance gives immediate feedback. Equal waves and Undo let the reader retrace an experiment without reconstructing the numbers.

This applies the outline's principles of direct manipulation, recognition over recall, visible state, and recovery. The specific hypothesis is that a reader can notice the broken crossing and explain it by changing one parameter. It has not been tested with participants.

In Explorable Explanations, Bret Victor proposes letting readers investigate assumptions inside the explanation. Here the question is attached to an action: break the crossing, inspect the result, and restore equality.

Three representations, one field

Engraving shows the zero contour. Vibration colors the positive and negative regions and reverses their signs over time. Sand shows an illustrative attraction toward the same zero contour. Click the plate and the probe, slice, and numeric reading all refer to the same coordinates.

Motion starts only when requested and can be paused. Switching representation preserves the wave numbers and balance. Changing a parameter begins a new sand distribution. The inspector keeps its location as the representation changes.

Native sliders, wave-number fields and step buttons provide keyboard and pointer alternatives. The probe supports clicking and arrow keys; no interaction depends on dragging. This follows the intent described in W3C's guidance on dragging movements. These choices are not a claim of full accessibility conformance.

Give a discovery somewhere to stay

Pins retain the wave numbers, balance, representation, and ink palette in browser storage. A thumbnail opens a fixed reference engraving while the active plate remains editable. Restoring a pin is undoable. Export creates a self-contained SVG or PNG with the current plate and its caption.

Sand particle positions and animation progress are transient. Pins restore the study's settings, not the exact instant of a moving simulation. Exporting a sand or vibration view captures its currently drawn pixels; engraving exports remain vector linework. Storage can be unavailable in private or restricted browsers, so the interface reports a failed save and offers export.

The mathematics and its boundary

The image supplies the cosine model. On a square from −1 to 1, the application draws the zero set of:

F(x,y) = [cos(nπx)cos(mπy) − b cos(mπx)cos(nπy)] / (1+|b|)

At b = 1, exchanging x and y reverses the sign, and both diagonals are nodal. Integer wave numbers are constrained to be different so an equal-wave plate does not collapse into the identically zero field. The common normalization changes displacement scale, not the zero set.

Actual Chladni demonstrations described by Harvard use sand to reveal nodal lines on vibrating plates. The model here reproduces the supplied mathematical figures. It does not solve a metal plate with specified material properties, thickness, support, or forcing. No physical frequency is inferred.

The sand animation uses bounded drift down the squared field toward its zeros. It explains gathering visually; it is not a granular dynamics simulation. Vibration is deliberately slowed and has no audio.

Questions for a reader study

  1. Predict what will happen when balance moves from 1.00 to 0.82. Change it, then explain which lines moved.
  2. Choose a point near a nodal line. Find the corresponding crossing on its slice.
  3. Pin two studies. Return to the first, then recover the plate you were exploring before the restore.
  4. Repeat the experiment on a phone and with a keyboard. Record missed controls and lost context.

The visual hierarchy, linked probe, and comparison shelf are design hypotheses. Browser checks establish that interactions work; they do not establish that readers understand the model.