A sphere visiting a two-dimensional world is seen by its residents as a dot that swells into a disc, shrinks, and vanishes — the whole visitor, reported one slice at a time. This entry runs that famous thought experiment live, then turns the same machinery on us: exact cross-sections and shadows of four-dimensional balls and polytopes passing through our three-dimensional space, the strange bookkeeping of volume in dimensions five, six, ten — and what tabletop gravity, the LHC, and a neutron-star merger actually say about whether the extra dimensions are there.
r(w) = √(R²−w²) · V_n = π^{n/2}/Γ(n/2+1)Nothing in her physics is wrong — she lacks a direction, not intelligence. If a 4-D object ever crossed our space we would see exactly this movie one dimension up: a 3-D solid appearing from nowhere, morphing, and vanishing, with a perfectly good conservation-law alibi in the bulk. Views 2 and 3 show that movie, computed exactly.
The slice radius law never changes, but the volume profile (1−w²/R²)(n−1)/2 sharpens ruthlessly with n: a 10-ball keeps 49% of its volume within |w| < 0.2R, a 100-ball 96%. Almost the entire object crosses in a brief moment near w = 0 — and almost all of it hugs the surface. This concentration of measure is real mathematics with real consequences, from statistical mechanics to why machine learning in high dimensions is hard.
Cell-first, a tesseract crossing our space is just a cube that sits there for a while — utterly unremarkable. Corner-first, the same rigid object is a morphing parade of polyhedra. A 2-D scientist watching a tilted cube has the same problem: the slice sequence depends on the angle of approach, and no single slice ever shows the thing itself.
A slice shows true sizes but only one w at a time; a shadow shows every part at once but distorts sizes by depth. Neither is the object — exactly as a CT slice and an X-ray of the same skull disagree, and are both honest. Everything anyone has ever shown you of a tesseract is one of these two pictures.
A dimension is just an independent direction — one more number needed to say where something is. The ladder never asks permission to continue: the mathematics of n = 4, 5, 6… is exactly as solid as n = 3. Whether physical space has more directions than three is a separate, experimental question — and the panels below the stage report what the experiments actually say.
No 3-D eye will ever see a 4-D object whole — our retinas are 2-D and our visual cortex reconstructs 3-D, full stop. But "cannot see" is not "cannot know". There are exactly three faithful pictures, each a rigorous mathematical operation with a declared distortion, and every view above is one of them. Anything else you have ever been shown — glowing "hyperspace" tunnels included — is decoration.
Cut the object with the hyperplane w = const and show the intersection: true sizes, true shapes, but only one w-level at a time. This is the CT scan of higher dimensions — and the brane picture of Fig. 22.2: a 3-sphere greeting a 2-D world as a growing and shrinking disc. Slices here are computed exactly, edge by edge.
Map every point down one dimension, optionally with perspective along w: the whole object at once, but sizes lie by depth. The cube-in-a-cube tesseract is this and nothing more — the inner cube is the far cell. This is the X-ray, and it is how every "picture of a tesseract" ever drawn actually works.
Record the slice at every w and stack the record: the Flatlander's stop-motion reconstruction of her visitor, and exactly how a radiologist rebuilds a 3-D skull from 2-D scans. Time plays the role of the missing axis — which is why "is time the fourth dimension?" is a natural question with a careful answer: it is a fourth coordinate, but with the wrong sign in the metric. Entry XXIII covers that story.
Class: analytic — everything rendered on this page is exact, closed-form geometry; nothing is keyframed or approximated. The four computations:
Pythagoras, verbatim, in any dimension: the cross-section of an n-ball at bulk offset w is an (n−1)-ball of this radius. One line of mathematics drives the whole of views 1 and 2 — and it is the same line for n = 3 and n = 100.
Each of the solid's edges either crosses the hyperplane or doesn't; the crossings are found by linear interpolation and their convex hull is the slice — exact for any convex polytope. Audited live by the sweep integral ∫V(w)dw = exact 4-volume and by Euler's V−E+F = 2 on every frame.
The n-ball volume, computed by the exact two-step recursion (no Γ approximation). It peaks at n = 5 and dies toward zero; the slab and shell fractions in view 2 come from integrating the slice profile, checked against scipy's incomplete beta function.
Perspective projection along w — identical in form to how a camera flattens z — after rotating in genuinely four-dimensional planes. A double rotation turns two orthogonal planes at once, something with no 3-D counterpart; the projection matrix is checked orthogonal to 10⁻¹² at load.
The ladder's rule — every drag doubles each element and promotes a copy — compresses to one row of binomial arithmetic per dimension. Nothing here is conjecture; these are counts you can do on your fingers, and the page's slicer touches every one of the tesseract's 32 edges each frame.
| object | k | vertices 2ᵏ | edges k·2ᵏ⁻¹ | squares | cubes | tesseracts |
|---|---|---|---|---|---|---|
| point | 0 | 1 | — | — | — | — |
| segment | 1 | 2 | 1 | — | — | — |
| square | 2 | 4 | 4 | 1 | — | — |
| cube | 3 | 8 | 12 | 6 | 1 | — |
| tesseract | 4 | 16 | 32 | 24 | 8 | 1 |
| penteract | 5 | 32 | 80 | 80 | 40 | 10 |
| hexeract | 6 | 64 | 192 | 240 | 160 | 60 |
The four regular 4-polytopes the slicer carries: the 5-cell (4-simplex, five tetrahedra — the 4-D tetrahedron), the tesseract (eight cubes), its dual the 16-cell (sixteen tetrahedra), and the 24-cell (twenty-four octahedra) — the strangest of them, self-dual and with no 3-D analogue at all. In five dimensions and above the zoo collapses: only the simplex, hypercube, and cross-polytope families survive. Three is not the richest dimension for regular solids; four is.
Everything above is mathematics — unconditionally true whether or not nature uses it. The physical question is separate: does space have more than three directions? Serious physics has proposed yes, repeatedly, for a century — always with the same escape clause for why we don't see them, and (so far) always with null results where it was testable.
Write general relativity in five dimensions and electromagnetism falls out of the extra components of the metric — gravity and light unified by geometry alone. Klein's fix for invisibility: the fifth direction is a circle of subatomic circumference. Every point in space is secretly a tiny loop, far too small to notice — the template every later theory borrowed.
Superstring theory is only consistent in 9+1 dimensions, M-theory in 10+1; the surplus six or seven are curled into Calabi–Yau spaces whose shape sets particle physics. Braneworld models (ADD 1998, Randall–Sundrum 1999) instead pin all matter and light to a 3-brane — the literal Fig. 22.2 picture — while gravity alone leaks into the bulk, predicting Newton's law fails at short range.
Torsion-balance tests find gravity perfectly inverse-square down to 52 μm (Eöt-Wash 2020): any large extra dimension is smaller than a dust mote. The LHC finds no KK gravitons or micro black holes. And GW170817's gravitational waves arrived with exactly the amplitude 3+1-dimensional propagation predicts over 130 million light-years — D = 4.02 ± 0.1: gravity is not leaking anywhere. Extra dimensions remain legal, but they are cornered.
The geometry on this page is Tier A: slices, shadows, volumes and counts are computed from first principles and audited at load. The honesty lines worth drawing:
Every frame is a 3-D (then 2-D, on your screen) picture of a 4-D thing: an intersection or a projection, each with its declared distortion. No rendering, here or anywhere, shows "what a tesseract looks like" — there is no honest answer to that request, and this page does not pretend to one.
A 4-D object has no 3-D optics: nothing says the slice of a hyperball is violet or glossy. Materials exist to make shape legible; the w-tint on shadow edges is a depth code, not a colour anything "has". The geometry underneath the paint is the exact part.
View 1 uses the brane as Abbott and Thorne meant it — a perfect intuition pump. Real braneworld models are dynamical: branes have tension, gravity warps the bulk, and matter is confined by field physics, not geometry alone. None of that dynamics is simulated on this page; the panels report it as proposals with experimental bounds.
52 μm, TeV colliders, and GW170817 close windows; they cannot close the idea. A dimension curled below ~10⁻¹⁹ m, or a bulk only gravity at cosmological wavelengths could probe, remains untouched. As with the topology of Entry XXV, the honest verdict is a bound, not a verdict.
arXiv:hep-ph/9803315arXiv:hep-ph/9905221arXiv:2002.11761arXiv:1801.08160