3 THE MODEL
A grating is N equally spaced slits, pitch d. Waves from all slits add in phase only when the path difference between neighbours is a whole number of wavelengths:
d·sinθ = mλ (m = 0, ±1, ±2 …)
The N-slit amplitude is a ratio of sines; its principal peaks reach N² in intensity and narrow as 1/N. Dispersion (angle per wavelength) is dθ/dλ = m/(d·cosθ) > 0.
5 THE LINEAGE
Two slits give a soft cosine fringe (the-double-slit, Young 1804). Add more slits and the fringes sharpen into lines; at Fraunhofer's 260 wires they became a spectrum. The grating is the many-slit limit of the double slit, and the finer ruler that the-dispersion (a prism) can only sketch — but with the sign of bending reversed.
7 THE WITNESS
A live re-check of the engine: it re-derives the first-order angles for red and blue and confirms red bends further (real dispersion). If window 6 drops the λ term, the two angles collapse and this badge flips.
re-checking…
4 DATA IN in ↓
0 THE PANEL lit
Live grating: each visible wavelength sent to θ=arcsin(mλ/d) for every order that fits.
8 DATA OUT out ↓
A proven spectrum: red is thrown wider than blue in every order; peaks are 1/N thin; orders can overlap for a broad source.
booting…
1 THE ADVERSARY wall
d·sinθ=mλ is unavoidably wave.2 THE GRAVEYARD
“A grating bends red less than blue, like a prism.”
→ Opposite. dθ/dλ = m/(d cosθ) > 0, so longer λ bends more. Prism dispersion is normal (blue more); grating dispersion is reversed.
“You always see infinitely many orders.”
→ Bounded by |m| ≤ d/λ. Beyond it sinθ>1 — no solution.
“More slits move the lines.”
→ They don't move the peaks; they narrow them (~1/N) and raise resolving power R=mN.
6 THE TAMPER
The disclosed planted void: replace mλ with a constant — drop the wavelength. All colours then map to one angle and the spectrum dies. The Witness (7) catches it live.