The imaginary component: what i is, why i² = −1, and how growth becomes rotation
The number line only lets you go forwards and backwards. The complex plane adds a second direction — up and down — and i is simply the name of “one step in that new direction.” Once you see multiplication by i as a quarter-turn, everything else follows: i² = −1 stops being a mysterious rule, and Euler’s formula eiθ = cos θ + i sin θ turns out to be compound growth that has learned to turn.
Part 1 — Why does i² have to be −1?
Start with the honest problem. On the ordinary number line, squaring any number — positive or negative — gives something positive or zero:
So no real number squares to −1. Equations like x² + 1 = 0 simply have no answer on the line. Mathematicians had two choices: declare such equations unsolvable forever, or invent a new direction and see whether the resulting system stays consistent. They chose the second — and consistency is exactly what forces i² = −1. There are three ways to see why, and they all agree.
① By definition — it’s the job description
i is defined as a number whose square is −1, written i = √−1. That isn’t circular trickery; it’s how every number system grew. “−3” was once just as scandalous (“you can’t have fewer than nothing”), and “2/3” required inventing fractions. Each time, we kept the old rules of arithmetic and asked: what new object would make the unsolvable solvable? For x² = −1, the answer is an object with exactly one defining property: square it and you get −1.
② By geometry — two quarter-turns must face you backwards
Here is the deeper reason the definition works. Multiplying by −1 flips the number line 180°: 3 becomes −3. Now suppose multiplying by i means rotating 90° counter-clockwise (the diagrams in Part 2 justify this). Then multiplying by i twice is a 180° rotation — which is exactly what multiplying by −1 does. Two operations that do the same thing to every number must be the same operation:
So i² = −1 isn’t an arbitrary decree. Any consistent “quarter-turn number” is forced to square to −1.
③ By consistency — the alternatives break arithmetic
Suppose instead we declared i² = +1. Then i would just be ±1 in disguise — nothing new. Suppose i² = 0. Then (1 + i)(1 − i) = 1 − i² = 1, but also expanding carefully gives contradictions with the distributive law once you try to divide. Only i² = −1 keeps addition, multiplication, and division all working — and it pays off immediately: every polynomial equation now has solutions. That payoff is the real test, and i passes it.
A complex number is then just a pair of instructions — go this far along the real axis, then this far up the imaginary axis:
The word “imaginary” is unfortunate historical baggage (Descartes meant it as an insult). The imaginary component is no less real than the “negative” component of a bank balance: it measures displacement along a genuine, usable direction — the vertical one.
Part 2 — The complex plane and its four major directions
Draw the real numbers horizontally and the imaginary numbers vertically. Every complex number a + bi is a point — equivalently, an arrow from the origin. The four unit arrows, pointing right, up, left, and down, are the compass points of this plane. Multiplying any number by one of them rotates it; multiplying by i in particular rotates it exactly one compass point counter-clockwise.
1 = 1 + 0i
The starting direction. Purely real, no imaginary part. Multiplying by 1 changes nothing — rotation by 0°.
1 × i⁰ — distance 1, angle 0°
i = 0 + 1i
One quarter-turn from 1. Purely imaginary. This is the new direction: 1 × i = i. Nothing on the number line points here.
i × i⁰ — distance 1, angle 90°
−1 = −1 + 0i
Two quarter-turns from 1 — and this is i². Apply ×i again: i × i lands here. A half-turn is the same as negating, which is the geometric proof that i² = −1.
i² = −1 — distance 1, angle 180°
−i = 0 − 1i
Three quarter-turns from 1: i³ = i² · i = −i. It is also one quarter-turn clockwise from 1 — multiplying by −i rotates the other way.
i³ = −i — distance 1, angle 270° (or −90°)
The cycle closes: i⁴ = 1
Four quarter-turns make a full turn, so the powers of i repeat forever with period 4: 1 → i → −1 → −i → 1 → i → … Keep this cycle in mind. In Part 4, growth steps will take tiny fractions of a quarter-turn at a time — and that is where circles and Euler’s formula come from.
Try it: multiply by i, over and over
Start at 3 + 1i. Each click multiplies by i using the rule (a + bi)·i = −b + ai — watch the arrow turn 90° and the coordinates swap and negate.
After 4 clicks you are back where you started — the period-4 cycle, felt rather than just stated. After 2 clicks the value is exactly negated: that is i² = −1 doing its work on a general number, not just on 1.
Part 3 — The familiar story: compound growth and e
Put $1 in an account paying 100% interest per year. If the interest is credited once, you end with $2. Credited twice (50% every six months), you end with (1 + ½)² = $2.25 — a little more, because the first interest itself earns interest. Credited n times a year, each credit adds 1/n of your current balance:
As n grows the total creeps upward but never explodes — it settles at a specific number, e ≈ 2.71828. This limit defines e, and its continuous version is the most important growth formula in mathematics:
The chart below is the “raw” growth story — the one every later diagram will be overlaid against. Orange shows what happens inside a single year of compounding as you raise the crediting frequency: the balance climbs along (1 + 1/n)k and the endpoint creeps toward e.
Key observation for what follows: with real growth, every step multiplies the balance by (1 + 1/n) — a factor bigger than 1 pointing in the same direction. The path never turns; it just stretches along the number line. Now ask the question this whole page exists to answer: what changes if the “1/n” we add each step points up instead of right?
Part 4 — Now add the imaginary component
Make one tiny change to the compounding factor. Instead of growing by a real 1/n each step, grow by an imaginary 1/n — i/n:
Each step now multiplies by a number that is mostly 1 (so the size barely changes) but tilted slightly upward (so the direction rotates a little). Multiplication in the plane does two things at once — it scales by the factor’s length and rotates by the factor’s angle. The factor (1 + i/n) has length √(1 + 1/n²), just above 1, and angle arctan(1/n), a small turn. Repeat it n times and the small turns add up to exactly 1 radian (about 57.3°) while the extra length washes out to 1.
How to read each diagram below: every figure shows the two versions of the same formula side by side, for the same n and the same step counts. On the left, the raw compound-growth line chart from Part 3 — (1 + 1/n)k climbing along toward e. On the right, the same formula with the imaginary component added — (1 + i/n)k plotted in the complex plane, with the orange dots on the axis marking where the plain real steps land, so the two can be compared directly. Watch the teal path bend further as n increases.
Step 1 — n = 1: one giant, crude turn
factor per step: 1 + i/1A single step multiplies by (1 + i) — equal parts “stay” and “turn.” Its angle is a full 45° and its length is √2 ≈ 1.414, so the one teal point lands at 1 + i, far outside the unit circle. The real version makes the same single leap to (1 + 1) = 2. Neither is close to its limit yet — but notice the geometry is already different in kind: the orange curve on the left just climbs, the teal point on the right has left the axis entirely.
real → 2.000 (limit e = 2.718)
complex → 1.000 + 1.000i
length = 1.414 (limit 1) · angle = 45.0° (limit 57.3° = 1 rad)
Step 2 — n = 2: two half-size turns
factor per step: 1 + i/2Two steps of (1 + i/2). Each factor has length ≈ 1.118 and angle ≈ 26.6°, so after two steps the total turn is ≈ 53.1° and the point sits at 0.75 + 1.00i. The path now visibly bends at the midpoint instead of jumping in one diagonal. The real twin, (1.5)² = 2.25, has inched closer to e.
real → 2.250 (limit e = 2.718)
complex → 0.750 + 1.000i
length = 1.250 (limit 1) · angle = 53.1° (limit 57.3° = 1 rad)
Step 3 — n = 4: the staircase starts to curve
factor per step: 1 + i/4Four steps of (1 + 0.25i), each turning only ≈ 14°. The teal staircase now hugs a curve: the endpoint is ≈ 0.629 + 0.938i, with length ≈ 1.13 — the unwanted growth is shrinking, while the accumulated turn (≈ 56.1°) is closing in on 1 radian. The orange line chart, by contrast, keeps bunching exponentially toward (1.25)⁴ ≈ 2.441.
real → 2.441 (limit e = 2.718)
complex → 0.629 + 0.938i
length = 1.129 (limit 1) · angle = 56.1° (limit 57.3° = 1 rad)
Step 4 — n = 8: recognisably an arc
factor per step: 1 + i/8Eight steps of ≈ 7.1° each. The teal points trace a clear arc of total turn ≈ 57.0°, ending near 0.579 + 0.893i with length ≈ 1.06. Two trends are now unmistakable: the endpoint’s angle is converging to exactly 1 radian, and its length is converging down to 1 — growth is draining away, pure rotation remains.
real → 2.566 (limit e = 2.718)
complex → 0.579 + 0.892i
length = 1.064 (limit 1) · angle = 57.0° (limit 57.3° = 1 rad)
Step 5 — n = 32: almost the circle
factor per step: 1 + i/32Thirty-two steps of ≈ 1.8° each. The staircase is nearly indistinguishable from a circular arc, ending within about 0.01 of the target eⁱ = cos 1 + i sin 1 ≈ 0.540 + 0.841i on the unit circle (dashed). Meanwhile the real version has reached (1 + 1/32)³² ≈ 2.677, close to e — it converged by stretching; the imaginary version converged by bending.
real → 2.677 (limit e = 2.718)
complex → 0.549 + 0.855i
length = 1.016 (limit 1) · angle = 57.3° (limit 57.3° = 1 rad)
Step 6 — n = 200 → ∞: Euler’s formula
factor per step: 1 + i/200At n = 200 the teal path lies on the unit circle for all practical purposes, landing at ≈ 0.541 + 0.843i. Let n → ∞ and the approximation becomes exact: the length factor (1 + 1/n²)^(n/2) → 1, the total angle n·arctan(1/n) → 1, and the endpoint is exactly cos 1 + i sin 1. That limiting statement, for a general total angle θ, is Euler’s formula: eiθ = cos θ + i sin θ.
real → 2.712 (limit e = 2.718)
complex → 0.542 + 0.844i
length = 1.003 (limit 1) · angle = 57.3° (limit 57.3° = 1 rad)
Part 5 — The limit: Euler’s formula
Nothing in Part 4 depended on the total turn being 1 radian. Replace i by i·θ — tilt each step so the turns add up to θ radians instead — and the same limit gives the general statement:
Read it geometrically: eiθ is the point you reach by taking infinitely many infinitesimal “grow-and-turn” steps whose turns total θ. Because the growth part washes out to a factor of exactly 1, you end up on the unit circle, at angle θ — whose coordinates are, by definition, (cos θ, sin θ). Exponential growth and circular motion are the same process, viewed along different axes.
The famous special case
Set θ = π — a half-turn, travelled by infinitely many tiny turns. Euler’s formula gives eiπ = cos π + i sin π = −1 + 0i, i.e.
Euler’s identity. It is not a coincidence bolted together from five constants; it is the statement “a half-turn lands you at −1,” written in exponential language — the same fact as i² = −1, one level up.
Why this matters beyond beauty
- Rotation becomes multiplication. To rotate any signal, vector, or waveform by θ, multiply by eiθ — one operation instead of a matrix.
- Waves become exponentials. cos θ and sin θ are just the real and imaginary parts of eiθ, so oscillations can be added, differentiated, and filtered with ordinary algebra. This is the working language of AC circuits, audio, radio, and quantum mechanics.
- Growth and turning unify. ex grows; eix turns; e(a+bi)x spirals — one formula covers bank accounts, pendulums, and decaying oscillations.
The one-sentence summary: i is a quarter-turn; i² = −1 because two quarter-turns make a half-turn; and Euler’s formula is what compound growth looks like when every growth step is also a tiny turn.
The lab: drag n and θ, watch growth become a circle
The interactive plane below plots (1 + iθ/n)k step by step for your chosen n and total angle θ — compare it with the unit circle and the target point eiθ. The companion real chart shows (1 + x/n)n converging to ex with x = θ, so you can see the two limits together. Push n to the top (576) and the teal staircase melts onto the circle.
The orange curve is (1 + θ/n)k over the same n steps; the dashed line is its limit eθ. Same n, same θ — the real version stretches, the imaginary version turns. That single swap, real 1/n → imaginary i/n, is the entire difference between an exponential and a circle.
Real limit: eθ · Imaginary limit: eiθ = cos θ + i sin θ
All diagrams on this page are computed from the formulas shown — the step plots evaluate (1 + 1/n)k and (1 + i/n)k exactly for each finite n, and the dashed targets are e and ei = cos 1 + i sin 1 ≈ 0.540 + 0.841i.
Footnote: most of this page was generated by Meta’s Muse assistant from a zero-shot prompt. It contains only limited cosmetic tweaks I made afterwards for taste. This is pretty remarkable, considering that I gave the same task to all the other LLMs that major providers offer with a budget of daily free tokens - and all of them messed up somewhere, requiring several rounds of iterative refinement with lots of handholding while still providing an inferior end result. One model that I won’t name (but it was from one of the top two providers) exhausted its daily tokens quota twice, and the result was still unsatisfactory, with factual errors and divergence from instructions given. Looks like Muse seems really well-suited for making visual explainers of this kind.