Three tasks that look as easy as the rest
By now you can bisect an angle, copy an angle, drop a perpendicular, and inscribe a hexagon — a whole workshop of constructions, each one proved exact. So three more requests from the ancient Greeks sound like just another afternoon's work. Trisecting an angle: given any angle, build the two rays that cut it into three equal parts. Doubling the cube: given a cube, build the edge of a second cube with exactly twice the volume. Squaring the circle: given a circle, build a square with exactly the same area. Each is stated in one short sentence, and bisection — cutting an angle into two equal parts — you already do with ease.
And yet these three resisted every geometer for more than two thousand years. Archimedes, the Persian and Islamic mathematicians, the Renaissance masters — all tried, all failed, and many produced beautiful approximate answers or solutions using extra tools. What nobody managed was an exact construction under the strict rules of this rung: unmarked straightedge, compass, finitely many steps. The honest reason this matters is that 'we couldn't find one' is a very different statement from 'there cannot be one' — and only in the 1800s did the second, far stronger claim get proved.
The one idea that decides all three
Everything turns on the bridge you met in guide 4. Start from a unit segment on a coordinate grid; a length is a constructible number exactly when some finite construction produces it. The two moves give you addition, subtraction, multiplication, division — and one extra gift, the square root. That is the whole toolkit: starting from the rationals you may, at each step, adjoin the square root of a number you already have. Anything you can build is a number reachable by a finite tower of square roots — and crucially, nothing else is.
The hidden engine underneath is a single algebraic fact: each time you intersect two lines you only solve a linear equation, and each time a line meets a circle, or a circle meets a circle, the worst you ever solve is a quadratic. A quadratic introduces at most a square root and nothing deeper. So no matter how long and clever your construction, you never escape the world of square roots. This is the lever that pries each impossible problem open — and it is also the honest reason the full proofs sit just beyond this rung: turning 'finite tower of square roots' into a precise, watertight criterion needs the field theory of the 1800s (degrees of field extensions, the work of Wantzel, Gauss, and Lindemann).
Each construction step solves at worst a quadratic: line meets line -> linear equation (no new roots) line meets circle -> quadratic equation (one square root) circle meets circle -> quadratic equation (one square root) So every constructible length lives in a TOWER of square roots: Q -> Q(sqrt a) -> Q(sqrt a, sqrt b) -> ... (finitely many) The 'degree' of such a number over Q is a power of 2: 1, 2, 4, 8, ... A number whose degree is NOT a power of 2 == not constructible.
Doubling the cube and trisecting the angle: the cube-root wall
Take doubling the cube first, because it is the cleanest. Let the given cube have edge 1, so volume 1; the new cube needs volume 2, so its edge x must satisfy x^3 = 2. That edge is the cube root of 2 — written 2^(1/3). The question is now purely arithmetic: is 2^(1/3) reachable by a finite tower of square roots? It is not. The cube root of 2 satisfies a cubic equation that cannot be broken into quadratics; in the language above its 'degree' over the rationals is 3, and 3 is not a power of 2. A construction can only ever land on numbers of degree 1, 2, 4, 8, and so on. Degree 3 is forbidden, so the edge cannot be built, and the cube cannot be doubled.
Trisecting an angle hides the very same cube-root wall behind a trigonometric disguise. The honest claim is not that no angle can be trisected — a right angle splits into three 30-degree pieces with ease, since you can construct 30 degrees directly. The claim is that no method works for every angle. The clean witness is 60 degrees. Trisecting it means constructing 20 degrees, which means constructing cos(20 degrees). Plug 60 degrees into the identity cos(3 theta) = 4 cos^3 theta - 3 cos theta: with cos(60 degrees) = 1/2 you get 8 x^3 - 6 x - 1 = 0, where x = cos(20 degrees). That cubic has no rational root and refuses to factor through quadratics — degree 3 again. So cos(20 degrees) is not constructible, and a 60-degree angle cannot be trisected with these tools.
Squaring the circle: a different, deeper wall
The third problem fails for a reason that is related but genuinely deeper. Give the circle radius 1, so its area is pi. A square of equal area needs side s with s^2 = pi, that is s = sqrt(pi). Since constructible numbers are closed under square roots, sqrt(pi) is constructible exactly when pi itself is. So the whole question collapses to one famous number: is pi a constructible number?
Here the cube-root story is not enough, because the obstruction is bigger. Every constructible number is a root of some polynomial with whole-number coefficients — such numbers are called algebraic. But in 1882 Ferdinand von Lindemann proved that pi is transcendental: it is the root of no polynomial with whole-number coefficients at all. A transcendental number cannot sit in any tower of square roots, cube roots, or roots of any kind, because it escapes every polynomial equation. So pi is not even algebraic, let alone constructible, and the circle can never be squared with straightedge and compass. This is the only one of the three whose proof reaches outside the degree argument entirely.
What this victory actually means
Pause on how strange and wonderful this is. A question about ink, paper, and two wooden tools was answered by algebra — by translating each geometric task into 'which numbers can I reach' and then proving a target number lies out of reach. The drawing board could never settle these questions, because no amount of drawing can prove that every possible construction fails. Only a proof about the structure of numbers can rule out the infinitely many constructions you will never try. This is the same spirit you will meet again at the summit of this whole ladder, where deep theorems tie geometry to algebra and analysis.
And resist the cheap reading that impossibility is a defeat. Knowing the exact edge of a tool's power is itself a triumph — it is the difference between superstition and understanding. Anyone who hands you a folded-paper trisection or a marked-ruler cube-doubling has not broken these theorems; they have changed the game, which guide 1 told you is always allowed. Within the compass-and-straightedge rules the verdict is final, proven, and beautiful: not 'we gave up', but 'we understood the rules so completely that we could prove where they end'.