A game with exactly two moves
You have already met Euclid's postulates as the rules a proof may lean on. The first three of them are not really statements about the world — they are permissions to draw. 'Through any two points you may draw a line.' 'You may extend a line as far as you like.' 'Given a centre and a point, you may draw the circle through that point.' Read that way, the postulates quietly hand you two physical instruments and tell you precisely what each is licensed to do. A geometric construction is any figure you can produce using only those licensed moves, in some finite number of steps.
The two instruments are the straightedge and the compass, and the most important thing to absorb is how little each one can do. The straightedge is a ruler with no marks on it: it can draw the straight line through two points you already have, and nothing more. It cannot measure a length, it cannot transfer a length, it cannot draw a line parallel to another 'by eye'. The compass can draw a circle (or an arc) centred at one point and passing through a second point — that is its entire repertoire. Neither tool produces a single new point on its own. New points appear only where two of these lines and circles cross.
What 'exact' really means here
When we say a construction gives a midpoint or a perpendicular, we do not mean 'close enough to look right'. We mean mathematically exact — perfect in the idealised world where lines have no thickness and points have no size. The pencil mark on your page is, as ever, a friendly stand-in for the true object. A construction is correct when you can prove, from the postulates and earlier theorems, that the point produced is the exact one claimed. The drawing persuades the eye; the proof persuades the mind, and only the proof counts.
This is why constructions belong here, beside proof, rather than in an art class. Every construction in this rung comes in two halves: the recipe (the sequence of straightedge and compass moves) and the justification (an argument, often resting on the congruent-triangle tests you already know, that the recipe really does what it claims). You will see in the next guide that to prove a construction bisects a segment, you build two triangles and show they are congruent — the construction and its proof are one and the same idea wearing two faces.
The collapsing compass — and why it does not matter
Here is a subtlety that catches almost everyone. The compass Euclid actually licenses is a collapsing compass: you may draw a circle with a given centre through a given point, but the moment you lift the compass off the paper it snaps shut and forgets the radius. A real metal compass holds its opening, so you can stab it down somewhere new and draw a circle of the same radius elsewhere. The idealised tool cannot — every circle must be pinned to a centre and a point it actually passes through. So it seems the postulates forbid the most natural move of all: 'carry this length over there'.
The rescue is the very first proposition Euclid proves about constructions, and it is a small marvel. The compass-equivalence theorem shows that anything a fixed, length-holding compass can do, a collapsing compass can also do — by a clever detour through a few extra circles. In effect, copying a length from one place to another is achievable with the weaker tool; it just takes more steps. Because of this theorem we are allowed, ever after, to use the convenient rigid compass freely and pretend the collapsing problem never existed. Honesty demands we know it is a theorem, not a free assumption.
From moves to numbers
Here is the bridge that makes this whole rung more than a craft. Drop a coordinate grid onto the plane and start, as the classical game does, with just two points: call them (0, 0) and (1, 0), so that the segment between them has length 1. Now ask: which lengths can you build from there using only the two moves? A length is called a constructible number if some compass-and-straightedge construction, starting from a unit segment, produces a segment of exactly that length.
The surprise is how rich this set is. Lining segments up end to end gives you addition and subtraction; the copying move you met above lets you transport lengths; using similar triangles (which you can also construct) gives multiplication and division. So every fraction is constructible. And one more move sneaks in for free: from a length L you can always construct the square root of L. That single fact — that constructible numbers are closed under square roots but not under cube roots or anything wilder — is exactly the fault line along which the impossible problems will eventually crack.
Two moves on the page -> What they do to lengths
------------------------------- -----------------------
lay segments end to end -> a + b and a - b
similar triangles -> a * b and a / b
semicircle + perpendicular -> sqrt(a)
Constructible from {0, 1}: +, -, *, /, and sqrt (any finite mix)
NOT reachable this way: cube roots, pi, ... (the impossible three)The promise and the honest limits
With just these two tools you can do a startling amount, and the rest of this rung walks through it: bisect a segment or an angle, copy an angle, drop or erect a perpendicular, draw a parallel, inscribe a circle in a triangle and circumscribe one around it, and build regular polygons — the equilateral triangle, the square, the regular pentagon, and Gauss's astonishing regular 17-gon. The right tools, used with care, reach much further than they have any right to.
And yet there are three things you cannot do, no matter how clever you are or how long you try: trisect an arbitrary angle, double the cube (build a cube with twice a given cube's volume), and square the circle (build a square with the same area as a given circle). These are not unsolved — they are proved impossible, and the proofs run exactly through the number idea above: trisection and doubling demand a cube root, squaring the circle demands pi, and none of those is reachable by a finite tower of square roots. The full arguments need algebra beyond this rung, so we will be honest about that when we reach them; for now, hold the shape of the answer.