Off the Page, Into the Room Introduction to Three Dimensional Geometry

Every point on a flat page can be pinned down with exactly two numbers: how far across, and how far up. That was enough for years of geometry, because a page has only two directions to measure in. Step off the page into an actual room, though, and a third question appears immediately: how far above the floor? A ball thrown across a room, the tip of a ceiling fan, the corner of a table - none of these can be located with just two numbers, because a room is not flat. Fixing a point in space needs three mutually perpendicular measuring directions instead of two, and the meeting point of a floor and two adjacent walls gives exactly that. Call this meeting point the origin O, and call the three directions the x-axis, y-axis, and z-axis. Every pair of these axes sweeps out a flat coordinate plane - the XY-plane, YZ-plane, and ZX-plane - and together the three planes are all the machinery three-dimensional geometry needs. A point that used to need two numbers on a page now needs three: an ordered triplet (x, y, z).

In flat 2D geometry, two perpendicular axes cut the page into 4 quadrants, and the signs of the two coordinates tell you instantly which quadrant a point sits in. Three coordinate planes do the same job in space, except each plane independently splits space into a positive side and a negative side, and three independent splits combine as 2 x 2 x 2 = 8 regions, called octants. These eight octants are labelled with Roman numerals I through VIII, and every possible combination of signs for x, y, and z corresponds to exactly one of them: (+,+,+) is octant I, and every other sign pattern has its own fixed label. Reading off the octant from a point's coordinates is pure sign-checking: the point (-3, 1, 2) has a negative x, positive y, positive z, which the standard octant table places in octant II, while (-3, 1, -2) - same x and y, only the sign of z has flipped - lands in a completely different region, octant VI. Nothing about the size of the numbers matters for this, only their signs.

Given any point P in space, how do you actually read off its three numbers? Drop a perpendicular from P straight down onto the XY-plane, landing at a foot-point M; then drop a further perpendicular from M onto the x-axis, landing at L. The three measured lengths OL, LM, and MP are exactly the x, y, and z coordinates of P - three edges of an invisible rectangular box sitting between the origin and the point. Equivalently, draw three planes through P, each parallel to one of the three coordinate planes; these planes meet the x-axis, y-axis, and z-axis at points A, B, and C, and the distances OA, OB, OC give the same triplet (x, y, z). Both descriptions are the same box viewed two different ways, and the correspondence runs in both directions: starting instead from any triplet (x, y, z), the same construction locates exactly one point P in space. A few positions are worth knowing by sight: the origin itself is (0,0,0), any point sitting on the x-axis alone has the form (x, 0, 0), and any point sitting in the YZ-plane alone has the form (0, y, z), because two of its three perpendicular distances have collapsed to zero.

Finding the straight-line distance between two points in space uses exactly one idea, applied twice: Pythagoras' theorem. Take two points P(x1, y1, z1) and Q(x2, y2, z2), and build the rectangular box that has PQ as its longest diagonal. Walking along the box's edges from P, one edge covers a gap of (x2 - x1), a second edge covers (y2 - y1), and together these two edges are the legs of a right triangle sitting flat on the box's base; by Pythagoras, the diagonal of that base has length root[ (x2 - x1)^2 + (y2 - y1)^2 ] - exactly the familiar 2D distance formula. Now a second right triangle appears: this base diagonal is one leg, the box's vertical edge (z2 - z1) is the other leg, and PQ itself is the hypotenuse. Applying Pythagoras once more, PQ^2 equals the base diagonal squared plus (z2 - z1)^2, which unpacks to PQ = root[ (x2 - x1)^2 + (y2 - y1)^2 + (z2 - z1)^2 ]. For example, the distance between P(1, -3, 4) and Q(-4, 1, 2) is root[ (-4-1)^2 + (1+3)^2 + (2-4)^2 ] = root[ 25 + 16 + 4 ] = root 45 = 3 root 5 units. When one point is the origin itself, the formula quietly simplifies to OQ = root( x^2 + y^2 + z^2 ).

Once distances between points can be measured, the distance formula becomes a general-purpose shape-detector. Three points are collinear exactly when the longest of the three distances between them equals the sum of the other two - because that is precisely what it means for one point to sit directly between the other two on a single straight line. Checking P(-2, 3, 5), Q(1, 2, 3), and R(7, 0, -1): PQ works out to root 14, QR to 2 root 14, and PR to 3 root 14, and since PQ + QR = root 14 + 2 root 14 = 3 root 14 = PR exactly, the three points are confirmed collinear. The same side-lengths can test for a right angle using the converse of Pythagoras - if the two shorter sides squared add up to the longest side squared, the angle between those two shorter sides is a right angle. This does not always work out, and it is worth checking rather than assuming: for A(3, 6, 9), B(10, 20, 30), and C(25, -41, 5), computing all three squared distances gives AB^2 = 686, BC^2 = 4571, and CA^2 = 2709, and since CA^2 + AB^2 (=3395) does not equal BC^2 (=4571), triangle ABC is not right-angled at all, despite looking like a plausible candidate before the arithmetic was actually done.

Distances in space are not only for checking fixed points - they can define entire sets of points at once, through a single equation. Suppose A is (3, 4, 5) and B is (-1, 3, -7), and P(x, y, z) is required to satisfy PA^2 + PB^2 = 2k^2 for some constant k. Expanding both squared distances in full and adding them together, most of the individual coordinates survive but combine neatly: PA^2 + PB^2 works out to 2x^2 + 2y^2 + 2z^2 - 4x - 14y + 4z + 109, so the condition becomes 2x^2 + 2y^2 + 2z^2 - 4x - 14y + 4z = 2k^2 - 109. This single equation, not a single point, describes every position P could take while satisfying the original distance condition - it is called the locus of P. The same idea, run with a simple equal-distance condition PA = PB instead of a sum of squares, always collapses down to a flat plane rather than a curved surface: setting the distances from A(3, 4, -5) and B(-2, 1, 4) equal and squaring both sides eventually simplifies all the way down to 10x + 6y - 18z - 29 = 0, the plane of points exactly halfway between A and B. Coordinate geometry in three dimensions is, ultimately, this same trick - turning a geometric description into an algebraic equation - repeated for shapes far more complicated than a single point.