[Beginner] — why a ratio of two sides depends only on an angle, and how that one fact lets you measure things you cannot reach.

Before you start

  • You know similar triangles and the AA criterion — triangles and polygons.
  • You know Pythagoras’ theorem.
  • Helpful: the equilateral and perpendicular constructions, which section 4 uses to derive exact values — constructions.

What you will be able to do after this lesson:

  1. Explain why , and are well defined — that is, why the ratio does not depend on the size of the triangle.
  2. Derive the exact values for , and from two constructed triangles, rather than memorising a table.
  3. Prove and from the definitions.
  4. Solve angle-of-elevation and depression problems, including the two-observation case where the distance is unknown.

Study route: read 1–5, attempt the prediction in section 5, then run the lab. Do the practice task before moving to graphs.


1. Why this exists

You need the height of a tree. You cannot climb it and you cannot lay a tape up it. You can stand somewhere, measure your distance from the base, and measure the angle from horizontal up to the top.

Similar triangles already told us that a stick and its shadow can give the tree’s height, if the sun cooperates. Trigonometry removes the dependency on the sun: it tabulates, once and for all, what the ratio of sides is for every angle. Then a single angle measurement plus a single distance gives the height.

The whole subject rests on one observation, which is worth stating before any notation:

In a right triangle, once you fix one of the non-right angles, the shape is fixed. Every triangle with that angle is a scaled copy of every other. So the ratio of any two sides is determined by the angle alone — it does not matter how big the triangle is.

That is the AA similarity criterion, applied to right triangles. Everything below is bookkeeping on top of it.

2. Terminology

Fix one non-right angle . The three sides then get names relative to that angle:

                    /|
                   / |
      hypotenuse  /  |  opposite
                 /   |     (opposite theta)
                / θ  |
               /_____|
              adjacent
TermPlain-English definitionWarning
HypotenuseThe side opposite the right angleAlways the longest; never changes name
OppositeThe side facing Swaps if you switch to the other angle
AdjacentThe side next to that is not the hypotenuseAlso swaps
opposite / hypotenuse—
adjacent / hypotenuse—
opposite / adjacent—
Angle of elevationThe angle up from horizontal to a line of sightMeasured from the horizontal, not the vertical
Angle of depressionThe angle down from horizontalEqual to the elevation seen from the other end — alternate angles

The mnemonic SOH-CAH-TOA encodes the three definitions. It is a memory aid, not a reason; the reason is section 1.

WARNING

“Opposite” and “adjacent” are defined with respect to the angle you chose. Switch to the other non-right angle and they exchange. This is the single most common source of wrong answers: of one angle equals of the other, and using the wrong one silently gives a plausible number.

3. The ratios are functions of the angle

Take a right triangle with and legs of some size. Double every side. Triple them. The angles are unchanged — scaling preserves angle — so all these triangles are similar, and corresponding side ratios are equal:

So is a number, not a property of a particular triangle. That is what licenses writing as a function. Block 1 of the lab checks this over several scale factors.

4. Exact values, derived not memorised

Two triangles are constructible with compass and straightedge, and between them they give every “nice” angle.

The triangle — half a square

Take a unit square and draw a diagonal. The two triangles are isosceles right triangles with legs and, by Pythagoras, hypotenuse . The non-right angles are equal and sum to , so each is :

The – triangle — half an equilateral triangle

Take the equilateral triangle of side constructed in constructions, and drop a perpendicular from the apex. It bisects the base (SSS congruence), giving a right triangle with hypotenuse , short leg , and by Pythagoras the third side .

The bisected apex angle is ; the base angle is still :

Notice and . That is not a coincidence — it is the opposite/adjacent swap of section 2, and in general:

which is where the “co” in cosine comes from: the sine of the complementary angle.

Two identities that come free

Let the hypotenuse be , opposite , adjacent . Pythagoras says . Divide through by :

And directly from the definitions:

Both are Pythagoras and division, nothing more. means , a notation that exists only to avoid writing brackets everywhere.

5. Elevation, depression, and the unknown-distance case

One observation. Standing from the base, measuring elevation to the top, with eye height :

Forgetting is the standard error. The triangle sits on your eye, not on the ground.

Two observations. Now suppose you cannot measure — the base is inside a fenced compound. Take the elevation from where you stand, walk directly away a measured distance , and take a second elevation (necessarily smaller). Then with the height above eye level:

Two equations, two unknowns. Eliminating :

This is the workhorse of practical surveying, and it needs no access to the base at all.

TIP

Predict before running the lab. In that last formula, what happens as and get closer together? What does that mean physically — and what does it tell you about how far apart your two observation points should be? Decide before opening the answers.

Worked example — runnable

Runnable example: save as trig_ratios.py in any empty directory and run python3 trig_ratios.py. Standard library only; writes no files.

"""Trigonometric ratios: why they are well defined, and what they measure."""
import math
 
EPS = 1e-12
 
 
def ratios(opposite, adjacent):
    """sin, cos, tan for the angle whose opposite and adjacent are given."""
    hyp = math.hypot(opposite, adjacent)
    return (opposite / hyp, adjacent / hyp, opposite / adjacent)
 
 
def height_one_observation(distance, elevation_deg, eye_height=0.0):
    return distance * math.tan(math.radians(elevation_deg)) + eye_height
 
 
def height_two_observations(step_back, alpha_deg, beta_deg, eye_height=0.0):
    """Height when the distance to the base cannot be measured."""
    ta = math.tan(math.radians(alpha_deg))
    tb = math.tan(math.radians(beta_deg))
    if abs(ta - tb) < 1e-12:
        raise ValueError("the two elevations are indistinguishable")
    return step_back * ta * tb / (ta - tb) + eye_height
 
 
if __name__ == "__main__":
    print("Block 1 - the ratio depends on the angle, not the size")
    base_opp, base_adj = 3.0, 4.0            # the 3-4-5 triangle
    for scale in (1, 2, 7, 100, 0.01):
        s, c, t = ratios(base_opp * scale, base_adj * scale)
        print(f"  scale {scale:>6}: sin={s:.10f} cos={c:.10f} tan={t:.10f}")
        assert abs(s - 0.6) < EPS and abs(c - 0.8) < EPS and abs(t - 0.75) < EPS
 
    print()
    print("Block 2 - exact values, derived from the two constructed triangles")
    print("  from half a unit square (legs 1, 1, hypotenuse sqrt2):")
    s45, c45, t45 = ratios(1.0, 1.0)
    print(f"    sin45={s45:.10f}  1/sqrt2={1/math.sqrt(2):.10f}")
    print(f"    tan45={t45:.10f}")
    assert abs(s45 - 1 / math.sqrt(2)) < EPS and abs(t45 - 1.0) < EPS
 
    print("  from half an equilateral triangle of side 2 (legs 1, sqrt3):")
    s30, c30, t30 = ratios(1.0, math.sqrt(3))     # angle 30: opposite 1
    s60, c60, t60 = ratios(math.sqrt(3), 1.0)     # angle 60: opposite sqrt3
    print(f"    sin30={s30:.10f} (= 1/2)      cos30={c30:.10f} (= sqrt3/2)")
    print(f"    sin60={s60:.10f} (= sqrt3/2)  cos60={c60:.10f} (= 1/2)")
    assert abs(s30 - 0.5) < EPS
    assert abs(c30 - math.sqrt(3) / 2) < EPS
    assert abs(s60 - math.sqrt(3) / 2) < EPS
    assert abs(c60 - 0.5) < EPS
    # the complementary relation
    assert abs(s30 - c60) < EPS and abs(s60 - c30) < EPS
    print("    sin(theta) = cos(90 - theta): confirmed for 30 and 60")
 
    print()
    print("Block 3 - the two identities")
    for angle in (0, 17, 30, 45, 60, 73, 89):
        r = math.radians(angle)
        pyth = math.sin(r) ** 2 + math.cos(r) ** 2
        assert abs(pyth - 1) < EPS
        if angle != 90:
            assert abs(math.tan(r) - math.sin(r) / math.cos(r)) < EPS
    print("  sin^2 + cos^2 = 1 and tan = sin/cos hold for every angle tried")
 
    print()
    print("Block 4 - measuring a tree you cannot climb")
    # A tree whose true height we fix, so the recovered value can be checked.
    TRUE_HEIGHT, EYE, DISTANCE = 18.0, 1.6, 25.0
    elevation = math.degrees(math.atan((TRUE_HEIGHT - EYE) / DISTANCE))
    print(f"  true height {TRUE_HEIGHT} m, eye {EYE} m, standing {DISTANCE} m away")
    print(f"  -> elevation would read {elevation:.4f} degrees")
    recovered = height_one_observation(DISTANCE, elevation, EYE)
    print(f"  recovered height = {recovered:.6f} m")
    assert abs(recovered - TRUE_HEIGHT) < 1e-9
 
    forgot_eye = height_one_observation(DISTANCE, elevation)
    print(f"  forgetting eye height gives {forgot_eye:.6f} m - short by exactly {EYE} m")
    assert abs((TRUE_HEIGHT - forgot_eye) - EYE) < 1e-9
 
    print()
    print("Block 5 - when the distance to the base cannot be measured")
    STEP = 15.0
    alpha = math.degrees(math.atan((TRUE_HEIGHT - EYE) / DISTANCE))
    beta = math.degrees(math.atan((TRUE_HEIGHT - EYE) / (DISTANCE + STEP)))
    print(f"  elevations {alpha:.4f} and {beta:.4f} degrees, {STEP} m apart")
    two_obs = height_two_observations(STEP, alpha, beta, EYE)
    print(f"  recovered height = {two_obs:.6f} m  (base never measured)")
    assert abs(two_obs - TRUE_HEIGHT) < 1e-9
 
    print("  sensitivity: how a 0.1 degree error in beta moves the answer")
    for err in (0.0, 0.1, -0.1):
        h = height_two_observations(STEP, alpha, beta + err, EYE)
        print(f"    beta error {err:+.1f} deg -> height {h:8.4f} m  ({h - TRUE_HEIGHT:+.4f})")
 
    print()
    print("trig_ratios: passed")

Expected output:

Block 1 - the ratio depends on the angle, not the size
  scale      1: sin=0.6000000000 cos=0.8000000000 tan=0.7500000000
  scale      2: sin=0.6000000000 cos=0.8000000000 tan=0.7500000000
  scale      7: sin=0.6000000000 cos=0.8000000000 tan=0.7500000000
  scale    100: sin=0.6000000000 cos=0.8000000000 tan=0.7500000000
  scale   0.01: sin=0.6000000000 cos=0.8000000000 tan=0.7500000000

Block 2 - exact values, derived from the two constructed triangles
  from half a unit square (legs 1, 1, hypotenuse sqrt2):
    sin45=0.7071067812  1/sqrt2=0.7071067812
    tan45=1.0000000000
  from half an equilateral triangle of side 2 (legs 1, sqrt3):
    sin30=0.5000000000 (= 1/2)      cos30=0.8660254038 (= sqrt3/2)
    sin60=0.8660254038 (= sqrt3/2)  cos60=0.5000000000 (= 1/2)
    sin(theta) = cos(90 - theta): confirmed for 30 and 60

Block 3 - the two identities
  sin^2 + cos^2 = 1 and tan = sin/cos hold for every angle tried

Block 4 - measuring a tree you cannot climb
  true height 18.0 m, eye 1.6 m, standing 25.0 m away
  -> elevation would read 33.2649 degrees
  recovered height = 18.000000 m
  forgetting eye height gives 16.400000 m - short by exactly 1.6 m

Block 5 - when the distance to the base cannot be measured
  elevations 33.2649 and 22.2936 degrees, 15.0 m apart
  recovered height = 18.000000 m  (base never measured)
  sensitivity: how a 0.1 degree error in beta moves the answer
    beta error +0.0 deg -> height  18.0000 m  (+0.0000)
    beta error +0.1 deg -> height  18.2194 m  (+0.2194)
    beta error -0.1 deg -> height  17.7845 m  (-0.2155)

trig_ratios: passed

The last three lines are worth more than the rest. A tenth of a degree — finer than most hand-held instruments resolve — shifts the answer by about m on an m tree, roughly . The formula is exact; the measurement is not, and the arithmetic amplifies the error. Move the two observation points closer together and this gets rapidly worse, which is what the prediction in section 5 was about.

Common pitfalls and traps

  • Swapping opposite and adjacent. They are named relative to the chosen angle. Redraw and relabel for the angle you are actually using, every time.
  • Degrees versus radians. Python’s math.sin takes radians. math.sin(30) is the sine of radians, which is — a perfectly plausible-looking wrong number. Every angle in the lab passes through math.radians first.
  • Forgetting eye height. The lab quantifies this: the answer is short by exactly the eye height, every time.
  • Assuming the ground is level. measures height above the observer’s eye level, not above the base of the tree, if the two are at different elevations.
  • Using . It is undefined — the adjacent side has length zero. In floating point math.tan(math.radians(90)) returns about rather than raising, which is a silent trap.
  • Treating an exact formula as an exact answer. Block 5’s sensitivity check is the honest part of the lesson. The mathematics carries no error; the protractor does.

Check your understanding

  1. In a right triangle, the side opposite is and the hypotenuse is . Find , and exactly.
  2. Without a calculator, find . Explain the result.
  3. A ladder m long leans against a wall at to the ground. How far up the wall does it reach, and how far is its foot from the wall?
  4. From the top of a cliff m high, the angle of depression to a boat is . How far is the boat from the base of the cliff?
  5. Why is never greater than ?
Answers — open only after an attempt
  1. The adjacent side is . So , , .
  2. . It happens because and are complementary: , since the opposite and adjacent swap.
  3. Height up the wall m. Distance of the foot m.
  4. The angle of depression from the cliff equals the angle of elevation from the boat — alternate angles between two horizontals. So , giving m.
  5. Because , and the hypotenuse is the longest side of a right triangle. A fraction whose denominator is at least as large as its numerator cannot exceed , with equality only in the degenerate case where the triangle collapses.

And the prediction from section 5: as , the denominator and the computed height blows up. Physically, two nearly-equal elevations carry almost no information about the distance — you have barely moved relative to the object. So the two observation points must be far enough apart that the elevation difference is comfortably larger than your measurement error. Block 5’s sensitivity numbers show what “comfortably” has to mean.

Practice — independent task

Implement tower_height(observations, eye_height) where observations is a list of (distance_walked, elevation_degrees) pairs taken by walking directly away from an object in a straight line, with the first entry at distance .

  1. With two observations, use the formula from section 5.
  2. With more than two, do better than picking two: each pair gives , which is linear in . Fit by least squares over all observations, and return both the estimate and a measure of spread.
  3. Assert that with noise-free synthetic data your estimate recovers the true height to within 1e-9, for at least three different true heights.
  4. Then add Gaussian noise of to the elevations, using a fixed seed, and report how the error behaves as the number of observations grows from 2 to 20.
  5. Answer in a comment: does the error shrink like , like , or not at all? Say which you observed.

Edge cases: two identical elevations (raise, do not divide by zero); an elevation of or ; observations taken while walking towards the object, so the step is negative.

Done when: the noise-free assertions pass, your noisy experiment is reproducible from its seed, and your answer to step 5 is based on numbers you produced rather than on what you expected.

Tradeoffs, limits and extensions

These definitions only cover . A right triangle cannot contain an obtuse angle, so nothing above says what means. Extending the definitions to every angle — including negative and reflex — requires the unit circle, which is the first section of graphs and identities. The extension is designed so that all the identities here survive unchanged.

Right triangles are a special case. For a triangle with no right angle, none of SOH-CAH-TOA applies. The general tools are the sine and cosine rules in sine and cosine rules, and both are derived by dropping a perpendicular to create right triangles — so this lesson remains the foundation.

Accuracy in practice. Surveying instruments quote angular accuracy in seconds of arc for this reason. The trigonometry is exact; the answer is only as good as the angle, and as block 5 shows, the amplification factor depends on the geometry you chose.

Before moving on

You are done with this lesson when you can:

  • Explain why is a number rather than a fact about one triangle.
  • Reconstruct the exact values for , , by drawing the two triangles, without a table.
  • Derive from Pythagoras in one line.
  • Solve a two-observation elevation problem and say how sensitive the answer is to angle error.

Recap for later lookup: , , ; ; ; ; exact values from half a square () and half an equilateral triangle (, ); two-observation height .

Next: Graphs and Identities — extending these three ratios from acute angles to every angle, using the unit circle.

↑ Back to Geometry and Trigonometry