sin x is close to x
With x in radians, as . So for small x, sin x and x are nearly the same number, and . This is the first small-angle approximation.
On a graph, the sine curve leaves the origin running along the line y = x. They stay close for a while and then part: by x = 1 the curve is at sin 1 = 0.841, well below the line.
The gold curve y = sin x and the dashed line y = x, from x = −1.5 to 1.5. Near the origin they almost coincide; toward the ends the curve bends away below the line on the right and above it on the left.
cos x is close to
Near 0, cos x is close to 1, but throws away how it changes. The double-angle formula , with , gives . For small x, , so .
The parabola has its maximum, 1, at x = 0, where it is flat, and it falls away on both sides. Cosine has the same shape there: flat at its maximum, then falling.
The gold curve y = cos x and the dashed parabola . Both are flat at height 1 at x = 0. At x = 1 they are 0.540 and 0.5, and by x = 1.5 they are 0.071 and −0.125.
tan x is close to x
. For small x the numerator is close to x and the denominator is close to 1, so .
tan x runs above the line y = x on the right, and sin x runs below it: at x = 0.5, sin x = 0.479 and tan x = 0.546. Both are close to x, from opposite sides.
The gold curve y = tan x and the dashed line y = x, from x = −1.2 to 1.2. They leave the origin together; on the right the curve climbs above the line, on the left it falls below it.
How close, and for which x
At x = 0.1, sin 0.1 = 0.0998334, so is out by 0.000167, about 0.17%, and the two agree to three decimal places. tan 0.1 = 0.1003347, out by 0.000335, about 0.33%. cos 0.1 = 0.9950042 against 1 − 0.005 = 0.995, out by 0.0000042.
At x = 0.5 the errors grow to 0.021 for sine (4.3%), 0.046 for tangent (8.5%) and 0.0026 for cosine (0.29%). At x = 1 they are 19%, 36% and 7.5%.
Each is within 1% for |x| below about 0.244 radians for sine, 0.173 for tangent and 0.662 for cosine, which are about 14°, 9.9° and 38°. So the approximations are for angles of a few degrees, measured in radians. Cosine’s stays within 1% the longest.
Radians only
Every one of these rests on , and that limit holds only with x in radians. In degrees, fails badly: 0.1° is 0.001745 radians, so sin 0.1° = 0.001745, not 0.1. An angle given in degrees has to be converted, by multiplying by , before any of the approximations is used.
Using them together
For small x, can be estimated by replacing each part. With 2x in place of x, , so the numerator is about . , so the denominator is about . The quotient is about . Check: at x = 0.1 it is 1.98669, and at x = 0.05 it is 1.99667.
Here would not do: the numerator would come out as 0. When an answer is a difference such as 1 − cos x, the approximation must keep the first term that does not cancel, which is the term.
The usual mistakes
Using the approximations in degrees. They hold only for x in radians.
Writing . The plus sign bends the parabola upward, but near 0 cosine falls away from 1; it is never above 1.
Writing . A straight drop falls at gradient −1 from x = 0, but cosine leaves x = 0 flat.
Writing . tan 0 = 0; it is the denominator, cos x, that is close to 1.
Using them for angles that are not small. At x = 1 radian, about 57°, is out by 19%.
A surveyor’s small angles
In the application below, an angle of 0.02 radians gives a mast’s height through , and an angle of 0.00125 radians gives the drop of a level sight line through , where would give no drop at all.
Worked example: A Surveyor's Tiny Angles: A Mast's Height and the Drop of a Long Sight Line
Question A surveyor stands 800 m from a radio mast and measures the angle to its top as 0.02 radians. (a) Use tanθ ≈ θ to find the height of the mast, and say how far out the approximation is. (b) From the same spot the surveyor sights along a level line for 8 km. Taking the Earth as a sphere of radius 6400 km, the ground falls below that line by 6400(1 − cosθ) km, where θ = 86400 radians. Use cosθ ≈ 1 − θ22 to find the drop, in meters.
1.The height is 800tanθ with θ = 0.02 radians. The angle is small, and for a small angle in radians tanθ ≈ θ.
The last column shrinks fast: for a small angle in radians tanθ and θ agree to more and more places. 2.So the height is about 800 × 0.02 = 16 m.
The height is 800tanθ with θ = 0.02, which is about 800 × 0.02 = 16 m. 3.(a) The mast is 16 m tall. The exact value is 800tan 0.02 = 16.0021 m, so the approximation is about 2 mm short over 16 m, which is well inside what a tape would settle.
(a) The mast is 16 m tall. The exact value is 800tan 0.02 = 16.0021 m, so the approximation is about 2 mm short. 4.For part (b), θ = 86400 = 0.00125 radians, smaller still. With cosθ ≈ 1 − θ22, the bracket becomes 1 − cosθ ≈ θ22 = 0.0012522 = 7.8125 × 10−7.
For part (b) the angle is 86400 = 0.00125 radians. The last two columns show 1 − cosθ and θ22 closing on each other. 5.The drop is 6400 × 7.8125 × 10−7 = 0.005 km.
1 − cosθ ≈ θ22 = 7.8125 × 10−7, so the drop is 6400 × 7.8125 × 10−7 = 0.005 km. 6.(b) The drop is 5 m over 8 km. Check with the exact cosine: 6400(1 − cos 0.00125) = 0.0050000 km, the same to the nearest millimeter, because the next term of the approximation is of size θ424, about 10−13.
(b) The ground drops 5 m over 8 km, and the exact cosine gives the same to the nearest millimeter.
Answer: (a) 16 m, about 2 mm short of the exact 16.0021 m; (b) a drop of 5 m over 8 km
Common mistakes
- Reading 0.02 as a number of degrees. The approximations tanθ ≈ θ and cosθ ≈ 1 − θ22 hold for radians only: in degrees the height would come out as 800tan 0.02° = 0.28 m, which is not a mast.
- Using cosθ ≈ 1 in part (b). It is true that cos 0.00125 is 1 to six decimal places, but the whole answer lives in what is left after the 1 is taken away: the drop would come out as 0. When the answer is a difference, the approximation must keep the first term that does not cancel.