Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I’m looking for a proof of $ \displaystyle \lim_{x \to 0} \frac{\sin(x)}{x} = 1 $ that does not use other trigonometric functions or any first-order approximation to the sine function. Is this possible? The other proofs that I’ve seen on this website don’t really fit these stringent requirements, so I was hoping to see a different kind of demonstration altogether. Thanks!!

share|improve this question
The limit should read $ \displaystyle \lim_{x \to 0} \frac{\sin(x)}{x} = 1 $. –  Haskell Curry Feb 23 '13 at 20:37
What definition of sin(x) are you taking here? –  anonymous Feb 23 '13 at 20:39
Oh, so you don't want to use $\tan $ and $\cos$, for instance...Otherwise, this would do: proofwiki.org/wiki/Limit_of_Sine_of_X_over_X/Geometric_Proof –  1015 Feb 23 '13 at 20:39
@Azmoti: It’s not an exact duplicate because the OP seems to want to avoid Taylor series. :) –  Haskell Curry Feb 23 '13 at 20:43
A functional approach is taken in "A New Approach to Circular Functions, II and lim (sin x)/x" by Gerson B. Robison in Mathematics Magazine, Vol. 41, No. 2 (Mar., 1968), pp. 66-70 –  Byron Schmuland Feb 23 '13 at 20:51

4 Answers 4

up vote 8 down vote accepted

An interesting proof that sees no trigonometry or Maclaurin series until the very final conclusion (at the cost of doing Riemann integration):

Define a function $ \theta: [-1,1] \to \mathbb{R} $ as follows: $$ \forall a \in [-1,1]: \quad \theta(a) \stackrel{\text{def}}{=} \int_{0}^{a} \sqrt{1 - x^{2}} ~ d{x}. $$ For each $ a \in [-1,1] $, we can interpret $ \theta(a) $ as the signed arc-length of the parametrized curve $$ \left\{ \left( x,\sqrt{1 - x^{2}} \right) \in \mathbb{S}^{1} ~ \Big| ~ \text{$ x $ is between $ 0 $ and $ a $} \right\}. $$

Observe that $ \theta $ has the following properties:

  • $ \theta(0) = 0 $ and

  • $ \theta: [-1,1] \to \mathbb{R} $ is a strictly increasing continuous function.

There is thus an inverse function $ \theta^{-1}: \text{Range}(\theta) \to [-1,1] $, and we define the sine function (or at least part of it) by

$$ \sin \stackrel{\text{def}}{=} \theta^{-1}. $$

Hence, \begin{align} \lim_{a \to 0} \frac{\sin(\theta(a))}{\theta(a)} &= \lim_{a \to 0} \frac{{\theta^{-1}}(\theta(a))}{\theta(a)} \\ &= \lim_{a \to 0} \frac{a}{\theta(a)} \\ &= \lim_{a \to 0} \frac{a}{\displaystyle \int_{0}^{a} \sqrt{1 - x^{2}} ~ d{x}} \quad (\text{By definition.}) \\ &= \lim_{a \to 0} \frac{1}{\sqrt{1 - a^{2}}} \quad (\text{By l’Hôpital’s Rule.}) \\ &= 1. \end{align} Therefore, $$ \lim_{\theta \to 0} \frac{\sin(\theta)}{\theta} = 1. $$

share|improve this answer
Cannot l’Hôpital’s Rule be used directly on $\sin(x)/x$? –  enzotib Feb 23 '13 at 20:54
@enzotib: If you use l’Hôpital’s Rule on $ \sin $, you get $ \cos $, which the OP wants to avoid. :) –  Haskell Curry Feb 23 '13 at 20:55
@enzotib, We aren't allowed to use l’Hôpital’s Rule because that would assume that $\frac{d}{dx} \sin x = \cos x$. However, that result follows from $\lim_{x\to 0} \frac{\sin x}{x} = 1$, so by assuming $\sin\prime = \cos$, we are assuming the limit we want to prove. –  George V. Williams Feb 23 '13 at 20:56
Not necessarily. We typically take the definition of sin(x) as a power series. From this definition, it's straightforward to prove that sin'(x) = cos(x) without proving that sin(x)/x tends to 1. –  anonymous Feb 23 '13 at 21:46
@anonymous, you cannot take sine as a power series without first knowing the derivative of sine. –  CogitoErgoCogitoSum Feb 24 '13 at 1:22

Taking the definition $$\sin(x)=\frac{1}{2i} (e^{ix} - e^{-ix}) $$

Than you get from the exponentialfunction the identity $$\sin(x)=\sum_{k=0}^\infty (-1)^n \frac{x^{2n+1}}{(2n+1)!}=x-\frac{x^3}{3!}+\frac{x^5}{5!}\mp \cdots$$

So we have $$\lim_{x\rightarrow 0 } \frac{\sin(x)}{x}=\lim_{x\rightarrow 0} 1-\frac{x^2}{3!} + \frac{x^4}{5!}\mp \cdots=1$$

Else you could define $$\sin(x)=x\prod_{k=1}^\infty \left(1-\frac{x^2}{k^2 \pi^2 } \right)$$ Than $$\lim_{x\rightarrow 0 } \frac{\sin(x)}{x} =\lim_{x\rightarrow 0} \prod_{k=1}^\infty \left(1-\frac{x^2}{k^2\pi^2}\right)$$

share|improve this answer
I think you could, instead of expanding the exponential function, directly take the limit of $e^{ix}/x$, with the aid of L'Hôpital. –  awllower Feb 24 '13 at 3:43

If you're not defining $\sin$ via Taylor series, the obvious other rigorous alternative would be to define it (near $0$) as the inverse function of $\arcsin$, where we take $\arcsin y = \int_0^y \frac{dt}{\sqrt{1-t^2}}$. Then

\begin{eqnarray} \int_0^y \frac{dt}{\sqrt{1-t^2}} &<& \int_0^y \frac{dt}{\sqrt{1-y^2}} = \frac{y}{\sqrt{1-y^2}} \\ \int_0^y \frac{dt}{\sqrt{1-t^2}} &>& \int_0^y \, dt = y \end{eqnarray}

So $\sqrt{1-y^2}<\frac{y}{\arcsin y}<1$, and thus $\lim_{y \to 0} \frac{y}{\arcsin y} =1$ by the squeeze theorem.

After making the change of variables $x=\arcsin y$ in this last limit (justified because $\arcsin y$ is continuous and monotone), it follows that $\lim_{x \to 0} \frac{\sin x}{x}=1$.

share|improve this answer
why it is so obvious that your function is injective ? –  Dominic Michaelis Feb 23 '13 at 21:09
@DominicMichaelis: It's the integral of something positive. –  Micah Feb 23 '13 at 21:10

I'm assuming that OP uses the geometric (slightly non-rigorous) definition of the trigonometric functions, i.e., $(\cos(t),\sin(t))$ is the point on the unit circle forming angle $t$ with the positive $X$ axis).

The requested limit (known as the first fundamental limit) can't be computed by any method that relies on the fact that $\sin'(x)=\cos(x)$ (in fact, this limit is required in the proof of that fact), and I think this is the point of this exercise. Here is a geometric proof, legitimate along these lines.

share|improve this answer
You just hit the nail on the head, Ittay. However, if we can give a formal definition of ‘angle’, then the problem of semi-rigor is solved. One can define ‘angle’ using arc-length, which is what I did above, but of course, I needed to resort to Riemann integration. –  Haskell Curry Feb 24 '13 at 3:54
and you just drove the nail all the way into the wood @HaskellCurry :) –  Ittay Weiss Feb 24 '13 at 4:03

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.