# Why is uncertainty divided by $\sqrt{3}$? [closed]

Why do we have to divide uncertainty of the measurement by $\sqrt{3}$?

For example we have the uncertainty of the measurement with a ruler being the smallest scale of 0.01cm, and it is not our uncertainty but we first divide it and so it makes our final uncertainty value.

Why is that this way?

## closed as unclear what you're asking by AccidentalFourierTransform, ZeroTheHero, John Rennie, Bill N, stafusaNov 28 '18 at 0:05

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

## 1 Answer

I have found out the answer.

When you do a measurement (one measurement) you have many uncertainty sources. But if you want to have the combined uncertainty, you don't add like $1 + 1$, because that would give you uncertainty of $2$, but that doesn't have to be the case. The true value of the measurement may lie in between the uncertainty regions but it may lie on the left or on the right of the range.

As we don't know where exactly is the measurement we have to treat that situation like a rectangular distribution. We have $a=x-\Delta x$ and $b=x+\Delta x$, where $\Delta x$ is the uncertainty.

$$E(X) = \int_a^b xp(x)\mathrm{d}x = \int_a^b \frac{x}{b - a}\mathrm{d}x = \left.\frac{1}{2}\frac{x^2}{b - a}\right|_a^b = \frac{1}{2}\frac{b^2 - a^2}{b - a} = \frac{b + a}{2}$$

\begin{align} \sigma^2 &= \int_a^b \bigl(x - E(x)\bigr)^2 p(x)\mathrm{d}x \\ &= \int_a^b \biggl(x - \frac{b + a}{2}\biggr)^2\frac{1}{b - a}\mathrm{d}x \\ &= \left.\frac{1}{3(b - a)}\biggl(x - \frac{b + a}{2}\biggr)^3\right|_a^b \\ &= \frac{1}{3(b - a)}\Biggl[\biggl(\frac{b - a}{2}\biggr)^3 - \biggl(\frac{a - b}{2}\biggr)^3\Biggr] \\ &= \frac{1}{3}\biggl(\frac{b - a}{2}\biggr)^2 \end{align}

So we see that the standard deviation of this distribution is exactly our uncertainty divided by $\sqrt3$. Without that our uncertainty would be bigger than it needs to be. And so the final uncertainty of the single measurement is then: $u(x)=\Delta x/\sqrt3$

• Except if the ruler has tick marks every $b-a$ and you are saying all you know is that the true value is uniformly in this range, the uncertainty is $(b-a)/\sqrt{12}$. This is called discretization error and the $\sqrt{12}$ is a standard result. Saying it's $\sqrt{3}$ off from $(b-a)/2$ is, well, bizarre, since $(b-a)/2$ is meaningless. – user10851 Dec 8 '15 at 3:41