It depends ultimately on the underlying distribution, but for most we can assume an approximately normal distribution. For a normal distribution, the confidence intervals are approximately 68.27% for $\displaystyle \mu \pm 1\sigma$ and 95.45 for $\displaystyle \mu \pm 2\sigma$. As the mean $\displaystyle \mu=50$ and standard deviation $\displaystyle \sigma=10$, this gives the range 40 to 60 as $\displaystyle \mu \pm \sigma$, so we should expect about $\displaystyle 0.6827 \cdot 1000 = 683$ in this range. Similarly, 30 to 70 is $\displaystyle \mu \pm 2\sigma$, so we expect about $\displaystyle 0.9545 \cdot 1000 = 955$ within that range. These formulae are ultimately derived from numerical integration of the probability density function of the standard normal distribution (normal distribution with mean $\displaystyle \mu=0$ and variance $\displaystyle \sigma^2 =1$), which is
$\displaystyle \dfrac{1}{\sqrt{2\pi}}e^{-\tfrac{x^2}{2}}$.
$\displaystyle \int_{-n}^{n}\dfrac{1}{\sqrt{2\pi}}e^{-\tfrac{x^2}{2}}\, dx$