In the book of Analysis on Manifolds by Munkres, at page 152, it is given that
However, I cannot see how do we we have that $[c_i, d_i] \subset [a_i- \lambda, b_i + \lambda]$. I mean I have tried forming counterexamples, but every case what obeying the above condition, but still I cannot see the logical implication.
So, why do we have $[c_i, d_i] \subset [a_i- \lambda, b_i + \lambda]$, and how can we show it ?
Edit:


Note that "$1/N$ is less than the small of $\delta$ and $\lambda$.", In the figure below, note that the interval $[a_i - 1/N,a_i + 1/N]$ contains both neighboring elements of the form $m/N$. That is, $$ a_i - \lambda < a_i - 1/N \leq c_i $$ and similarly, $$ b_i + \lambda > b_i + 1/N \geq d_i $$
Edit:
Note that the maximality of $c_i$ contradicts with that if $c_i + 1/N \leq a_i$, so we have $c_i \leq a_i \leq c_i + 1/N$.