Suppose that $p$ is a prime and $m,n$ are positive integers. If $p^{n-1} + \cdots + p + 1$ divides $\sum\limits_{k=1}^m p^{j_k}(p^{i_k - 1} + \cdots p + 1)$, why is it that $n$ divides $\sum\limits_{k=1}^m i_k$, where the $i_k$ and $j_k$ are non-negative integers?
This is a fact that a proof presented in some notes for my number theory course boils down to, but I can't quite see what argument could work here. I should add that it's possible that it's only true for certain values of $p$ and $n$.
Any help would be appreciated. Thanks in advance!
Multiplying both $p^{n-1} + \cdots + p + 1$ and $\sum\limits_{k=1}^m p^{j_k}(p^{i_k - 1} + \cdots p + 1)$ by $p-1$, we obtain that $p^n-1$ divides $\sum\limits_{k=1}^m p^{j_k}(p^{i_k} - 1)$. In general, when $j_k$ are distinct we have to take into account not only the numbers $i_k$, but $j_k$ too. But even when all $j_k$ are equal, the claim can fail. For instance, when $p=2$, $n=2$, $m=3$, $j_1=j_2=j_3=0$, and $i_1=i_2=i_3=1$ we have $p^n-1=3$, $\sum\limits_{k=1}^m p^{j_k}(p^{i_k} - 1)=3$, but $n=2\not|3=\sum\limits_{k=1}^m i_k$.