This question is regarding a detail I am unsure about in a problem in the Dummit and Foote abstract algebra textbook that I am currently working on. The problem itself is that if $H \leq K \leq G$ I need to show $|G:H| = |G:K|\cdot|K:H|$. Within this problem I am not sure about the uniqueness of how the cosets are counted in the case of both sides being finite.
If $H \leq K \leq G$ as defined and I let $x_i$ for $i = 1, \dots, |G:K|$ be the coset representatives of $K$ in $G$ and $y_j$ for $j=1,\dots,|K:H|$ be the coset representatives of $H$ in $K$, why is it that each pair of cosets, $x_iy_j$ is a unique coset representative of $H$ in $G$?
Suppose $x_{i_0}y_{j_0}$ and $x_{i_1}y_{j_1}$ are two such cosets representatives where $x_{i_0}y_{j_0} = x_{i_1}y_{j_1}$. What makes it necesarry for $i_0=i_1$ and $j_0=j_1$?