I develop in depth the machinery of $(\mathcal L, n)$-models originally introduced by Shelah and, independently in a slightly different form by Kripke. This machinery allows fairly routine constructions of true but unprovable sentences in $\mathsf{PA}$. I give two applications: 1. Shelah's alternative proof of the Paris-Harrington theorem, and 2. The independence over $\mathsf{PA}$ of a new $Π^0_1$ Ramsey theoretic statement about colorings of finite sequences of structures.