Galois theory has always struck me as rather mysterious, perhaps because its modern formulation is shrouded in concepts that did not yet exist during Galois' time (e.g., fields, groups, vector spaces). Even after taking a course on Galois theory, I have little idea of how Galois might have arrived upon his ideas in the first place.
Question: What was Galois' original line of reasoning that led him to the unsolvability of the quintic?
There is a great answer here that explains some of the differences between Galois' original work and modern Galois theory. However, it doesn't really explain how Galois' proof worked, or how he discovered it in the first place.
From what I gathered, Galois was interested in symmetric functions, which are rational functions in some indeterminates $t_1, \dots, t_n$ that are invariant under permutations of these indeterminates. The most important examples of symmetric functions are the coefficients of the polynomial $(x - t_1) \cdots (x - t_n)$.
Here is my guess as to how Galois proceeded. He considered the permutations of $t_1, \dots, t_n$ as a 'group' acting on the set of all rational functions. The symmetric functions are precisely those rational functions that are fixed by every permutation. Naturally, he began to consider the rational functions that are fixed by 'subgroups' of the permutation group (e.g., he may have considered the rational functions fixed by the subgroup of even permutations). From this, he eventually realized that there is a sort of correspondence between these subgroups and the operation of 'adjoining' certain formulas (e.g., he may have discovered that the subgroup of even permutations corresponds to adjoining the square root of the discriminant). This is essentially the fundamental theorem of Galois theory. Furthermore, he noticed that the well-known quadratic/cubic/quartic formulas are each given by successively adjoining roots of symmetric functions, which corresponds to a chain of subgroups of the permutation group. This led him to study the subgroups of the permutation group on five indeterminates, and he eventually realized that there is no chain of subgroups corresponding to a quintic formula.