[xentac]

You are correct. They aren't a linear history. In fact, the key is the @{#} syntax (man git-rev-parse). Each stash commit is a single commit pointed to by a ref/stash reference. Because of the reflog, we can see what ref/stash used to point at, that is @{1}, @{2}, etc.

[__david__]

It's actually (potentially) more than a single commit. If you have a dirty index when you stash then that will also be committed along with your working directory changes. If you use the new(ish) --include-untracked option to stash, a 3rd commit will be added that has the untracked files in it. The main stash commit (that holds your working directory changes) is a fake merge of all the commits, just so there are pointers back to the other 2 commits.