Northrop Frye Book

Often engaging, never convincing.

Watkins' premise is that it's possible to use the comparative method to reconstruct specific formulaic constructions, metrical styles, and other figures of poetic speech of Proto-Indo-European, and it's important not to let the book's mountain of (variably relevant, variably credible) philological trivia and more than occasional deliberate obscurantism hypnotise you into not noticing that this premise is false. The thing about the comparative method in linguistics isn't just that it lets us take two words in two related languages and let us reconstruct a common proto-form that we can automatically date to the time they were one language; it's that in reconstructing the proto-language it lets us construct a set of rules by which we can confidently distinguish true cognates from borrowings (and have a rough impression for when they were borrowed)'something that isn't possible even in principle when it comes to any aspect of poetics.
Obviously when dealing with significant parallels in substantial bodies of work across relatively shallow time depths, it's usually possible to say something sensible about them, but Watkins spends more time spuriously projecting trivial two-word fragments (like the overexposed κλέος ἄφθιτον) across thousands of years, and when he does try to apply his method to something more significant'"the" titular dragon-slaying myth'the best he can do is to come up with hypergeneric formulas ("HERO SLAY ADVERSARY") which, for all their insipid vacuity, don't even fit the evidence properly and require endless variation.

Did (to focus on a specific) the Proto-Indo-Europeans have a dragon-slaying myth? Probably; almost everyone did, Indo-European or not. Are literally any of the details recoverable using Watkins' methods? Clearly not.