It's commonly posited that all PIE roots consist of two groups of consonants, neither of which can be empty. For example, the root *h₁ed- has the groups *h₁ and *d.
However, I'm not aware of any language that shows a reflex of initial *h₁. All the derivations I can think of would work just as well if the root was **ed-, with no initial consonant.
So: is there a language which shows a definite reflex of initial *h₁, which requires us to reconstruct *h₁ed- instead of **ed-? Or is the initial *h₁ posited just to make the root structure more elegant?