On pg. 54 (and again on pg 402 for another purpose) I cite the supposed Hittite form tarh- 'conquer' as an example of the preservation of a consonantal reflex of medial *h2 in Hittite, but Alwin Kloekhorst 2008:835–9 has shown that there is no such verb, but only a tarhu-/taruh- reflecting a u-present *terh2u-.
Further on pg. 54 I state that medial *h3 was loss without exception in Anatolian, but the question of the reflex of medial *h3 needs to be reconsidered. The meaning of CLuv. ta-tarh- as established by Kloekhorst is 'break' not 'conquer' and therefore might be compared with the family of Grk. τιτρώσκω ᾽I wound'. This suggests that *h3 may have had a consonantal reflex after a sononant.