In his valuable article on the Indo-European vowels in Albanian (Lg. 26.379–88), Stuart A. Mann recommends a principle of etymological method that is characterized by admirable, not to say Utopian, rigor. He considers ‘complete correspondence of whole words’ (379) to be the only sure evidence on which to base etymology, and regards this as an improvement upon ‘continental neo-grammarian practice’ (379 fn. 2). As an antidote to the reckless overworking of the theory of Wurzelerweiterung, Mann's principle is certainly to be hailed. It would, of course, result in wholesale pruning of our etymological dictionaries, which may or may not be desirable. Would, for instance, Welsh ebol ‘colt’ < have to be rejected as a cognate of OLat. equos ‘horse’, since complete correspondence is lacking and a simplex *eb < * cannot be attested? Or would Mann apply his principle in its strictest form only to such procedures as the reconstruction of the original vocalic system of a language, such as he attempts in his article? In the latter event, his conservatism doubtless guarantees a certain dependability of results. But once the principle of whole-word correspondence is imposed, it goes without saying that the words used in establishing correspondences must themselves be verified philologically and not merely lexicographically. The bane of Celtic linguistics has been a flair, at certain periods, for creative lexicography. The Welsh Dictionary (1803) of the notorious William Owen Pughe, for instance, contains—not the extant Welsh vocabulary of his day, although some of that was inevitably incorporated in the work— but, in large measure, fictitious words which are the author's contribution to his mother tongue. Continental scholarship and even native authorities, relying on the work of Pughe and his followers, proceeded to draw etymological and phonological conclusions that had no basis in linguistic fact, being derived from Pughe's ghost words. Recent Welsh lexicographical practice has fortunately succeeded in getting rid of most of these phantoms. As a matter of fact, predeeessors of Pughe were often far less fantastic than he; as early as 1632 John Davies provided an extremely reliable work (except for its etymologies), based on a careful analysis of the vocabulary of Welsh literature.