Louisa Stark has made a convincing and promising start at a genetic relation for Mayan and Araucanian.' Apart from the impressive number of probable cognates, the correspondences which do not consist of identities are striking; this is the sort of phonological correspondence that we should expect for such a relation. In the consonantism there are also developments that fit in well with Uru-Chipaya. The loss of glottalization is noteworthy. But even more striking is the frequency, as in Uru-Chipaya also, of different reflexes of consonants (especially in the continuants) in root-final position from what is found in root (word) initial.2 Both Araucanian and UruChipaya give the appearance of having altered the finals of earlier Mayan-like monosyllables, and then of having added on suffixes. This is what we might expect of a Mayan-like language that had moved into a Quechua-like Sprachbund.