Detailed linguistic work, involving both internal and comparative reconstruction, traditionally has been an important source of information about the cultural history of nonliterate peoples. Such work, in turn, depends on descriptive materials obtained in the field, but, as native languages evolve or disappear, it becomes increasingly necessary to resort to philological methods more familiar from the study of literate cultures. As Ives Goddard reminds us, we must then dust off older, often prestructuralist descriptions of American Indian languages, which have fallen into oblivion but which, in spite of all sorts of problems of interpretation, are sometimes the best because the oldest sources of information. A case in point is the kinship terminology of the Uto-Aztecan language Tubatulabal, formerly (and perhaps still) spoken in Kern County, California. Along with much else, these words are simply missing from Charles F. Voegelin's brief vocabulary, which is the most recent published lexical work on the language. However, even though this work is full of errors both of omission and commission, since its appearance, no Uto-Aztecanist has, to my knowledge, looked for Tubatulabal vocabulary in older sources for this language, which go back to the beginning of the century.