IN THIS PAPER I am assuming that I am addressing those who have not done fieldwork and that those who have are simply reading with a critical eye. I am assuming also that the informants to be selected are for research in dialect geography and not in social dialectology. William Labov, Roger Shuy, and others active in that specialty have dealt with it elsewhere; I must concern myself with what is within my own experience. The experience of all of us who work with regional speech rests upon one very basic assumption. This is that the speech of a person reflects the speech of the community of which he is a part, specifically of a segment of that community of which he is a part. A man's idiolect is his own, of course, never entirely like anyone else's, but nevertheless it differs only in very small measure from the idiolects of those with whom he associates in his own speech community. Here is where our research differs from that of the social scientist, for in one informant we can have a computerized summary of the language of that community. I say "we can have," for not everyone you pick up on the street or encounter in a house-by-house survey is a satisfactory informant. When is an informant satisfactory ? When is one acceptable ? A deductive answer can be found, of course, in the clear statement provided by Hans Kurath in chapter 2 of the Handbook of the Linguistic Geography of New England.' With that statement in the back of our heads, let us lead up to the answer inductively. I would like to try to find an answer in terms of generalizations from a number of brief case studies drawn from actual experience. Remember that our basic assumption is that a person's speech is a microcosm of the speech of his language community, that is, of those people who have common speechways, who respond in the same characteristic way to a given need to communicate. If we want to study only the older segment of the speech community, we naturally want an older person; if we want the younger segment, we need a younger person. Similarly with respect to educated and uneducated segments or to further arbitrary divisions along a continuum of educational background from illiteracy to the Ph.D.