AT ISSUE IN THIS BRITISH ARTICLE ARE THE THEORETICAL ASSUMPTIONS OF SYSTEMATIC EMPIRICISM WHICH ENTAIL THE PRODUCTION OF CATEGORIES AND GENERALIZATIONS AS EVIDENCE FOR THE EXISTENCE OF AN OBJECTIVE ORDER IN REALITY. ANTHROPOLOGISTS AS EMPIRICISTS CLAIM THAT THEORIES ARE CONSTRUCTED FROM FACTS AND PROVED BY THEM AND THAT THE PROCESS OF OBSERVATION, HYPOTHESIS FORMATION, AND THEORY TESTING IS THE FUNDAMENTAL BASIS OF THEIR DISCIPLINE AS A SCIENCE. THIS PAPER MAKES THE CONTRARY SUGGESTION THAT EVEN APPARENT EMPIRICAL KNOWLEDGE OF THE WORLD IS THEORETICAL, CONSTRUCTED IN CONCEPTS AND LANGUAGE WHICH ARE THEMSELVES THOUGHT OBJECTS AND NOT CONCRETE OBJECTS. THIS MEANS THAT ETHNOGRAPHIC DATA ARE REPRESENTATIONS OR INSERTIONS IN THEORETICAL CONSTRUCTIONS. THE IMPLICATIONS FOR SOCIAL ANTHROPOLOGY ARE NOT THAT THE STRATEGIES OF FIELDWORK SHOULD NECESSARILY BE CHANGED, BUT THAT THE DATA THUS OBTAINED SHOULD NOT BE SIMPLY ORDERED OR COLLATED AND TREATED AS ITEMS IN THEORY CONSTRUCTION. ON THE CONTRARY, ETHNOGRAPHIC DATA SHOULD BE 'READ' FROM A SPECIFIC THEORETICAL PERSPECTIVE INDEPENDENT OF ANY CONCRETE INQUIRY. IN ITS EARLY STAGES, THE SCIENCE OF ANTHROPOLOGY WAS A LARGELY UNDIFFERENTIATED DISCIPLINE ENCOMPASSING THE 'WHOLE SCIENCE OF MAN' AND ALMOST COMPLETELY IGNORING SOCIAL ASPECTS. RELEVANT FACTS WERE SOUGHT AS ITEMS OF PROOF IN THE GRAND SCHEMES OF MANKIND, WITHOUT INTEREST IN SOCIETIES AS HOLISTIC ENTITIES. THE IMPORTANT ELEMENT IN ANTHROPOLOGICAL WORK AS IT EVOLVED THROUGH THE EFFORTS OF RIVERS, MALINOWSKI, RADCLIFFE-BROWN, AND DURKHEIM REMAINED THE OBSERVER'S SELECTION PROCESS IN DATA COLLECTION, CHARACTERIZING THE FUNCTIONALIST-EMPIRICIST APPROACH THROUGH RELIANCE UPON THE KNOWING SUBJECT. CRITICISM OF THIS PRACTICE BY BACHELARD AND ALTHUSSER HAS LED SOME ANTHROPOLOGISTS TO CONSIDER INSTEAD THE MARXIST CONCEPT OF THE MODE OF PRODUCTION AND TO EXAMINE SOCIETIES FOR EXAMPLES OF EXPLOITATION OR CLASS STRUGGLE. THE TASK OF THE ETHNOGRAPHER THEN BECOMES THE EXAMINATION OF THE CURRENT SITUATION IN WHICH BOTH OBSERVER AND OBSERVED ARE INVOLVED, RATHER THAN THAT OF DESCRIBING A 'SOCIETY.' THE DATA ARE PRODUCTS OF THE CURRENT SITUATION AND MUST BE READ IN THIS CONTEXT. THE TASK IS NOT TO EXPLAIN THE REAL WORLD BUT TO EXPLAIN WHY IT APPEARS AS IT DOES. FOR THIS, A PROCESS OF CONCEPTUAL 'READING' IS PROPOSED INVOLVING AN ATTEMPT TO CONSTRUCT THE TOTALITY OF REPRESENTATIONS WITHIN WHICH THE PARTICULAR REPRESENTATIONS ARE PRODUCED, BE THEY TEXT, ARTIFACT, STATISTICAL RECORD, OR PHOTOGRAPH. NOTES AND REFERENCES ARE PROVIDED. (MRK)