Following some of the more recent sociological literature which has been critical of research into riots, the topic of this thesis addresses itself to a hitherto neglected aspect of riots. It is an initial exploratory effort into the ecological dimensions of official statistics, utilizing the relevant temporal and spatial conceptualizations suggested by the sociological disaster literature. The data sources were the offense and arrest records of the District of Columbia Metropolitan Police Department and the fire data on the Daily Communication Log of the District of Columbia Fire Department. The offense and fire data were treated as partial indicators of the situation reported to the police with the arrest data as partial indicators of the response made by the police to the riot. The data were conceptually organized along three dimensions. First, the type of criminal violation was classified into six general categories: crimes against persons, crimes against property, traffic violations, crimes without victims, crimes related to fires, and miscellaneous crimes. For a more detailed analysis, the total crimes falling into any one of these general categories were subclassified into more detailed subcategories within each general category. Second, the spatial dimension was trichotomized into three locational specifications: the riot areas of major destruction, the corridor areas of sporadic destruction, and the non-riot areas of minimal or no riot destruction. Third, the temporal dimension was dichotomized into the total riot period of organized response and a representative normal time period, so that the latter could serve as a benchmark against which to compare the former. Two specific questions were posited: what degree of difference existed between the defined riot period and the representative normal time period in terms of crimes and spatial location as reflected by the official statistics and what kinds of differences were evident. Three specific hypotheses were evaluated: (1) the offense and fire data hypothesis which suggested that the degree of association between the offenses reported and the selected riot-normal time period varies directly with the degree of concentrated riot damage, (2) the arrest data hypothesis which suggested that the degree of association between the police's response and the selected riot-normal time period varies directly with the degree of concentrated riot damage, and (3) the comparative hypothesis which suggested that the degree of association between the police's response and the selected riot-normal time period is less than the degree of association between the offenses reported and the selected riot-normal time period . Utilizing the lambda proportionate reduction in error statistic, the data were inconclusive relative to the first hypothesis and generally failed to support the second and third hypotheses, although the magnitude of the data indicated that there were some differences. The nature of the differences indicated that the incidence of fires and burglary violations increased substantially, while larceny, false fire alarm reports, and the degree of violence in crimes against persons decreased in the reported offenses during the riot. The police response was dominated by arrests for disorderly conduct and curfew violations with burglary arrests ranking second. While there were decreases in larceny and traffic arrests, the latter were still substantially represented during the riot and no meaningful numbers of arson arrests were made. Further, it was concluded that substantial numbers of offenses reported and arrests made occurred in the non-riot areas. It was concluded that the disaster literature provided relevant conceptualizations for the analysis of the spatial and temporal dimensions of riots, that further analysis of these dimensions is warranted, and that other dimensions of the disaster approach appear to be useful when applied to riots.