My purpose here is simple—if not particularly lofty. I intend to characterize in general terms the activities that we engage in when we do social science history research, to describe somewhat abstractly how we conduct inquiry in social science history. This examination of our scholarly activities is provoked by several convictions: First, the philosophy of science—even an epistemology of social science history to the extent that there is one—is by and large not helpful to the conduct of inquiry. Second, the efforts in philosophy (and elsewhere) to conceptualize the conduct of inquiry in philosophy using examples from the physical sciences seem especially inapplicable to social science history. Third, when we do become self-conscious about what we are doing, we concentrate on methods and techniques of data collection and analysis and we create the impression that our scholarly research is fully described by technical decisions and activities.

“There simply are no reliable historical statistics on church membership,” writes a prominent sociologist of religion, Glock (1959: 39), “and it is extremely doubtful that accurate statistics can be produced through manipulating the unreliable ones.” Problems have so regularly hampered the collection of religious statistics in the United States that historians as well normally hold them in disrepute (Landis, 1957). “Nothing is more elusive in church history,” Littell (1971: 36) has written, “than honest statistics.” Commager (1950: 166) goes further: “Church statistics,” he charges, “attain an unreliability that would be a penal offense in a corporation.”

This article examines the extent of the potential market surplus by geographic region in the northern United States from Vermont to Minnesota on the eve of the Civil War. Our interest is in measuring the extensiveness and pervasiveness of market commitment, how dependent it was upon the transport system, and at what stage in the settlement life cycle it emerged. This study is designed to complement a similar one by McInnis (1984) for the area just north of the American border, an area then known as Canada West.

It is now commonplace to acknowledge that self-sufficiency never really existed in North American farming. From the earliest years of settlement farmers had some degree of orientation to the market. Even on the frontiers of most recent settlement, farms generally could not be characterized as being of a wholly subsistence nature. Nor was there some mystic moment when agriculture shifted from self-sufficiency to commercialization. The real issue is the extent of commercialization at any time and place. One way of examining that is to determine the extent to which specialization and resort to the market was feasible, through assessing the magnitude of marketable surpluses. How much did farms typically produce over and above the consumption needs of their own households, and how did that vary according to circumstances?

The study of the socioeconomic differentiation of urban residential patterns has occupied generations of social scientists. Sociologists, economists, and political scientists adduce a strong relationship between residential patterns and human behavior. Moreover, the residential patterns established early in the city’s development are also believed to influence the evolution of future social and spatial organization (Hurd, 1903; Hoyt, 1939; Hoover and Vernon, 1959). Economists and political scientists have been preoccupied with the rapid suburbanization of the urban population since the end of World War II and the fragmentation of the metropolitan area into independent political jurisdictions (ACIR, 1965; Margolis, 1961; Oates, 1971; Tiebout, 1956; Pack and Pack, 1978).

Characteristically, a problem of contemporary social science is obtaining from the members of a sample of n subjects information concerning k variables that are of interest to the investigator. The usual methods of solving this problem are the use of questionnaires, interviews, and/or observation schedules. An immense amount of literature is available on these data collection instruments, which need not be reviewed here. But the instruments rarely succeed entirely in solving the problem; all are subject to the further difficulty that for some fraction of the sample and some variables the required information proves to be unobtainable.