¹ For the sake of clarity, I capitalize “state” when it refers to the concept and leave it lower case when it refers to one of the fifty state governments.

² One recent work on U.S. unemployment compensation, conducted within the paradigm of State studies, has an explicit research design focused on state governments, for the project recognizes that long before the federal government's activities in unemployment compensation, state governments had either implemented or tried to implement such programs. It is founded on the explicit understanding that “state level processes were central to the shaping of U.S. public social provision” (Amenta et al., 1987: 140). State studies tend to conceptualize the State in terms of a centralized system. A federal, highly decentralized system such as that of the United States poses what may be a false problem, one which inheres in the concept of a State study: Why is the system not centralized? To avoid such a false, or perhaps more accurately, premature question, Amenta et al. pose their problem as a comparison between nation states, thus facilitating their conceptualization without being trapped by the question of “Why no centralized unemployment compensation?”

³ The local governments, too, often have been wrongly conceptualized as powerless, essentially trivial, and purely dependent on state government, a conceptualization in conformance with their constitutional status. The result: a vast underestimation of the political (and economic) power of local government in the nineteenth and early twentieth centuries (Frug, 1980). Recent work has shown convincingly that the formal nature of Dillon's Rule, which in the late nineteenth century articulated the dependent nature of all local government, had little impact on local government behavior. Teaford (1984) has analyzed legislation to show that in most cases, local governments and cities got exactly what they wanted from legislators. Partly because it was so intensely local, and partly because of electoral control over taxes, the local State has always been underfunded, thus fiscally constrained by its own taxpayers, not state governments. McDonald's (1987) study of San Francisco makes clear that the limitations of local government “imposed” by the state government were in fact done so at the request of local government. My work on finance, for instance, has shown that local finance was very much a local political tool (Monkkonen, 1988; see also McDonald and Ward, 1984).

⁴ Above this should be the numbers arrested. Here I omitted the data: when reported, the arrests often exceed known murders, probably due to multiple arrests in response to the same offense. Note that arrests do not include name lists, hence may not overlap with the newspaper and coroner reports. In a newspaper-constructed data base, almost all of the mentioned homicides resulted in arrest.

⁵ Should we take this as evidence of an “autopoeitic” system, that is, a system that operates in relative independence from its environment (Teubner, 1984)? Perhaps. The behavior of the state courts exhibited somewhat less annual variation than did the murder rates in New York City, suggesting that the court system had a more stable, internally coherent process while the homicide rates fluctuated more randomly. Regressions of New York City homicides and state-level convictions as pure functions of time also suggests that convictions had a slightly smoother linear drift (R ² for the former is .05, the latter, .15). These statistics confirm to a very slight degree the independence of local criminal courts from actual crime. In that one can interpret R ² as proportion of variance explained, one could interpret the statistics to mean that internal court processes accounted for 10 percent of the court's behavior. My sense is that the 10 percent of convictions for which simple linear drift accounts is an appropriate proportion of court behavior determined by “autopoeisis” and that the remainder comes from the external environment.