Put briefly: reasonably consistent annual data about homicides were not systematically gathered and published prior to 1930, when the FBI began issuing its Uniform Crime Reports. There were vital statistics collected by individual states from about 1900, but these were incomplete. Recently, sociologist Douglas Eckberg has made a significant breakthrough in correcting and reestimating these vital statistics, so that now we have the first even remotely believable national time series of homicides per capita available. 1 And Roger Lane's pioneering research on Philadelphia established estimates for the period 1839-1901, cumulated in seven year intervals, which will probably prove to be prescient. 2
In a project designed to move back systematically as far as possible, I have been assembling an annual homicide series on New York City probing back to the late eighteenth century. 3 It is desirable to go back even to the seventeenth century in order to facilitate comparison with England and Europe, where research has shown high medieval and somewhat lower early modern homicide rates. 4 New York City is an attractive study site because of its size, age, unity of city and county boundaries, and historically consistent political borders. There was only one major political boundary shift: in 1898 the city and county expanded from Manhattan to incorporate the formerly separate city of Brooklyn, as well as the Bronx, Staten Island, and Queens. The population denominator has been adjusted in my data series to capture this shift. The English takeover from the Dutch occurred in 1664, ensuring a long governance of the city under similar legal systems.
The further back in time, of course, the more likely the series is to be broken or fragmented or biased by unknown losses. Without developing