The Problem of the Census

With eleventh census not yet completed, the Census Bureau had to begin to make arrangements for the next. If the ever increasing length of time necessary to collect data for the census was not slowed, the Bureau was in danger of being forced to run two census simultaneously. In order to avoid this, the bureau needed an efficient method to tabulate data.

The Census Bureau held a competition to find a more efficient method to determine the nation's population. The efficency of each system was judged by the time necessary to input and compute the data collected from four areas of St. Louis. Hollerith was faced with two main competitors, William Hunt and Charles Pidgin. Hunt, who had recently resigned as the head statistician of the Population Division, entered the "slip" system. Pidgin entered the "chip" system. Each system made use of color pieces, slips in the former and chips in the latter, which coded for various data. The methods used by Hunt and Pidegon transcribed the data in 144 hours, 25minutes and 100 hours, 25 minutes, respectively. Hollerith completed the task in 72 hours, 27 minutes. The data was then compiled by each machine. Pidgin's system took 44 hours, 41 minutes; Hunt 55 hours, 22 minutes. Hollerith's system again clearly out ran all of is competitors by scoring a time of 5 hours, 28 minutes.⁴ Hollerith's machine won the contest.