Limited available evidence suggests complex seasonality in early-twentieth-century U.S. prisoner statistics. Prisoner censuses for July 1, 1922 and Jan. 1, 1923 allow a limited, half-year comparison of prison populations. Significant under-reporting in the 1923 counts for local penal facilities makes comparison for those facilities meaningless. Prison populations in federal and state facilities, which appear to have comparable coverage, fell 3% in the second half of 1922. In the 1923 census report, commitments to federal and state facilities were estimated to be 3% lower in the second half of 1923 compared to the first half. Because commitments to state and federal prisons are typically for longer than a year, the average daily number of prisoners during a year typically differs by less than 10% from the count of prisoners on a given day.^
Seasonality in jail populations is different from that for federal and state penal populations. Because jail stays are typically much shorter than prison stays, jail populations tend to be much more seasonal than prison populations. Interactions between changes in commitments and changes in typical lengths of stays can generate complicated seasonal effects in jails. Estimates from the prisoner census of 1923 indicates a 17% increase in commitments to local penal institutions (generic jails) in the second half of 1923, compared to the first half of 1923. However, reported population data for county and city jails in 1933 shows that the jail population increased 8.6% across the first half of the year. That increase consisted of 7% and 30% increase in male and female populations, respectively. The gender balance in the population increase suggests a seasonal effect rather than a trend effect. On a mid-year basis, jail commitment seasonality and jail population seasonality apparently moved in opposite directions. Given the wide variations in prisoners’ length of stays in jails, jail commitments and jail populations are not closely correlated.