Quantifying Union Prisons

What the Official Records Can and Cannot Tell Us about Civil War Prisons
Jack Furniss & William Kurtz
July 1862 Prison Returns

The Nau Center is in the very early stages of a digital project looking to add to what historians know about military prisons in the Union and the Confederacy. This venture hopes to use digital methods to make a vast collection of data more accurate, manageable, and viewable in order to enrich our understanding of the Civil War prison system.* While it has not always featured centrally in our telling of the war’s narrative, for those who passed through it the prison system held abiding significance, as well as playing a central role in political and military negotiations during the conflict and figuring prominently in conflicting postwar memory traditions.

Investigating some of the numbers seemed a logical place to start. Do we know how many prisoners were held in Civil War prisons and how many died? The figures most regularly quoted by historians originated in volume five of James Ford Rhodes’s mammoth history of the United States in the Civil War era, published in 1904. Rhodes explained that he got his numbers from General F. C. Ainsworth, Chief of the Record and Pension Office. Ainsworth reported 214,865 Confederates in northern prisons, with 25,976 deaths (12 percent), and 194,743 Union inmates in Confederate prisons, with 30,218 deaths (15.5 percent). Ainsworth’s figures are widely quoted by historians but it is not clear how he calculated his totals.

We decided to see how Ainsworth’s numbers compared to the amalgam of all the monthly data for Union prisons given in Series II, Volume VIII, pages 986-1004 of the Official Records (OR). We have compiled these charts into a spreadsheet, allowing us more easily to manipulate and interrogate the data. We quickly discovered that the OR records offer no easy means to corroborate or question Ainsworth’s numbers. We found that what the OR provides is a record of 320,155 instances of imprisonment.† This number is far higher than Ainsworth’s number of individual prisoners primarily because it includes transfers between prisons. From the monthly returns, it is impossible to know exactly who was transferred and how many times. Hence we cannot be sure how many men would be counted twice or more within the total number including transfers. Nonetheless, we can make some reasonable assumptions. When we tally up the totals for transfers to other destinations (Table 2) and compare those to the overall imprisonment totals for each camp (Table 3) we can see that certain prisons – Louisville, Nashville, and Fort McHenry for instance – served primarily as hubs where prisoners arrived before being moved on to another permanent location. It seems reasonable to assume then that many prisoners would be transferred just once, making them appear as a duplicate, but not more, in the records. When we remove the 139,820 transfers listed in the OR, we get an estimate of 180,335 POWs, a number 34,530 below Ainsworth’s total. For deaths, we calculated 20,338, more than five thousand fewer than Ainsworth’s number. Coming in well under Ainsworth’s estimates actually corresponds with gaps that we know exist in the OR.

The OR dataset is broken down by each prison for every month, but the records only begin in July 1862 (when the Commissary General of Prisoners ordered prisons to start sending returns to the War Department), and there are a good number of months thereafter that still have no data for certain prisons. One challenge for us will be to amass and amalgamate accurate records for the first year of the war. We know that while some informal exchanges and a limited number of battles kept prisoner numbers relatively low, many soldiers still entered the prison system in 1861. When the formal Dix-Hill exchange system came into practice in summer 1862, the prison population dropped precipitously. A year later the Union ended exchanges due to the Confederacy’s refusal to recognize black soldiers as prisoners of war. As Table 4 shows, the number of captives then soared dramatically.

In the OR data we do have, from July 1862 onward, there are still significant problems to address. Sometimes, explanations prove straightforward. Camp Butler, for instance has no returns from November 1862 to January 1863, but we know the site stopped being a site for POWs during this period, serving purely as a base to train recruits. Other gaps are not so easily explained. No records exist for Camp Douglas in March 1863, when we know the prison was in operation and indeed that deaths peaked around this time due to overcrowding, freezing temperatures, and smallpox. Several other prison camps similarly have blanks in months of known operation. Another issue is prisons that we know to have existed but never made it into the OR data or which appeared and then vanished entirely. The OR begins with just twelve sites, but, by late summer 1864, it listed twenty-five prisons. Even so, we know that as the War Department’s prison system evolved, other locations not officially listed at all–Riker’s Island, New York, or Fort Jefferson in the Dry Tortugas, being two examples–nonetheless served at least briefly as prisoner of war camps. Finally, as the introduction in the OR itself states, the returns sent in by prison camps often failed to distinguish between prisoners of war and imprisoned citizens, further muddying attempts to extrapolate precise figures on POWs.

When we move past the question of the total number of individual POWs, we realize that having a digitally accessible version of the OR data still tells us a great deal about the overall prison system. With monthly information on deaths, escapes, releases, sickness, and the number of citizens held, for each individual prison, we can begin to aggregate, interrogate, and display the Union prison system. As some of the tables below show, we can see how populations waned and waxed over time, how certain prisons served specialist purposes, and we can take snapshots of the prisoner population at particular moments in the war. As we attempt to fill the gaps in the OR, we will be more able to apply it to big historiographical questions. Does the data support the idea that prisoner experience can be usefully generalized, or did everything – from the possibility of escape to the likelihood of death – depend on the particular prison? Can the OR’s monthly death and sickness figures support Charles W. Sanders Jr.’s argument that a deliberate, revengeful policy of lowering rations and worsening conditions in Union prisons brought a commensurate increase in suffering for Confederate POWs? When we examine the number of citizens recorded as inmates, do we find justification for the claims made by wartime Democrats that Union prisons swelled with political prisoners in the months before elections? How does this relate to perennial debates about the Lincoln administration’s record on civil liberties?

While currently focused on the Union, this Nau Center project will ultimately try to build a similar database for Confederate prisons. There is no collated breakdown of the records of individual Confederate prisons as the OR provides for the Union. Inevitably, for the Confederacy even more so than for the Union, no perfect set of records can be reconstructed. But we believe we can get closer, and in doing so we can learn a great more about the Confederacy and about the lives of many thousands of Union soldiers.

The mammoth task of collecting and assessing missing data will be aided by the pioneering work already done by historians on both individual prisons and the systems as a whole. Delving into the archival records ourselves and building on the work of others, we aim to develop the most detailed picture of the Civil War prison systems possible. This will be made digitally visible during the process, with an interactive website that will allow users to watch a visualization showing the geographical and chronological development as individual camps filled and new ones were constructed as the war progressed. The journeys and stories of representative individuals through the system will also be available to allow visitors to the site to remain cognizant of how the people caught within this vast bureaucracy experienced it.  

We are in the very early days of a major undertaking and this blog post primarily lays out initial findings and an intended direction of travel. We hope this project will quickly become collaborative, linking with other scholars and institutions in both the collection and presentation of information on Civil War prisons. So if you are reading this blog and want to get involved, please do reach out to us at wkurtz@virginia.edu.

Jack Furniss is a PhD candidate in the Corcoran Department of History at the University of Virginia. His dissertation, “States of the Union: The Political Center in the Civil War North,” uses the careers of Northern governors to recapture and reevaluate the role of centrist politics in governance and ideology during the Civil War era. 

* Our project currently examines only those prisoners of war who became inmates of Federal and Confederate prisons. It does not include those who were captured and then paroled on the battlefield or prisoners who died before they reached a prison.

† This number is itself an estimate due to imperfect data in the OR. We calculated it by adding the number of prisoners listed as "joined" for each month to the initial population figure. There were some instances where records for a prison were missing for a month, or even several months, before restarting. In these cases, even if we knew from other studies that the prison remained in active operation, we conservatively added to our total the minimum number of prisoners that we could definitely verify had been added to that prison's population in the intervening months.