September 20, 2011
Recounting the Dead
By J. DAVID HACKER
Even as Civil War history has gone through several cycles of revision, one thing has remained fixed: the number of dead. Since about 1900, historians and the general public have assumed that 618,222 men died on both sides. That number is probably a significant undercount, however. New estimates, based on Census data, indicate that the death toll was approximately 750,000, and may have been as high as 850,000.
The notion that we’ve drastically undercounted the Civil War dead is not a new idea: in fact, Francis Amasa Walker, superintendent of the 1870 Census, estimated that the number of male deaths was “not less than 850,000.” So how did the lower number come to be the accepted count — and why does it matter that it was wrong?
Efforts to identify, rebury and count the dead began as soon as the war ended. A precise count proved impossible, however: both armies lacked systematic procedures to identify the dead, wounded and missing in action, as well as an official means to notify relatives of a soldier’s death. Men went missing; battle, hospital and prison reports were incomplete and inaccurate; dead men were buried unidentified; and family members were forced to infer the fate of a loved one from his failure to return home after the war.
Instead, postwar counts of the Union dead drew from regimental muster-out rolls and battle reports. An 1866 report compiled under the direction of Provost Marshal General James B. Fry estimated that 279,689 men in the Union forces died in the war. The estimated death toll increased to 360,222 by the late 19th century, partly as a result of widows and orphans bringing forward information when applying for pensions and survivors’ benefits.
But a direct count of the Confederate dead proved impossible. The destruction of the Confederate army and many of its records limited investigators to partial counts. The Fry report documented just 133,821 Confederate deaths from incomplete returns. That number didn’t change much: since Confederate widows and orphans were ineligible for federal benefits, the estimate was never supplemented with information from survivors.
Francis Amasa Walker’s interest in estimating the number of war-related deaths was a result of the 1870 Census returns. The final Census count put the population at 38,558,371, up just 22.6 percent from the count in 1860. All previous 19th-century censuses had documented decennial growth rates between 32.7 percent and 36.4 percent, a near-constant rate of increase that 19th-century Americans had come to expect and celebrate as a measure of the nation’s strength, progress and future prosperity.
The 31-year-old superintendent was understandably defensive. City boosters in Philadelphia and New York had charged the 1870 enumeration with excessive coverage errors, and President Grant had taken the unusual step of ordering a recount of those cities. Although the second counts failed to turn up many additional residents, the Census remained suspect. After all, if past growth patterns had continued, the population should have been 41.5 million. Had the Census somehow missed 3 million people?
Walker acknowledged that the 1870 census was far from perfect, but he refused to concede that it was more deficient in its coverage of the population than preceding censuses. Instead, he reasoned, the war was to blame. The disappointing growth rate, he countered, was the result of the “notorious and palpable effects of the war, which hampered the growth of the black population, checked immigration, limited marriages and births and led to the direct loss of close to a million men.”
Although the Surgeon General’s Office had at that point documented 304,000 Union deaths, Walker noted that the number was based only on those men who died during their terms of service. About a third of the 285,000 men discharged for disabilities and many of the remaining 2 million men who survived the war, he argued, subsequently died as a result of diseases and wounds contracted while in the Army. “Tens of thousands were discharged to die; tens of thousands died within the first few months after discharge,” he wrote. “Tens of thousands more lingered through the first or second year.” Together with the losses calculated by the Surgeon General’s Office, Walker concluded that “500,000 will surely be a moderate estimate for the direct losses among the Union armies.”
Walker’s estimate of Confederate losses was necessarily rougher. He started with a guess at the number of men participating — about half of the aggregate number participating on the Union side — and his assumption that Confederate soldiers’ longer average terms of service and relative lack of nourishing food, medicine and skilled physicians resulted in a greater risk of death. “Without attempting to deal at all nicely with this subject,” he argued, “it is difficult to see how anyone could, upon reflection, place the losses of the confederate armies at less than 350,000 men.”
Unfortunately, Walker did not pursue the line of inquiry further. After his reappointment as superintendent for the 1880 Census, he had to explain the overly rapid growth of the South’s population between 1870 and 1880 and defend the Census from charges of fraud in the form of over-counting. After a field investigation by the Census geographer Henry Gannett failed to turn up any evidence of fraud, suspicion returned to the 1870 census. Gannett charged that many of the 1870 enumerators were appointed for their Republican political connections, not for their local knowledge or ability to conduct a census. The inevitable result, he concluded, was a large undercount.
This time Walker agreed. Having been successful in pushing through many costly reforms for the 1880 census, one of which was to shift enumeration responsibilities from federal marshals answerable to the Justice Department to a much larger field force selected for their qualifications and answerable to the Census Office, Walker must have felt some measure of justification from Gannett’s report.
But with the census discredited — a crude calculation by the 1890 census office subsequently indicated that the 1870 Census had undercounted the South’s population by 1,260,078 (10 percent of the region’s and 3 percent of the nation’s population) — the opportunity for a more comprehensive examination of the war’s human cost was lost to the political winds. The estimate of 360,222 Union deaths stood.
The count of Confederate dead was, however, heavily debated. William F. Fox, a private citizen and Union army veteran whose 1889 book on regimental losses remains a classic reference work for Civil War historians, relied on battle reports and unofficial estimates to obtain a total of 94,000 Confederate battle deaths. He complained, however, that records were incomplete, especially during the last year of the war, and that battlefield reports likely under-counted deaths (many men counted as wounded in battlefield reports subsequently died of their wounds). In 1900 Thomas L. Livermore, who, like Fox, was a private citizen and Union army veteran, put the number of Confederate non-combat deaths at 164,000, using the official estimate of Union deaths from disease and accidents and a comparison of Union and Confederate enlistment records.
Livermore’s estimate assumed Union and Confederate troops suffered an equal risk of death from disease, a conservative assumption that Walker had explicitly rejected. Despite acknowledging that his estimate of disease mortality likely undercounted Confederate deaths and his concern that Fox’s estimate of battle deaths could “be accepted only as a minimum,” Livermore combined the two estimates to arrive at a total of 258,000 Confederate deaths, a total that remains unrevised more than a century later.
So why should we now doubt that number? For one thing, Fry, Walker, Fox, Livermore and other early investigators were limited by the quality of the data available. Using new quantitative sources, we can now make a more comprehensive and accurate estimate of war-related deaths. With one exception, microfilm copies of the original manuscript returns have been preserved for all censuses since 1850 (the 1890 Census manuscripts were lost in a fire). Census microdata samples created from these returns at the Minnesota Population Center make it possible to estimate undercounts by age and sex in censuses back to 1850 and to construct a Census-based estimate of male deaths caused by the war.
Census undercounts are estimated using multiple censuses and a demographic method known as back projection. The results confirm that, indeed, the 1870 Census was the most poorly enumerated. It was not nearly as bad as Walker feared and as 1890 census officials charged, however: the net undercount was 6.5 percent in 1870, compared to 6.0 percent in 1850, 5.5 percent in 1860, and 3.6 percent in 1880.
War-related losses are estimated by comparing sex differences in mortality during the 1860s with sex differences in mortality in the 1850s and 1870s. The results indicate that the war was responsible for the deaths of about 750,000 men (using less conservative assumptions, the total may have been as high as 850,000). Although that estimate is 100,000 fewer than the 850,000 deaths suggested by Walker, it is closer to his guess than it is to traditional estimate of 618,222 deaths, which has been cited uncritically for too long. If the Census-based estimate is correct, the traditional estimate is about 20 percent too low.
Although there are limitations to using Census data to estimate of Civil War mortality — civilian deaths are too few to be measured accurately, and deaths cannot be reliably divided into Union or Confederate subtotals — the method provides a more complete assessment of the war’s human cost. In addition to the men who died during their terms of service, the Census-based estimate of male mortality includes men who died between the date of their discharge and the 1870 Census from diseases and wounds contracted during the war, as well as non-enlisted men who died in guerilla warfare and other war-related violence. It excludes, however, men dying from war-related causes who would have died under the normal mortality conditions of the late 19th century. This final group, included in all direct counts of the Civil War dead, represents about 80,000 men.
So what? Above a certain count, do the numbers even matter? Well, yes. The difference between the two estimates is large enough to change the way we look at the war. The new estimate suggests that more men died as a result of the Civil War than from all other American wars combined. Approximately 1 in 10 white men of military age in 1860 died from the conflict, a substantial increase from the 1 in 13 implied by the traditional estimate. The death toll is also one of our most important measures of the war’s social and economic costs. A higher death toll, for example, implies that more women were widowed and more children were orphaned as a result of the war than has long been suspected.
In other words, the war touched more lives and communities more deeply than we thought, and thus shaped the course of the ensuing decades of American history in ways we have not yet fully grasped. True, the war was terrible in either case. But just how terrible, and just how extensive its consequences, can only be known when we have a better count of the Civil War dead.
On The Web: http://opinionator.blogs.nytimes.com/2011/09/20/recounting-the-dead/