Abstract

Accounts from victims and observers, including new research in the U.S. National Archives and the Bavarian National Archives, suggest that American soldiers committed crimes against persons—especially rape and various forms of assault—and against property in Europe after World War II more often than statistics about charges and prosecutions at the time indicated. More importantly, previously unexamined statistical summaries of crimes committed by American troops, as recorded by the U.S. Provost Marshal, provide unprecedented quantitative information about these crimes in the European Theater of Operations (eto) during the first postwar year, May 1945 to June 1946. The absolute number of crimes decreased as the number of troops declined, but the rate of crime (number per 10,000 troops) increased during the same period.

American soldiers engaged in all manner of criminal activity as they made their way across Europe after World War II. But in comparison with the well-established criminality accompanying the Soviet occupation in the East, the rate of violent and nonviolent American criminality, and its impact on occupied and liberated European societies, has been difficult to determine because of the lack of adequate statistics. At the time, civilians in France, Italy, the Benelux countries, and especially Germany made frequent reference to such American criminal activity as looting, petty thievery, drunkenness, rape, and murder. U.S. military reports confirm American soldiers’ predilection for theft and black-market racketeering, often couched in the euphemisms “fraternization” and “harassment”; some of them even suggest more violent inclinations. These depictions, however, had little effect on the early historiography of American soldiers in Europe. The first historical narratives were shaped mainly by the accounts of witnesses and victims, along with a few crime statistics. Scholars writing between the 1940s and 1960s acknowledged the looting and drunkenness but downplayed the violence, preferring to portray American soldiers as liberators and valiant conquerors of Nazism, war weary and prone to excessive drinking but not rapists and murders.1

Beginning in the 1970s, a small but dedicated literature questioned the earlier, positive narratives, arguing that the American proclivity for violent crime was worse than previously suggested. Mirroring scholarly attention to the behavior of the Soviets, it focused primarily on American-perpetrated rape, yielding numerous new reports of horrendous acts of violence, while also confirming American soldiers’ more well-known misbehavior. These findings align with the broader works of Bessel, Mazower, and others who recently suggested far more pervasive criminal violence in Europe in the transition from war to peace. Nevertheless, quantitative analysis of American crime has remained difficult to obtain, focusing, understandably, on rape and sex crimes alone.

The estimated number of American-perpetrated rapes in Germany and France in 1945—around 14,500—as derived from prosecutions by the Judge Advocate General (jag), do not align with that based on the reports of observers and witnesses, which, in Kennett’s words, indicate “a disturbing pattern.”2

These “low” estimates of rape by Americans have had important consequences for historical interpretation. Even at their highest, they pale compared to the 1 to 2 million rapes believed to have been committed by Red Army soldiers during roughly the same period, lending Americans the better reputation. In fact, the existing quantifiable data have become the empirical linchpin for critiquing recent allegations that Americans engaged in more violence than once thought. In 2007, MacDonogh used these data to argue that the U.S. zone had fewer sexually violent incidents than the Soviet zone because “the Americans were more attractive to the Germans…they were taller and more athletic.” Similarly, Walters cited the (jag) data in his dismissal of Gebhardt’s “far too high” estimate of 190,000 American rapes in the ten years after the war—based on figures for live American–German births—as “motivated more by some Leftist anti-American agenda than by proper historical inquiry.”3

This division in the existing literature indicates a pressing need for a re-examination based on new criminal statistics regarding American crime and its effect on European societies liberated or occupied by American soldiers. This article presents and analyzes newly discovered statistical summaries of investigations and charges against American soldiers in the European Theater of Operations (eto) just after the war (from May 1945 to July 1946), as issued by the Provost Marshal (the head of the military police [mp]). This analysis reveals that the rate of rape and other assaults by American soldiers was much higher than previously suggested. Moreover, the high rate of rape was one part of a broader, more pernicious trend in American-perpetrated crime. The monthly rate of crime was high immediately following the war, decreasing during the summer before increasing again, even as the number of American soldiers in Europe fell dramatically.4

These findings permit a confident calculation of American criminal acts in Europe after the war. A higher rate of American criminality, especially criminal violence, has important implications for our interpretation of Western Europe’s social and cultural transition from war to recovery. It lays the foundation for new comparative analyses of American postwar occupations in various European contexts and in Japan, as well as of occupations by other the Allied powers in Europe—the Soviets, the British, and the French.

Crimes Committed by American Soldiers

Nearly every scholar who has examined American criminality and violence in Europe during and after World War II has uncovered new accounts of brutality and violence. This study is no different. The following two examples are indicative of the reports by victims that were discovered during archival research for this article. On April 15, 1945, as Allied troops swept across Germany, American soldiers entered the homes of Ilse Reuhl and Lina Hachenburger in the town of Butzbach and the rural district of Hanau (Landkreis Hanau) north of Frankfurt. Reuhl was twenty-three years old at the time, living with her parents and her fourteen-year-old sister. She made her statement first to German authorities, who then relayed it to the U.S. military government (mg). According to Reuhl, the soldiers demanded entry, claiming to be “American police.” Once in the house, they forced the family into the living area and took turns raping the daughters in front of their parents. Hachenburger’s account was virtually identical. She was home alone hiding with her young children when the soldiers knocked on her door. When she failed to respond, they broke in and ransacked the house before forcing the children to watch as each of the five soldiers raped their mother.5

Similar accounts of American criminal violence are ubiquitous in the archival records. Klaus-Dietmar Henke mapped a dramatic rise in American-perpetrated rape in Germany from March to June 1945, the months spanning the final conquest of Germany and the end of the war. In his view, this “rash” of violence can be attributed to the progress of the war, but the frequency with which American soldiers abused the people that they were supposedly liberating indicates other factors at play. Roberts found numerous accounts of rape and assault by American soldiers in Normandy during the months following the June 1944 invasion. Lilly also uncovered many individual descriptions of violent assaults, rapes, and gang rapes in Britain and France, as well as in Germany. Together, Roberts’ and Lilly’s findings suggest a more pervasive problem that conquest and the mere presence of large numbers of liberating/occupying soldiers cannot explain.6

Rape is just one of the more egregious examples of what appears to have been a pervasive American propensity for criminal violence, at least according to observers and victims. In the days following the capture of Nuremberg on April 21, 1945, “numerous reports of looting and cases of rape” were made to the first mg detachment to govern the city. The Third Infantry Division’s mps confirmed the problem. In a single day, they investigated three rapes, two incidents of “fraternization,” and a “summary matter” involving soldiers “creating a disturbance on civilian premises.” Similarly, judging from a military report describing the first weeks of U.S. occupation in the Bremen Enclave, “looting…by invading troops” was so severe that it threatened the “security situation” in the newly occupied city.7

Although Henke’s data and the above reports show rape and violence to have been distinct features of the period during and shortly following the establishment of U.S. control, reports of significant violence were also in evidence later in the occupation. An attack in the Friedberg train station in Hesse on June 22, 1945, two months after the defeat of German forces in the region, is indicative. According to German witnesses, a group of drunken American soldiers who entered the station late in the evening began to harass the German railway police and German soldiers. Holding the Germans at gunpoint, they tore their identification papers and beat them. They then forced nearly all of the women and girls outside, “dragging some by the hair,” and repeatedly raped them, their screams echoing through the station.8

These descriptions of violent crimes form part of a broader picture of criminality and disorder that emerges from the documentary record. For instance, the citizens of Kulmbach in northeastern Bavaria complained vehemently in July 1945 when American soldiers randomly and drunkenly fired their weapons at night. The U.S. commander of the mg detachment noted that the shooting caused “consternations among the patients” in a local hospital, “many of whom [were] in a poor mental condition.” According to the mg detachment in Nuremberg, American soldiers stole and vandalized art treasures and absconded with vital building supplies.9

Persistent American criminality caused mg officers (mgos) and detachments in Germany to view American soldiers as an impediment to orderly reconstruction. On October 25, 1945, Lt. John E. Westburg, assistant public safety officer for rural district Hanau, lamented that the “number and gravity of the incidents [of U.S.-perpetrated crime] is steadily becoming worse.” American crime was so serious that it undermined mg efforts to ensure civilian law and order: “The administration of police and protective measures and the enforcement of civilian law and order is becoming increasingly difficult owing to the continued misbehaviour of US troops.” A U.S. mg report issued on November 9, 1945, from Bremen listed twenty-nine incidents by U.S. soldiers, including sixteen assaults on civilians, a drunken shooting that injured a child, and “molestation of German girls on the street.” In that same month, Maj. Melvin Mawrence in rural district Munich even took the extreme, and admittedly rare, step of deploying German police under his command against rowdy Americans taking advantage of their extra-territorial protections.10

The German police also frequently complained about crime by U.S. soldiers. In late August 1945, the German chief of police in Darmstadt compiled a list of civilian complaints about livestock theft, armed robbery, and, in one case, an American truck deliberately running over a pedestrian. In November, the German chief of police in Bremen lodged a similar complaint about his inability to stop American soldiers from committing, among other things, “unprovoked beatings and assaults, armed robbery, house-breaking, and rape” because of their extra-territoriality immunity. The daily diaries and monthly logs of U.S. mg detachments in regions throughout Germany are rife with such reports. In rural district Dillingen, for example, the mg commander Maj. Claude F. Baker was forced to assign an officer the full-time task of handling German grievances.11

The higher levels of U.S. military command were acutely aware of the spiraling crime rates among U.S. soldiers. Beginning in late 1945, they evinced an increasing concern about the damaging effects on German attitudes toward the occupation. Acknowledging this growth in concern, Ziemke attributed this “epidemic” of crime in August 1945 to the green, poorly disciplined, and often idle, soldiers who replaced the veteran troops. Although the problem started earlier, the Provost Marshal’s office was alarmed by the increasing rate of American crime throughout the first year of the occupation, even as the number of soldiers rapidly declined. In April 1946, the Provost Marshal argued that the heavy involvement of U.S. soldiers in the black market, and its attendant crimes, constituted a “menace to security” that challenged “the success of Military Government.”12

Quantifying the Crimes of American Soldiers

Despite the volume of reports by observers, victims, and military personnel that this article and previous studies have unearthed, the extent and frequency of American crimes has been difficult to establish. Important questions about documentary data have thus far remained unresolved—for instance, whether the reported events were merely the tip of an iceberg of pervasive criminality or whether, by focusing on the most salient incidents, these reports may have inadvertently exaggerated the quantity of criminality. Recent research in the social sciences repeatedly demonstrates that even a few sensational reports of crime can increase the perceived frequency and severity of criminal activity. U.S.-perpetrated crimes in Europe certainly occurred often enough to seed negative perceptions among the population. Why should we think that neither the German public nor the mgos would be susceptible to inflated perceptions about it?13

Previous attempts to quantify the extent of American criminality have faltered because they had access only to the limited jag data. At minimum, these prosecutions would have been confined to instances in which a definite suspect had been identified and a prima facie case against that individual had been established. In the absence of a suspect or sufficient evidence, many of the crimes that came to the attention of the mps never reached court. Moreover, the crimes known to mps were almost surely only a fraction of the larger number of crimes that went undetected or unreported—the so-called “dark number” in criminological studies.14

In addition to the overall dilution in the number of crimes that actually reached the courts, certain crimes were less likely to be prosecuted than others. Crimes that were difficult to prove in court—in particular, rape—are liable to have been underrepresented. Lilly reports that 733 American soldiers in Germany and France were prosecuted for rape by jag in 1945, and Henke located another 400 to 1,000 soldiers who appear to have been prosecuted for sexual crimes, judging from the euphemisms in the charges and the reductions in charges. The race of alleged perpetrators might also have figured into legal outcomes. According to Lilly’s analysis, African-American defendants were significantly over-represented among jag prosecutions.15

Notwithstanding these constraints, several scholars have attempted to estimate the number of rapes throughout the eto in 1945. Given Radzinowciz’s well-known hypothesis that roughly 5 percent of rape victims report the crime, Lilly calculated the number of rapes committed by American soldiers in Germany and France from January to September 1945 as 14,660. He based his total on the higher of two figures for incidents of rape versus the number of victims (552 versus 484 for Germany, and 181 versus 125 for France) that he found in jag records for the period in question—hence, 552+181, or 733, divided by .05 = 14,660.16

As noted above, Gebhardt used births of children fathered by Allied soldiers to estimate the number of rapes in occupied Germany, thus producing a higher estimate. For the period 1945 to 1955, she found evidence for 1,870 German babies fathered by American soldiers through rape. She also used Radzinowicz’s hypothesis to infer that only 5 percent of rapes resulting in pregnancy came to light, venturing further that rapes resulting in pregnancy represented 19 percent of the total number of rapes. From the product of these two estimates (.05 × .19 = .01), Gebhardt surmised that only 1 percent of all American-perpetrated rapes resulted in a reported live birth, thus calculating that approximately 190,000 rapes (1,870 * 1/.01 = 187,000) occurred during the first decade after the war.17

Gebhardt’s method raises a few questions. At the risk of oversimplification, her estimate that approximately 9,500 rapes occurred during any six-month period roughly agrees with Lilly’s estimate for April through September 1945. However, it is not clear how she handles the rapidly decreasing number of American soldiers in the eto from 1945 to 1946. If Gebhardt’s method were applied to the large number of troops still present in Europe during that period, plus an allowance for the nine-month gestation period, her estimate might be higher, although how much higher is impossible to know. Furthermore, Gebhardt’s estimate of the percentage of rapes resulting in pregnancy appears to be inflated. Prospective studies of pregnancy after a single act of unprotected intercourse indicate that, at the height of the menstrual cycle, the percentage of pregnancies is only about 10 percent. Across the entire menstrual cycle, the percentage of pregnancies per rate of intercourse is about 3 percent. Finally, Wiegrefe and others undermine Gebhardt’s method of estimating incidents of rape from live births because of the difficulty of distinguishing children of rape from children of consensual relations. Moreover, Grossmann found that almost 90 percent of pregnancies resulting from rape were aborted; the legal prohibitions against abortion were set aside in favor of granting reprieve to women violated by people whom the Nazis deemed racially inferior. Although, as she notes, these figures pertained primarily to Berlin, the pattern may well have been similar elsewhere. If she is correct, live births would dramatically underestimate the rate of rape.18

Collating the Provost Marshal’s Summaries

The U.S. Provost Marshal recorded month-by-month statistical summaries of investigations and criminal charges against U.S. military personnel in the eto. The mps and the criminal investigation branch (cid) provided most of the data for these reports, which are scattered across files in the collection examined and not always uniform in their tabulations and classifications of crime. Nevertheless, we have integrated these summaries to create a consistent set of statistics on U.S. crime covering a fourteen-month period from May 1945 to June 1946.19

Table 1 shows month-by-month tabulations of the number of incidents involving U.S. soldiers investigated by mps or cid (upper portion) and charges at arrest (lower portion) across the entire eto. At the war’s end in May 1945, the eto stretched from the United Kingdom across France and the Benelux countries into Germany, Austria, a portion of Czechoslovakia, and a slice of Italy that was temporarily occupied by U.S. troops. The tabulated figures therefore do not include the major body of U.S. troops in Italy, which were under the separate command of the Mediterranean Theater of Operations. The number of troops in the eto at the start of each month is listed in the bottom row.20

Table 1

Frequency of Crime Reports and Charges: U.S. Military Personnel in the European Theater

codecrime reportedmay-45jun-45jul-45aug-45sep-45oct-45nov-45dec-45jan-46feb-46mar-46apr-46may-46jun-46
Homicide    23 36 37 37 28 45 28 17 21 33 27 
Rape    25 25 39 29 26 10 11 18  31 35 
Assault    93 130 115 146 88 76 95 50  51 61 
Other crimes against persons     
Stealing and fraud    112 168 185 207 141 117 55 30  35 32 
U.S. property offenses    68 84 71 52 53 91 40 36  40 33 
1–4 Crimes against persons    147 191 191 218 148 134 134 85  115 123 
5–6 Crimes against property    180 252 256 259 194 208 95 66  75 65 
1–6 Total serious crimes    327 443 447 477 342 342 229 151  190 188 
Other incidents    103 170 132 168 140 155 80 71  95 108 
1–7 All incidents    430 613 579 645 482 497 309 222  285 296 
 Smith report    586 611 596 697 509 489 318     
 
code charge at arrest may-45 jun-45 jul-45  sep-45    jan-46 feb-46 mar-46 apr-46 may-46 jun-46 
Homicide 72 57 39  13    27 30 12 18 17 15 
Rape 284 74 55  15     15 19 
Assault 244 120 143  75    51 51 18  30 41 
Other crimes against persons 11 24 21      
Stealing and fraud 311 235 106  88    52 18 16  21 19 
U.S. property offenses 111 96 112  52    57 41 22  19 12 
1–4 Crimes against persons 611 275 258  103    85 88 35  62 75 
5–6 Crimes against property 422 331 218  140    109 59 38  40 31 
1–6 Total serious charges 1,033 606 476  243    194 147 73  102 106 
Other charges     71    92 56 40  48 57 
1–7 All charges 1,033 606 476  314    286 203 113  150 163 
      Smith Report 290 314 315 312 252 286 194   
 US Troop Strength (000’s) 3,021 2,812 2,488 2,164 1,795 1,424 936 660 553 493 407 368 329 290 
codecrime reportedmay-45jun-45jul-45aug-45sep-45oct-45nov-45dec-45jan-46feb-46mar-46apr-46may-46jun-46
Homicide    23 36 37 37 28 45 28 17 21 33 27 
Rape    25 25 39 29 26 10 11 18  31 35 
Assault    93 130 115 146 88 76 95 50  51 61 
Other crimes against persons     
Stealing and fraud    112 168 185 207 141 117 55 30  35 32 
U.S. property offenses    68 84 71 52 53 91 40 36  40 33 
1–4 Crimes against persons    147 191 191 218 148 134 134 85  115 123 
5–6 Crimes against property    180 252 256 259 194 208 95 66  75 65 
1–6 Total serious crimes    327 443 447 477 342 342 229 151  190 188 
Other incidents    103 170 132 168 140 155 80 71  95 108 
1–7 All incidents    430 613 579 645 482 497 309 222  285 296 
 Smith report    586 611 596 697 509 489 318     
 
code charge at arrest may-45 jun-45 jul-45  sep-45    jan-46 feb-46 mar-46 apr-46 may-46 jun-46 
Homicide 72 57 39  13    27 30 12 18 17 15 
Rape 284 74 55  15     15 19 
Assault 244 120 143  75    51 51 18  30 41 
Other crimes against persons 11 24 21      
Stealing and fraud 311 235 106  88    52 18 16  21 19 
U.S. property offenses 111 96 112  52    57 41 22  19 12 
1–4 Crimes against persons 611 275 258  103    85 88 35  62 75 
5–6 Crimes against property 422 331 218  140    109 59 38  40 31 
1–6 Total serious charges 1,033 606 476  243    194 147 73  102 106 
Other charges     71    92 56 40  48 57 
1–7 All charges 1,033 606 476  314    286 203 113  150 163 
      Smith Report 290 314 315 312 252 286 194   
 US Troop Strength (000’s) 3,021 2,812 2,488 2,164 1,795 1,424 936 660 553 493 407 368 329 290 

May, June, July 1945

The numbers shown for May through July 1945 were compiled from a set of tables entitled “Crimes Investigated by MP CI Personnel.” A footnote in each statistical summary states, “In the above statistics, the accused is charged with the major crime committed,” indicating that the listed numbers refer to individuals charged with a crime. Consequently, each individual charged was likely counted once, even when facing multiple charges. These numbers are displayed as “charges.”21

The categories of crime listed in Table 1 were based on finer sub-categories in the summaries. The totals shown were aggregated into categories labeled “homicide” (murder and manslaughter), “rape” (rape and assault with intent to rape), “serious assault” (assault with a deadly weapon), and “other crimes against persons” (riot). These categories were aggregated into a larger category of “crimes against persons.” The remaining categories were “stealing and fraud” (robbery, burglary, and housebreaking) and “U.S. property offenses” (government property), which, as later summaries indicated, entailed all manner of theft and misappropriation of U.S.-owned property. These data were aggregated into “crimes against property.”

Table 2

Rate of Crimes per 10,000 U.S. Military Personnel, Mid-Month Estimates

codecrime reportedmay-45jun-45jul-45aug-45sep-45oct-45nov-45dec-45jan-46feb-46mar-46apr-46may-46jun-46
Homicide    0.12 0.22 0.31 0.46 0.46 0.86 0.62 0.44 0.60 1.07 1.28 
Rape    0.13 0.16 0.33 0.36 0.43 0.19 0.24 0.46  1.00 1.65 
Assault    0.47 0.81 0.97 1.83 1.45 1.45 2.11 1.29  1.65 2.88 
Other crimes against persons    0.03 0.00 0.00 0.08 0.10 0.06 0.00 0.00  0.00 0.00 
Stealing and fraud    0.57 1.04 1.57 2.59 2.33 2.24 1.22 0.77  1.13 1.51 
U.S. property offenses    0.34 0.52 0.60 0.65 0.87 1.74 0.89 0.93  1.29 1.56 
1–4 Crimes against persons    0.74 1.19 1.62 2.73 2.44 2.56 2.98 2.19  3.72 5.82 
5–6 Crimes against property    0.91 1.57 2.17 3.25 3.20 3.98 2.11 1.70  2.42 3.07 
1–6 Total serious crimes    1.65 2.75 3.79 5.98 5.64 6.54 5.09 3.89  6.14 8.89 
Other incidents    0.52 1.06 1.12 2.11 2.31 2.96 1.78 1.83  3.07 5.11 
1–7 All incidents    2.17 3.81 4.91 8.09 7.95 9.51 6.87 5.73  9.21 14.00 
 Smith report    2.96 3.80 5.05 8.74 8.39 9.35 7.07     
 
code charge at arrest may-45 jun-45 jul-45 aug-45 sep-45 oct-45 nov-45 dec-45 jan-46 feb-46 mar-46 apr-46 may-46 jun-46 
Homicide 0.25 0.22 0.17  0.08    0.52 0.67 0.31 0.52 0.55 0.71 
Rape 0.97 0.28 0.24  0.09    0.10 0.16 0.13  0.48 0.90 
Assault 0.84 0.45 0.61  0.47    0.98 1.13 0.46  0.97 1.94 
Other crimes against persons 0.04 0.09 0.09  0.00    0.04 0.00 0.00  0.00 0.00 
Stealing and fraud 1.07 0.89 0.46  0.55    0.99 0.40 0.41  0.68 0.90 
U.S. property offenses 0.38 0.36 0.48  0.32    1.09 0.91 0.57  0.61 0.57 
1–4 Crimes against persons 2.09 1.04 1.11  0.64    1.63 1.96 0.90  2.00 3.55 
5–6 Crimes against property 1.45 1.25 0.94  0.87    2.08 1.31 0.98  1.29 1.47 
1–6 Total serious charges 3.54 2.29 2.05  1.51    3.71 3.27 1.88  3.30 5.01 
Other charges     0.44    1.76 1.24 1.03  1.55 2.70 
1–7 All charges 3.54 2.29 2.05  1.95    5.47 4.51 2.91  4.85 7.71 
 Smith report       1.46 1.95 2.67 3.91 4.16 5.47 4.31  
codecrime reportedmay-45jun-45jul-45aug-45sep-45oct-45nov-45dec-45jan-46feb-46mar-46apr-46may-46jun-46
Homicide    0.12 0.22 0.31 0.46 0.46 0.86 0.62 0.44 0.60 1.07 1.28 
Rape    0.13 0.16 0.33 0.36 0.43 0.19 0.24 0.46  1.00 1.65 
Assault    0.47 0.81 0.97 1.83 1.45 1.45 2.11 1.29  1.65 2.88 
Other crimes against persons    0.03 0.00 0.00 0.08 0.10 0.06 0.00 0.00  0.00 0.00 
Stealing and fraud    0.57 1.04 1.57 2.59 2.33 2.24 1.22 0.77  1.13 1.51 
U.S. property offenses    0.34 0.52 0.60 0.65 0.87 1.74 0.89 0.93  1.29 1.56 
1–4 Crimes against persons    0.74 1.19 1.62 2.73 2.44 2.56 2.98 2.19  3.72 5.82 
5–6 Crimes against property    0.91 1.57 2.17 3.25 3.20 3.98 2.11 1.70  2.42 3.07 
1–6 Total serious crimes    1.65 2.75 3.79 5.98 5.64 6.54 5.09 3.89  6.14 8.89 
Other incidents    0.52 1.06 1.12 2.11 2.31 2.96 1.78 1.83  3.07 5.11 
1–7 All incidents    2.17 3.81 4.91 8.09 7.95 9.51 6.87 5.73  9.21 14.00 
 Smith report    2.96 3.80 5.05 8.74 8.39 9.35 7.07     
 
code charge at arrest may-45 jun-45 jul-45 aug-45 sep-45 oct-45 nov-45 dec-45 jan-46 feb-46 mar-46 apr-46 may-46 jun-46 
Homicide 0.25 0.22 0.17  0.08    0.52 0.67 0.31 0.52 0.55 0.71 
Rape 0.97 0.28 0.24  0.09    0.10 0.16 0.13  0.48 0.90 
Assault 0.84 0.45 0.61  0.47    0.98 1.13 0.46  0.97 1.94 
Other crimes against persons 0.04 0.09 0.09  0.00    0.04 0.00 0.00  0.00 0.00 
Stealing and fraud 1.07 0.89 0.46  0.55    0.99 0.40 0.41  0.68 0.90 
U.S. property offenses 0.38 0.36 0.48  0.32    1.09 0.91 0.57  0.61 0.57 
1–4 Crimes against persons 2.09 1.04 1.11  0.64    1.63 1.96 0.90  2.00 3.55 
5–6 Crimes against property 1.45 1.25 0.94  0.87    2.08 1.31 0.98  1.29 1.47 
1–6 Total serious charges 3.54 2.29 2.05  1.51    3.71 3.27 1.88  3.30 5.01 
Other charges     0.44    1.76 1.24 1.03  1.55 2.70 
1–7 All charges 3.54 2.29 2.05  1.95    5.47 4.51 2.91  4.85 7.71 
 Smith report       1.46 1.95 2.67 3.91 4.16 5.47 4.31  

The categories of “crimes against persons,” as opposed to “crimes against property,” appear in many of the summaries from subsequent periods. It is debatable whether robbery should be viewed as a crime against property or against persons, because it entails intimidation or assault as part of a theft. However, the statistical summaries that we located consistently categorized robbery as a crime against property; therefore, we applied this convention to our tabulations. With regard to the geographical location of the investigations and charges, the reports covered all U.S. forces in the eto, including the United Kingdom, France, the Benelux countries, and Germany, as well as other areas of the former Third Reich. An analysis of crimes committed by U.S. soldiers solely in the occupied areas of Germany was not possible.22

August through December 1945

The statistics for these months were compiled from three different sources: (1) Two statistical summaries under the heading “Office of the Theater Provost Marshall [sic], Criminal Investigation Division,” were entitled “Tabulation of Offenses Committed by U.S. Military Personnel during Period from August to December 1945”; one of them covered the “Occupational [sic] Zone,” which included the American-occupied portion of Germany, Austria, and possibly residual forces in Czechoslovakia and Italy, and the other the “Liberated Area,” the re-designated “Communication Zone.” They tabulated reports of incidents that were investigated but not charges. Table 1 shows the sum of these figures for the two areas. (2) A single page entitled “Tabulation of Offenses Investigated for the Month September 1945” showed separate counts for reports of investigations and arrests. (3) A single table entitled “Crimes Committed by U.S. Military Personnel” for August 1945–February 1946,” shows monthly totals for, respectively, reports of crimes and charges but nothing else. In Table 1, these totals, which are listed in rows labeled as “Smith Report,” are based on the report’s signature block identifying “Eugene Smith, Lt Col CMP, Chief CI Div” as the author. They are primarily useful for tracking charges in the months of August, October, November, and December 1945, for which no finer compilation has yet been found.

Added to the categories for serious crimes in the May through July summaries were figures for less serious offenses. Some of them were breaches of military regulations, most notably awol (absent without leave), “careless discharge of a firearm,” “desertion,” “drunk and disorderly” conduct, and “miscellaneous.” These figures are reported as totals in rows entitled “other incidents” and “other charges.”23

January through June 1946

The numbers for these months were compiled from a set of monthly summaries entitled “Tabulations of Offenses Investigated,” showing figures for both investigations and charges. Separate tabulations of offenses in the Occupied Zone and Liberated Area for January, February, and March 1946 confirmed the totals for the entire eto. The summary for April was incomplete; the second page was missing from the file, leaving only homicides.24

Troop Levels

The monthly numbers of military personnel in the eto, which are listed along the bottom row of Table 1, derived from three sources: First, the number of troops in May 1945 and June 1945 were found in a U.S. Army history of the eto. They were sourced from the Adjutant General’s Office (ago), Machine Records Branch. Second, the number of troops for August 1945 to March 1946 came from the table “Theater Troop Strength,” located in the same set of records as the statistical summaries. The table was attributed to the Machine Records Branch, otag (Office of the Adjutant General). Third, the number for June 1946 is reported in a brief history of the U.S. Army in Germany. Finally, the troop levels listed for July 1945, April 1946, and May 1946 are estimates computed from linear interpolations of the numbers listed for the adjacent months. These interpolations are italicized in Table 1.25

Analyzing the Provost Marshal’s Statistical Summaries

Examination of Table 1 shows the number of troops in the eto fell from 3,021,483 on May 1, 1945, to 290,000 on June 1, 1946, and to approximately 133,000 on July 1, 1946. Table 2 presents the number of incident reports and charges per 10,000 troops in each month to show accurate chart-crime trends and to account for this reduction. Given that troop numbers declined rapidly each month, the denominator for computing the monthly rate is an estimate of the mid-month number, based on averaging the reported troop levels at the start of one month and the start of the next.26

Frequency and Rate of Reported Offenses and Charges

Figure 1’s top row of panels depicts the absolute frequency of reported offenses and charges for the three major crime groupings—all crimes against persons (left-hand panel), crimes against property (center panel), and minor offenses plus other incidents (right-hand panel). To provide a more detailed picture of crimes against persons, the lower row of panels in Figure 1 displays the absolute number of reports and charges of homicide, rape, and assault. In a corresponding fashion, Figure 2 shows the rate of reported offenses and charges per 10,000 troops as shown in Table 2. Figure 2 is organized in exactly the same fashion as Figure 1. In both figures, each panel shows the month-by-month data points with trend lines fitted using linear regression. The legend lists the estimated slope and correlation coefficient (r) for reported offenses and charges, respectively. In addition to the linear trend lines, second-order polynomial lines are also plotted to test any apparent curvature.

Fig. 1

Absolute Frequency of Crime Reports and Charges

Fig. 1

Absolute Frequency of Crime Reports and Charges

Fig. 2

Rate (Number/10,000 Troops) of Crime Reports and Charges

Fig. 2

Rate (Number/10,000 Troops) of Crime Reports and Charges

Crimes against Persons

For crimes against persons, the absolute frequency of reported offenses and charges generally fell after May 1945. However, the decreases in reported offenses and charges did not always match the progressive decrease in troop levels. In fact, a substantial upward trend in the rate of reports (p .001) is in evidence between May 1, 1945, and July 1, 1946, which explains a large proportion of the variance, r2 = .78. By contrast, the upward trend for the rate of charges is weaker and does not attain statistical significance—r (7) = .51; p = .16—explaining only a modest proportion of the variance, r2 = .26. Although the estimated slope for charges is positive, its 95 percent confidence interval includes the value of 0.

Inspection of the lower panels in Figure 1 suggests that, consistent with the overall trend for crimes against persons, the absolute number of charges for homicide, but even more so for rape and assault, were at their highest in May 1945, after which they fell quickly. However, as Figure 2 shows, the relative rate of reports of homicide, rape, and assault shows a strong upward linear trends—smallest r (8) = .80; p .01—which explains a large portion of the variance, smallest r2 = .64. The specific trend for homicide charges shows an upward linear trend—r (8) = .81; p .01—which explains a substantial portion of the variance, r2 = .64. Any curvature in this trend is negligible; the polynomial fit explains only a small, additional proportion of the variance, added r2 = .04. By contrast, the rate of charges for rape shows substantial curvature. Specifically, the rate drops from .97 in May 1945 to .09 in September 1945, but rises from .10 in January 1946 to .90 in May 1946. The linear trend is virtually flat, while the polynomial fit explains most of the variance, r2 = .84. Finally, the rate of charges for assault shows an upward linear trend in combination with some early curvature.

Crimes against Property

The absolute frequency of reported offenses and charges for crimes against property generally declined (see Figure 1). However, relative to troop levels, the rate of crimes against property appeared constant throughout the months under examination (see Figure 2). For both variables, regression analyses yields low, nonsignificant correlation coefficients (ps .10), and the confidence intervals for the slopes of the regression lines include the value of 0.

Minor Offenses plus Other Incidents

The earliest statistics concerning minor offenses and other incidents were not available until August 1945. Thereafter, the absolute frequency of both reported offenses and charges for this category showed modest declines. However, when expressed relative to troop levels, reports and charges show a significant upward trend, which explains a large proportion of the variance, r2 = .71. The upward trend for the rate of charges does not quite reach the criterion of statistical significance—r (7) = .786; p = .063—but it explains an appreciable proportion of the variance, r2 = .61. The estimated slope for charges is positive, but its 95 percent confidence interval includes the value of 0.

Clearance Rate

The trends in Figure 2, significant or otherwise, suggest that the rate of reported offenses grew faster than the rate of charges. To test whether this divergence reflects a growing disparity between receipt of a report and the ability of the police to identify and charge a suspect, we computed the clearance rate, expressed as the ratio of charges to reports. To illustrate the results, Figure 3 shows the ratio of charges to reported offenses, expressed as a proportion, for each of the six crime categories plotted previously. The best-fitting straight line for the ratios over months is also shown.

Fig. 3

Clearance Rates for Six Listed Crime Categories

Fig. 3

Clearance Rates for Six Listed Crime Categories

Figure 3 reveals that the clearance could be calculated only for the months in which the number of reported offenses and the number of charges were both available. Unfortunately, no statistical summaries for reported offenses could be found for the period from May to July 1945, in which the charges for crimes against both persons and property were highest. For the subsequent months, the clearance rates appeared to be steady; any upward or downward trends in them were not statistically significant; all ps .10. For example, the ratio for homicides shows an upward trend of only .02 per month, and the ratio for rapes shows a downward trend of .01 per month. Even if these trends were statistically significant, they would have explained only 7 percent and 9 percent of the variance, respectively.

In addition to being relatively stable across months, the clearance rate appears consistent across crime categories. The mean clearance rates for crimes against persons, crimes against property, and minor offenses plus other incidents were, respectively, .57 (sd = .09), .55 (sd = .05), and .55 (sd = .09). Among crimes against persons, the mean ratios for homicide, rape, and assaults were, respectively, 0.64 (sd = .24), 0.51 (sd = .13), and 0.57 (sd = .11). The largest apparent difference is that for homicide (.64) versus rape (.51). This difference, however, is not statistically reliable—F (1, 5) = 1.50, p .20—and the 95 percent confidence interval for this difference includes the value of 0.

The “Dark Number” Problem

The statistical summaries in this article almost certainly suffer from the problem of the “dark number,” which plagues virtually every criminological study. Even in the best of circumstances, victims frequently fail to report crimes; even when victims do report them, the police do not always record and investigate them, at least to the satisfaction of the victims. Throughout the eto but especially in Germany, the U.S. military did not always investigate U.S.-perpetrated crimes. German authorities and civilians frequently complained of U.S. inaction in response to their reports. For instance, when an ex-German soldier with surname Brater returned home in August 1945 to discover that his wife Hedwig had reported being terrorized and assaulted (although not raped) by American soldiers two months earlier, he complained to local mg authorities. According to a detailed report made to local German police, Brater “begged” authorities to help him to gain restitution and compensation for her injuries. Similarly, from August onward, Darmstadt’s and Bremen’s German chiefs of police collected lists of unaddressed complaints against American soldiers.27

Even when reports of crimes were investigated, the identification of perpetrators was frequently impossible, mainly due, at least in part, to the transient nature of deployment. Griffith, an mgo in Bavaria during the early occupation, later wrote that one of his primary duties was “parading weeping rape victims” past their potential American attackers, whom the women could never identify.28

A variety of methods for estimating the dark number have been attempted, including anonymous surveys of a population for victims and their rates of reporting to police, all of which have the attendant problems of self-reporting. Nonetheless, their widespread use in recent decades has confirmed that victims consistently under-report crimes to police. A study of surveys from thirty-one nations reveals the percentage of crimes reported to the West German police in 1989 to have been 52 percent for property offenses, 21 percent for assault and threatening behavior, and 11 percent for all forms of sexual assault against women, including rape. Corresponding surveys of nations in what was once the “liberated zone” of the eto yielded similar results. Specifically, for 1989, the percentages of sexual assault reported in Belgium, France, and the Netherlands were 15 percent, 15 percent, and 13 percent, respectively.29

No population survey has yet been found for American-perpetrated crime in Germany or elsewhere in the eto from 1945 to 1946, but a range of estimates can be obtained by working backward from the number of detected crimes and suspected reporting rates. As previously described, Lilly and Gebhardt resorted to this strategy, basing their estimates of rape on Radzinowicz’s hypothesis of a 5 percent reporting rate for sexual assault. Although the 5 percent rate is controversial, the 1989 figures described above suggest that it may be reasonably accurate in the postwar context. Social conditions in developing countries provide a modern facsimile of the strained conditions in postwar Germany. Palermo, Black, and Peterman’s recent study of rape reports to police in twenty-four developing countries in South American, Africa, and Central and Southeast Asia showed significant variation: The percentage of women making formal reports ranged from 0.1 percent to 17 percent. Yet, in the majority of countries examined, the rate of reporting was less than 4 percent. The more recent findings from Germany, other European nations, and developing countries appear to serves as ample justification for assuming similar reporting rates under postwar conditions and thus for treating Radzinowicz’s 5 percent as a viable median for estimations.30

Cognisant of the risks of extrapolation, we attempted to estimate the magnitude of the actual number of offenses in each category for the 1945 to 1946 period according to the following framework:

  • (1) 

    We used the results of the 1989 victim surveys to set the reporting percentages at 52 percent for property offenses, 22.5 percent for assaults/threats, and 10 percent for sexual offenses. To accommodate the distinct possibility that reporting percentages in the postwar period might be far lower, we set a “mid- range” by halving the foregoing rates to 26 percent, 11.25 percent, and 5.0 percent, and we set a “low range” by halving them again to 13 percent, 5.12 percent, and 2.5 percent. For the mid-range figures, the 5 percent point for sexual offenses matches Radzinowicz’s hypothesis. (2) The range of reporting rates specified above is also intended to cover reports of crimes that further investigation determined to have been false alarms or unverifiable.31

  • (3) 

    We assumed the number of recorded incidents to have been a relatively constant proportion of the total number of crimes during the period under study. This assumption may underestimate the proportion of crimes to reports in the early stages of the occupation. There is evidence that mps and mgos paid more and more attention to U.S. crime as time passed, particularly in late 1945, when the cid and the Provost Marshal proposed even tighter restrictions on U.S. soldiers given the persistence of crime. By June 1946, the Provost Marshal was performing daily checks of soldiers’ financial accounts, confining soldiers to bases, and engaging Germans instead of Americans to guard materiel vital to the occupation—food, coal, and other items of value on the black market—that Americans were liable to steal.32 More proactive policing may have been responsible, at least partly, for the increased rate of reported crime from August 1945 onward. However, quantifying the results of increased policing would at this point be entirely speculative. The simplifying assumption of constant policing and therefore consistent proportions of total crimes to reports is safer for the present approximations.

  • (4) 

    As shown in Table 2, the monthly number of recorded incidents was not available until August 1945. Since the clearance rate from August 1945 to June 1946 was relatively constant, around 55 percent for all crime categories, the number of reported incidents outnumbered charges by 182 percent (1/.55). The frequency of charges was multiplied by 1.82 in order to obtain an estimate of recorded incidents for May to July 1945 (no figures were available for April 1946).

  • (5) 

    The categorization of crimes by the American authorities from 1945 to 1946 roughly corresponded to that in the more recent German surveys of crime rates and reporting addressed previously. Clearly, modern crimes against property do not include the theft and misappropriation of U.S. government property by American personnel. Moreover, the number of lootings, minor thefts, minor assaults, and loutish conduct may not have been considered serious enough to count even when they did come to the attention of the mps. But for purposes of estimating the magnitude of American crime, this rough equation of crime categories seems as safe as attempting a more refined definition, which would require additional guesses and unresolved questions.

Based on these suppositions, we calculated a range of low, middle, and high estimates for actual incidents. The results of these calculations are plotted in Figure 4. The estimated frequencies are plotted on log10 scales to focus on their magnitudes as they changed over time. In general, estimated property crime and assault showed a similar pattern. Starting in May 1945, their estimated frequencies numbered in thousands, and, as time progressed, they dropped into the hundreds. In the case of rape, the May estimates ranged from 5,000 to 20,000. Thereafter, the estimates of rape declined into the hundreds with the high estimate hovering around 1,000 per month.

Fig. 4

The Dark Number

Fig. 4

The Dark Number

The monthly rape estimates, especially from May to July, suggest that Lilly’s total of 9,680 for the entire period from January to September is an underestimate. This discrepancy in estimates is consistent with the possibility that the jag court records that Lilly used represent only cases that came to trial and thus did not count all criminal incidents reported to the police. Moreover, as Henke suggests, many American soldiers ostensibly arrested for rape were ultimately charged with a different offense, further distorting the data. These arguments provide grounds for regarding Gebhardt’s 1 percent figure as reasonable. For caution’s sake, the present computations of the dark number for rape assume a low-range reporting percentage of 2.5 percent, which produced an estimate of 1,400 rapes for June 1946. Had the number of rapes stabilized later in 1946, the present estimates and Gebhardt’s estimates would appear to converge somewhere between a 1 percent and 2.5 percent rate of detection. No matter where the actual figure falls on the scale of estimates, many rapes appear to have gone undetected by authorities.33

Crime in the Eto: Perception and Reality

These analyses of the Provost Marshal’s summary statistics yield three major trends. First, the absolute number of charges for crimes against persons, especially rape and other forms of assault, was high in May 1945 before progressively declining in subsequent months. Second, the absolute number of charges for crimes against property was also highest in May 1945, and it too began to decrease during the following months, though less dramatically than crimes against persons. Third, although the absolute number of reports and charges for all crimes fell as troop levels dropped, the rate of both reported offenses and subsequent charges of all types steadily increased. Hence, there appear to have been two waves of American-perpetrated crime. The first one, which Henke and Lilly detected in their study of the jag data, was large, as indicated in the absolute number of crimes against persons and property. Its decline preceded the second wave, marked by a rise in the rate of crime among American soldiers still present in the eto from August 1945 to July 1946.34

These statistical findings help to explain the emergence of a pervasive and widespread negative view of U.S. soldiers among U.S. mgos and Germans during the first year of the occupation. Apart from any misgivings about the precision of the foregoing estimates, the pattern of detected crimes certainly illuminates how the perception of American criminality evolved. The early spike in all varieties of crime included a sufficient number of direct experiences and observations to facilitate an unflattering prototypical representation of American soldiers. The classic “availability heuristic” explains this process: People remember and generalize bad experiences far more readily than they do good ones, and once they become established, negative prototypes tend to endure. Consequently, even though the absolute frequency of crimes declined with the reduction in troop numbers, the resurgent rate of detected crimes by the remaining American soldiers reinforced the negative reputation of U.S. troops in the minds of the American authorities and the general population.35

The availability heuristic is a well-established psychological phenomenon that finds confirmation in the virtually universal fear of crime. According to O’Connell and Whelan’s study, people are predisposed to believe that crime is worse than it is. Even moderate levels of crime can excite exaggerated reactions. O’Connell and Whelan conducted their research in Dublin during the 1990s when crime rates were low. Because the crimes committed by American soldiers in Europe were relatively frequent, negative perceptions had an extraordinary opportunity to take hold within the population. Nor would direct experience with crime have been required. News, especially bad news, can spread quickly. Modern research reveals not only that a population can turn a few scattered, grisly events into a terrible “trend” in no time but also that rumors do not need the aid of modern technology to move rapidly. In postwar Europe, the U.S. military and mgos were keenly aware of the dangers that rumors posed, especially in Germany’s U.S. Zone. The Military Government Legal Code that the United States used to govern occupied territories criminalized the “dissemination of any rumour calculated to alarm or excite the people.” Civilians feared gangs of displaced persons (dps), Soviet incursion, and American exploitation. Canoy and Seipp, who each analyzed the rumors circulating within Germany at the time, concluded that fear was almost certainly a strong contributor to them. But the new data about American-perpetrated crime examined in this article suggest a greater rate of violent crime than previously thought and thus a more substantial basis for rumors about criminal activity on the part of American soldiers.36

Broader Implications

The fact that the rate of American criminal violence in the eto was higher than previously estimated has important implications for our understanding of Western Europe’s social and cultural transition from war to peace. Studies of postwar violence and trauma have already eroded the idea that the end of the war marked a “zero hour” (Stunde Null)—a caesura between war and peace in Europe, as well as between the Third Reich and German postwar society. Bessel maintains that parts of Eastern Germany, particularly Saxony, were still feeling the effects of a substantial civilian crime wave well after the end of the war. Resistance against the Soviets continued in parts of the Ukraine and the Baltic states until the early 1950s. The Soviet administration of its German sector was disorderly and disorganized. According to Slaveski, for nearly two years after the war, Red Army soldiers and Soviet officials from Moscow engaged in a violent struggle for control of the occupation; German civilians were caught in the middle.37

The transition from war to peace was not cut and dry in the West either. In the western zones of Germany, the necessary rebuilding and a massive refugee crisis combined with the high levels of American crime to create long-lasting psychological trauma. This scenario casts doubt on the divide between a disorderly East and a comparatively peaceful West—the American “heroes” versus the villainous Soviets—as endorsed by MacDonogh, among others. A climate of war continued well after May 1945 in Western Europe, exemplified by the relations between American soldiers and civilians wherever the U.S. military went. The negative experiences of occupation/liberation by American forces were felt not just in U.S.-controlled Germany but throughout Western Europe.38

The supposedly better behavior of American soldiers vis à vis their Soviet counterparts is further negated by other American occupations. For instance, Kovner wrote about how prostitution in Japan, which had always been regulated, was transformed by U.S. occupation. Prior to the arrival of the victorious American soldiers, internal propaganda had instilled a widespread fear of rape among the Japanese population. Many families hoped to protect the nation by throwing women at the soldiers as sexual sacrifices. Japan remained in this “moral panic”—a phenomenon typical in societies under severe strain—during the occupation. Women from all classes were thrust into sex work. Yet, according to Kovner, “although some servicemen did force themselves on Japanese women, sexual violence never approached the level predicted before the occupation began.” The bulwark of unregulated sex workers and brothels around American installations may have forestalled a more extensive rate of rape. Sexual exploitation therefore acquired a veneer of legality, while black marketeering, theft, and other crimes associated with the new sexual economy grew to support it. Although violent criminality, particularly sex crimes, may not have manifested in Japan exactly as it did in the eto and Germany, the criminal and otherwise unsavory exploitation of the local population by occupation soldiers certainly occurred in parallel fashion.39

Studies of British and French occupation in Germany, such as they are, suggest that the behavior of soldiers in occupied territory was endemic. In the case of the Allied occupation in Germany, however, a quantitative accounting of British and French troop behavior is necessary to test this hypothesis. Unfortunately, none is available for these two zones. Gebhardt addresses rape in all four Allied Zones of Germany, but, as previously noted, scholars have expressed significant reservations about her methods. Her estimated figures for British and French rapes—50,000 and 45,000, respectively—stand in need of better grounding, and her figure of 860,000 for the totality of Allied rapes falls short of nearly all prior estimates of Soviet-perpetrated rapes alone. Better calculations of British and French troop crimes of all types would be worthwhile.40

In contrast to the limited exploration of British, French, and even American troop crime, scholars have devoted considerable energy to quantifying Soviet crimes in Eastern Europe, particularly rape. Naimark, Jacobs, and Johr, among others, have attempted to estimate the number of rapes by Soviet soldiers in Eastern Europe and Germany. Naimark noted that although we will never know how many rapes were committed, the number most likely lies between 1 and 2 million. Johr’s number of 1.9 million achieves a greater consensus than the lower estimates; in Grossmann’s words, “this may be a horrifically accurate estimate.” As Grossmann explains, the Red Army’s mass rape of German women now routinely figures in many narratives as revenge for Nazi Germany’s vicious war of extermination and conquest launched against the Slavic East. The Soviets fed this narrative of savage retribution via the bodies of German women by arguing that they had “more reason for revenge” than did the other Allies. Yet, Naimark’s analysis of the causes of mass rape suggests more subtle and complicated interactions between soldiers and civilians than simplistic appeals to group hatred, or even patriarchal aggression, can explain. Naimark also implicated the Russian drinking culture as well as Russian ideas about honor and masculinity. Moreover, Soviet soldiers also raped Polish and other Slavic women. Even if they did not do so as often or as ferociously as they pursued Germans, their actions point to a deeper logic governing the violent criminality of occupation soldiers.41

The Soviet and American contexts were different. The American soldiers fighting in Europe were not retaliating against an invasion, and they were not fighting against an enemy seeking to eliminate them on racial principles. In France and the Benelux countries, they were liberators, not conquerors. Yet, they still committed violent crimes at consistently high levels throughout the eto. Moreover, as shown previously, the U.S. military grew rapidly concerned about American soldiers abusing their authority. According to Ziemke, American soldiers believed that “they were entitled to civilian property.” They may also have believed they were entitled to civilian bodies. Notwithstanding the obvious contextual circumstances that lead to crime during occupation, our findings suggest that further investigation will uncover other factors entailed by military occupation that contribute to violent criminality.42

American soldiers in post-World War II Europe were all too often predators who attacked, raped, and robbed at will. Liberators they were, but also men willing to punish and exploit hapless civilians in their charge. The U.S. military statistics examined in this article force a radical re-appraisal of the benign portrait of American soldiers as liberators and enlightened occupiers in post–World War II Europe. In spite of numerous reports, the severity of soldiers’ violent crimes remained hidden for a long time. A less than flattering portrait of the U.S. soldier emerged in the 1970s, but only recently have scholars attempted to add more substance to it. Unfortunately, the lack of quantitative data has hindered their efforts. The data in this article, however, reveal that American soldiers were far more violent toward the liberated and occupied populations than previously supposed. Prior studies based on jag prosecutions correctly identified their criminal tendencies, but they vastly underestimated the rate of their crimes. Notwithstanding the problems associated with the dark number, it now appears safe to conclude that U.S. soldiers raped and assaulted civilians with frightening abandon. Future studies will have to adopt the standpoint that the violence of soldiers during occupation was closer to normal than to exceptional. The quantitative data presented in this article, and the implications thereof, lay the foundations for the important research to come.

Notes

1 

For an example of euphemisms in the U.S. military reports, see Headquarters Third Infantry Division, “Report of Military Police Operations in Nurnberg, Germany, During Period 19–22 April 1945,” April 23, 1945, Entry 427, Box 5402, National Archives at College Park, Maryland (hereinafter nacp); “Functional History of Military Government in the Bremen Enclave,” n.d. (presumed late 1945), RG 260, Entry A1 140, Box 575, 53–54, nacp. Oliver J. Frederiksen, The American Military Occupation of Germany: 1945–1953 (Frankfurt, 1953), 115–118; Jan Foitzik, Sowjetische Militäradministration in Deutschland (SMAD) 1945–1949 (Berlin, 1999), 52–73; Norman M. Naimark, The Russians in Germany: A History of the Soviet Zone of Occupation (Cambridge, Mass., 1995), 69–140; Harold Zink, The United States in Germany (Westport, 1957), 239–240. For the military’s use of euphemisms, see Klaus-Dietmar Henke, Die amerikanische Besetzung Deutschlands (Munich, 1995), 201.

2 

Closer examination of violent American criminality began with Susan Brownmiller, Against Our Will: Men, Women and Rape (London, 1975), 64. For an example of prior, careful comparison between the Soviet occupation and the U.S. Zone, see Naimark, Russians in Germany, 106, 501 (n. 172), who noted numerous U.S. military reports about American-perpetrated rape, suggesting that the problem of criminality by occupation soldiers was not confined to the East. For continuing violence after the war, see Richard Bessel, Germany 1945: From War to Peace (New York, 2009); Mark Mazower, Dark Continent: Europe’s Twentieth Century (New York, 1999). Lee B. Kennett, GI: The American Soldier in World War II (New York, 1987), 217. The figures for rape come from J. Robert Lilly, Taken by Force: Rape and American GIs in Europe During World War II (New York, 2007), 12, who provides numbers for rape trials in Germany and France, the highest of which are 11,040 in Germany and 3,620 in France.

3 

Miriam Gebhardt, Als die Soldaten kamen: Die Vergewaltigung deutscher Frauen am Ende des Zweiten Weltkriegs (Munich, 2015); Atina Grossmann, “A Question of Silence: The Rape of German Women by Occupation Soldiers,” October, LXXII (1995), 46; Giles MacDonogh, After the Reich: The Brutal History of the Allied Occupation (New York, 2009), 240; Naimark, Russians in Germany, 133; Guy Walters, “Did Allied Troops Rape 285,000 German Women? That’s the Shocking Claim in a New Book. But is the German Feminist behind It Exposing a War Crime—or Slandering Heroes?” Daily Mail, 26 March 2015, available at http://www.dailymail.co.uk/news/article-3011930/Did-Allied-troops-rape-285-000-German-women-s-shocking-claim-new-book-German-feminist-exposing-war-crime-slandering-heroes.html (accessed September 9, 2015).

4 

These statistics are compiled from a variety of Provost Marshal reports—“Provost Marshal Reports,” RG 498, UD 747, Container 4255, nacp. Across disciplines, a variety of terms are used to the label the ratio of the number of events relative to a population number for a defined interval—“incidence,” “per capita ratio,” and “rate per n thousands.” For present purposes, we adopt a criminological convention of “rate,” defined herein as the number of crimes per 10,000 troops per month.

5 

Lina Hachenburger, “Violation of Mrs. Lina Hachenburger, Born June 30th 1913 at Griedel,” April 16, 1945, RG260, Entry A1 1424, Box 480, nacp; Ilse Reuhl, “Rape of Ilse Reuhl, Born April 15th, 1922, and of her sister,” 18 April 1945, RG260, Entry A1 1424, Box 480 nacp.

6 

Henke, Die amerikanische Besetzung Deutschlands, 199–202; Lilly, Taken by Force; Mary Louise Roberts, What Soldiers Do: Sex and the American GI in World War II France (Chicago, 2013), 195–254.

7 

Headquarters Third Infantry Division, “Report of Military Police Operations in Nurnberg, Germany, During Period 19–22 April 1945,” 53–54.

8 

Biedenkapp, “Report,” July 2, 1945, RG 260, Entry A1 1424, Box 480, part 1, nacp.

9 

Charles H. Andrews, “Daily Report. From 0001 14 July 1945 to 2400 15 July 1945,” July 16, 1945, OMGUS Reports No. 88, Bavarian State Archives (hereinafter BayHStA); “MG Detachment H4 B3, 3rd ECA Regiment, Kulmbach Daily Reports,” July 15, 1945, OMGUS Reports No. 79, BayHStA.

10 

Military Government Detachment E2C2 Bremen, “Weekly Military Government Reports,” November 9, 1945, RG 260, Entry A1 140, Box 582, 5–6, nacp; Melvin Mawrence, “Initial Military Government Historical Report for the Office of Military Government for Landkreis München,” November 3, 1945, OMGUS Reports No. 182, BayHStA; John E. Westburg, “Criminal Offenses by US Troops,” October 25, 1945, RG 260, Entry A1 1424, Box 480, nacp.

11 

Darmstadt Police President, “Crimes, Grievances and Troubles,” August 27, 1945, RG 260, A1 1424, Box 495, nacp; “Functional History of Military Government in the Bremen Enclave, October–November 1945,” n.d. (presumed late 1945), RG 260, Entry A1 140, Box 575, 1–3, nacp; Claude F. Baker, “War Diary Det. GI G3,” July 31, 1945, OMGUS Reports No. 195, BayHStA.

12 

Earl F. Ziemke, The U.S. Army in the Occupation of Germany, 1944–1946 (Washington, D.C., 1975), 336. Office of Theater Provost Marshal, “Report of Operations. 1 January–31 March 1946,” n.d. (presumed April 1946), RG 498, Entry UD 747, Container 4255, 21, nacp.

13 

Michael O’Connell and Anthony Whelan, “The Public Perception of Crime Prevalence, Newspaper Readership and ‘Mean World’ Attitudes,” Legal and Criminological Psychology, I (1996), 179–195. For other studies in this vein, see Carrie B. Fried and Amiee Maxwell, “Rape Rumors: The Effects of Reporting or Denying Rumors of Sexual Assaults on Campus,” Journal of Applied Social Psychology, XXXVI (2006), 2766–2784; Susan J. Smith, “Crime in the News,” British Journal of Criminology, XXIV (1984), 289–295.

14 

For the present purposes of estimating the number of crimes, we ignore whether a prosecuted suspect was found guilty or innocent. There was usually little doubt that a crime occurred; the verdict concerns only whether a particular individual was held responsible. For the “dark number,” see D. Kim Rossmo and Rick Routledge, “Estimating the Size of Criminal Populations,” Journal of Quantitative Criminology, VI (1990), 293–314; Wesley G. Skogan, “Dimensions of the Dark Figure of Unreported Crime,” Crime Delinquency, XXIII (1977), 41–44.

15 

Henke, Die amerikanische Besetzung Deutschlands, 199–202; Lilly, Taken by Force, 12.

16 

Lilly, Taken by Force, 12; Leon Radzinowicz, Sexual Offenses: A Report of the Cambridge Department of Criminal Science (London, 1957), xv.

17 

Gebhardt, Als die Soldaten kamen, 31.

18 

Gebhardt, ibid., 31; Grossmann, “Question of Silence,” 61; Klaus Wiegrefe, “Postwar Rape: Were Americans as Bad as the Soviets?” Der Spiegel, 2 March 2015, available at http://www.spiegel.de/international/germany/book-claims-us-soldiers-raped-190-000-german-women-post-wwii-a-1021298.html (accessed September 9, 2015). For the frequency of fertilization, see Allen J. Wilcox et al., “Likelihood of Conception with a Single Act of Intercourse: Providing Benchmark Rates for Assessment of Post-coital Contraceptives,” Contraception, LXIII (2001), 211–215.

19 

“Provost Marshal Reports.”

20 

For a brief description of the European Theater of Operations (eto), see United States Army, United States Army in World War II, CMH Publication 11-9 (Washington, D.C., 1992), 70–71, available at http://www.history.army.mil/books/wwii/11-9/11-9c.htm (accessed February 4, 2015).

21 

“Crimes Investigated by MP CI Personnel,” nacp, RG 498, UD 747, Container 4255.

22 

“Provost Marshal Reports,” RG 498, UD 747, Container 4255, nacp.

23 

One of the Provost Marshal’s statistical tabulations (“Summary of Serious Incidents Received 16 April–30 June 1946,” RG 747, Container 4255, nacp) has a note about “other incidents”—“predominantly vehicle accidents, [plus] escapes, accidental drowning, train and plane accidents, illegal border crossing, explosions, discovery of bodies, fires by spontaneous combustion, and incidents not classified as crimes or disorders.”

24 

“Tabulation of Offenses Investigated.”

25 

“Theater Troop Strength,” RG 498 UD 747 Container 4255, nacp; Forrest C. Pogue, United States Army in World War II: European Theater of Operations, the Supreme Command (CMH Publication 7-1)(Washington, D.C., 1954), Appendix E, 542; U. S. Army Europe, Major Events in U.S.AREUR and European History, 2015, available at http://www.eur.army.mil/organization/timeline.htm (accessed June 1, 2015).

26 

Benjamin James Harris, “United States Zone Constabulary: An Analysis of Manning Issues and Their Impact on Operations,” unpub. M.A. thesis (Kansas State University, 2006), 1.

27 

“Hedwig Brater, Okarben, Grossgasse 11,” August 13, 1945, RG260, Entry A1 1424, Box 480, nacp. For examples, see Military Government Detachment E2C2 Bremen, “Weekly Military Government Reports,” August 1945 to May 1946, RG260, Entry A1 140, Box 582, nacp; Darmstadt Police President, “Weekly Reports for August to December 1945,” RG 260, A1 1424, Box 495, nacp; for the “dark number,” Matthieu de Decastelbajac, “Brooding Over the Dark Figure of Crime: the Home Office and the Cambridge Institute of Criminology in the Run-up to the British Crime Survey,” British Journal of Criminology, LIV (2014): 928–945; Rossmo and Routledge, “Estimating the Size of Criminal Populations,” 293–314; Skogan, “Dimensions of the Dark Figure of Unreported Crime,” 41–44.

28 

William E. Griffith, “Denazification Revisited,” in Michael Ermarth (ed.), America and the Shaping of Germany Society, 1945–1955 (Providence, 1993), 155.

29 

For property crimes in Belgium, France, and the Netherlands, the reporting rates were 54%, 65%, and 58%, respectively. For assaults and threats, the corresponding figures were 30%, 37%, and 39%. See Jan van Dijk, John van Kesteren, and Paul Smit, Criminal Victimisation in International Perspective, Key findings from the 2004–2005 ICVS and EU ICS (The Hague, 2008), 263–266.

30 

For debates about the 5% reporting figure, see Bonnie S. Fisher et al., “Reporting Sexual Victimization to the Police and Others: Results from a National-Level Study of College Women,” Criminal Justice and Behavior, XXX (2003), 6–38; Lilly, Taken by Force, 117–118; Michelle A. Mengeling et al., “Reporting Sexual Assault in the Military: Who Reports and Why Most Servicewomen Don’t,” American Journal of Preventive Medicine, XLVII (2003), 17–25; Tia Palermo, Jennifer Bleck, and Amber Peterman, “Tip of the Iceberg: Reporting and Gender-Based Violence in Developing Countries,” American Journal of Epidemiology, CLXXIX (2013), 602–612; Radzinowicz, Sexual Offenses, xv; Kate B. Wolitzky-Taylor et al., “Reporting Rape in a National Sample of College Women,” Journal of American College Health, LIX (2011), 582–587 (esp. 583).

31 

Van Dijk, van Kesteren, and Smit, Criminal Victimisation in International Perspective, 264–266; Radzinowicz, Sexual Offenses, xv.

32 

Office of Theater Provost Marshal, “Report of Operations. 1 April–30 June 1946,” n.d. (presumed July 1946), RG 498, Entry UD 747, Container 4255, 23, 39, nacp.

33 

Henke, Die amerikanische Besetzung Deutschlands, 199–202; Gebhardt, Als die Soldaten kamen, 31.

34 

Henke, Die amerikanische Besetzung Deutschlands, 199–202; Lilly, Taken by Force, 118.

35 

Amos Tversky and Daniel Kahneman, “Availability: A Heuristic for Judging Frequency and Probability,” Cognitive Psychology, V (1973), 207–233.

36 

O’Connell and Whelan, “Public Perception of Crime Prevalence,” 179–195. Military Government Detachment E2C2 Bremen, “Weekly Military Government Reports: August 1945 to May 1946,” RG260, Entry A1 140, Box 582, nacp; United States, Armed Service Forces, Military Government Handbook Germany, Section 2M: Proclamations, Ordinances, and Laws Issued by Allied Military Government in Germany, Ordinance I, Article II, Section 25, 7; “Refugee Riot Laid to a False Rumor,” New York Times, 29 April 1946. For rumors, see Jose Raymund Canoy, The Discreet Charm of the Police State: The Landpolizei and the Transformation of Bavaria, 1945–1965 (Boston, 2007), 113–116; Chip Heath, “Do People Prefer to Pass along Good or Bad News? Valence and Relevance of News as Predictors of Transmission Propensity,” Organizational Behavior and Human Decision Processes, LXVIII (1996), 79–94; Daniel Kahneman, “A Perspective on Judgment and Choice: Mapping Bounded Rationality,” American Psychologist, LIII (2003), 697–720; Michael A. Kamins, Valerie S. Folkes, and Lars Perner, “Consumer Responses to Rumors: Good News, Bad News,” Journal of Consumer Psychology, XI (1997), 165–187; Adam Seipp, “Refugee Town: Germans, Americans, and the Uprooted in Rural West Germany,” Journal of Contemporary History, XLIV (2009), 685.

37 

Bessel, “Establishing Order in Post-war Eastern Germany,” Past Present, 210 (2011), 146, 150–152. Filip Slaveski, The Soviet Occupation of Germany: Hunger, Mass Violence, and the Struggle for Peace, 1945–1947 (New York, 2013), 28–60, discusses the conflict between the Red Army and Komendants of the Soviet Military Administration in Germany, specifically soldiers resisting efforts to restrain their criminal activity. The Stunde Null argument is still endorsed by, among others, Curtis F. Morgan, Jr., James F. Byrnes, Lucius Clay, and American Policy in Germany, 1945–1947 (New York, 2002), 63. For a simple breakdown of the Stunde Null argument, see Patrick Thaddeus Jackson, Civilizing the Enemy: German Reconstruction and the Invention of the West (Ann Arbor, 2006), 115. See Mazower’s rejection of the argument in Dark Continent, 213.

38 

Lilly, Taken by Force; MacDonogh, After the Reich, 240; Roberts, What Soldiers Do.

39 

Sarah Kovner, Occupying Power: Sex Workers and Servicemen in Postwar Japan (Stanford, 2012), 52–54, 152–154.

40 

Gebhardt, Als die Soldaten kamen, 31.

41 

Naimark, Russians in Germany, 107–115, 133; Ingeborg Jacobs, Freiwild: Das Schicksal deutscher Frauen 1945 (Berlin, 2008), 5; Barbara Johr, “Die Ereignisse in Zahlen,” in Helke Sander and idem (eds.), Befreier und Befreite: Krieg, Vergewaltigung, Kinder (Munich, 1992), 54–59; Brownmiller, Against Our Will, 64.

42 

Ziemke, U.S. Army in the Occupation of Germany, 147.

Author notes

Thomas J. Kehoe is Learning Designer (Faculty of Health, Arts, and Design), Office of the Pro-Vice Chancellor (Learning Innovation), Swinburne University of Technology. He is the author of “Fighting for Our Mutual Benefit: Understanding and Contextualising the Intentions behind Nazi Propaganda for the Arabs during World War Two,” Journal of Genocide Research, XIV (2012), 137–157; with Frederik Juliaan Vervaet, “Honor and Humiliation in Apuleius’ Apologia,” Mnemosyne: A Journal of Classical Studies, LXVIII (2015), 605–640.

E. James Kehoe is Professor of Psychology, University of New South Wales. He is the author of, with Scott D. Gayton, “A Prospective Study of Character Strengths as Predictors of Selection into the Australian Army Special Forces, Military Medicine, CLXXX (2015), 151–157; with Gayton, “Character Strengths and Hardiness of Australian Army Special Forces Applicants, Military Medicine, CLXXX (2015), 857–862.