Fall From Grace: Arming America and the Bellesiles Scandal (Part 3)
C. Was Homicide Rare?
Bellesiles claims that, in step with low gun rates, homicide rates were
low until the Civil War. Bellesiles claims that “[w]hites rarely assaulted
other whites in the colonies and almost never killed one another.”118 These
claims are not only unsupported by the evidence he offers, but also false.
Randolph Roth, who has studied homicide rates throughout early America,
exposes this error in his review in the William and Mary Quarterly.119 Roth
points out that homicide rates during much of the seventeenth century were
actually higher than they are today. In other places and times in early
America, rates were similar to those today:
The homicide rate for adult European colonists in New England
before King Philip’s War was as high as the rate in the United
States today, 7-9 per 100,000 adults per year. Before the Pequot
War, the rate was higher still: roughly 110 per 100,000 adults per
year, or 11 to 14 times the rate today. A number of those colonists
were murdered by Native Americans, but the homicide rate was
still very high if one discounts those murders, as Bellesiles does.120
How does Bellesiles make such a basic error? In part, he just presents
false counts in the records he cites or makes claims that could not possibly
be supported by the evidence on which he relies. For example, Bellesiles
claims that “in forty-six years Plymouth Colony’s courts heard five cases of
assault, and not a single homicide,” 121 citing the standard published version
of seventeenth-century records of Plymouth Colony courts.122
There are many homicide cases heard in Shurtleff’s Records of the
Colony of New Plymouth Colony in New England, and they are relatively
easy to find. One need only look in the indices to find the murder and
manslaughter prosecutions. As Randolph Roth writes:
The records cover 1633-1691, with some gaps. Bellesiles does not
state which 46 years he studied, but every contiguous period of 46
years contains homicides. The 11 homicides are in 1:96-97; 2:132-
34; 3:70-72, 73, 82, 143, 205, 5:159, 167-68, 264-65, 6:82, 113,
141-42, 153-54; 7:305-07. A probable homicide appears in 2:170-
71, and 3 suspicious deaths that may have been homicides in 3:202-
03, 217-18, 4:32-33, 5:141. The 3 multiple murders during King
Philip’s War are in 5:204-06, 209, 224. Three additional murders in
Plymouth Colony appear in William Bradford, Of Plymouth
Relative to other crimes, homicide prosecutions appear to be common.
Bellesiles misses every homicide prosecution in these records.
Nearly as stunning is Bellesiles’s claim: “[D]uring Vermont’s frontier
period, from 1760 to 1790, there were five reported murders (excluding
those deaths in the American Revolution), and three of those were
politically motivated.” 124 The source he cites for this count is the Vermont
Superior Court records. He presumably meant the Vermont Supreme Court,
since Vermont had no Superior Court in that period. But he could not
possibly have used these Supreme Court records to count murders for
thirty-one years in Vermont, from 1760 to 1790. As Roth explains about the
Vermont Supreme Court:
[T]hat court did not open until December 1778, and its minutes
from September 1782 to August 1791 have been missing since the
early twentieth century. In fact, Vermont, together with the rest of
New England, had an elevated homicide rate during the American
Revolution, and 70 percent of known adult homicides and probable
homicides in Vermont, 1760-1790, were committed with guns.125
Thus, Bellesiles could not have counted Vermont murders during 1760-
1790 in the source he cites because that source did not exist for more than
half of the period and is lost for most of the rest of the period. Where did
Bellesiles come up with his numbers for thirty-one years of Vermont data?
We may never know.
These are not the only problems with Bellesiles’s accounts of murder.
His counts in his main table of homicide data (Table 6)126 do not add up. He
relates that he has 735 cases of homicide and that he drew 501 cases from
one source and “an additional 184 cases” 127 from a list of newspapers. But
this still leaves Bellesiles exactly fifty cases short of his total of 735 cases.
Where did the other fifty cases come from? Readers are left to speculate.
Finally, Bellesiles’s unsupported claim that homicide rates rose after
the Civil War128 is much too simple a story. Just as the gun culture and the
romance of the gun were supposedly taking over (in the decades after the
Civil War), homicide rates were actually plummeting throughout much of
the country, while in the Reconstruction South murder was rising.129 The
relationship between guns and homicides over time is so complex that it
cannot be reduced to the easy formula put forward in Arming America that
high gun ownership and high homicide rates go together.
D. Were Privately Owned Guns Mostly in Poor
While it is not surprising that government-owned guns might be rusting
away in armories during peacetime, Bellesiles claims that guns in private
hands were also mostly old or broken. For example, he claims that 53% of
the guns in frontier probate inventories were listed as broken or defective:
“An examination of more than a thousand probate records from the
frontiers of northern New England and western Pennsylvania for the years
1765 to 1790 revealed that only 14 percent of the inventories included
firearms; over half (53 percent) of these guns were listed as broken or
otherwise defective.” 130 Bellesiles makes a similar claim about the guns
listed in Providence, Rhode Island, probate inventories: “More than half of
these guns are evaluated as old and of poor quality.”131
Neither claim is true. Justin Heather and I have completed a careful
analysis of data from four of the six counties in Bellesiles’s 1765-1790
frontier sample (those from Vermont) and a partial analysis of inventories
from the other two counties (those from Western Pennsylvania). So far the
rate of guns “listed” as old or broken is less than 15%, not the 53% that
Bellesiles claims.132 Bellesiles’s own website report on guns in frontier
Vermont now shows very few listed as old or broken.133
As to the Providence, Rhode Island, data, Bellesiles has dropped the
claim from the hardback edition of Arming America that the guns in the
inventories were evaluated as old or broken and now claims that the
majority of guns are so low-valued that he reappraises them as old or
broken.134 There are a number of problems with this claim. Most important,
historians should not reappraise 300-year old guns that they have never
seen based solely on evidence of their monetary value. Bellesiles does not
provide a sufficient basis for his reappraisal. He does not reappraise a few
very low-valued guns. Rather, he appraises the median-priced gun in
Providence as old or broken. The best evidence we have for what a typical
gun cost in Providence, Rhode Island, is the very probate data showing that
guns cost about one pound.135 This is consistent with other data, as I show
in the next Section. A new military-quality weapon in a time of war might
go for two to three times that amount, but that does not mean that an
ordinary working gun or fowling piece in a time of peace would go for
more than about a pound. In addition, Bellesiles should have at least
disclosed the fact that he made such a reappraisal in his original
publication. Instead, he claimed this reappraisal only after his error was
Finally, as to the frontier data on dysfunctional guns, Bellesiles says
that they are listed as such. It is not possible to change this claim based on a
reappraisal. Of the estates that Heather and I examined, 83-91% of them
listed guns that were not described as old or broken.136 This does not, of
course, indicate that most of these guns were of military quality or even
suitable for battle. Many were undoubtedly fowling pieces, better suited for
hunting birds. But this is solid evidence that many Americans owned
E. How Expensive Were Guns?
Michael Bellesiles claims that guns were too expensive for widespread
private ownership, a claim that has often been repeated by positive
reviewers.137 Bellesiles writes that “a flintlock cost £4 to £5.” 138
Of course, everything was expensive in colonial America for a
populace that was very poor by today’s standards. Reviewers apparently
failed to note that Bellesiles provides no source for his claim about what
guns cost. Yet good evidence exists, and it conflicts with Bellesiles’s claim.
First, there are auction data. In North Carolina auctions in 1774, a
simple “gun” sold for less than £1 (median price: £0.8).139 This was
roughly the same as a table, a chair, a dictionary, a great coat, or a saddle.140
Comparing the cost of buying a simple shotgun or pistol at Wal-Mart today
to buying these other items would suggest that guns were not relatively
more expensive then than they are today.
We also have extensive probate data from the colonial period, most of
which shows median prices for guns not listed as old or broken from just
under £1 to about £1.5.141 Further, with median probated estate sizes in
1774 of more than £200,142 a gun at about £1 was a relatively minor
expense. Even if one rightly assumes that probated estates are skewed
toward the wealthier decedents, an analysis of the effect of wealth shows
that guns were listed in substantial portions of estates above the very
poorest.143 Only for estates below £10 did fewer than thirty percent of
inventories list guns. And, whatever the cost, people bought guns before
other seeming essentials. In the earlier colonial period, Gloria Main and
Anna Hawley both found more guns than tables or chairs or stools.144 When
men could afford to buy a gun, they did.145 This suggests either that they
were very useful tools or that they had an important social meaning (for
example, to reinforce their owners’ masculinity or provide peace of
Randolph Roth mentions a newspaper ad from 1785 for 3000 new
British muskets at only $3 apiece, a very low price compared to other
common items. Here is Roth’s account:
Gun dealers, for their part, knew that they had to appeal to farmers,
gardeners, and small-game hunters who fired shot as well as to
militiamen who had to own military-grade, ball-firing weapons. For
instance, when Joseph and William Russell of Providence, Rhode
Island, advertised the sale in 1785 of 3,000 “EXCELLENT NEW
BRITISH MUSKETS” for three dollars each, they hastened to add
that at least 600 were “neat Fowling Pieces.” 146
Almost all of the existing evidence suggests that in a world where
nearly everything was expensive, guns were not particularly so. They were
within the reach of most families, especially if the families thought them
more important than having a table or a chair, as many apparently did, since
guns were roughly as commonly listed in probate inventories as these
seeming essentials. Part of Bellesiles’s confusion may stem from looking at
the prices of new military weapons in a time of war, and not accounting for
condition, temporary shortages, or type of weapon. A more typical price for
an ordinary used gun in colonial America would have been £1, not the £4 to
£5 asserted in Arming America.
F. How Effective Were Guns, Bladed Weapons, and
Arming America’s accounts of military actions, militia ineffectiveness,
and battle weaponry show similar problems in the use of evidence, though
Bellesiles’s overall view of the militia is a standard one. As to the
ineffectiveness of militia compared to regular army troops, Bellesiles offers
an extreme, unnuanced version of the standard view, but his view is widely
shared. To present a more balanced analysis of the historical record would
take greater expertise on the history of militia than I have and more space
than one section of a review. But to give one example, George Washington,
who according to Bellesiles was unrelentingly negative about militia,147
actually had an ambivalent view of militia—as is evident in Mark
Kwasny’s excellent analysis, Washington’s Partisan War.148
Yet even where Bellesiles is more or less correct, he takes his evidence
further than it will bear. As Clayton Cramer has discussed, he quotes
Washington out of context on the poor state of militia reporting for duty149
without noting that Washington was only referring to a few troops out of a
large number about which he was not complaining.150 What Washington
clearly treated as exceptional is taken by Arming America as the norm.
In his account of Lexington and Concord, Bellesiles systematically
understates the effectiveness of guns and the militia, emphasizing hand-to-hand
combat. He goes to great lengths in Arming America to replace the
“myth” of the American with a gun with a new myth of Americans often
relying on an axe as a weapon. For example, Arming America claims: “At
Menotomy, Massachusetts, the Americans fell on the British with a
vengeance; the combat was almost entirely hand-to-hand, axes against
bayonets.” 151 Justin Heather has gone through the accounts that Bellesiles
cites and neither claim is true—it was not “almost entirely” hand-to-hand
combat, and there is no mention of Americans using axes. Heather finds
that guns were a very important part of the battle.152 Indeed, the idea that the
Americans fought the British with axes is questionable even without
checking sources, since axes were unwieldy for hand-to-hand combat.153
In Bellesiles’s fervor to establish the shortages of guns and the
unfamiliarity of American militia with guns, he misstates evidence. For
example, he writes:
Pikemen were present at nearly every encounter in King Philip’s
War, as there were not enough guns to go around. Nonetheless, in
October 1675, the Massachusetts General Court ordered that,
“whereas it is found by experience that troopers & pikemen are of
little use in the present warr with the Indians . . . It is ordered by the
court . . . that all pikemen are hereby required . . . to furnish
themselves with fire armes.” But they could not locate sufficient
guns, leading one Massachusetts soldier to recall in 1681, “I
thought a pike was best for a young soldier, and so I carried a pike,
and . . . knew not how to shoot off a musket.” 154
The quotation is used by Bellesiles to support three propositions:
gunlessness, reasonable reliance on edge weapons (in this case, a pike), and
unfamiliarity with guns. The account, however, was not from a
Massachusetts soldier recalling his days of gunlessness during King
Philip’s War. Instead, the quotation is from John Dunton, an English
bookseller on a five-month vacation to America in 1686,155 who wrote a
letter back to England about the unusual habits of American settlers:
But from Love, I must make a Transition to Arms; and cou’d you
think that [I] . . . wou’d ever make a Souldier? Yet so it fell out: For
’tis their Custom here for all that can bear Arms, to go out on a
Training Day: But I thought a Pike was best for a Young Souldier,
and so I carry’d a Pike; and between you and I, Reader, there was
another Reason for it too, and that was, I knew not how to shoot off
a Musquet. But ’twas the first time I ever was in Arms; which tho’ I
tell thee, Reader, I had no need to tell to my Fellow-Souldiers, for
they knew it well enough by my awkward handling of them. For I
was as unacquainted with the Terms of Military Discipline, as a
wild Irish Man [who did not know his right hand from his left] . . . .
But we were even here, for tho’ they understood Arms better than I,
yet I understood Books better than they.156
Unlike American settlers, this bookish English visitor knew nothing about
arms. Dunton observes American familiarity with guns, and the fact that he
was armed by others suggests no shortage of firearms. As to his preference
for pikes, Dunton explains his reasons, which mostly do not apply to the
Americans he is writing about. Last, Bellesiles uses the word “recall” as if
Dunton is speaking about his past experiences in King Philip’s War, rather
than his current experience in arms for the first time. Bellesiles also
mistakenly shifts the date (1686) to five years closer to King Philip’s War,
and he uses the source to support his contention that there were gun
shortages during that war.157 Bellesiles somehow turns a tale of American
familiarity with guns, reliance on guns, and well-armed units into the
G. Were Guns Kept in the Home?
In one of the book’s stranger arguments, Bellesiles argues that, by law
and in fact, privately owned guns were not kept in the home, but rather
were stored in central armories.158 This has profound implications for his
thesis, because if guns were not kept in the home, they were not generally
available for homicide, vermin control, target practice, war, or defense
against Native Americans or criminals. How could the trusty musket (or
rifle) be kept over the mantelpiece if by law it was centrally stored?
Further, Bellesiles claims:
[L]egislators feared that gun-toting freemen might, under special
circumstances, pose a threat to the very polity they were forced to
defend. Colonial legislatures therefore strictly legislated the storage
of firearms, with weapons kept in some central place, to be
produced only in emergencies or on muster day, or loaned to
individuals living in outlying areas.159
Bellesiles cites a long string of statutes in support of his unusual claim,160
but as Clayton Cramer points out, these statutes do not state that privately
owned guns must or should be centrally stored in armories.161 Either they
say nothing about Bellesiles’s fanciful claim, or they provide for the central
storage of gunpowder, which was explosive and dangerous to keep in large
quantities in the home.162
One class of data that seems to support the widespread use and keeping
of guns in the home is the accidental firearm death data that Randolph Roth
has collected. Roth concludes that accidental firearm deaths in New
Hampshire and Vermont between 1783 and 1824 were suffered at rates
slightly higher than today’s annual rate of four per million persons.163 The
occurrence of so many accidents in what was essentially peacetime supports
the notion that guns were kept in the home (and therefore actually used),
not centrally stored.
H. Were Guns Common in Travel Accounts?
Arming America also relies on travel accounts to demonstrate the
unimportance and ineffectiveness of firearms and the importance of axes as
weapons.164 Bellesiles concludes:
Generally stated, an examination of eighty travel accounts written
in America from 1750 to 1860 indicates that the travelers did not
notice that they were surrounded by guns and violence. . . . That
absence of discussion about guns in travelers’ accounts is
intriguing . . . .165
There are a number of problems with Bellesiles’s use of this body of
evidence. First, guns are frequently mentioned in the very travel accounts
that Bellesiles cites, as Clayton Cramer and others have pointed out.166
Second, Bellesiles uses the travel accounts to push the ineffectiveness of
firearms and the relative effectiveness of axes. Bellesiles’s statements in
support of the relative importance of the axe over the gun are not supported
by the cited sources and at least one of the views attributed to a traveler
cannot be found in the cited source.167
The traveler whose account gets the fullest treatment in Arming
America is Frederick Gerstaecker.168 But Bellesiles, pushing his pro-axe
theme, puts words into Gerstaecker’s mouth. Bellesiles wrote: “He
[Gerstaecker] noted that they [Americans] were very ‘expert’ at the use of
axes, ‘which they begin to wield as soon as their arms are strong enough to
use them,’ adding that axes made very good weapons.” 169 Gerstaecker did
note that American frontiersmen were “particularly expert with the axe,
which they begin to wield as soon as their arms are strong enough to lift
it.” 170 He also explained that Americans use the axe “for a variety of
purposes—building houses, laying roofs and floors, forming the chimneys
and doors, the only other tool used being the auger.” 171 Gerstaecker did not,
however, add “that axes made very good weapons,” as Arming America
Arguing that guns were not needed for personal use, Bellesiles
paraphrases the immigrant account of Ole Rynning: “Rynning advised his
Norwegian readers to bring ‘good rifles with percussion locks,’ as such
good guns were far too expensive in America and could be sold there for a
solid profit. Guns thus had an economic value, but if thought requisite for
self-protection, it remained an unstated assumption.” 173 As Clayton Cramer
has pointed out, Rynning actually urges immigrants to bring “‘good rifles
with percussion locks, partly for personal use, partly for sale.’” 174 If
Bellesiles had just quoted the four words after those he did quote, his
readers could have seen for themselves that Rynning believed guns should
be brought for personal use; this was not an “unstated” assumption.
Some of the very travel accounts that Bellesiles quotes for the “absence
of discussion about guns in travelers’ accounts” 175 and the proposition that
“travelers did not notice that they were surrounded by guns and
violence” 176 contain strong statements that guns were all around them.
Baynard Rush Hall, writing under the pen name Robert Carlton, for
example, describes his love of rifles and their use in Indiana territory:
Let none think we western people follow rifle shooting, however,
for mere sport; that would be nearly as ignoble as shot gun
idleness[.] The rifle procures, at certain seasons, the only meat we
ever taste; it defends our homes from wild animals and saves our
corn fields from squirrels and our hen-roosts from foxes, owls,
opossums and other “varments.” With it we kill our beeves and our
hogs, and cut off our fowls’ heads: do all things in fact, of the sort
with it, where others use an axe, or a knife, or that far east
savagism, the thumb and finger. The rifle is a woodsman’s lasso.
He carries it everywhere as (a very degrading comparison for the
gun, but none other occurs), a dandy a cane. All, then, who came to
our tannery or store came thus armed; and rarely did a customer go,
till his rifle had been tried at a mark, living or dead, and we had
listened to achievements it had done and could do again.177
This passage shows not only the wide use of guns, but the passion for guns
that Bellesiles argues was absent in early America. In many of the travel
accounts that Bellesiles cites, the settlers or travelers describe the
ubiquitousness of guns and the skill of Americans in using them.178 The
accounts support widespread gun ownership on the frontier and suggest that
guns, not axes or bladed weapons, were the primary weapons for combat
I. How Central Are the Errors to the Thesis of
One of the oddest claims to surface recently is that the problems with
Arming America touch only the probate data (as if contrary evidence could
just be ignored) or touch only the quantitative data (as if there was not a
public scandal long before the quantitative errors were discovered). To
address such a belief is one of the reasons that I wrote this Review. Too
much attention has focused on the probate data. The probate data are
important to the book’s thesis, though they are discussed on only about
thirteen pages of the book,180 plus some additional footnotes. They were the
original impetus for the book.181 In early positive reviews of Bellesiles’s
work in the press and in scholarly articles, the probate data were the most
frequently mentioned statistical source material.182
Indeed, just to read an account of what the book was about from the fall
of 2000 is to realize how much people have recently tried to recast it.183 It is
revealing to see what Bellesiles himself said in his interviews with the press
from the early months before the probate data were revealed to be false,
such as in the Emory Report,184 Salon,185 or Playboy.186
Consider these questions and answers from a taped interview in
PLAYBOY: You suspected the image we have of a musket over
every fireplace. When did you first begin to notice the missing
BELLESILES: . . . . I was studying probate records, the most
complete records, the most complete record of property ownership
in early America. They contain lists of absolutely everything that a
person owned—scraps of metal, broken glasses, bent spoons,
broken plows. . . . While studying these probate records, I realized I
was not seeing guns. They were supposed to be in every single
home. When I looked at the frontiers of western Pennsylvania and
northern New England, I found guns in only 10 percent of the
probate records, and half of those guns were not in working order.
Since then, I’ve read 11,150 probate records, samples over a 100-
year period, and I have found guns in 13 percent of the probate
records. Prior to 1850, the gun is just not there.
PLAYBOY: What else did you look at?
BELLESILES: States kept inventories of weapons. . . . [A]ll the
governments regularly took a census of firearms. They sent the
constables door-to-door to ask, “What guns do you have? What
condition are they in?” . . .
PLAYBOY: How many guns did the states find in the census?
BELLESILES: It depends on the state. In the Colonial period, there
were only enough guns for about one and a half to two percent of
the populace. . . .
PLAYBOY: Who was allowed to own guns?
BELLESILES: Only white male Protestant property owners. Not
indentured servants. Not slaves. Not Indians. Not Catholics. . . .
PLAYBOY: What did a gun cost in the 18th century[?]
BELLESILES: A functional gun would cost five to six pounds,
which is equivalent to a year’s wages for an unskilled laborer,
about half a year’s wages for a skilled artisan. . . .
. . . .
PLAYBOY: The current gun debate is mired in homicide rates. If
there were no gun culture in the Colonial era, how did we die?
BELLESILES: Scholars of violence who have looked at homicide
found that there was little interpersonal violence in America prior
to the 1840s. . . . When I was doing my research, I found county
court records that did not show a homicide in a 50-year period.187
If one reads these claims in light of what has been revealed since, one sees
one error after another. Bellesiles claims:
(1) Probate records list “absolutely everything that a person
owned—scraps of metal, broken glasses, bent spoons, broken
plows”—when it is generally accepted that probate records are
(2) In “the frontiers of western Pennsylvania and northern New
England, I found guns in only 10 percent of the probate records”—
rather than the roughly 40% of inventories that actually listed
(3) “[H]alf of those guns were not in working order”—rather than
fewer than 15% actually listed as not working;190
(4) “[A]ll the governments regularly took a [door-to-door] census
of firearms”—when none of Bellesiles’s gun censuses are in fact
full censuses of arms in all hands and apparently none were done
(5) “[T]here were only enough guns for about one and a half to
two percent of the populace”—when the best estimate is that about
54% of adult males owned firearms in their probate estates, and all
published estimates are much higher than Bellesiles’s;192
(6) “Only white male Protestant property owners” were “allowed
to own guns”—when Catholics were rarely barred from gun
ownership and women and poor white freemen were never barred
in any source Bellesiles cites for propositions such as this;193
(7) “A functional gun would cost five to six pounds”—when
ordinary guns usually cost about £1;194
(8) “[T]here was little interpersonal violence in America prior to
the 1840s”—when homicide rates were as high or higher than
(9) County court records “did not show a homicide in a 50-year
period”—when Bellesiles missed 100% of the homicide
prosecutions in the 46 years of Plymouth records that Bellesiles
says had no prosecutions for homicide.196
Every one of these claims is false, and they are a pretty fair sampling of the
errors discussed in this Review. If I had pulled the corresponding claims out
of the book, I might have been accused of selectivity; yet seeing them one
after another in a taped interview suggests just how central these myths are
that Bellesiles advances. These are the sorts of claims that were praised on
the book’s release, but have now been exposed as false.
To support his claim of low gun ownership, Bellesiles himself cited the
probate data and the militia data.197 And in April 2000, the New York Times
called the probate data “Mr. Bellesiles’s principal evidence.” 198 When
Charlton Heston tried to dismiss the probate data as irrelevant and
incomplete, he was rightly criticized for not wanting to face facts and for
The actual probate data can’t be easily put aside. They clearly undercut
the book’s thesis on many of its main points about early America—the
number of guns in private hands, the era when gun ownership was first
widespread, the condition of guns, where guns were kept, the price of guns,
the familiarity of Americans with guns, the relative desire for guns, the
gender breakdown of gun ownership, and the change in gun ownership over
time. These are not just isolated facts; they go to the role and social
meaning of guns in early America.
But what if the probate data could somehow be made to disappear? The
sad fact is that we would still have the worst historical scandal in decades.
The errors in the probate data may be the easiest to see, but they are not the
only serious ones. There was a scandal before Justin Heather and I exposed
Arming America’s probate errors, and there will be a scandal now that our
position is widely accepted, and the focus is returning to other parts of the
Unless one goes through all the book’s comments on a particular topic
and the evidence cited to back them up, one can’t really see just how
systematic the errors are. Randolph Roth has done this for Bellesiles’s
homicide data; Robert Churchill has done this for the gun censuses and
militia counts; Justin Heather and I have done this for the probate data;
Heather has done this for the stories about axes, bayonets, and edge
weapons; Clayton Cramer has done this for several types of sources,
including the gunsmith information, militia statutes, and substantial
portions of the travel accounts. When one goes through an entire body of
evidence, some errors are big and some are small, but the overall effect is
shocking, indeed unprecedented for a Bancroft-Prize-winning book. Nearly
every sentence that Bellesiles wrote about probate records in the original
hardback edition of Arming America is false.200 Nearly everything that
Bellesiles says about homicide is either false or misinterpreted, as is most
of what he wrote about the relative merits of the axe over the gun.201 When
the sources do not support the main premise of Arming America, Bellesiles
sometimes misreports their content in a way that fits his thesis, as he does
in over 200 instances mentioned in this Review.202 Using Arming America,
one could build a wonderful course for graduate students about historical
methods—each student checking a different body of sources. Indeed, Eric
Monkkonen is teaching such a course at UCLA this year.
There are two recurrent characteristics of Bellesiles’s problems
throughout the book and the dispute: (1) innumeracy, and (2) a failure to
reconcile his findings with the existing literature. Bellesiles thinks that
counting is important, indeed crucial to the book: “Without such efforts at
quantification, we are left to repeat the unverifiable assertions of other
historians, or to descend into a pointless game of dueling quotations—
matching one literary allusion against another.” 203 Yet he created no
database for any of his data.204 He just made tick marks on a legal pad—in
the case of the probate data, over 11,000 of them.205
It is clear from Bellesiles’s responses to criticism, moreover, that he
does not understand how someone could prove his probate data false
without checking all of it.206 For the last year, Knopf and Bellesiles have
published a book whose most prominent data are not only false, but known
to be mathematically impossible. The math, which has been verified by
Randolph Roth, could be done by an average middle schooler; it is just
computing a mean from several means.207 Why Knopf has not investigated
this problem is unclear.
Bellesiles’s innumeracy slides into his more general failure to reconcile
his findings with the existing literature. Bellesiles claims that South
Carolina had the lowest homicide rates in the country, while other
historians wrote the opposite.208 Bellesiles doesn’t explain why he got
different results. In this case, it’s because he treated explicitly partial data as
if they were complete and then compared these data to the state population.
Bellesiles finds low counts of guns in probate records, but there are actually
fairly high counts in the sources he cites. Again, he has made no attempt to
figure out why his numbers are so different from everyone else’s. Bellesiles
claims that probate inventories list every item in an estate, but the scholars
he cites say the opposite. Again, he has made no attempt to reconcile his
When I first contacted Bellesiles privately in November 2000 with
serious problems with his probate counts, I started with the Providence,
Rhode Island, data. Bellesiles published (and confirmed in correspondence)
that he used the published Records of the Town of Providence.209 I offered
to lend him a copy to facilitate his checking. Resolving differences in
counts ideally should be a matter of cooperation among scholars. It was not
hard to see that he had counted women as men and intestate estates as
having wills; an hour in the library would have shown that. Bellesiles wrote
back that he would recount Providence, but that it was not “a top
priority.” 210 It was as if he were not surprised that he had miscounted most
of the estates in Providence, or at least not curious whether he had done so.
In January 2001, when I first publicly presented the paper that I wrote
with Justin Heather, Bellesiles responded to the criticism in a way that he
repeated throughout the scandal—he mentioned all the hostile e-mail
invective that he had received from gun lovers and attacked the quality of
work of everyone who disagreed with him, including Alice Hanson Jones,
whom he praises in the acknowledgements to the book.211 Bellesiles
claimed that the deceased Jones, a giant in the field for whom the prize in
economic history is named, confused the word “gown” for the word “gun”
and avoided the poorer estates in her sample.212 Of course, he provided no
evidence for these claims. I have checked enough of Jones’s estates against
the original records to know with relative certainty that she made no
systematic reading errors. In one form or another, Bellesiles has quietly
backed off on all the main claims that we showed were erroneous in that
first January 2001 draft of our probate study.
Arming America is an impressive book, especially to those not versed
in the materials that Bellesiles wrote about. It is extremely well-written for
a book that covers so many apparent specifics of gun ownership and use.
Superb historians praised it on its release. Yet even from the beginning,
there were those who found disturbing differences between Arming
America and its sources. As time has passed and other scholars have
entered the debate, these errors—which once looked like such serious
defects that they could not be true—have been confirmed. So far Bellesiles
has not successfully defended any challenged portion of the book. Nor has
he or any of his dwindling corps of defenders been able to point to any
specific errors that Bellesiles’s academic critics have yet made.
Undoubtedly, those whose scholarship has uncovered errors in Arming
America have made mistakes—everyone does from time to time. What is
unprecedented in such a prominent book is how many errors it contains and
how systematically the errors are in the direction of the thesis.
The book and the scandal it generated are hard to understand. How
could Bellesiles count guns in about a hundred Providence wills that never
existed, count guns in San Francisco County inventories that were
apparently destroyed in 1906, report national means that are mathematically
impossible, change the condition of guns in a way that fits his thesis,
misreport the counts of guns in censuses or militia reports, have over a 60%
error rate in finding guns in Vermont estates, and have a 100% error rate in
finding homicide cases in the Plymouth records he cites? We may never
know the truth of why or how Arming America made such basic errors, but
make them it did.
As scholars, we must content ourselves with correcting errors and
searching for the realities of gun ownership, use, and social meaning.
Beyond that, we might try to figure out how to avoid a repetition of this
The historical profession will survive the Bellesiles scandal. If people
had gone to the library when questions were raised over a year ago, then
much of the acrimony could have been avoided. The errors in the
Providence materials (e.g., counting women as men and counting guns in
about a hundred wills that never existed) are just as clear and just as easy to
check as those of Stephen Ambrose or Doris Kearns Goodwin. But
Ambrose and Goodwin did not claim that they were political martyrs. They
knew that people would eventually check the source books and see for
themselves; they knew there was no point in denying the claims of error.
Bellesiles took a different tack. I was surprised when he did not take the
usual scholarly approach of grudgingly admitting his errors—either when I
contacted him privately or when I later presented my scholarship publicly.
Perhaps Bellesiles acted differently than Ambrose and Goodwin did
because his errors are so much more serious. They go to issues at the heart
of the book—how many guns there were, what condition they were in, who
owned them, how they were used, and how much they cost. Even if
Bellesiles withdrew the probate data, there would still be other problems—
problems that scholars other than Justin Heather and I are examining with
great care. Only by looking closely at the militia counts, gun censuses,
battle stories, travelers’ accounts, and every other type of source that
Bellesiles relied on can the historical profession evaluate Arming
America—and the new mythology of relative gunlessness in early America
that it tried to create.
118. Id. at 81; see also id. at 353 (claiming that there were only five murders in Vermont
from 1760 to 1790).
119. Roth, supra note 41, at 234.
120. Id. at 235.
121. BELLESILES, supra note 3, at 82.
122. 1-10 RECORDS OF THE COLONY OF NEW PLYMOUTH IN NEW ENGLAND (Nathaniel B.
Shurtleff et al. eds., Boston, William White 1855-1861).
123. Roth, supra note 41, at 234 n.31 (citations omitted).
124. BELLESILES, supra note 3, at 353.
125. Roth, supra note 41, at 236 (citations omitted).
126. BELLESILES, supra note 3, at 450.
128. See id. at 434, 436.
129. See ERIC H. MONKKONEN, MURDER IN NEW YORK CITY 9-10 (2000). Randolph Roth is
finding the same pattern as Monkkonen in many areas outside New York City, except in the
South, where homicide was increasing. Randolph Roth, Toward Better Ways To Count Guns,
Panel Presentation Before the Social Science History Association (Nov. 2001).
130. BELLESILES, supra note 3, at 13.
131. Id. at 109.
132. Lindgren & Heather, supra note 84.
133. Bellesiles, supra note 85.
134. Compare BELLESILES, supra note 93, at 109, with BELLESILES, supra note 3, at 109.
135. See 6, 7, 13 & 16 EARLY RECORDS OF THE TOWN OF PROVIDENCE, supra note 80.
136. See Lindgren & Heather, supra note 24 (manuscript at 25, 28 tbl.3, 42 tbl.8, 49);
Lindgren & Heather, supra note 84.
137. See, e.g., Paul Finkelman, Taking Aim at an American Myth, 99 MICH. L. REV. 1500,
1501 (2001) (book review).
138. BELLESILES, supra note 3, at 106.
139. 3 JONES, supra note 23, at 1691-720.
141. See, e.g., id.; Lindgren & Heather, supra note 84.
142. See Lindgren & Heather, supra note 24 (manuscript at 32 tbl.7).
144. Id. (manuscript at 6-10 & nn.9-24) (citing MAIN, supra note 22, at 288-89 tbls.C.3-4;
and Hawley, supra note 22, at 28).
145. Id. (manuscript at 6-10 & nn.9-24, 32 tbl.7).
146. Roth, supra note 41, at 232 (quoting Advertisement, BOSTON GAZETTE, Mar. 7, 1785).
147. BELLESILES, supra note 3, at 159, 193-95.
148. MARK V. KWASNY, WASHINGTON’S PARTISAN WAR, 1775-1783, at 16 (1996)
(“Washington presented a more complex attitude toward the use of the militia in the
Revolutionary War than the traditional description allows.” ); see also id. at 17, 83, 110, 135, 185.
149. See BELLESILES, supra note 3, at 159. Bellesiles wrote:
Colonel Washington reported on the militia to Governor Dinwiddie: “Many of them
[are] unarmed, and all without ammunition or provision.” In one company of more than
seventy men, he reported, only twenty-five had any sort of firearms. Washington found
such militia “incapacitated to defend themselves, much less to annoy the enemy.”
150. Washington wrote:
I think myself under the necessity of informing your Honor, of the odd behaviour
of the few Militia that were marched hither from Fairfax, Culpeper, and Prince William
counties. Many of them unarmed, and all without ammunition or provision. Those of
Culpeper behaved particularly ill: Out of the hundred that were draughted, seventy-odd
arrived here; of which only twenty-five were tolerably armed.
I proposed to the unarm’d, that as they came from home (at least with a shew) of
serving their country; and as they were, from the want of arms, incapacitated to defend
themselves, much less to annoy the enemy, or afford any protection to the Inhabitants;
that they shou’d (during their short stay here) assist in forwarding the public works; for
which I offered them 6d. per day extraordinary. But they were deaf to this and every
other proposition which had any tendency to the interest of the Service.
As such a conduct is not only a flagrant breach of the law, and a total contempt of
Orders, but will be such a precedent (shou’d it pass without impunity) as may be
productive of the most dreadful consequences. I therefore flatter myself, your Honor
will take proper notice of these men. I have written to their County Lieutenant on this
Letter from George Washington to Robert Dinwiddie (June 27, 1757), in 2 THE WRITINGS OF
GEORGE WASHINGTON FROM THE ORIGINAL MANUSCRIPT SOURCES, 1745-1799, at 78, 78-79
(John C. Fitzpatrick ed. 1931); see Cramer, Firearms Ownership, supra note 27, at 51-52.
151. BELLESILES, supra note 3, at 174.
152. Justin Lee Heather, Weapons of War in Colonial America: A Situational Hierarchy
(2001) (unpublished manuscript, on file with author).
154. BELLESILES, supra note 3, at 117 (quoting Letter from John Dunton to George Larkin
(Mar. 25, 1686), in JOHN DUNTON’S LETTERS FROM NEW-ENGLAND 56, 140 (William H.
Whitmore ed., Boston, T.R. Marvin & Son 1867)).
155. William H. Whitmore, Preface to JOHN DUNTON’S LETTERS FROM NEW-ENGLAND,
supra note 154, at i, xxii-xxiii.
156. Letter from John Dunton to George Larkin, supra note 154, at 140-41.
157. Compare supra text accompanying note 154, with Letter from John Dunton to George
Larkin, supra note 154.
158. BELLESILES, supra note 3, at 73.
160. Id. at 472-73 n.10.
161. Clayton E. Cramer, Primary Historical Sources, at http://www.claytoncramer.com/
primary.html (last visited Apr. 18, 2002).
163. Roth, supra note 41, at 232-33.
164. BELLESILES, supra note 3, at 305-22.
165. Id. at 306.
166. See, e.g., Cramer, Firearms Ownership, supra note 27, at 37; Heather, supra note 152, at
167. See infra notes 168-172 and accompanying text.
168. BELLESILES, supra note 3, at 311-15 (discussing FREDERICK GERSTAECKER, WILD
SPORTS IN THE FAR WEST (London, Routledge 1854)).
169. Id. at 313 (quoting GERSTAECKER, supra note 168, at 241).
170. GERSTAECKER, supra note 168, at 241.
172. BELLESILES, supra note 3, at 313; see also Heather, supra note 152 (manuscript at 24)
(discussing Gerstaecker’s travel account).
173. BELLESILES, supra note 3, at 341 (emphasis added).
174. Cramer, Firearms Ownership, supra note 27, at 147-48 (quoting OLE RYNNING, OLE
RYNNING’S TRUE ACCOUNT OF AMERICA 99 (Theodore C. Blegen ed. & trans., 1926)) (emphasis
175. BELLESILES, supra note 3, at 304.
177. ROBERT CARLTON [BAYNARD RUSH HALL], THE NEW PURCHASE, OR, SEVEN AND A
HALF YEARS IN THE FAR WEST 107-08 (James Woodburn ed., Princeton Univ. Press 1916)
178. WILLIAM N. BLANE, AN EXCURSION THROUGH THE UNITED STATES AND CANADA,
DURING THE YEARS 1822-23, at 145 (London, Baldwin, Craddock & Joy 1824) (“Every boy, as
soon as he can lift a rifle, is constantly practicing with it, and thus becomes an astonishingly
expert marksman.” ); FORTESCUE CUMING, SKETCHES OF A TOUR TO THE WESTERN COUNTRY
(1810), reprinted in 4 EARLY WESTERN TRAVELS, 1748-1846, at 46 (Reuben Gold Thwaites ed.,
1904) (“The inhabitants of this country in common with the Virginians, and all the back woods
people, Indians as well as whites, are wonderfully expert in the use of it: thinking it a bad shot if
they miss the very head of a squirrel.” ); 1 CHARLES AUGUSTUS MURRAY, TRAVELS IN NORTH
AMERICA 118-19 (London, Richard Bentley 1839) (“ [N]early every man has a rifle, and spends
part of his time in the chase.” ); 2 ISAAC WELD, TRAVELS THROUGH THE UNITED STATES OF
NORTH AMERICA, AND THE PROVINCES OF UPPER AND LOWER CANADA, DURING THE YEARS
1795, 1796, AND 1797, at 150 (London, John Stockdale 1807) (comparing Canadian hunters to
U.S. hunters, and stating that “[t]he people here, as in the back parts of the United States, devote a
very great part of their time to hunting, and they are well skilled in the pursuit of game of every
description” ). Bellesiles cites these reports at BELLESILES, supra note 3, at 542-44 n.5. I am
indebted to Clayton Cramer for identifying these accounts.
179. See Cramer, Firearms Ownership, supra note 27, at 131-51. Bellesiles cites Trabue’s
account for the proposition that “North Americans often perceived the ax as the equal of a gun.”
BELLESILES, supra note 3, at 76. Trabue’s account, properly considered, shows reliance on
firearms rather than axes. WESTWARD INTO KENTUCKY: THE NARRATIVE OF DANIEL TRABUE
44-46, 111 (Chester Raymond Young ed., 1981).
180. See supra note 106 and accompanying text.
181. BELLESILES, supra note 3, at 13.
182. Supra notes 108-112 and accompanying text.
183. A Chicago Tribune review (which is quoted on the back of the paperback edition of
Arming America) is fairly typical:
Bellesiles, a professor of history at Emory University with a specialty in the culture of
violence, argues . . . that early Americans had little use for guns and owned them hardly
at all, and that gun ownership did not become widespread until a combination of
government subsidy and clever marketing forced guns upon a heretofore unwilling
population. This is a book guaranteed to make a lot of people angry.
In many ways, “Arming America” is the best kind of non-fiction. Bellesiles is
trying to do a big thing—explain how the U.S. became so enamored of the firearm—
and he goes about it with imagination and the care of a good historian. He stumbled on
his thesis, he writes, when examining early American probate records for a study of
frontier economics. In more than 1,000 records from New England and western
Pennsylvania from 1765 to 1790—records that included property down to broken
teacups—only 14 percent listed guns, and of those, more than half noted that the guns
were in useless condition. “That was the beginning of this project,” he writes, “a ten-year
search for a word that isn’t there.”
What follows is more than 600 pages, copiously footnoted, that absolutely
devastate the myth of the gun in early America. Bellesiles starts with the guns
themselves. Guns in the 17th and 18th Centuries were so complicated, delicate,
inaccurate and expensive that they were little more than status playthings for the rich.
Dan Baum, Targeting America’s Gun Culture: A New Book Shoots Down the Conventional
Wisdom About the History of Our National Passion for Firearms, CHI. TRIB., Sept. 3, 2000, at C1.
184. Terrazas, supra note 48.
185. Bowman, supra note 30.
186. James R. Petersen, Arming America: When Did We Become a Gun Culture?, PLAYBOY,
Jan. 2001, at 69, http://www.guncite.com/gun_control_bellesiles_plby.html.
187. Petersen, supra note 186.
188. See supra note 34 and accompanying text.
189. Lindgren & Heather, supra note 84.
190. See supra note 132 and accompanying text.
191. See supra Subsection II.B.1.
192. See supra note 77 and accompanying text; see also Lindgren & Heather, supra note 24
(manuscript at 25 & n.64).
193. See Cramer, supra note 161 (presenting scanned copies of militia and other statutes).
194. See supra note 141 and accompanying text.
195. See supra note 120 and accompanying text.
196. See supra notes 121-123 and accompanying text.
197. See supra note 113 and accompanying text.
198. Ramirez, supra note 29.
199. Bowman, supra note 30.
200. For example, five of the six sentences on probate records on page 13 of Arming America
are false; of the twenty-one sentences about probate records on pages 109-10, seventeen are false,
two are misleading, and only two sentences (having little to do with the thesis) are true.
BELLESILES, supra note 3, at 13, 109-10; supra Subsection II.B.3.
201. See supra Sections II.C, II.F.
202. Over 170 of these involve basic misreadings of probate inventories or wills in
Providence, Rhode Island, and Vermont, confirmed in print by Robert Churchill or Randolph
Roth. See Churchill, supra note 39; Roth, supra note 41. See generally infra Appendix.
203. BELLESILES, supra note 3, at 262.
204. E-mail from Michael Bellesiles to author (Sept. 19, 2000) (on file with author).
206. Alison O. Adams, Silenced: Is Uncivil Discourse Quelling Scholarship on Controversial
Issues?, EMORY ACAD. EXCHANGE, Dec. 2001-Jan. 2002, at http://www.emory.edu/ACAD_
EXCHANGE/2002/decjan/silenced.html. Adams wrote:
Although Bellesiles was quoted in the Chronicle of Higher Education saying that
Lindgren’s criticisms were “valid,” he emailed the Academic Exchange, “I have never
understood Lindgren’s logic of mathematical impossibility. Since neither he nor I have
the numbers, which were lost in the Bowden [Hall] flood [in 2000], I am at a loss to
grasp his omniscience.”
207. If there are more than 201 inventories from Bellesiles’s sixteen Southern counties (at his
mean of 18.3% guns) for the twenty-six years 1765-1790, then Bellesiles’s national mean of
14.7% of estates listing guns is mathematically impossible, since there are only 1200 inventories
from the only region below the mean—the frontier, at 14.2% with guns. In fact, there are
thousands of Southern inventories in his sample, not the 201 or fewer that could mathematically
support his mean. We have shown his mean to be false with actual data both from Maryland and
from Charleston, South Carolina. Lindgren & Heather, supra note 24 (manuscript at 53-54). An
analogy might make the logic clearer. If someone tells you that they have a 3.9 GPA with thirty
grades, but the first ten grades you check are Bs, you know that the 3.9 GPA is false. You don’t
have to check all the grades to prove that the GPA is false. Similarly, you do not need to recount
all twenty-six years of data in Arming America to show that its national mean is false, just six
months of data in one large Southern county, Charleston, South Carolina.
208. Roth, supra note 41, at 237 (explaining that the historian whose evidence Bellesiles cites
to support low homicide rates in South Carolina actually concluded from that evidence that it was
a “homicidal place” ).
209. In Arming America, Bellesiles disclosed that he obtained his Providence data from three
volumes of the published records: “This data is drawn from Horatio Rogers et al., eds., The Early
Records of the Town of Providence, 21 vols. (Providence, RI, 1892-1915), vols. 6, 7 and 16.”
BELLESILES, supra note 3, at 485 n.133.
He confirmed that his Providence data were drawn from the published records in
correspondence: “Finally, I am sorry to hear that you come up with different numbers from
Horatio Rogers, et al., eds., The Early Records of the Town of Providence (21 vols. Providence,
R.I., 1892-1915). I used these books at the Huntington Library six years ago and have not yet
come across my notes.” E-mail from Michael Bellesiles to author, supra note 95.
210. E-mail from Michael Bellesiles to author, supra note 95.
211. BELLESILES, supra note 3, at 582.
212. Posting of Michael A. Bellesiles, email@example.com, to H-OIEAHC@h-net.msu.edu
(Jan. 9, 2001) (copy on file with author). Bellesiles wrote:
[Jones’s] sample set does not constitute a complete run for every county in the years
covered, and I noticed that the shorter probate inventories were generally the ones
ignored. I was also struck by how often the word “gun” appeared, when in the
eighteenth century that word generally referred to cannon. I turned to the original files,
where I read words like “gown” that were recorded as “gun.”
Id. Bellesiles has never been able to provide even one example of Jones confusing gowns for guns
or of her supposed missampling in any county.
comments powered by Disqus
Clayton E. Cramer - 8/28/2002
If the evidence had been obscure, it might have been tolerable that historians were so easily taken in. What depresses me is how widespread the evidence was that Bellesiles was, at least, wrong. I've been reading a number of recent histories of Bacon's Rebelion; it is apparently well known that Governor Berkeley complained what misery it was to govern a colony when six out of seven men are "Poore Endebted... and Armed." At one point, Berkeley complained that Bacon had 2000 armed men behind him, and 600 men with guns actually showed up at the capital to back up Bacon--and this without any use of the public arms.
Historians operate on the "union card" principle: if you are a professor, you know what you are talking about; if you aren't, they ignore you (even if you have an MA in History and can document what you are saying). I am reminded of the famous lines from one of the late 19th century plays (by George Bernard Shaw?) A paraphrase: "What shall we do with this private? He is an idiot." "Make him a general, and then everything he says will make sense."
John G. Fought - 8/28/2002
Thank you for your careful and systematic work
on the whole range of factual errors in the book,
and for putting them into a clear relationship to
its argument. Everyone involved, with perhaps one
exception, is in your debt.
- Colorado Students Strip Naked in Protest of ‘Censorship’ of AP History Classes
- They should give this definition of History to all first year undergrads on their first day
- Field Report: What I learned by attending a workshop on Korean history
- Historians suggest ways California can integrate gay history into the school curriculum
- Now it’s Andrew Bacevich’s turn to do a MOOC