Counting Guns: Response to Bellesiles

James Lindgren jlindgren at WORLDNET.ATT.NET
Mon Jan 15 00:56:02 PST 2001

 In most academic disputes, it is hard for a reader to tell who is right
and who is wrong.  This is not one of those disputes.  This time, you
just need to spend an hour looking at the records.  The inaccuracies in
Michael Bellesiles' probate study are so massive that it would take only
an hour with the Providence records (volumes 6,7, and 16 of Rogers,
Early Records of the Town of Providence) to see that his counts are way
off.  These records are available in many good research libraries or on
CD-ROM from for $38, $8 more than the cost of Arming

 In doing our careful replication of the only 2 sources of probate
records that he both cited and published that he used (Providence and
Alice Hanson Jones' American Colonial Wealth), Justin Heather and I
found large discrepancies.  When in November I notified Michael
Bellesiles that the Providence counts were far off, I asked him to
recount Providence so as to reconcile our counts.  We offered to lend
him a copy of the records.  He replied that he would do so, but that it
was not a top priority for him.  I can only assume that he has still not
recounted, since he has not retracted the Providence findings yet.

 Hoping that he would recount Providence is why we gave him so much time
before going public and in part why I have been so slow to respond
online.  I keep hoping that he will be curious enough to check the
Providence data and candid enough to admit overwhelming errors.  It
would be better for all of us (particularly his defenders on internet
lists) if the retraction comes from Bellesiles himself rather than
effectively from us or his supporters.

 His strategy of responding by attacking on the internet rather than
responding to the evidence in a replication seems a plausible short-term
strategy, but a poor long-term strategy.  Ultimately, enough people are
going to check the Providence records and he will lose his credibility.
Since he has cited no evidence that actually supports his probate counts
in the 17th and 18th centuries, all he has is his credibility.

 Academic disputes are ultimately resolved by serious scholars looking
at the evidence, not by briefs posted to discussion lists.  Discussion
lists are great for getting information out, but not good for resolving
disputes if people don't actually look at the evidence.

 When I first wrote Michael Bellesiles for his data in late August, I
had not the slightest idea that his data could be inaccurate.  I just
wanted to control for other common objects; I assumed that his gun
counts were impeccable and that he had a database.  I was interested
mostly because it was about probate (which I teach and practiced many
years ago).

 It turns out, of course, that he has no database and never did.  His
data was collected by making marks on a legal pad, which would mean at
least 11,000 tick marks.  Further, the probate tables in both his 1996
J. of American History article and in Arming America have no cell counts
or sample sizes, just percentages.  For example, when Bellesiles says
that in the 1765-1790 period, 16.1% of probate inventories in Northern
coast urban counties listed firearms, he doesn't tell us his count
(sample size).  Is it 16.1% of 1,000 inventories or 5,000 inventories or
what?  Further, which were the Northern urban counties among his list of
counties? A replicator should not have to guesss.  It is rare for
published work using many separate samples not to disclose the size of
those samples.

 Since on this list, we did not go through our evidence (we just posted
our abstract), let me suggest some of the problems with Bellesiles'
discussion of the Providence records.

Bellesiles claims that most of the guns in the Providence inventories
listing guns “are evaluated as old and of poor quality.” (AA,p. 109) In
fact, fewer than 10% of the guns are so listed.  Anyone who skims the
records can see that Bellesiles' claim is inaccurate and our count is

Bellesiles claims that there are only males in the supposed 186
Providence estates with itemized inventories, when there are 17 women
leaving such inventories.  In just volume 16, see the estates of Mary
Borden (p.60), Sarah Clemance (p.420), Abigail Hopkins (p.410), Joanna
Inman (p.236), Mary Inman (p.146), Tabitha Inman (p.238), Ann Lewes
(p.429), Rachal Potter (p.346), Elizabeth Towers (p.278), Hannah Wailes
(p.165), Anna Whipple (p.370), Susanna Whipple (p.174), Mary Whiteman
(p.70), and Lydia Williams (p.341).  These are obviously women, not men.
Yet Bellesiles tells us, "These 186 probate inventories from 1680 to
1730 are all for property-owning adult males . . . ." (AA,p.109)

Bellesiles claimed that no women in all the inventories he looked at
owned guns.  In Providence, Freelove Crawford (6:117, described as a
widow) owned 5 guns (listed just after 5 knives). As we document in our
paper, 18% of female estates in the Jones database owned guns, including
female estates among the first five in the following counties: Worcester
(MA), Anne Arundel (MD), Southampton (VA), Chesterfield (VA), and
Halifax (NC).

Bellesiles claims that all 186 Providence estates had both wills and
inventories that "scrupulously recorded every item in the estate"
(AA,p.109) when fewer than half did. This is very easy to check just by
looking at the good subject matter index.  Further, dozens of times the
records noted that the decedent died intestate.  See, eg, volume 7 (pp.
32,53,45,65,69,106,109,112, 139,142, 145,152,157,179,205) and vol. 16
(pp. 9,12,14,17,28,31,33,37, 45,62,63,73,92,97,120,121,
124,156,159,167,175, 197,199,228,241,246,248, 279,286,312,316,
332,343,358,366, 373,377,380,425, 428,430,441,446,448,457,
462,467,468).  It is simply ahistorical to assume that old records are
perfectly complete and that everyone died with a will, and to purport to
count 186 wills when half are simply not there and never were (since
they died intestate).

Bellesiles claims that “a great many inventories" list “one of ye
Queenís Armes,” (AA,p.109) when only one inventory did (O.Browne, 6:188
and 16:2).  In all, Bellesiles misclassified over 60% of the Providence
estates on criteria that he thought important enough to mention.

We have data, a database, precise counts, evidence to support our claims
about probate records, and citations to our sources.  Bellesiles has no
database, no counts, and no cites to ANY sources of probate records that
actually support his claims of low gun ownership and poor gun condition.

As this email is already long, I will respond later to his precise
posting.  And perhaps before I post further responses, he will look into
his heart and see that his current strategy of attacking the sources he
cites and giving no evidence for his claims will ultimately fail.

 I have had nothing to do with any attempts to contact Bellesiles'
university administration (as I hope he knows).  If anyone ever raised
such an idea to me, I would try to talk them out of it. Bellesiles of
all people should understand that sometimes published work gets erratic

 The truth will ultimately come out because most historians are moved
more by evidence than by prejudice or appeals to their vanity as careful
scholars.  The unsupported attacks on Alice Hanson Jones and the
unwarranted attempts to tar serious scholars with the actions of other
people will be washed away by the simple evidence--that guns are very
common in 17th and 18th century probate inventories and are mostly
listed in good condition.

 Here is a link to our study, Counting Guns in Early America.  I expect
a new draft in a few days. You can download it, open Adobe Acrobat 4,
and open the file.
The link is:

We also want to thank many of the historians and law professors on these
discussion lists for their extraordinarily kind comments and helpful

Best Regards,

James Lindgren
Professor of Law
Director, Demography of Diversity Project
Director of Faculty Research
Northwestern University School of Law

Chair-Elect, AALS Section on Social Science
Co-Founder, AALS Section on Scholarship

Ph.D. Student, Sociology (concentration in Social Statistics)
University of Chicago

More information about the Conlawprof mailing list