in Better Angels, steven pinker drew rather heavily on the work of manuel eisner, an historian of crime (see here and here for more on eisner’s work). it was from eisner that pinker got much of the data for this chart showing the decline in violence (homicide) rates in europe over the course of the middle ages:

pinker - fig. 3.3

in turn, eisner and other historians of crime today give a lot of credit to a fellow named james buchanan given who seems to be (to have been?) something of a pioneer in the history of crime in medieval europe. i’ve seen his Society and Homicide in Thirteenth-Century England, which was published in 1977, frequently referred to by these crime historians.

anyway, what i’ve never seen anyone mention is that given compared the homicide rates for different regions of thirteenth century england!


yes, you read that right! we’ve got data — albeit kinda rough data — for homicide rates from various regions of england in the 1200s. and it’s not even my birthday!

here from Society and Homicide [pgs. 35 & 37-38 and pgs. 150-152. links added by me. note that i’ve renumbered given’s footnotes here since the same numbers were often repeated on consecutive pages — would be confusing in this quote then]:

“Obviously these rates are, at best, only approximations. As has been pointed out in Chapter 1, the population estimates on which they are based are vague. Despite their crudity, these estimated homicide rates are nevertheless interesting. If the population estimates made by the author are used as the basis of calculation, it is found that the homicide rate varied from a high of 64/100,000 per annum reported at the 1232 eyre of Warwick to a low of 4/100,000 per annum reported at the 1227 and 1248 eyres of Bristol. Of the rural areas, Warwick consistently had the highest homicide rates, with an overall rate of about 47/100,000 per annum for the 25 years covered by the three eyres. Norfolk had the lowest rate, 9/100,000 per annum for the 23 years covered by the eyres. If the estimates based on J. C. Russell’s figures are used as a basis for calculation, some difference appears. Although the highest homicide rate still remains that of the 1232 Warwick eyre, it is much reduced, being only 30/100,000 per annum. And the overall rate for Norfolk is found to have increased to 15/100,000 per annum. If we assume that the counties in question had the same population in the thirteenth century as they did in 1801, the homicide rates found are still high. The highest, however, is that for the 1276 eyre of Bedford, 18.9/100,000, and the lowest is 6.8/100,000 for the 1227 eyre of Kent.

“Because the population estimates upon the basis of which these homicides rates have been figured are very imprecise, homicide rates have been calculated on yet a fourth, and considerably different, basis. Instead of using population as a basis for estimating homicide rates, I have used the number of settlements within the country. Homicide rates have been calculated in terms of the number of homicides per twenty settlements per annum. For example, in the four years covered by the 1202 Bedford eyre there were 22 homicide reported. Since there were about 146 settlements in Bedfordshire, this means that for every twenty settlements in the county the figure was 0.8 for homicides commited every year. Similarly, in the eleven years covered by the 1268-69 Norfolk eyre, 399 homicides were reported. Since there were 698 settlements in Norfolk, this means that for every twenty settlements there were 1.1 killings every year. With calculations on this basis, the 1232 Warwick eyre and the 1276 Bedford eyres show the highest rates, of 1.6 homicides committed every year for every twenty settlements. The 1241 Oxford eyre now shows the lowest homicide rate, with only 0.5 in every twenty settlements each year….

“In this chapter an attempt will be made to sketch the different ways in which violence manifested itself in the various agrarian societies contained within the borders of the five counties whose eyre rolls have been analyzed for this study. These counties have been divided into eight regions: rural Bedfordshire, the plains of northern Oxfordshire,[1] and Felden Warwickshire,[2] all three common-field regions containing large nucleated villages practicing communal agriculture and characterized by the prevalence of impartible inheritance and large numbers of unfree peasants; the Chiltern Hills,[3] where settlements were more scattered and individual freedom more common; rural Kent, where virtually all the peasants were free, settlement scattered, partible inheritance practiced, and agrarian activities unregulated by the village community; rural Norfolk, where partible inheritance also prevailed and the peasants were also rather free of seigneurial control, but where settlement was predominantly in large, tightly knit villages that controlled the agrarian activities of their residents; and the woodland regions of the Weald of Kent[4] and the Forest of Arden,[5] where settlements were relatively recent and very scattered and the peasantry largely free from the control of lords….

Murder appears to have been far more frequent in the counties of Kent and Warwick than anywhere else (see Table 2, pg. 36). Warwick was clearly the most violent. Kent was probably the next most violent.[6] The shire with the lowest homicide rate was Norfolk, which had a rate of only about 9/100,000 per annum. Bedford and Oxford came between these two extremes.

[1] The hundreds of Bampton, Banbury, Bloxham, Bullingdon, Chadlington, Ploughley, and Wootton have been included in the Oxford plains.
[2] The hundred of Kineton has been included in Felden Warwickshire.
[3] The hundreds of Binfield, Dorchester, Ewelme, Langtree, Lewknor, Pirton, and Thame have been included in the Chiltern Hills.
[4] The hundreds of Blackburn, Barkely, Cranbrook, Marden, East Barnfield, Rolvenden, Tenterden, and Selbrittenden have been included in the Weald.
[5] The hundreds of Barlichway, Hemlingford, Knightlow, and the Liberty of Pathlow have been included in the Forest of Arden.
[6] Although Table 2 does not indicate that Kent had homicide rates too noticeably higher than those of Oxford and Bedford, it should be remembered that those homicides committed in the Cinque Ports of Kent were not recorded in the eyre rolls. These additional killings, if their numbers were known, would push the homicide rates for Kent higher than those indicated in the table.”

here is given’s table 2:

given - table 2

to sum up the averages according to his estimates, we’ve got:

(Bristol – 4/100,000)
Norfolk – 9/100,000
(London – 12/100,000)
Oxfordshire – 17/100,000
Bedfordshire – 22/100,000
Kent – 23/100,000
Warwickshire – 47/100,000

to put these figures into perspective, the homicide rate for bristol is approximately what you find in albania or burundi today; the rate for london is like the rate for nigeria or nicaragua today; the rate for bedfordshire like the democratic republic of congo or brazil today; and the rate for warwickshire is something like belize or ivory coast (which i can’t believe are worse than the drc…). as given says [pg. 40]:

“[M]urder was a frequent phenomenon in medieval England. As has been pointed out above, the number of homicides in every twenty settlements oscillated between a high of almost 1.6, reported at the 1276 Bedford eyre and the 1232 Warwickshire eyre, and a low of 0.5, reported at the 1241 Oxfodrd eyre. In other words, there was a good possibility that there would have been a homicide in every settlement in these counties once every twenty to forty years. Therefore, it is possible that every person in England in the thirteenth century, if he did not personally witness a murder, knew or knew of someone who had been killed.

here is a map that i’ve made indicating the distribution of these averages of the homicide rates across england:

given thinks that the rate for kent ought to be higher since he had no data for the cinque ports which were located there (see footnote 6 above). i’m glad that these towns weren’t included, because i don’t think they’d tell us much about the regular population of kent. established in part as military towns, you’d think that they’d have attracted a rather rough crowd, not all of them from kent. in other words, a good portion of them would’ve been a self-sorted group — from who knows where.

oxfordshire (17), bedfordshire (22), and kent (23) all seem to be in a similar range. these counties are all in the lowland zone of england, and oxfordshire and bedfordshire were both heavily manorialized (per given), so — according the theory — we would expect to find a lot of outbreeding in these regions. and while it did not have large numbers of manors but, rather, lots of freeholds, kent had had probably the earliest secular laws prohibiting cousin marriage in england (from the 690s), so we shouldn’t be surprised if inbreeding had been avoided there for a long time as well.

norfolk (9) had extraordinarily low homicide rates. that county did have some manors, but not loads of them. and i have been guessing that they were, in fact, slight inbreeders given the closeness of their extended families — i’ve been guessing that they were some of my inbetweeners. unfortunately, i have not had any data on their mating patterns! perhaps they were extreme outbreeders. perhaps not. definitely need to find out more about the people in norfolk (east anglia)! they will be a test case.

what excited me about my little map there is that the population of warwickshire — in england’s intermediate zone, almost in the highland zone — had such high homicide rates. highlanders are normally inbreeders (maybe) — and so the population in warwickshire should’ve been more inbred than the lowlanders down in oxfordshire, etc. — and they, therefore, should also be more violent (according to the theory). so i thought that these numbers, maybe, fit the theory pretty well.

now i’m not so sure. i think there might be an even more interesting explanation!

much of the data for warwickshire comes from the forest of arden, that “desert inaccessible under the shade of melancholy boughs.” the arden was heavily colonized during the medieval period starting in the eleventh century, although there are some indications that there had been a few earlier settlements by the anglo-saxons before the conquest (see A Study of Medieval Colonization in the Forest of Arden, Warwickshire). much of the eleventh century settlement of the forest of arden was done by independent individuals and their families establishing their own homesteads — the area was not, at least initially, structured along the lines of manorialism. quite the reverse [pg. 4]:

“The pattern of settlement in the Arden remained that of a forest, slowly cleared and settled by individuals or families rather than by communities. A traditional open-field system had never existed in the Arden: much of the arable land had always been enclosed, and where open fields were present, their pattern was highly irregular. Enclosure continued throughout the seventeenth century, usually undertaken by gentry in co-operation with yeomen and richer husbandmen. Medieval Arden had had more freeholders and lighter labour services than the south of the country….”

the question then is: who were these individuals who settled the forest? where did they come from? if they came from further west than, or even from the north of, the arden, you’d think they’d have emanated from more inbred groups — if they came from the lowlands to the south or east, more outbred groups. i don’t have an answer for you (although it might be in this article which i don’t have access to just now).

what i was thinking, though, was that these were clearly a self-sorted group of people — individualists — who were happy to strike out on their own to seek their fortune in the world. really, they sound a bit like the type of people who settled the american west! perhaps, just like the individuals who settled the west, this was a bit of a rough crowd, and therefore they were more prone to explosive types of violence. don’t know. just speculating. (~_^)

and/ooorrr…maybe many of them came from even further afield than just areas neighboring the arden. in reading through A Study of Medieval Colonization in the Forest of Arden, Warwickshire, i noticed that an awful lot of the names mentioned in the rolls and registers of the time from arden appear to be NORMAN names!: Herbert, son of Dolfin (sounds norman to me); Thomas de Hawkeshawe; Henry de Ladbroke (norman?); William de Bereford; Hugh de Benetford; the Archers (how much more norman could you get?!); Roger Gerin; Philip Duruvassal; William de Barnvile. do these names sound norman to you? because they do to me, although i could be wrong.

if many of the settlers of the arden were norman, maybe this explains the high homicide rate. remember what gregory clark had to say about the normans [The Son Also Rises – pgs. 254-257]:

“Norman surnames are also significantly overrepresented in English armies in the years 1369–1453, more than three hundred years (ten generations) after the Norman Conquest. This was the period of the Hundred Years’ War, the long struggle between the French and English crowns for control of the English-held territories in France. The evidence on the composition of armies comes from surviving muster rolls, which list soldiers engaged in English armies in France, Scotland, Wales, and elsewhere.

“What is surprising, however, is the heavy concentration of Norman-derived surnames at all ranks of the armed forces. Even among the lowest ranks of the army, the archers, Norman surnames still show up at three or four times the frequency predicted by their population share. Archers were skilled workers, with wages comparable to artisans, but did not rank particularly high on the social scale. The preponderance of Norman surnames among them thus does not stem from the relatively high social status of these names: to the contrary, this should have led to Norman surnames’ being underrepresented in these ranks. Instead it seems to suggest that even ten generations after the conquest, the descendants of the Norman conquerors still had a taste and facility for organized violence. This hypothesis is supported by the share of knights and esquires in these armies with Norman surnames. This was 3–11 percent, much greater than the share of Norman surnames found in the more pacific realm of Oxford and Cambridge at the same time.

This particular concentration of Norman surnames in the realm of violence is not contemplated in the general theory of social mobility advanced here and thus represents an unexplained anomaly.

well, that’s all i’ve got for you today — a bunch o’ speculations. hope you don’t mind! (~_^)

i shall endeavor to find out more about the east anglians. i also wish that there were more homicide data for other regions of medieval england — maybe there are! i shall have to keep an eye out for those, too.

previously: outbreeding, self-control and lethal violence and kinship, the state, and violence

(note: comments do not require an email. the forest of arden.)