A.-M. Guerry's Moral Statistics of France: Challenges for Multivariable Spatial Analysis

1.3 MB·English

by  Michael Friendly

StatisticalScience
2007,Vol.22,No.3,368–399
DOI:10.1214/07-STS241
(cid:13)c InstituteofMathematicalStatistics,2007
A.-M. Guerry’s Moral Statistics of
France: Challenges for Multivariable
Spatial Analysis
Michael Friendly
8
0
0
2
Abstract. Andr´e-MichelGuerry’s(1833)EssaisurlaStatistiqueMorale

n de la France wasoneofthefoundationstudiesofmodernsocialscience.
a
Guerry assembled data on crimes, suicides, literacy and other “moral
J
statistics,” and used tables and maps to analyze a variety of social is-
8
2 sues in perhaps the first comprehensive study relating such variables.
Indeed, the Essai may be considered the book that launched mod-
]
E ern empirical social science, for the questions raised and the methods
M Guerry developed to try to answer them.
Guerry’sdataconsistofalarge numberofvariables recordedforeach
.
at of the d´epartments of France in the 1820–1830s and therefore involve
t both multivariate and geographical aspects. In addition to historical
s
[ interest, these data provide the opportunity to ask how modern meth-

1 ods of statistics, graphics, thematic cartography and geovisualization
v can shed further light on the questions he raised. We present a variety
3
of methods attempting to address Guerry’s challenge for multivariate
6
spatial statistics.
2
4
Key words and phrases: History of graphics, crime mapping, biplot,
.
1 multivariate visualization, moral statistics.
0
8
0
: 1. INTRODUCTION with the nearly simultaneous work of Adolphe
v
Quetelet(Quetelet(1831),1835)inBelgium,Guerry’s
Xi On July 2, 1832 a slim manuscript was presented Essai (published in 1833) established the scientific
totheAcad´emieFranc¸aisedesSciencesbythe29year
r study of “moral statistics” in Europe and became
a old lawyer Andr´e-Michel Guerry titled Essai sur la
the launching pad for much of modern social sci-
Statistique Morale de la France. Guerry’s findings
ence: criminology and sociology in particular.
were both startling and compelling. His presenta-
Guerry’s results were startling for two reasons.
tion, in tables and cartes figuratives, of statistical
First he showed that rates of crime and suicide re-
dataoncrime,suicideandothermoralaspects,mea-
mained remarkably stable over time, when broken
sured only recently in France, broke new ground in
down by age, sex, region of France and even sea-
thematic cartography and data visualization. Along
son of the year; yet these numbers varied system-
atically across d´epartements of France. This regu-
Michael Friendly is Professor, Psychology Department, larity of social numbers created the possibility to
York University, Toronto, Ontario, M3J 1P3 Canada conceive, for the first time, that human actions in
e-mail: [email protected]. the social world were governed by social laws, just
as inanimate objects were governed by laws of the
This is an electronic reprint of the original article
physical world. By extension, these laws could be
published by the Institute of Mathematical Statistics in
Statistical Science, 2007, Vol. 22, No. 3, 368–399. This uncovered by the careful collection and analysis of
reprint differs from the original in pagination and social “facts,” meaning numbers. Second, he over-
typographic detail. turned some widespread beliefs about the nature
1
2 M. FRIENDLY
and causes of crime and its relation to other observ- three,” a:b as c: ?, to make proportional compar-
ablefactors,suchaseducationandpoverty.Overhis isons; but by these means political arithmeticians
lifetime, he completed three major works on moral were able to establish a basis for comparisons over
statistics, winning the Montyon prize in statistics geographic region, time, age and other categories
twice from the Acad´emie Franc¸aise des Sciences. (Klein (1997)).
His last work (1864) contemplated multivariate ex- TheBillsofMortality werebasedonparishrecords
planations of relations among moral variables, at a of christenings and deaths, recorded nearly weekly
time well before the development of correlation and and with at least a modicum of uniformity. In 1710,
regression. Yet Guerry’s contributions to statistics, John Arbuthnot (1710), a Scottish minister and
graphics and the rise of modern social science in the physician to Queen Anne, calculated the ratio of
early 1800s are neither well known nor widely ap- male to female births from these records for 1629–
preciated outside criminology and sociology. 1710 and observed that the ratio was consistently
This paper recounts Guerry’s work, the questions greater than 1 (see Figure 1). He used this law-
he asked and the methods he used to answer them ful regularity to argue that divine providence, not
in relation to his place in the history of data visu- chance, governs the sex ratio, in perhaps the first
alization and statistics. To do so, we first describe application ofprobability tosocialstatistics andthe
thecontext in whichheworked andwhich ledto the first formal significance test.
rise of the moral statistics movement in Europe. A By the mid-1700s, the importance of measuring
second section describes his life and works and the and analyzing population distributions and the idea
methods he introduced to the study of moral statis- thatethicalandstatepoliciescouldencouragewealth
tics. Guerry worked with voluminous data on social through population growth was established, most
variables, distributed over time and space notably by Johann Peter Su¨ssmilch (1741), who ad-
(d´epartements of France, counties of England), and vocatedexpansionofgovernmentalcollectionofpop-
finely categorized along numerous dimensions (age, ulation statistics. Data on the social character of
sex and status of accused, detailed breakdown of human populations was still lacking, however.
types of personal and property crime, motives for IntheperiodleadinguptoandthroughtheBour-
suicide, etc.). The final sections attempt some re- bonRestorationfollowingNapoleon’sdefeatin1815,
analyses of Guerry’s data to address the challenges crime was a serious concern, particularly in Paris,
posed for modern multivariate and spatial analysis. which witnessed an explosive growth in population,
along with widespread inflation and unemployment,
1.1 The Rise of “Moral Statistics” and Modern
and the emergence of an impoverished, dangerous
Social Science
class of petty criminals (les mis´erables); see Beirne
The empirical and quantitative study of factors (1993a) and Chevalier (1958). Then, as now, there
affecting human society such as education, crime were two basic schools of thought regarding crimi-
and poverty that gave rise to modern social sci- nal justice policy and much debate about the treat-
ence began between 1825 and 1835, with the work ment of prisoners. A liberal, philanthrope position
of Andr´e-Michel Guerry and Adolphe Quetelet. But advocatedincreasededucation,religiousinstruction,
the roots of this endeavor and the very possibility improved diet (bread and soup!) and better prison
of observation-based laws governing human popula- conditions as the means to reduce crime and recidi-
tions go back much further. vism. Hard-line, conservatives feared attempts at
The systematic study of social numbers, at first prisonreform,doubted theefficacy of campaigns for
concerning population data and dynamics, began in publiceducationandviewedsuggestionstoabandon
the 1660s with John Graunt’s (1662) and William the harsh punishment of convicts under the ancien
Petty’s (1665) analyses of the London Bills of Mor- r´egime withgravesuspicionifnotalarm.Buttheev-
tality. This work showed how such numbers could idencemarshalledtosupportsuchrecommendations
inform the state about matters related to popula- was fragmentary, restricted and often idiosyncratic.
tion growth, age-specific mortality, ability to raise See Whitt (2002, pages xxvi–xxxi), Beirne (1993b)
an army, the consequences of plagues, wealth, taxes andPorter(1986,pages27–30)formorebackground
and soforth. “Political arithmetic,” as it was called, onthesocialcontextinwhichtheEssai waswritten.
was based only on the simple ideas of standardiz- Thischangedin1825whentheMinistryofJustice
ing raw numbers by relevant totals and the “rule of in France instituted the first centralized, national
GUERRY’SMORAL STATISTICS OF FRANCE 3
Fig. 1. Arbuthnot’s data on the male/femalesex ratio. The average, shown by the upper dashed lineis1.07; the thicker line
shows aloesssmooth. Theprobability of ratios greater than 1over82 years is(1)82=2×10−25 under the nullhypothesis that
2
the probability of a male birth is 1/2.
system of crime reporting, collected quarterly from During the same period, a wealth of other data
every d´epartement and recording the details of ev- on moral and other variables becameavailable: data
ery criminal charge laid before the French courts: on age distributions and immigrants in Paris began
age, sex and occupation of the accused, the nature with the 1817 census; Alexander Parent-Duchatelet
1
of the charge and the outcome in court. Annual (1836) provided comprehensive data on prostitutes
statistical publications of this data, known as the in Paris, by year and place of birth; the Ministry of
Compte G´en´eral de l’Administration de la Justice
War began to record data on conscripts who could
Criminelle en France, began in 1827 under the ini-
read and write; information on wealth (indicated
tiative of Jacques Guerry de Champneuf, the direc-
by taxes), industry (indicated by patents filed) and
tor of affaires criminelles in the Ministry of Justice
even wagers on royal lotteries became available for
2
(Faure, 1918, pages 293–294).
the d´epartements of France in various bulletins of
the Ministry of Finance, 1820–1830. Thus,the stage
1This extended a model begun in Paris in 1821 with an- was set by this “avalanche of numbers” (Hacking
nual publications of the Recherches Statistiques sur la Ville (1990)) for someone to try to make sense of com-
de Paris et leD´epartement de laSeine underthedirectionof
peting claims about the causes of crime from com-
Jean Baptiste Joseph Fourier (1768–1830) toward the end of
prehensivedataanddetailedanalysis.Andr´e-Michel
hislife.Thesevolumesdetailedbirths,marriagesanddeaths,
butalsoprovidedextensivetabulationsandbreakdownsofin- Guerryhappenedtobeintherightplaceattheright
mates of Parisian insane asylums, and of motives and causes time, but more importantly, he had a passion for
of suicides. See Hacking(1990, pages 73–77).
numbers and his quest for taming it would nearly
2Guerry de Champneuf was said to be related to Andr´e-
occupy his entire professional life.
Michel, by a contemporary reviewer of the Essai (Caunter
(1833)); Hacking (1990, page 77) calls him a cousin. There
is no evidence to support a family relation. The similarity
of names caused some confusion among American sociolo- two into the siamese twin, M. de Guerry de Champneuf, or
gists, starting with M. C. Elmer (1933), who conjoined the M. de Champneuffor short.
4 M. FRIENDLY
2. GUERRY’S LIFE, WORK AND METHODS early studies concerned physiological characteristics
(e.g., pulse rate) of inmates at insane asylums and
Unlike Quetelet, a brilliant academic politician
prisons. Toward the end of his career, he invented
andeffectiveself-promoterwhoachievedprominence
a calculating or tabulating device (the ordonnateur
in academic and social circles throughout Europe
statistique)toaid thework onthedatafromhislast
andwhoselifehasbeenwidelybiographed,Guerry’s
workandmagnumopus(Guerry(1864)),thedetails
fame in his lifetime, like his life itself, was more 5
of which remain shrouded. Over his career, he pro-
modest. Aside from brief, bare-bones entries in the
duced the three major works on moral statistics de-
GrandDictionnaireUniversel (Larousse(1866))and
scribed below. General discussion of his methods of
similar sources (Carr´e de Busserolle (1880);
analysis follows in Section 2.4.
Vapereau (1858)), the primary sources on Guerry’s
life are the seven-page necrology by Alfred Maury, a 2.1 1829: Statistique Compar´ee de l’E´tat de
long-timefriendandfellowmemberoftheAcad´emie l’Instruction et du Nombre des Crimes
des Sciences Morales et Politiques, read at his fu-
Guerry’s first publication on moral statistics was
neral in April, 1866, and notices on Guerry’s work
a large, single-sheet set of three shaded maps com-
byHypolyteDiardandErnestVinet.Thesewereini-
paringcrime andinstruction titled Statistique Com-
tially publishedseparately in the month of Guerry’s par´ee de l’E´tat de l’Instruction et du Nombre des
death and then printed together in Diard (1867).
Crimes produced together with the Venetian ge-
Secondary sources include Whitt’s (2002) preface
ographer Adriano Balbi (Balbi and Guerry (1829)),
to the translation of the Essai, Beirne (1993b) and
shown here in Figure 2. The data on crime from the
a scattering of brief mentions, often in relation to Compte G´en´eral of 1825–1827 were combined with
Quetelet, by criminologists, sociologists (Lazarsfeld data from the census to give measures of population
(1961); Isambert (1969)) and historians (Hacking per crime (number of inhabitants to give one con-
(1990); Porter (1986)). demned person) for 81 d´epartements; the data on
Guerry was born in Tours on December 24, 1802; instruction are based on the number of male chil-
hisbirthcertificatelistshisfather,MichelGuerry,as dren in primary schools in 26 educational districts
a public works contractor, and the indications from (cours royales and acad´emies) in France, also in the
historical sources are that his family circumstances form of inhabitants per student.
were comfortable though neither wealthy nor highly
connected. He studied law, literature and physiol-
this graphical form, using sectors of fixed angle, but varying
ogy at the University of Poitiers, and was admitted radii to show frequency of some events, typically for cyclic
to the bar in Paris, where he became a Royal Advo- phenomena. Guerry’s plate shows six such diagrams, all for
cat (Diard (1867)). In 1827 he began to work with dailyphenomena.Fouroftheseshowdirectionofthewindin
8 sectors, for 3-month periods; two show births and deaths,
the Compte G´en´eral in the course of his duties with
respectively, by hour of the day. Guerry says that these rep-
the Ministry of Justice. He became so fascinated by
resentjustapartofhisoriginal,muchlargerdiagrams,which
these data that he abandoned active practice in law were not at first designed to bepublished.
to devote himself to their analysis, a task he would 5The ordonnateur statistique is simply mentioned in pass-
pursue until his death in 1866. Sadly, no details of ingbyLarousse;Maury(page5)saysthemachinewasoffered
3 byGuerry’sheirstotheConservatoiredesArtsetM´etiers. It
his personal or family life are available.
is possible that this device briefly came into the hands of
One early statistical work (Guerry (1829)) exam-
Maurice d’Ocagne, Professor at the ENPC and principal de-
inedtherelationbetweenweatherandmortalityfrom
veloper of nomography, but almost certain that he observed
variousdiseases,andcontainedgraphsofadmissions it when he discovered the collection of calculating machines
tohospitalsandpolarareadiagramsofthevariation held by the Conservatoire. Ocagne presented several lectures
of weather phenomena, by month and hour4; other at the Conservatoire in February and March 1893 titled “Le
CalculSimplifi´eparlesProc´ed´esM´ecaniquesetGraphiques,”
but there is no mention of Guerry in the papers printed in
3Guerryhadnosiblings,henevermarriedandhadnochil- the Annales du Conservatoire (d’Ocagne (1893)). There is a
dren. Nothing is yet known about his personal life in Paris. curious connection with IBM here: IBM France introduced
However,hisfamily haddeep roots in thearea around Tours thetermordinateur toreplacethedeprecatedfranglais term,
that have now been traced back to the early 1600s (Friendly computeur; Ocagne also studied another collection of early
(2007c)). calculating devices belonging to General Sebert, purchased
4These 1829 polar area charts predate those by Florence later by IBM France, which may also have acquired others.
Nightingale (1857), who is widely credited as the inventor of The archivists at IBM can findno records of these.
GUERRY’SMORAL STATISTICS OF FRANCE 5
Fig. 2. Guerry and Balbi’s1829 StatistiqueCompar´eedel’E´tatdel’InstructionetduNombredesCrimes. Topleft:crimes
against persons; top right: crimes against property; bottom: instruction. In each map, the d´epartements are shaded so that
darker is worse (more crime or less education). The legend at the lower left gives the data on which the maps were based.
Source: Courtesy of BNF; Palsky (1996, Figure 19).
The exact source of the data on instruction is un- lowerlevelofinstructioninthesouthofFrancecom-
clear. The legend for the maps cites the Ministry of pared with the north; he referred to this as the
Instruction for 1822, but Guerry’s printed commen- contrast between la France obscure and la France
tary on this work (republished in Guerry (1832)) ´eclair´ee. Quite shortly, Baron Charles Dupin, per-
cites Balbi’s 1822 Statistique du Royaume de Portu- haps inspired by this observation, thought to por-
gal... as the first document that published a table tray these data on a map of France (Dupin (1826),
of data on the level of public education in France. 1827), using shades of varying darkness to depict
What is clear is the impact they had on Guerry degrees of ignorance. This invention—the first mod-
and others. In 1823 the geographer Conrad Malte- ernstatistical map—was thestarting pointof atrue
6
Brun (1775–1826) in commentary on Balbi’s ta- graphical revolution that Guerry would extend to a
ble remarked that there appeared to be a much more general social cartography with the compara-
tive analysis of social issues. The legacy of this rev-
6Journal des D´ebats, 21 jul 1823, pages 3–4. olution is commonplace today, in maps of disease
6 M. FRIENDLY
incidence, poverty, child mortality, income distribu- (condamn´es). Here, he argued persuasively that the
7
tion and so forth (Friendly and Palsky (2007)). number of indictments was a more useful indicator
Guerry and Balbi’s Statistique Compar´ee... in of the numberof crimes committed becauseit is less
1829 was the first use of shaded maps to portray likely to be influenced by the factors that affect ju-
crimerates. Theirpresentation is alsonotableinthe ries: the nature of the crime, severity of punishment
history of statistical graphics as the first to com- and the place where the accused is judged. More-
bine several moral variables in a single view, al- over, although an indictment by no means implies
lowing direct comparison of crimes against persons the guilt of the accused, it does reasonably imply
andagainstpropertywithdataoninstructionacross that a crime was committed; conversely, a person
the d´epartements of France. They suggested that, might be acquitted for a variety of reasons, but that
(a)surprisingly,personalcrimesandpropertycrimes does not mean that a crime did not take place.
seemed inversely related overall, but both tended to The Essai published in 1833 contained numerous
be high in more urban areas; (b) a clear demarca-
tables giving breakdowns of crimes against persons
tion could be seen in instruction between the north
and property by characteristics of the accused, fre-
and south of France along a line running northeast
quencies of various subtypes of crime in rank order
8
from Geneva (Ain) to Coˆtes duNord ; (c) the north
forbothmenandwomen(formen,themostpopular
of France not only showed the highest levels of ed-
personal crime was assault and battery; for women,
ucation, but also of property crime. At the very
infanticide)andfrequenciesof crimesbyage groups.
least, this work testified to the importance of de-
To go beyond simple description, Guerry classi-
tailed data, sensiblypresented, toinformthe debate
fied the crimes of poisoning, manslaughter, murder
on the relations of crime and education.
and arson according to the apparent motive indi-
2.2 1833: Essai sur la Statistique Morale de cated in court records (for poisoning, the motive
la France was most frequently adultery; for murder, it was
hatred or vengeance). This quest to examine mo-
Over the next three years Guerry would occupy
tives and causes is most apparent and impressive in
himself with the extension and refinement of these
his analysis of suicide, a topic of considerable de-
initial results,with extensive tabulation ofnewdata
bate in both the medical community (which con-
from diverse sources and with answers to method-
sidered it in relation to madness and other mal-
ological questions that might lead to challenges to
adies) and the legal community (which considered
his conclusions.
whether it should be a crime or at least within the
Onthemethodologicalside,hediscussedhowthese
purview of the justice ministry). “What would be
measures should be defined to ensure comparability
useful to know would be the frequency and impor-
across France and what we would now term valid-
ity of the indicators used. For education, for exam- tance of each of these causes relative to all the oth-
ple, he considered the reported levels of instruction ers.Beyond this,itwould benecessary to determine
(number of male children in primary school) to be whether their influence ...varies by age, sex, edu-
suspect due to variations in local reporting; after cation, wealth, or social position” ((Guerry, 1833,
1827, better and more uniform data became avail- page 131, WR trans.)). To this end, he carried out
ablefromtheMinistryofWar,whoseexams fornew perhaps the first content analysis in social science
recruits gave numbers for those who could read and by classifying the suicide notes in Paris according
write. to motives or sentiments expressed for taking one’s
Asecondmajorquestionheaddressedwaswhether life. This approach to the study of suicide would
crimes should be counted by the number of indict- later be adopted by Durkheim (1897), but without
ments (accus´es) or by the number of convictions muchcredittoGuerryandothermoralstatisticians.
TheEssai alsocontainedacollectionofbargraphs,
7See www.worldmapper.org for a collection of over 300 highlightingcertaincomparisons(crimesagainstper-
world cartograms, where territories are resized according to sons occurred most often in summer months, while
thesubject of interest. those against property were most frequent in the
8This sharp cleavage between France du Nord et Midi
winter;suicidesbyyoungmalesweremostoftencar-
or France obscure vs. France ´eclair´ee would become reifined
ried out with a pistol, while older males preferred
as the “Saint-Malo–Geneva line” and generate much debate
hanging).Aswell,todiscussgeographicaldifferences
about causes and circumstances through the end of the 19th
century. and relate these moral variables to each other, he
GUERRY’SMORAL STATISTICS OF FRANCE 7
Fig.3. Guerry’s1833mapoflevelsofinstructioninFrance(Plate III).Theoriginalcontainsthenumbers(percentofyoung
men who can read and write) for each d´epartement in rank order listed below the map. Source: Author’s image collection.
prepared six thematic maps of France, adding ille- and write. Figure 4 shows all six of Guerry’s maps,
gitimatebirths(infants naturelles),donationstothe reproduced from his data using modern software.
poor (number of gifts and bequests) and suicide to Guerry’s Essai was received with considerable en-
those of personal crime, property crime and educa- thusiasminEuropeanstatistical circles,particularly
tion presented earlier, but based on more complete inFranceandEngland.InFranceitwasawardedthe
data and better indicators. For ease of comparison, PrixMontyon in1833andthepublicationincludesa
thesevariableswereallexpressedinaformsuchthat laudatoryreportonitscontentstotheAcad´emiedes
“more is better,” for example, population per crime Sciences. Guerry was also elected to the Acad´emie
or percent able to read and write. In preparing the desSciencesMoralesetPolitiquesandatsomepoint
maps, these variables were first converted to ranks was awarded the cross of chevalier of the Legion
and then the d´epartements were shaded according of Honor (Diard (1867)). The Essai was reviewed
to rank, so that darker tints were applied to the quite favorably in the Athenaeum (Caunter (1833))
d´epartements that fared worse on a given measure and the Westminster Review (Anonymous (1833)),
(more crime, less education). which gave a lengthy discussion of Guerry’s findings
Thesemapsaregenerally moredetailed andfinely and praised the book as one of “substantial inter-
drawn than those produced in 1829. Figure 3 shows est and importance.” Henry Lytton Bulwer’s (1834)
anexample,themaplabeled“Instruction,”butshow- France, Social, Literary, Political devoted 26 pages
ingthepercentofmilitaryconscriptswhocouldread to Guerry’s results,calling the work “remarkable on
8 M. FRIENDLY
Fig. 4. Reproduction of Guerry’s six maps. Color coding, as in Guerry’s originals, is such that darker shading signifies
worse on each moral variable. Numbers for each d´epartement give the rank order on that variable.
many accounts.” Guerry displayed the maps in sev- almost invariably in tables, is concerned with the
eral expositions in Europeand, in 1851, had two ex- numerical exposition of facts; thelatter presents the
hibitions in England—an honored public one in the successive transformation of these facts, by calcula-
Crystal Palace at the London Exposition and a sec- tion,byconcentration andtheirreductiontoasmall
ond at the British Association for the Advancement numberof generalabstract results.Onecan seehere
of Science (BAAS) in Bath. a thorough explanation of the graphic method ap-
plied to moral and social data.
2.3 1864: Statistique Morale de l’Angleterre
Fifteen of the plates show data for the
Compar´ee avec la Statistique Morale de la
d´epartements of France (from 1825 to 1855) or the
France...
countiesofEngland(1834–1856)onaparticulartopic,
Guerry’s most ambitious work, and the capstone first for France, then for England: crimes against
of his career, did not appear for another 30 years, persons, crimes against property, murder, rape,
but it was well worth waiting for. The Statistique larceny byservants (vol domestique), arson,instruc-
Morale de l’Angleterre Compar´ee avec la Statistique tion and suicide (only for France). In each case, to
Morale de laFrance waspublishedinagrandformat ensure comparability of the numbers for the vari-
(56×39 cm,aboutthesizeofalargecoffee table); it ous crimes across d´epartements and counties, and
containsanintroductionof60pagesand17exquisite from one measure to another, Guerry standardized
colorplates.TheintroductionsetsoutGuerry’sview the rates for each map and country to “degree of
of the history of the application of statistics to the criminality,” with a mean of 1000 and common (un-
moralsciences.Init,heproposestoreplacetheterm specified) metric. Thus, one could easily see where
“moral statistics” or simply documentary statistics Paris/Seine or London stood on murder vs. suicide
with “analytical statistics.” The former, presented or compare one to the other on theft.
GUERRY’SMORAL STATISTICS OF FRANCE 9
Fig. 5. Guerry’s 1864 Plate 1: Crimes against persons in France. Source: Courtesy of Staatsbibliothek zu Berlin.
Each of these plates exemplifies the program of rank order on the variable, the highest (rank=1)
statistique analytique that Guerry had in mind, as shaded darkest and the lowest shaded lightest. A
illustrated by Figures 5 and 6. The map of England large variety of special symbols and annotations are
or France shows the geographic distribution, with used on the map to indicate noteworthy patterns or
counties or d´epartements shaded according to their circumstances, for example, up or down arrows to
10 M. FRIENDLY
Fig. 6. Guerry’s 1864 Plate 2: Crimes against persons in England. Source: Courtesy of Staatsbibliothek zu Berlin.
show increase or decrease over time in a geographic viations within these data. Thus, surrounding each
unit. The table below the map lists the ranks and map, he placed a variety of line graphs designed to
data values, expressed as “degree of criminality.” decompose or transform these overall facts or to re-
Each map is an overall summary for 30 years, late them to other factors. Most of these featured
for all accused and for all crimes in a given class. time series graphs of trends over time, often decom-
Guerry wanted also to show patterns, trends or de- posed into separate series by subtype of crime or