Файл: [D._Kim_Rossmo]_Geographic_Profiling(BookFi.org).pdf

© 2000 by CRC Press LLC

They applied two “navigational metrical tests” to the spatial and temporal

data associated with the crimes to confirm their deduction (Kind, 1987a).
The first test involved the calculation of the centre of gravity (spatial mean)
for the 17 crimes believed to be linked to the Yorkshire Ripper.

61

 The second

test consisted of plotting time of offence against length of day (approximated
by the month). The Advisory Team theorized the killer would be unwilling
to attack late at night if his return home was too far. The tests determined
the centre of gravity for the Ripper crimes lay near Bradford, and found the
later attacks were those located in the West Yorkshire cities of Leeds and
Bradford; these results supported the team’s original hypothesis that the killer
was local. The Advisory Group’s interim report recommended “a special team
of high-grade detectives be dedicated to enquiries in the Bradford area”
(Kind, 1987a, p. 390).

In January 1981, Peter William Sutcliffe was arrested by two patrol offic-

ers in Sheffield’s red-light district. Sutcliffe, a truck driver who later confessed
to being the Yorkshire Ripper, resided in Heaton, a suburb of Bradford. While
the resolution of the case occurred independent of the recommendations of
the Advisory Group’s report, their suggestion to focus on the Bradford area
was valid, and if made sooner, might have led to an earlier case closure and
helped save lives (Kind, 1987a).

Russian police used geography to set a trap for the USSR’s worst serial

murderer (Conradi, 1992). Dozens of women and children had been found
slashed and mutilated in the lesopolosa (forest strips) of the Rostov-on-Don
region over a period of 12 years (Cullen, 1993). Investigators believed the
offender travelled for employment and used local commuter trains to find
potential victims whom he lured into nearby woodland. Studies of aerial
surveillance maps were unsuccessful in determining the killer’s point of ori-
gin, and a computer biorhythm analysis of offence time, weekday, season,
weather, and location was inconclusive (Lourie, 1993). So in an effort to
better the odds of snaring the murderer along miles of rail line, Operation
Forest Strip deployed 360 officers to maneuver him into key target areas.
Uniformed officers guarded, in an obvious manner, all train stations except
three near the forested areas where it was felt the killer would likely strike.
These stations, and the surrounding lesopolosa, were surveilled by plain-
clothes officers. The strategy led to the identification and arrest in November
1990 of former schoolteacher Andrei Chikatilo. He was later convicted of 53
murders and executed by gunshot.

Newton and Newton (1985) applied “geoforensic analysis” to a series of

unsolved female homicides that occurred from 1983 to 1985 in Fort Worth,

61 

The analysis was based on 13 murders and 4 assaults, but the actual toll at the time was

13 murders and 7 attempted murders; the Byford Advisory Team missed one victim, and
included an unrelated murder.

© 2000 by CRC Press LLC

Texas. They found localized serial murder or rape forms place-time patterns
different from those in “normal” criminal violence. The unsolved Fort Worth
murders were analyzed with both quantitative (areal associations, crime site
connections, centrographic analysis) and qualitative (landscape analysis)
techniques.

Newton and Swoope (1987) utilized similar techniques in a retrospective

analysis of the Hillside Stranglers case. Geographic centres (spatial means)
were calculated for points of fatal encounter, body or car dump sites, and
victim residences. They found the centre of the body dump sites lay close to
the residence of murderer Angelo Buono, and that a search radius based at
this point decreased with the addition of each new crime location. A con-
spicuous void in the crime pattern became evident when the crime locations
were plotted on a map of Los Angeles; there was an area surrounding the
home of Buono within which no crimes occurred. Newton and Swoope point
out such a “coal-sack effect,” resulting from an offender avoiding criminal
activity close to home, has investigative significance.

Barrett (1990) documented several experience-based observations by

police regarding the connection between crime locations and area of offender
residence. If the murder and body dump sites are different, then the killer
generally lives in the area where the victim was attacked. Conversely, if the
victim was left at the murder scene then the killer is probably not local.

62

 A

crime scene close to a major road is an indication the murderer may not be
from the area, while a crime scene a mile or more from a major road suggests
the killer is local. A hidden body may mean the offender is more or less
geographically stable and wishes to reuse the dump site, while an unconcealed
body suggests the murderer is transient and unconcerned if police discover
the victim.

Detective Constable Rupert Heritage of the Surrey Police and Professor

David Canter, a psychologist at The University of Liverpool, have produced
offender profiles for several British criminal investigations. They note that
“one of the most successful components of the reports we have given to police
forces has been an indication of where the assailant might live” (Canter, 1994,
p. 283). Their first major inquiry was Operation Hart, set up to apprehend
the Railway Killer who raped and murdered several females in the Greater
London area from 1982 to 1986 (Lane & Gregg, 1992).

Blood typing reduced the register of over 5000 suspect names to 1999.

Number 1505 on this list was John Francis Duffy, a carpenter for British Rail.
The chronological pattern in the crime locations suggested a process of
learning and planning on the part of the offender. The profile reasoned the

62 

This observation is inconsistent with the earlier finding that disorganized killers, who

tend to live close to their crime sites, usually leave the victim’s body at the murder scene.
It may be that Barrett is referring to only organized offenders.

© 2000 by CRC Press LLC

earlier attacks were closer to the Railway Killer’s home and therefore he
probably lived in the region circumscribed by the first three offences — the
Kilburn/Cricklewood area of northwest London. This focus helped lead
police to Duffy who was subsequently sentenced to seven life terms (Copson,
1993). An element of luck played into the analysis as Canter (1994) later
admitted further research did not substantiate the “triangle hypothesis.”

In another case, Heritage and Canter located the home area of the Tower

Block Rapist in Birmingham (Canter, 1994). Responsible for a number of
sexual assaults on elderly women from 1986 to 1988, Adrian Babb hunted in
blocks of high-rise apartments situated in the Edgbaston and Druids Heath
districts, on the edge of the city centre. The attacks revealed a strong sense
of area familiarity, but at the same time, confidence in anonymity. This
suggested the offender resided close by, but not in the immediate neighbour-
hood. Furthermore, the rapist’s comfort in prowling through the victims’
high-rise apartment buildings, labelled “streets in the sky,” indicated he lived
in such a structure himself. The target area was a patchwork of distinct
territories divided by Birmingham’s busy arterial routes, and this pattern,
informed by victimology and demographic information, was used as the basis
for inferring the offender’s mental map. The analysis correctly predicted the
location of Babb’s home base in nearby Highgate.

It should not be forgotten that criminals also use maps.  Both David

Berkowitz, the Son of Sam, and Sylvestre Matuschka, the Hungarian train
wrecker, plotted their future crimes on maps.  The Zodiac Killer mailed a
code with a map marking where his next  bomb was planted to 

The San

Francisco Chronicle

.  While he shot and stabbed at least 8 victims in the Bay

area, murdering 6, Zodiac claimed 37 kills (Graysmith, 1976).  His confirmed
attacks occurred from 1966 to 1969, but his correspondence with newspapers,
police, and others stretched over 12 years.  A letter to the 

Chronicle

 newspaper

postmarked June 26, 1970, included a cipher and a map providing clues to
the location of a bomb intended for a school bus:  “The Map coupled with
this code will tell you where the bomb is set.  You have untill next Fall to dig
it up” (King, 1996, p. 302).  A further hint was provided on July 26, 1970:
“P.S.  The Mt. Diablo Code concerns Radians & # inches along the radians”
(King, p. 305).

Zodiac’s Phillips 66 service station roadmap focused on Mt. Diablo

(“Devil’s Mountain”) in Contra Costa County, across the Bay from San
Francisco.  Mt. Diablo was used to help plot latitude and longitude after the
Civil War.  A compass symbol was hand drawn in the middle of the map,
the centre of which contained the Naval Radio Station (Graysmith, 1976).
The bomb was never found, the mystery of the map never solved, and the
Zodiac Killer never apprehended.

© 2000 by CRC Press LLC

10.3 Offender Residence Prediction

As early as 1986, LeBeau recognized the investigative potential of geostatis-
tical techniques and crime pattern research for reducing the offender search
area in rape cases. Until the development of geographic profiling in 1990,
however, there was no tested systematic method beyond centrography for
approaching this problem. Taylor (1977) states that geographic patterns
should be viewed through the processes that produce them. Accordingly,
crime pattern theory was utilized as a heuristic for the construction of an
algorithmic model for locating offender residence (see Benfer et al., 1991).
The research of Brantingham and Brantingham interprets offenders’ activity
spaces in order to describe where crimes are most likely to occur. Geographic
profiling is, in effect, an attempt to invert this model, by using crime locations
as the basis for predicting most probable area of offender residence or work-
place. So while the two models have different purposes and inputs, their
underlying concepts and ideas are similar.

10.3.1 Criminal Geographic Targeting

In the simplest case, offenders’ residences lie at the centre of their crime
patterns and can be found through the spatial mean. The intricacy of most
criminal activity spaces, however, indicates that more complex patterns are
the norm. George Rengert (1996) proposes four hypothetical spatial patterns
for the geography of crime sites: (1) uniform pattern, with no distance-decay
influence; (2) bull’s-eye pattern, exhibiting distance decay and spatial clus-
tering around the offender’s anchor point; (3) bimodal pattern, with crimes
clustered around two anchor points; and (4) teardrop pattern, centred around
the offender’s primary anchor point, with a directional bias towards a sec-
ondary anchor point. Crime patterns are also distorted by a variety of other
real world factors — movement follows street layouts, traffic flows affect
mobility, variations exist in zoning and land use, and crimes cluster depen-
dent upon the nature of the target backcloth. The spatial mean is therefore
limited in its ability to determine criminal residence.

Many researchers have noted the importance of direction as well as

distance in the analysis of spatial patterns of criminal behaviour. Rengert and
Wasilchick (1985) found a directional bias towards burglars’ workplace and
recreation sites. Canter and Hodge (1997) noticed that while crimes of U.S.
and British serial killers generally grouped around their homes, they were
also biased towards other activity sites. In their Sheffield study, Baldwin and
Bottoms (1976) observed that crime disproportionately occurs in the city
centre, indicating offender preference for such a bearing. Nonparametric
assessments of spatial autocorrelation in the orientation of criminal travel

© 2000 by CRC Press LLC

suggests that directional information can assist in certain police investigations
(Costanzo, Halperin, & Gale, 1986).

Environmental criminology establishes a framework within which jour-

ney-to-crime research, centrography, and other geographic principles may
be combined to create a method for determining offender residence from
crime locations. Set theory provides a useful first approach for addressing
this problem (Taylor, 1977). The ATF/FBI research on serial arsonists found
that 70% set fires within 2 miles of their home. 

Figure 10.1

 shows a Venn

diagram for three hypothetical serial arsons.

63

 The medial circles surrounding

each crime location are defined by a radius equal to journey-to-crime dis-
tance 

d

, within the range of which percentage 

p

 of the offender’s arsons occur

(

d

 = 2 miles; 

p

≥

 0.70). The probability of the offender’s residence lying within

the area circumscribed by a single circle is therefore also 

p

. Because the crimes

are connected, the lune overlap areas between any two circles is more likely
to contain the offender’s residence. The highest probability is in the middle
region where all three of the circles intersect.

The residence of the offender is most likely within (in decreasing order

of probability): (1) the middle intersection; (2) the lunes; (3) the circles; and
(4) the background area. The areal probabilities for the four different spaces

Figure 10.1

    Journey-to-crime Venn diagram.

63 

This procedure is akin to constructing an intersection of investigative frames through

the use of geographic information (Kind, 1987b).