"There is a serious danger that fire policy will be developed on the basis of work carried out in the context of the market place rather than being underpinned by research which has been subjected to full process of academic rigour and peer review" Professor D Drysdale (European Vice-Chair, International Association of Fire Safety Sciences) and D T Davis (Chair of the Executive Committee, Institution of Fire Engineers). Fire Engineers Journal 61, 10, 6-7


















Useful links

Research Advice

About us

Contact us


Bethke, E. (2002) Geographic profiling of Peterborough arson cases of 2000 and 2001: preliminary data analysis and manual crime mapping, dissertation for BA (hons) Department of Forensic Science; APU, Cambridge. www.fitting-in.com/bethke.htm.



In this paper, the arson data of malicious fires in the years 2000 and 2001 in Peterborough’s Central Ward area was analysed with the main focus on geographical distribution of the crimes.  This was done by organising the data into various relevant subsections – months, day of the week and time of the day.  After having completed that task, the arson data could be plotted onto a street map of the search area to visualise the sites.  The next step entailed a detailed examination of these subsections and crime maps for any noticeable arson centres, i.e. “hot-spots”, and/or for any distinct arson patterns, i.e. crime series.  The areas main arson centres and high-risk times, as well as a few potential serial cases, were determined through this investigation.


The methodologies of computerised crime incident analysis were discussed in their applicability and usefulness in arson prevention and reduction initiatives.  In this context, the data requirements for the CrimeStat Analysis Programme were identified.  This led to the conclusion that the data was not detailed enough, making even the map plotting quite awkward.


As a final point, the organisation of police patrols around “hot-spot” areas and high-risk times was discussed, in conjunction with the hopefully increased deterring effects through police presence.






This paper was a product of many people’s guidance and assistance.  I truly appreciate the encouragement I received from everyone and would like to take this opportunity to express my gratitude.


Special thanks go to my Project Supervisor, Mr. David Hughes, and my link to the Fire and Rescue Service, Dr. David Baigent, for their continuous support and advice throughout the course of my research.


From the Huntingdon Fire Station, I want to thank ADO Kevin Smith for helping me to focus my research on a more valuable study area and for putting me in contact with the relevant fire personnel.


Without any data, there would be no project – so, I would like to thank Mr. Wayne Law and Mr. Keith Shillam at Fire Headquarters in Huntingdon for providing me with the necessary arson data and street maps of Central Ward, Peterborough.


Finally, I especially want to thank my family for tolerating all the time spent in my “world of arson” during the holidays, and to my friends for unfailingly managing to distract me from my work – it helped, honest!!!






Useful links

Research Advice


About us

Contact us




Committing arson is a very serious offence, which is why it is imperative to understand what exactly it involves and to be aware of the extent of its danger. 

Arson is known as the criminal and deliberate act of setting fire to another person’s property – be it a building, other outdoor structures or vegetation – frequently causing considerable damage.  This can come in the form of financial loss, physical injury and even death.  Malicious fires are the largest single cause of all major fires and its numbers have doubled over the past decade, costing just under £2.1 billion per year1.  Another difficulty concerning the detection of arson is that evidence dealt with at a fire scene is very different from that found at scenes of other types of crime, making arson one of the most under-investigated offences.  This is an extremely worrying dilemma, as fire-setting is not perceived as a high profile problem.  Different aspects of investigated arson cases fall into the jurisdiction of various agencies (i.e. fire service, police and insurers)2.  Through the lack of communication and cooperation between the different services, this results in the fact that no one agency is aware of the entire problem.


On the rare occasion that an arsonist is caught by the police and tried for the offence he/ she committed, the Criminal Damage Act (1971) currently deals with the case.  Any arson attack that endangers life is punishable by maximum sentence of life imprisonment.  Prior to this act, arsonists were dealt with under the Common Law.  In specific cases, involving mental impairment of the accused, the Mental Health Act (1983) is called upon.


Historically, arson has never been as big a problem as it is now.  Less than 0.5% of all attended fires were caused deliberately in the 1950s.  This has progressively risen, so that by the 1990s 40% of all fires were malicious3.  In 1999, the fire service attended 935,500 fires in the UK, which is an 8% increase from the amount of fires in 1998.  Of the total number of fires, 102,900 were caused deliberately.  This high amount of arson fires can be partially explained by the rise in vehicle fires4.  For the UK, the 1999 statistics show the following distribution of malicious fires:

·        Dwellings – 13,800 cases (4% increase)

·        Other buildings  - 17,900 cases (5% increase)

·        Road vehicles – 63,200 cases (30% increase)

·        Other outdoor structures – 7,900 cases


In Cambridgeshire, the statistics for 1999 show that the total number of fires was 3,981, of which 1,188 cases were identified as malicious fires4.  The amount of arson cases has again increased since then.  The 2001 statistics show the rise of deliberate fires to be a total of 2,275 arson cases in Cambridgeshire5.  Through intensive research, the main problem area in the Cambridgeshire County has been identified as being the Dogsthorpe Fire Station in Peterborough, which oversees the Central Ward area of the city.  In this area of Peterborough, 796 cases of arson were attended, making it 35% of the total of all malicious fires in Cambridgeshire5.


An important aspect of arson that should not be ignored is the person committing the act.  The arsonist must have a reason, motivating him/ her to set fire to property.  Different types of malicious fires have different motivations that lead the offender to commit the crime.  The motivation of an arsonist is usually categorised by the arsonist’s choice of target.  Criminal arson is the category of fires dealing with incidents committed for financial gain through insurance fraud or concealment of another crime that was committed prior to the fire-setting.  Other motives for malicious fires are revenge and vandalism.


This is the most important form of arson for the research done in this paper – vandalism.  These arson cases are most likely to be committed by juveniles, setting fire out of mischief with the intent to destroy property.  The fires are usually set in an area familiar to the offender, making school properties and surrounding residential areas common targets6.  As these cases involve smaller fires and the offenders are still young, a general “they-will-grow-out-of-it” mentality prevails.  This is a serious misconception, because not taking notice of smaller, apparently “petty” arson cases will allow the perpetrators to continue with their fire-setting habits.  As a result, more incidents will occur and the committed fires will become more vicious7.  Not being apprehended for “petty” arson at a younger age is often seen as a precursor to other, more severe serial offences, i.e. sexual or homicide offences6.


Another aspect of arson cases is actual target that was chosen.  Due to previous research, it has been established which localities are particularly at risk of falling victim to malicious fires3.  Properties in isolated areas and/or in areas, which have a high rate of criminal activity, are in danger of experiencing more deliberate fires.  The areas surrounding local authority housing estates and premises that attract large crowds (e.g. schools, shopping areas, etc.) are also at a higher risk of arson fires.  There is particular concern about the fact that schools are falling victim to an increasingly large amount of arson attacks. 


Prevention & Reduction

Having established that a risk exists, the next stage is to try to reduce possibilities of arson being committed.  The continued increase in the number of deliberate fires attended by the fire brigade calls for immediate plans to attempt a reduction of this rate.  As arson is a multi-agency problem, there are several organisations that can make a contribution to an efficient arson reduction initiative2.  Here, the lack of communication and cooperation between the different agencies may cause considerable troubles.  This should be made one of the overall priorities.  Only if the departments work together to prevent arson, can any significant results be achieved.


Over the past years, substantial effort has been placed on establishing fire service policies for preventative methods and the development of Arson Task Forces throughout the country was the first step to combating arson crimes.  To enable any form of action to be taken, it is necessary to be familiar with the details of the problem.  This can be achieved through the accumulation of informative data, representing the true extent of arson.


Currently, there are only three full-time fire investigation teams in England and Wales2, meaning that the majority of fire stations rely on their fire teams to perform multiple duties.  This is based on a very limited amount of knowledge and training in actual fire investigation.  Coupled with the inadequate training availabilities, there is little standardisation of fire investigation methods.  The solution to this problem would be the formation of a national model for fire investigation that is officially recognised.


It has been recommended by the Home Office that research should be undertaken to quantify the proportions of deliberate fires according to their motives2.  When used correctly, the results of such a study could help to target any intervention methods more effectively against the responsible groups.  General deterrence may be achieved by informing the public about the arson problem and making people aware of the negative effects it has on a community.  A more individual deterrence could result from educational programmes, focusing on schools and colleges.  This would be more effective in the long run, as it sparks awareness of the problem and the danger in a younger age group. 


Another method of receiving information about the arson problem is the geographic plotting of fire data.  The basis of which is described in more detail in the subsequent section of this paper.  

Geographic Profiling

For crime to be combated successfully, it is essential to thoroughly study previous incidents because they can give an insight into offenders’ motivations, planning strategies, chosen targets and chosen forms of crime.  The analysis of specific crime characteristic involves the collection and examination of data concerning a criminal incident, offender and/or target.  The overall goal is to provide management information for the prevention and reduction of crime.  Firstly, this analysis can be used to assist in the organisation of police resources to deter and prevent criminal activity, by supplying tactical and strategical support for police operations, etc.  Secondly, it can aid the evaluation of effectiveness displayed by programmes employed in policing and crime prevention, by supplying information about technical support systems and allowing optimisation of their use.


Another important factor to analyse is the location chosen to commit the crimes.  This aspect can be examined with the help of a technique known as geographic profiling.  Though it is a swiftly expanding field in the world of crime analysis, geographic profiling has long stood in the shadow of the ‘ever-popular’ offender profiling.  The technique can be divided into two subcategories.  The more sensationalised application of geographic profiling is as an investigative technique, concerned with the location of crime scenes that form part of a series of incidents committed by one offender.  Here it is used as an investigative tool for tactical analysis8, focusing on the development of police operations in congruence with the specific case at hand.  On a more general and ‘day-to-day’ basis, it can be employed as a prognostic technique for strategic crime analysis8.  This concentrates on the location of specific types of crimes committed in a certain area and the possible occurring shifts within that vicinity or to a different area, over time.    


As with most investigative techniques, geographic profiling is based on a number of assumptions about the behaviour of people in familiar and/or unfamiliar surroundings.  It is generally thought that a person prefers to move around in a known area, where they feel more at ease.  Naturally, this also applies to an offender’s behaviour, when committing a crime.  He / She is more likely to commit a crime in a neighbourhood they know well and feel secure in9.  Due to this factor, it is expected that the distribution of incidents will provide information about the area in which the offender most likely lives (home base), works and/or seeks entertainment in10.  This also includes the various travel-routes linking these different areas.  When dealing with a suspected series of offences, it is significant that crime scene characteristics linking the offender to the crimes are identifiable at all of the scenes in the series.  This is required so as to minimise the potential of overlapping two different cases.


Including geographic profiling into an investigation has a wide array of positive effects.  As mentioned above, it can assist in the organisation of police operations.  When investigating a serial crime, geographic profiling can help to limit the possible search area by analysing the given data and prioritising ‘favoured’ locations, which may be part of the environment known to the offender – his / her awareness space.  The pinpointing of these areas could, through the correct application of this information, lead to a quicker apprehension of the offender.  This form of utilisation is known as task force management.  This limitation of the search area is also useful to limit the search population.  When a geographic profile indicates a certain area as the awareness space of an offender, the investigation can be centred on the residents of this area.  This could also be used as an aid in correlation with already known information about potential suspects.  Both of these factors apply to the investigation of serial offences.  When dealing with the strategic analysis of data, the resulting information will be useful to saturate any affected areas with police patrols.  The provided trend information can help to show potential problem areas, and where to focus police attention.  Through increased police presence, deterring effects could be achieved to prevent crimes from being committed. 


When using geographic profiling, it should not be forgotten that it is one of many investigative tools.  The most efficient way to employ this technique is in conjunction with other methods.  It can provide very relevant and significant information in aid of crime solving, but geographic profiling is not an all-powerful wonder weapon against crimes9.  Two criticisms have been brought up about this technique.  It has been said that geographic crime analysis focuses the investigation too much on one single factor known about the offender – location selection.  The other limitation is that although objective measures are produced, the data that geographic profiling relies on is based on subjective interpretations.  In defence, any possibility to investigate and reduce crimes with should be tested. 


In this paper, the analysis of arson data will be undertaken with the help of geographic aids.  Initially, the Cambridge area was chosen for locality’s sake, but as seen above, the Peterborough region was soon preferred.  This was due to the obvious arson problem in that particular area.  For the purpose of this paper, the analysis will involve the correlation of all relevant, accurate data – so as to perform the preliminary examination of the arson incidents and the manual plotting of crime maps.  For various reasons, discussed later on, only a theoretical introduction into computerised crime mapping was possible.  This, however, does not discredit the results of the other work.

Data Collection

Before any form of analysis can take place, the relevant data needs to be collected from a reliable source.  It has to contain extensive information of arson cases in a certain station ground area, i.e. the date, time and of course location of the crime.  Two other very important requirements, which the data needs to fulfil, are that it must be accurate in its descriptions and that it must be from a recent time period.  If all these preconditions have been recognised, the information needed for and received from the analysis will be a representative sample of arson incidents.  In this way, the analysis can be more helpful in planning preventative measures.


As advised by my project supervisor, initial contact was attempted with the Cambridge Fire Station, Police Station and County Council.  Each of these agencies was communicated with for various requests within the framework of this paper.  The Cambridge Fire Station received a general written letter, asking for the data of arson fires attended by the fire brigade in the Cambridge area.  At the same time, the Cambridge Police was asked for any data of arson cases passed on to and recorded by the police in the mentioned area.  And finally, the Cambridge County Council was contacted via post, requesting a thorough ordinance survey map to enable the spatial analysis.  This course of action, unfortunately, remained without any significant results.  The requests were most likely dismissed, as no specific member of staff was contacted.


Another contact at APU recommended making any requests known to Assistant Chief Fire Officer Peter Warner.  The overall result was a meeting with Assistant Divisional Officer Kevin Smith at Huntingdon Headquarters, which led to the refocusing of the project on Peterborough, as it represented the worst affected area for arson incidents.  On successful receipt of the relevant fire data of the Central Ward area, the analysis could begin.  The received data consisted of general case information concerning all fires for the years 2000 and 2001 in the Central Ward division of Peterborough.  It was arranged in a table; giving the incident number, date of incident, location, code details, time of incident and day of the week.


As only the case information concerning arson fires is required, the received data had to be sorted through to exclude all non-malicious fires.  This was done with the aid of the code details, as these are the information gathered by the fire-fighters at the scene and consist of the target of the fire and the suspected cause.  Even this early stage of the analysis proved to be problematic, due to the fire codes sometimes being quite ambiguous.  As the case details shown in the arson data are based on personal interpretation of fire personnel at the scene, it is going to be subjective.  In defence, it can be said that these people are professionals, who know what possible evidence they are searching for at a scene and try to note what they observe without bias.  If any uncertainty about the evidence in fire cases is present, the case details noted by the fire-fighters will reflect this through the use of code “mixes” such as C2M, D3O, etc.  As this issue involves a potential criminal act, it is better to note the doubtful circumstances as ambiguous, than to assume an accidental cause.  Though this shows uncertainty, cases holding this type of coding should be investigated and passed on as potential arson cases.  This ambiguity could be settled by the plotting and profiling of the arson data.  If any distinct and recognisable patterns can be seen, an indefinite arson case may be identified as being part of a series.  Once the arson cases have been selected from the general records, it needs to be divided into subsections – month of the year, day of the week and time of the day – to allow for a better, clearer analysis.


Crime “Hot-Spots”

Even when making the preliminary examination of the data, there are specific factors that should be looked out for from the beginning of the investigation, and throughout.

As the aim of spatial crime analysis is to “hot-spot”/determine the locations, where arson fires occur predominantly, the main thing that should be focused on is the search for fires, sharing a common type of target or even a specific reoccurring target in a certain area, at a similar time of the day and/or on a specific day of the week.  Another interesting point to take note of is the occurrence of “petty” arson, i.e. fires covered by the codes B (vegetation), C (outdoor structures) and D (refuse), especially on weekdays before and/or after school hours or at weekends.  Specific areas that are affected by a higher percentage of these kinds of fires include schools, housing estates and residential areas, as these “smaller” arson incidents are generally attributed to children causing criminal damage.  Along a more “advanced” route of arson cases, the increase of vehicle fires should not be left uninvestigated.  Once again, attention should be paid to these types of fires taking place in certain street areas (e.g. isolated areas), at specific times of the day and/or on similar days of the week.  In general, the analyst has to look out for any sign of a recognisable pattern of arson cases.     


Following the guidelines for the data analysis and having organised the records into the above-mentioned subsections, the examination of the arson distribution can commence.  All the case data referred to in the subsequent data discussion of the paper can be found in Appendix A (p. i – x).  Any statistical information cited in the different subsections has been collected in Appendix B (p. xi – xx).


Central Ward Arson Data - 2000

In the year 2000, the Dogsthorpe Fire Station in Peterborough had to deal with a total number of 112 fires within their call-out area.  The amount of malicious fires was 86, making up 76.8% of the total fires.  Here, it should be mentioned that those cases with ambiguous code details were included in this count, as “suspicious”.


The first stage in the data analysis was to examine the cases by the months in which they occurred, identifying the various arson peaks throughout the year.  The months with the highest number of arson cases were October (total of 11 arson cases), February and July (total of 10 arson cases each).  The months September (total of 4 arson cases), November, December (total of 3 arson cases each) and January (1 arson case) were those with the lowest numbers.  The average total of deliberate fires that were set each month lies by around 7.2 cases, making March an average month – as far as arson cases are concerned.


Overall in 2000, vehicle arson was the most predominant form of malicious fires with 47 cases for the year, representing the highest single target of deliberate fires and supporting the noticeable rise of this crime.  The next two highest targets of malicious fires are refuse (total of 17 cases) and building fires (total of 11 cases).  This leaves outdoor structures (total of 6 cases) and vegetation (total of 4 cases) as the more infrequent target for arson attacks.


Concerning any detectable arson patterns, these were analysed with the help of a regular street map.  This was necessary to enable the estimation of distances between two or more locations, showing whether it would be possible to travel from one site to another in a given time period – i.e. the time period between two or more potential serial cases.  This preliminary information gathering is vital for the later computer analysis of specific data.  In the year 2000, only two possible series came to attention.  The first series took place in the evening of the 13th April and included two sites of deliberate vehicle fires.  At 18:31, one vehicle was set alight in Cromwell Road (near the Hobson’s Playgroup Centre).  About two hours and twenty minutes later at 20:52, a second car was set on fire in a parallel street to Cromwell Road, named Gladstone Street (Taverner’s Road end).  The distance between these two sites could definitely have been bridged within the given time period – it is almost too long for the two arson attacks to be connected, but every eventuality must be taken into account.  The second possible series of malicious fires also involves vehicle fires.  On the 6th October at 19:46, a car is set on fire in Russell Street.  Around half an hour later at 20:16, a similar incident occurs in the same street.  Again, the time period allows for the two fires to have potentially been set by the same offender.


The next subsection of the arson data is the division by days of the week.  In 2000, the days with the highest number of deliberate fires were Thursday (total of 16 cases), Monday and Saturday (total of 14 cases each).  The days Tuesday (total of 8 cases) and Wednesday (total of 10 cases) were the ones with the lowest number of arson cases.  The average total of deliberate fires that were set on one day lies by around 12.3 cases, making Friday the average day.


Having organised the data into the overview of the weekdays allows for the examination of arson “hot-spots” instead of arson patterns.  By doing this, it becomes apparent that vehicle fires are very much focused around certain locations.  On Mondays, the majority of deliberate vehicle fires take place in the Gladstone Street / Link Road area.  In comparison, most car fires on Wednesdays are attended in the Clarence Road / Cromwell Road area.  The malicious car fires on Thursdays and Fridays are also centred in the Cromwell Road area, including its side streets.  The rest of the arson cases are distributed quite evenly throughout the week – in chronological terms – and also throughout the Central Ward area – in geographical terms.


Organising the arson data by time of day permits the investigation of when certain targets of malicious fires are most at risk.  The time periods, with the highest numbers of arson cases in 2000, were 18:00 – 19:00 and 20:00 – 21:00 (total of 12 cases each); and 01:00 – 02:00, 16:00 – 17:00 and 19:00 – 20:00 (total of 8 cases each).  The average number of deliberate fires set per hour is about 3.6, meaning that the high peaks just mentioned are well above average.


In fact, the majority of arson cases occur in the evening hours, between 18:00 and 22:00hrs.  The more consistent numbers of malicious fires are committed in the very early morning, late afternoon and at night.  The least fires of all are set during the morning and the early afternoon, between 07:00 and 15:00hrs.


Another discovery can be made in connection with so-called “petty” arson cases.  They occur after 15:00hrs and with an increased frequency between 16:00 – 17:00 and 18:00 – 19:00.  These kinds of deliberate fires can be found mostly in the Lincoln Road, Cromwell Road and Gladstone Street areas, including their side streets.  These streets are residential areas, which lead to and from schools in the centre of Peterborough.  As mentioned above, “petty” arson is usually connected with children setting fires on their way home from school.  This correlates with the peak times for “petty” arson.  Even the schools, etc., fall victim to arson attacks.  The Hobson’s Playgroup Centre grounds (Cromwell Road) were the location given for 9 malicious fires; all cases have different targets, from burning refuse to the building itself being set on fire.  The fires occurred after 16:00hrs, throughout the evening and into the night (22:00hrs).  The cases were made up of four vehicle fires, three building fires and 2 refuse fires.  The Beeches Primary School (Beech Avenue, Cromwell Road) was also affected by one arson attack to the actual school building at 17:42 in the afternoon on the 29th July.


Central Ward Arson Data - 2001

The total number of fires that the Dogsthorpe Fire Station had to deal with was 138 for the year 2001.  Around 89.9% of these fires were caused deliberately, meaning the amount of arson cases is 124 altogether.  Though the number of total fires had increased over the passing year, the actual number of accidental fires had decreased, leaving a higher percentage of malicious fires.  Again, it should be noted that the cases with ambiguous code details were included in this count, as “suspicious”.


Dividing the arson data by month shows how this escalation of arson fires has affected the distribution of the cases.  The months with the highest numbers of deliberate fires are June (total of 35 cases), May and July (total of 18 cases each).  Those months with the lowest amounts of arson are March and October (total of 2 cases each).  The average number of arson fires per month was calculated to be 10.3 cases, making April the average month. 


Although vehicle fires are still the predominant form of arson at 61 cases for the year 2001, both refuse (total of 34 cases) and building fires (total of 22 cases) have doubled since the previous year.  The two more uncommon targets for arson attacks are once again outdoor structures (total of 4 cases) and vegetation (total of 2 cases), showing a decrease of malicious fires by half.


Through the increase of deliberate fires in general, a higher number of distinct arson patterns was recognisable. The first possible pattern that could be noticed took place mid- and end of May in Cromwell Road.  In total, five building arson cases occurred around and including the Haramby Community Centre within the matter of eight days.  No two were set on the same day; the first occurring on the 20th May at 03:05 in the early hours of the morning.  The very next day 21st May at 10:25, another fire was set.  After a couple of days break, the third arson attack ensued on the 25th May in the late afternoon, at 17:05.  The last two fires were started on the 27th May at 21:31 and the 28th May at 00:09.  Though all the fires occur at completely different times of the day, the regularity / frequency with which this particular property falls victim to building arsons should be closer investigated.  At the beginning of June, two malicious refuse fires were set alight in Link Road (between Gladstone St and Cromwell Rd).  Both fires occurred on the 4th June in the early morning hours.  The first one was called-in at 03:39 and the second one 12 minutes later at 03:51.  As these arson fires took place in the same street, there would have been no problems for the arsonist to arrive at the second site within the given 12-minute time period.  Towards the middle of June (11th), six cases of deliberate fires involving cars were reported to the fire station within the time period of two hours.  All fires occurred in different streets, but on closer examination it can be seen that the locations are in close vicinity to each other and could be reached by cycling or driving within the observed time frame.  In chronological order, the arson fires started at 00:16 on Clarence Road, 00:51 on Gladstone Street, 01:08 on Bamber Street, 01:43 on Lincoln Road, 02:01 on Searjeant Street and ended at 02:14 on Link Road.  Though the locations are slightly scattered within the Central Ward area, this large accumulation of vehicle fires in the early morning hours of a Monday is very suspicious and should definitely be further investigated.  On the 18th June, four more malicious vehicle fires occurred in the early hours of the morning.  Starting at 00:29 and 01:20 on Gladstone Street, going on to Cromwell Road at 01:54 and ending at 02:15 on Gladstone Street, all the sites of this series are in easy walking distance of each other and can be reached by moving in a south-to-north direction – in correlation to the given times when the fires occurred at the specific locations.  The next possibly related malicious fires took place in the early afternoon of the 26th June.  These fires involved ignited refuse and occurred on Cromwell Road at 14:39 and on Link Road at 14:49.  The two sites are not very far apart and can be easily reached by foot – even in the 10 minute time period.  According to the times and the targets of the fires, it is very likely that they were caused by children on their way home from school, walking in a north-to-south direction through the street network.  For the month of September, it can be said that all vehicle arsons took place during the night, meaning between 21:00 and 04:00hrs.  On the 13th September, three cases of deliberate car fires occurred at 01:59 on Cobden Avenue, 03:35 and 03:58 on Gladstone Street.  The latter two fires most definitely represent a series of vehicle arson; the first fire fits in due to the timing and location of the incident.  Moving in a south-to-north direction, the perpetrator could have reached the last two sites by foot, even within the given one and a half hours time period.  A couple of days later on the 15th September, another two vehicle fires were started at 02:29 on Craig Street and at 02:42 on Cromwell Court.  Again, these arson sites are within walking distance of each other, if the arsonist moved in a south-to-north direction.  Finally, on the 21st September, two more vehicle arsons were reported.  The first of which took place on Cromwell Road at 00:57, involving one car, and the second incident occurred six minutes later, at 01:03, on Russell Street.  However, the latter site involved three cars.  The distance between these two locations could be covered in the given time frame, when walking in a north-to-south direction through the street network.


Naturally, the division of the arson data into the days of the week also showed a slight shift, concerning the number of cases per day.  The days with the highest numbers of arson fires were Monday (total of 32 cases) and Saturday (total of 23 cases).  In comparison, those days with the lowest amount of malicious fires were Thursday (total of 9 cases) and Tuesday (total of 11 cases).  Calculating the average showed that 17.7 fires were set each day, making Sunday an average day for arson.

No distinct changes occurred with the arson “hot-spots” from the previous year.  Only minor shifts were noticeable.  On Mondays, vehicle fires happened mostly between 22:00 and 03:00hrs and were still focused around the area of Gladstone Street and its side streets.  Deliberate vehicle fires on Thursdays occurred more around the south-end of Gladstone Street.  Arson cases, on any other day of the week, have no definite centres.


In comparison with the arson data from the previous year, some changes occurred in the main times for arson fires.  In the year 2001, the time periods with the most deliberate fires were 02:00 – 03:00 and 03:00 – 04:00 (total of 11 cases each), including 00:00 – 01:00 and 01:00 – 02:00 (total of 10 cases each).  This summary of risk times reflects the general alterations, which occurred since 2000.  The average amount of arson fires set per hour was calculated to be 5.2 cases – proving that once again the main risk times are well above average.


The majority of malicious fires now take place in the very early hours of the morning and some at night.  The more constant amounts of arson cases occur in the late afternoon, the evening and during the night.  This is where a shift becomes noticeable.  The least number of fires are still set in the morning and early afternoon.


Crime Mapping

Another element needed for the preliminary spatial analysis of the arson data is a detailed street map of the affected area, showing the general localities – i.e. the residential estates, schools, parklands, etc.  Through the previously mentioned contact at Huntingdon Headquarters, it was possible to receive the required ordinance street map.  It then had to be enlarged for the purpose of plotting the various fire sites.

This stage of the analysis also proved to have its difficulties.  The ‘location’ details, which are the vital parts of the data, were quite general in their description of fire sites and sometimes even misleading.  Many of the locations only gave the street names of where the fire had occurred, without mentioning the house number.  If a crossroad of the street was noted in the details, it was unclear whether it was used because the fire actually occurred there or whether it was used because the road was a better-known area when referring to the location of the fire.  Due to this confusion, the plotting of the Central Ward maps was at times very awkward and done to the best possible degree under the circumstances.


All the information discussed in the following parts of this paper can be found in Appendix C (p. xxi – xxvii).


Central Ward Arson Maps - 2000

The map number 1 (Appendix C, p. xxii) was plotted with the aid of the first subsection of Central Ward Arson Data - 2000 – division by months.  The main arson centres that could be distinguished at first were the Cromwell Road and the Link Road areas.  Most of the deliberate fires affecting the Cromwell Road area occur in the winter, i.e. in the months from December to February.  Only few arson fires were committed in the summer months.  The fires in the Link Road area mainly take place in the middle of the year – May to August time.

Other sites that frequently fall victim to malicious fires are the Gladstone Street, Clarence Road and the Russell Street areas.  In the first half of the year (February to July), Clarence Road is particularly affected.  In comparison, deliberate fires set in the Russell Street area mostly occur during the latter half of the year, in the late summer and autumn months.  Most of these circumstances are mirrored in the Central Ward Arson Data - 2000 section.


Map number 2 (Appendix C, p. xxiii) displays the data arranged by day of the week.  Unfortunately, the plotted data on this street map is geographically spread over the entire Central Ward area, giving no detectable arson patterns.


The third map (Appendix C, p. xxiv) shows the data by the time of the day.  The arson patterns on this map strongly support the results from the ‘time of the day’ subsection in Central Ward Arson Data - 2000.  Most of the fires, according to the results of the plotted data, occurred in the evening and the late afternoon.  These were mainly in the areas of Cromwell Road, Gladstone Street and Clarence Road.  Another large amount of fires is set during the night and the very early hours of the morning.  A lot of these arson fires are located around the Link Road region.  Also, the morning and early afternoon are the times with the lowest number of arson cases. 


Central Ward Arson Maps - 2001

The fourth map (Appendix C, p. xxv) once again focuses on the arson centres in Central Ward arranged by month.  The main sites for deliberate fires are again Cromwell Road and Link Road, but now also Gladstone Street.  Cromwell Road suffers particularly in the spring months – especially the month of May is a high-risk time.  In the Link Road and Gladstone Street areas, the summer months are the commonest time for arson crimes – with a specific focus on the month of June. 

Other arson centres are the Lincoln Road, Clarence Road and Cobden Avenue areas, although here the fires are distributed throughout the whole year.  These results back up the arson series discussed in the ‘month’ subsection of Central Ward Arson Data - 2001.


On map number 5 (Appendix C, p. xxvi), the data was organised by day of the week.  The results in Central Ward Arson Data - 2001 already show that Monday is the day with the highest amount of malicious fires.  The plotted map depicts which areas the cases are mostly located in.  These areas are Cromwell Road, Link Road and Gladstone Street.  The locations of all other arson cases are spread throughout the Central Ward district.


Map number 6 (Appendix C, p. xxvii) shows the distribution of the arson cases through time of the day and confirms the previously detected shift of the high-risk times between the years 2000 and 2001.  The majority of the fires occur during the night and/or the early morning hours.  An also significant amount is shown to be committed in the evening and the late afternoon.  Once again though, the least deliberate fires are set in the morning and the early afternoon.


CrimeStat & Data Requirements

CrimeStat is a spatial statistics computer programme, which was devised to analyse location data belonging to crime incidents.  Its purpose is to provide a tool for the analysis of large crime data sets, allowing the plotting of different incident locations and selecting subsets of the data for a more specified investigation.  The computerised maps that are produced with the help of this technique, assist the crime analyst in identifying shifts of the crime sites over time, determine “hot-spots” and detect any other relevant crime patterns.  This programme was also designed to offer statistical information about the entered data, give descriptions of the spatial arrangements of crime occurrences and create geographic models of crime incidents.    


The first step to using CrimeStat is to set up a reference file.  This file consists of the basic information about the study area that contains all the data locations.  It requires the formation of a grid system, employing X- and Y-coordinates to define the parameters of the search region.  The most widely used coordinate system is the projected coordinates method.  Here, the search region is not portrayed as a part of the earth’s surface, but is instead projected onto a flat plane.  This will later influence the way in which the data is entered into the new programme file, as the coordinate systems in all files must be identical for the analysis to produce reliable results.  Other factors, which must be given, are the actual length of the street network belonging to the study area and whether direct or indirect distance calculations should be used.  These factors will depend on the local demographics of the search area.  Direct distances are the shortest distance between two points.  When using a projected coordinates system, the shortest distance between two points on a flat plane is a straight line.  Indirect distances are the approximate routes of travel on a rectangular street network.


After setting up a reference file, the relevant analysis data needs to be programmed into a so-called primary file.  This file will then hold all the information about the specific crimes that are to be examined.  Firstly, the identical settings must be entered, to again define the search area.  This means that the same coordinate system, distance measurements, etc. should be used.  Once this has been done, the data can be typed into the new file, using the appropriate X- and Y-coordinates referring to the correct locations on the grid system.  Then the necessary analysis programme can be run to identify the spatial distribution and/or the crime “hot-spots”.11


To enable the entry of the Peterborough arson data into CrimeStat’s file system, the above-mentioned requirements must be fulfilled.  A grid system would have to be superimposed on the Central Ward area, to allow the definition of the study area by use of coordinates.  This would help to create the necessary reference file needed to proceed with the analysis.  This, in itself, is not very problematic.  Unfortunately, as mentioned during the crime map discussion, the location details given for the arson sites are vague and misleading when manually plotting a map.  They would be completely insufficient when trying to enter them into the CrimeStat programme.  This would require the details noted for the separate cases to be made much more specific and exact. 


Once all the data is precise enough to enter into a primary file, there are two forms of incident analysis that can be applied.  The first possibility would be the examination of all data collectively, so as to receive “hot-spot” identification.  The other application would involve the selection of subsets, i.e. the crime patterns thought to belong to a series.  This analysis would concentrate on the spatial distribution of the arson series.  The relevant data that could be used for this test, would be the possible series discussed especially in the Central Ward Arson Data - 2000 section.


It makes no difference, whether the arson data is examined by the fire brigade or whether it is passed on to the police for crime analysis.  The use of crime data analysis and incident mapping could be extremely helpful not just in the investigation of arson crimes, but also in the development of preventative methods.  Mainly, its application could contribute to the strategic organisation of police patrols in areas affected by an excessive number of arson incidents, indicated by the resulting information.  Even though police presence should be constant, increased attention needs to be paid to the high-risk times and locations.


As can be seen in the Central Ward Arson Data - 2000 and 2001 chapters, there are definite high-risk times, when more deliberate fires are set.  Although a shift has taken place from the evening hours in 2000 to the very early hours of the morning and night time in 2001, this whole time-period from 18:00hrs to 04:00hrs should be treated as priority for specialised police patrols.  It is quite unlikely that arson incidents during the morning and the early afternoon (obvious low-risk times).


As well as focusing on certain times with higher arson rates, any surveillance or police patrols should be focused on specific street areas that experience larger amounts of arson cases.  In 2000, the main arson centres were Cromwell Road and Link Road, closely followed by Gladstone Street, Clarence Road and Russell Street.  The Central Ward Arson Map - 2000 shows that Cromwell Road, Gladstone Street and Clarence Road are mostly affected by arson cases in the evening hours and in the late afternoon.  Most of the malicious fires set in the Link Road area occur during the night and the early hours of the morning.  If police patrols would be organised by time-location correlation, a more efficient form of policing could be achieved.  The “hot-spot” centres in 2001 show the continuous problem areas (Cromwell Road and Link Road) with a shift, leading to the inclusion of Gladstone Street.  When arranging police patrols, attention should also be paid to the side streets along the main problem areas – as crimes do shift into the “back streets” on increased police presence12.

The actual effectiveness of crime mapping as an investigative and/or strategic technique can only be evaluated by its successful application.


1  Weiner, M. (2001) The Economic Cost of Fire. Home Office Research Study 229, October 2001.


2  Home Office. (1999) Safer Communities: Towards Effective Arson Control. Report of the Arson Scoping Study.


3  Lewis, A. & Dailey, W. (2000) Fire Risk Management in the Workplace: A Guide for Employers. 2nd Edition. London: Fire Protection Association (FPA).


4  Watson, L. et al (2000) Fire Statistics United Kingdom 1999. Home Office Publication, Issue 20/00, 8th November 2000.


5  Cambridgeshire Fire and Rescue Service. (2001) Statistics for Arson in Cambridgeshire: April 2000 / March 2001.


6  Rossmo, K. (2000) Geographic Profiling. Boca Raton: CRC Press LLC.


7  Humphries, M. (2001) “What Motivates An Arsonist?” Prison Service Journal, Issue 133.


8  Goldsmith, V. ed. (2000) Analyzing Crime Patterns: Frontiers of Practice. London: Sage Publications Inc.


9  Ramsland, K. (2000) “Geographic Profiling.” http://www.crimelibrary.com/ forensics/geog.htm, Dark Horse Multimedia Inc.


10 Godwin, M. (2000) “Geographic Profiling.” The Encyclopaedia of Forensic Science. Academic Press, pp. 991–998.


11 Levine, N. et al (1999) CrimeStat: A Spatial Statistics Program for the Analysis of Crime Incident Locations. Washington: The National Institute of Justice.


12 Bottoms, A. (1997) “Environmental Criminology.” The Oxford Handbook of Criminology. Oxford: Oxford University Press, pp. 585–650.






Baigent, D. (2001) One more last working class hero: a cultural audit of the UK fire service, Cambridge: Fitting-in and APUhttp://www.fitting-in.com/thesis.htm


Bolviken, E. & Egeland, T. (1995) “Arson, Statistics and the Law: Can the Defendant’s Proximity to a Large Number of Fires Be Explained By Chance?” Science and Justice, April / June 1995; Vol. 35, No. 2: pp.97–104.


Bottoms, A. (1997) “Environmental Criminology.” The Oxford Handbook of Criminology. Oxford: Oxford University Press, pp. 585–650.


Brantingham, P. & P. (1981) Environmental Criminology. London: Sage Publications.


Bundeskriminalamt. (2000) “BKA-Forum: Geographische Kriminalitaetsanalyse.” http://www.bka.de, Wiesbaden 20th March 2000.


Cambridgeshire Fire and Rescue Service. (2001) Cambridgeshire and Peterborough Fire Authority: Annual Report 2000 / 2001.


Cambridgeshire Fire and Rescue Service. (2001) Statistics for Arson in Cambridgeshire: April 2000 / March 2001.


Day, J. (2001) “Understanding the Characteristics of Fire-Setters in Comparison to other Types of Offenders.” Prison Service Journal, Issue 133.


Donahue, M. L. (2002) “The ATF Fire Research Laboratory: A New Forensic Resource for Fire-Scene Investigation.” Forensic Science Communications Vol. 4 No. 1, January 2002.


Duggan, L. & Shine, J. (2001) “An Investigation of the Relationship between Personality Disorder, Hostility, Neuroticism and Self-Esteem amongst Incarcerated Fire-Setters.” Prison Service Journal, Issue 133.


Evans, D. et al (1992) Crime, Policing and Place: Essays in Environmental Criminology. London: Routledge.


Godwin, M. (2000) “Geographic Profiling.” The Encyclopaedia of Forensic Science. Academic Press, pp. 991–998.


Goldsmith, V. ed. (2000) Analyzing Crime Patterns: Frontiers of Practice. London: Sage Publications Inc.


Home Office. (1999) Safer Communities: Towards Effective Arson Control. Report of the Arson Scoping Study.


Humphries, M. (2001) “What Motivates An Arsonist?” Prison Service Journal, Issue 133.


Leong, G. & Silva, J. (1999) “Revisiting Arson from an Outpatient Forensic Perspective.” Journal of Forensic Science, May 1999; Vol. 44, No. 3: pp.558–563.


Levine, N. et al (1999) CrimeStat: A Spatial Statistics Program for the Analysis of Crime Incident Locations. Washington: The National Institute of Justice.


Lewis, A. & Dailey, W. (2000) Fire Risk Management in the Workplace: A Guide for Employers. 2nd Edition. London: Fire Protection Association (FPA).


Meldrum, G. (2001) “Hammering Out Arson.” Fire Prevention, Vol. 350.


Povey, D. et al (2000) Recorded Crime Statistics: England and Wales, April 1999 to March 2000. Home Office Publication, Issue 12/00, 18th July 2000.


Prins, H. (2001) “Arson – A Brief Review.” Prison Service Journal, Issue 133.


Ramsland, K. (2000) “Geographic Profiling.” http://www.crimelibrary.com/forensics/ geog.htm, Dark Horse Multimedia Inc.


Ritchie, E. & Huff, T. (1999) “Psychiatric Aspects of Arsonists.” Journal of Forensic Science, July 1999; Vol. 44, No. 4: pp. 733–740.


Rossmo, K. (2000) Geographic Profiling. Boca Raton: CRC Press LLC.


Smith, K. (2001) Arson Reduction: Cambridgeshire Fire and Rescue Service. Huntingdon Fire Station Headquarters, August 2001.


Watson, L. et al (2000) Fire Statistics United Kingdom 1999. Home Office Publication, Issue 20/00, 8th November 2000.


Weiner, M. (2001) The Economic Cost of Fire. Home Office Research Study 229, October 2001.


Wood, B. (2001) “Arson: A Geographical, Demographical and Motivational Profiling Perspective.” Interfaces, April / June 2001; No. 26: p.3.








Useful links

Research Advice


About us

Contact us