Research Report Digest, Issue 9
January 2013
In NIJ’s Research Report Digest, you will find brief descriptions of studies in various criminal justice disciplines, such as crime and forensic sciences, and evaluations of technologies that are used in the law enforcement and corrections fields.
This issue includes reports based on NIJ-funded research that were added to the NCJRS Abstracts Database from July-Sept. 2012.
Digging into the Data
NIJ posts datasets from completed research. Check out the 25 most recent datasets.
Find research reports related to:
Crime
Determinants of Chicago Neighborhood Homicide Trends: 1980-2000
Author: Brian J. Stults
This study examined homicide trends in Chicago neighborhoods for 1980 to 2000 using three data sources from the Inter-University Consortium for Political and Social Research and the National Archive of Criminal Justice Data. Researchers assessed neighborhood variations in homicide trends in Chicago neighborhoods. They assessed negative structural characteristics and counterbalancing neighborhood levels of collective efficacy. Findings revealed significant variation in homicide trends across Chicago neighborhoods from 1980 to 2000. The analysis showed that homicide trajectories were consistently associated with early levels of concentrated disadvantage and change over time. Change in family structures was also predictive of homicide trajectory, but only in neighborhoods with high homicide levels at the beginning of the study period.
Read the complete report Determinants of Chicago Neighborhood Homicide Trends: 1980-2000 (pdf, 81 pages)
Evaluating the Implementation of a Family-Focused Prevention Program: Effectiveness of SAFE Children
Authors: David B. Henry, Patrick H. Tolan, Deborah Gorman-Smith, Michael E. Schoeny, Jack Zwanziger, and Sage Kim
This is an evaluation of the SAFE Children intervention. SAFE Children is a family-oriented delinquency prevention program for parents facing the challenges of raising children in inner-city communities. The program resulted in increased levels of academic achievement and parental involvement in school. Children and their families were randomly assigned to treatment and control groups. Program participants increased reading skills at a rate near national norms. In contrast, control students were just below the national average. Program families stayed involved in their children’s schooling over the 2½ years of the study. Control group families showed decreasing parental involvement. Program children in high-risk families showed decreased aggression over time, but high-risk controls had no change in aggression. Program children from high-risk families also had positive increases in measured concentration and social competence; control children showed no change in these developmental areas. Among high-risk families, parental supervision improved for those in the program, but was unchanged for high-risk control families. In a longer-term booster intervention and follow-up study, evaluators recruited and tracked 382 of the original 424 participants. The program consists of a reading tutoring program and a family-focused intervention (20 weeks duration) provided when the child is in first grade. Weekly multiple-family group meetings address issues of parenting, family relationships, child development and parental involvement in their child’s schooling.
Read the complete report Evaluating the Implementation of a Family-Focused Prevention Program: Effectiveness of SAFE Children (pdf, 69 pages)
Increasing Student and Community Safety Partnership: A Researcher-Practitioner Partnership between West Virginia University
Department of Geology and Geography, the West Virginia University Police Department and the Morgantown Police Department
Authors: Gregory Elmes and George Roedl
This project used daily reports to analyze and identify crime patterns across campus and municipal law enforcement jurisdictions. The partnership set up a Geographic Information System and improved collective crime-reduction measures through analyses of the data. The researchers used automated models to extract and code crime incident to help practitioners generate their own maps of crime locations. This improves efficiency and promotes continued analysis after the formal project ends. The partnership identified hot-spot areas in each police jurisdiction, as well as many cross-jurisdictional hot spots. The partnership set crime-reduction programs based on the crime clusters. Preliminary assessments suggest student victimization is random, based on crimes of opportunity and consistent with current environmental criminology theories. Campus and municipal crime rates of 10 priority offenses had a consistent decline during the 2010-2011 study period; however, crime rates of other offenses increased. The increase may be due to more offenses being reported or observed by law enforcement officers in hot-spot areas.
Violence
Child Custody Evaluators’ Beliefs About Domestic Abuse Allegations: Their Relationship to Evaluator Demographics, Background,
Domestic Violence Knowledge and Custody-Visitation Recommendations
Authors: Daniel G. Saunders, Kathleen C. Faller, and Richard M. Tolman
This study examined what child-custody evaluators believe about allegations of domestic abuse made by one parent against another.
These involved divorce proceedings in which the court made a custody decision. There were four study objectives. The first
was to explore the extent to which professionals who make court recommendations believe that domestic violence allegations
are false. The second was to examine the association between these beliefs and knowledge of domestic violence, as well as
custody recommendations. The third was to learn whether beliefs about false allegations relate to beliefs that false allegations
of child abuse are common, and that parents often alienate their children from the other parent. The fourth objective was
to explore the relationships between beliefs about false allegations and beliefs about social dominance and justice. Practice
implications include how beliefs about false allegations of domestic violence relate to other beliefs and to custody-visitation
recommendations.
Judges, private attorneys and custody evaluators were more likely than domestic violence workers and legal aid attorneys to
believe that mothers make false accusation. Domestic violence workers and legal aid attorneys gave the highest estimates of
the percentage of fathers making false domestic violence, while judges and custody evaluators gave the lowest percentages.
On average, evaluators estimated that one-fourth to one-third of child abuse allegations were false. On average, evaluators
estimated that 26 percent of mothers’ domestic violence allegations were false and 31 percent of fathers’ domestic violence
allegations were false.
Read the complete report Child Custody Evaluators' Beliefs About Domestic Abuse Allegations: Their Relationship to Evaluator Demographics, Background, Domestic Violence Knowledge and Custody-Visitation Recommendations (pdf, 176 pages)
GPS Monitoring Technologies and Domestic Violence: An Evaluation Study
Authors: Edna Erez, Peter R. Ibarra, William D. Bales, and Oren M. Gur
This study examined the use of Global Positioning System technology to monitor compliance with court-mandated “no contact” orders in domestic violence cases. The study also determined the effectiveness of GPS as a form of pretrial supervision in DV cases compared to other pretrial supervision conditions. The findings show the use of GPS affects behavior over both the short and long terms. The short-term impact was associated with no contact attempts; defendants enrolled in GPS monitoring had fewer program violations compared to those placed in traditional electronic monitoring (EM), which uses radio frequency technology. The long-term impact involves monitoring house arrest remotely, but without tracking. Seemingly, GPS tracking increases defendants’ compliance with program conditions compared to those who are monitored for presence at a particular location (usually the home) but are not tracked for all locations. The study used a quasi-experimental design, and the sample included more than 3,600 defendants across three sites. Defendants enrolled in the Midwest GPS program had a lower probability of being rearrested for a DV offense during the 1-year follow-up period compared to a comparison group of defendants who had been in a non-GPS condition (e.g., in jail, in an EM program or released on bond without supervision). In another study site, those placed on GPS had a lower likelihood of arrest for any criminal violation within the 1-year follow-up period. In a third site, however, no impact from participation in GPS monitoring was found. The authors speculate that the diversity of the defendants placed on GPS monitoring in the third site, and a different method used for selecting the sample of defendants, may explain the anomaly.
Read the complete report GPS Monitoring Technologies and Domestic Violence: An Evaluation Study (pdf, 245 pages)
Forensic Sciences
Application of Machine Learning to Toolmarks: Statistically Based Methods for Impression Pattern Comparisons
Authors: Nicholas D.K. Petraco, Helen Chan, Peter R. De Forest, Peter Diaczuk, Carol Gambino, James Hamby, Frani L. Kammerman,
Brooke W. Kammrath, Thomas A. Kubic, Loretta Kuo, Patrick McLaughlin, Gerard Petillo, Nicholas Petraco, Elizabeth W. Phelps,
Peter A. Pizzola, Dale K. Purcell, and Peter Shenkin
This project provides a scientific basis for the reliability and validity of impressions made by tools and firearms. The study focused on striation patterns left by tools and on cartridge casings from firearms. Impressions made by tools and firearms can be viewed as mathematical patterns composed of features. This study used the mathematics of multivariate statistical analysis to recognize variations in these patterns. In computational pattern recognition, this is called “machine learning.” The mathematical details of machine learning can yield the quantitative difference between identification and no identification. The project had three main tasks. The first was toolmark pattern collection and archiving. The second step involved setting up a database and Web interface for sharing toolmark data. Third, researchers identified and used multivariate machine-learning methods relevant to the analysis of collected toolmarks. This research produced objective and testable methods for associating toolmark impression evidence with the tools and firearms that produced them.
Read the complete report Application of Machine Learning to Toolmarks: Statistically Based Methods for Impression Pattern Comparisons (pdf, 101 pages)
Application of Raman Spectroscopy for an Easy-to-Use, on-Field, Rapid, Nondestructive, Confirmatory Identification of Body
Fluids
Author: Igor K. Lednev
Raman spectroscopy is a technique for observing vibrational, rotational and other low frequency phenomena. It has found a
variety of practical applications, ranging from real-time monitoring of gas anesthesia mixtures during surgery to determining
the chemical composition of historical documents.
This study showed the potential of Raman spectroscopy for the nondestructive, confirmatory identification of body-fluid traces.
It also developed a program for the automatic identification of body fluids in dry mixtures. The project confirmed that the
Raman spectroscopic signature of each body fluid is unique and can be used for identification. The study produced a library
of Raman signatures for blood, semen, vaginal fluid, saliva and sweat. It also involved developing software for the automatic
identification of an unknown sample. The method included body-fluid stains on various substrates and contaminated stains.
The project showed that mixed samples of body fluids can be identified through automatic mapping if two body fluids are not
thoroughly mixed. Dry traces of blood and semen were successfully identified on several substrates of practical importance.
Anonymous groups of donors contributed their body fluids. Donors disclose their age, race and sex to ensure the required diversity
and sample size. Samples of each body fluid were obtained from donors who represented five races. The developed software algorithms
compared experimental Raman spectra with the library of Raman signatures, providing quantitative measures of similarity.
Read the complete report Application of Raman Spectroscopy for an Easy-to-Use, on-Field, Rapid, Nondestructive, Confirmatory Identification of Body Fluids (pdf, 80 pages)
Enhancing Scene Processing Protocols to Improve Victim Identification and Field Detection of Human Remains in Mass Fatality
Scenes
Authors: Dennis C. Dirkmaat, Erin N. Chapman, Michael Kenyhercz, and Luis L. Cabo
This research provides protocols to improve the efficiency and effectiveness of forensic processing of large-scale crime scenes. It focuses specifically on mass fatality incidents. Field testing shows that comprehensive documentation of the location of evidence and human remains can be accomplished quickly. Results also showed that it is possible to make accurate estimates of the volume of evidence using various methods. These methods can be used to estimate the amount of evidence in unsearched areas. This helps to predict how many people will be needed at the scene. It also helps to estimate how long it will take to process and release specific portions of a scene. Better scene processing involves improved protocols for detecting, recovering and identifying human remains at large-scale scenes. This is important for reconstructing events related to a plane crash, bomb incident or other mass fatality incident.
Read the complete report Enhancing Scene Processing Protocols to Improve Victim Identification and Field Detection of Human Remains in Mass Fatality Scenes (pdf, 156 pages)
Establishing the Quantitative Basis for Sufficiency Thresholds and Metrics for Friction Ridge Pattern Detail and the Foundation
for a Standard
Authors: Randall S. Murch, A. Lynn Abbott, Edward A. Fox, Michael S. Hsiao, and Bruce Budowle
The primary purpose of this study was to develop a sound, quantitative basis for assessing the quality of fingerprint images. Researchers got fingerprint images from several sources. The largest was a set of 117,323 anonymized images from 2,575 people taken from the Federal Bureau of Investigation’s Criminal Justice Information Services division. Another source was 516 latent prints from the National Institute of Standards and Technology Special Database 27. Researchers also used prints obtained in the laboratory using latent lifting methods. Finally, they used a newly created database of digitally altered images of actual prints for use in determining drop-off points. The researchers tested their ideas on these actual images and then performed statistical analyses to test the validity of the results. The analyses led to developing quantitative thresholds for unbiased selection and for the use of Level 2 detail. The researchers developed software for use in extracting minutiae, ridges and extended feature representations of images; they also conducted data mining to identify new feature types that are statistically rare in fingerprint image databases. The researchers also developed a method for conducting high-speed computing when performing the data mining work, and developed an improved automated method for conducting image segmentation in which the fingerprint region is separated from the background of an image. This work has improved the objective level for assessing the quality of fingerprint images. However, more work is required.
Read the complete report Establishing the Quantitative Basis for Sufficiency Thresholds and Metrics for Friction Ridge Pattern Detail and the Foundation for a Standard (pdf, 53 pages)
Evaluating High Dynamic Range (HDR) Processing with Regards to the Presence of Individualizing Characteristics in Shoeprint
Impressions
Author: Kristin Rogahn
This study examined High Dynamic Range (HDR). This method for processing a series of photographs into one image captures the fullest range of highlights and shadows present in the original impression. HDR increases the span between shadows and highlights in an image by taking more than one picture of the same scene. The various shots heighten shadows, midtones and highlights. This approach then merges the images into one unified picture with tremendous tonal range. Researchers found that HDR does not produce a significant increase in details compared with viewing the same images in Photoshop. However, bracketing exposures increases the ability to capture more detailed images of footwear impressions than a single image alone. This allows the use of HDR software for rapid processing and comparison. Difficult lighting situations that lead to challenging photographic conditions are common at crime scenes. The quality of evidence collected in the field determines the ease of footwear analysis. It is important that the photographer accurately document the scene details despite vast differences between the brightest areas and the darkest shadows. Technicians should use the auto-bracketing feature at the crime scene when faced with difficult lighting situations, limited equipment or time constraints. This will provide a quick, straightforward and forgiving means of capturing the best image possible in the form of multiple source images.
Read the complete report Evaluating High Dynamic Range (HDR) Processing with Regards to the Presence of Individualizing Characteristics in Shoeprint Impressions (pdf, 17 pages)
Filling a Critical Need by Establishing a Fully Functioning, CODIS Dedicated Laboratory
Authors: Bill Gartside and Scott McWilliams
This report describes the launch of an automated laboratory dedicated to CODIS DNA sample profiling in Wyoming. Before this project, there was no equipment or laboratory space at the Wyoming State Crime Laboratory dedicated to this purpose. For this project, offender DNA samples are being collected and archived with Bode buccal collectors. The lab manages the process with an internally developed Excel-based information management system. The current first-pass success rate for convicted-offender samples is more than 95 percent. The lab uses conventional analysis for difficult samples. The lab staff significantly reduced the DNA sample backlog. The turnaround time from sample reception to database entry has been reduced from just over 2 years to less than 60 days.
Read the complete report Filling a Critical Need by Establishing a Fully Functioning, CODIS Dedicated Laboratory (pdf, 101 pages)
Forensic Investigation Techniques for Inspecting Electrical Conductors Involved in Fire
Authors: Richard J. Roby and Jamie McAllister
This research focused on distinguishing characteristics for energized and nonenergized wires subjected to various types of fire exposures. This improved understanding of the various electrical and thermal conditions that can produce beads on electrical wires. Based on preliminary studies conducted by the authors, they hypothesized that characteristic “arc-beads” could form on nonenergized wires as well as energized wires. They also hypothesized that beads form on wire because of its thermal state, not its energized state. These proposed hypotheses are different from the current state of the art in the field, which holds that beads only form on energized wire. The findings support the hypotheses. No trends or distinguishing visual or microscopic characteristics between energized and nonenergized wires were found in the samples reviewed. The researchers tested just over 190 wires under various fire conditions. Wire types included 12-gauge and 14-gauge solid conductors and 16-gauge and 18-gauge stranded conductors. Wires were tested in both an energized and nonenergized state. Energized wires were tested under load and no-load conditions. Under load conditions, the energized wires were plugged into a 119- to 120-volt power source with 9 to 113 amps of current. Under no-load conditions, the wires were plugged into the power supply but no current was flowing in the circuit.
Read the complete report Forensic Investigation Techniques for Inspecting Electrical Conductors Involved in Fire (pdf, 259 pages)
Implementation of a DNA Triage and Analysis System Dedicated to Increasing the Throughput of High Volume Crimes in a Forensic
Laboratory
Authors: Elizabeth Thompson, Mary Hong, Camille Hill, and Scott Scoville
The Orange County Crime Lab (OCCL) adopted a twofold approach to improve its screening and processing of property-crime DNA evidence. The Property Crime DNA Program featured the use of a sophisticated DNA case submission and triage system and a team approach to process biological evidence from property crimes using a newly created high-volume-crime DNA platform. Representatives from the lab and the Orange County District Attorney’s Office evaluated DNA analysis requests submitted to the lab from county police agencies. Forensic DNA analysts examined property-crime DNA analysis requests from a scientific basis and prosecutors evaluated the same requests from a legal perspective. DNA analysts were assigned to a team that was devoted exclusively to processing property-crime DNA samples using a customized scheduling system and an automated DNA processing platform. Team members cooperated in screening, analyzing and delivering property-crime-scene DNA sample results each week. Because DNA analysis was expected to take no longer than 15 workdays, analysts aimed to give law enforcement results for property crime cases within 25 workdays. This goal was not achieved within the project’s time constraints. Instead, the average turnaround time for providing DNA results to police investigators was 74 workdays.
Read the complete report Implementation of a DNA Triage and Analysis System Dedicated to Increasing the Throughput of High Volume Crimes in a Forensic Laboratory (pdf, 133 pages)
Reducing Uncertainty of Quantifying the Burning Rate of Upholstered Furniture
Author: Marc L. Janssens
The goal of this study was to develop a set of guidelines to be used by crime scene investigators for predicting the burning rate of upholstered furniture and for reducing the uncertainty of the predictions. The researchers conducted two series of full-scale furniture and room calorimeter tests. The first series of tests was designed to quantify ignition scenario and enclosure effects on the heat release rate (HRR) of upholstered furniture for three upholstered furniture burning rate models (Babrauskas, Babrauskas 2, and CBUF). The second series of tests involved the use of 22 sets of used upholstered furniture. In these tests, 27 full-scale room fire tests were conducted on at least one item in each set of furniture. The results of the second series of tests were used to assess the predictive capacity of the three upholstered furniture burning rate models. The assessment found the three models significantly under predicted the peak HRR of upholstered furniture. The assessment also found that the accuracy of the Babrauskas model needs to be improved. Besides testing the predictive capability of the three models, the study also investigated factors that affect the HRR of burning upholstered furniture. These include the size of the ignition source (small flame vs. large flame), the source, the ignition source location and the padding material used in the furniture. Implications for policy and practice are discussed.
Read the complete report Reducing Uncertainty of Quantifying the Burning Rate of Upholstered Furniture (pdf, 208 pages)
Spontaneous Ignition in Fire Investigation
Authors: James G. Quintiere, Justin T. Warden, Stephen M. Tamburello, and Thomas E. Minnich
This scientific description of spontaneous ignition and an associated theory provides fire investigators with the knowledge
needed to decide whether such ignition has occurred.
Spontaneous ignition is initiated through heat transfer or the inability to cool a hot material. The process starts as a chemical
reaction that is not yet combustion; the result of ignition can be smoldering or flaming. The key variables involved in this
process are the size of the material, the nature of the heat transfer and the particular chemical and physical properties
of the material. In any fire investigation, it is important for the investigator to recognize the signs of spontaneous ignition
and to learn how to decide whether it was possible in the particular fire pattern. This requires collecting samples for measurement.
It is also important for forensic laboratories to be able to perform tests that can determine whether spontaneous ignition
has occurred. Three scenarios for spontaneous ignition are considered in this report: (1) a cold material in hot surroundings,
(2) a material on a hot surface, and (3) a hot material in cold surroundings. About 200 investigators took part in a survey
about their experiences with spontaneous ignition. Most of the incidents of spontaneous ignition reported involved materials
infused with linseed oil, wet hay storage, and problems with clothing just taken from a dryer. Other cases involved potting
soil, mulch, and various other materials. The current study focused on linseed oil and cotton.
Read the complete report Spontaneous Ignition in Fire Investigation (pdf, 69 pages)
Thermal Properties Database
Authors: Arnaud Trouvé and Thomas Minnich
This report describes the process of creating a thermal-properties database, along with its features. The goal of the centralized database was to streamline the process required to determine the thermal properties and burning behavior of materials and objects. The database was developed in two components: a material property component and an object component. The material property component contains information on thermal properties, ignition temperatures, and critical heat flux at ignition for different representative materials (metals, plastics, woods and miscellaneous materials). This component contains cone calorimeter test data in which fuel sources are described using a material-science-based microscopic perspective. The object component of the database contains information on post-ignition fuel mass loss rates, heats of gasification, and heats of combustion of different representative flammable objects. This component contains furniture calorimeter test data in which fuel sources are described using an engineering-based macroscopic perspective. The database is a stand-alone tool intended for use in hand calculations and analysis by fire investigators, forensic scientists, fire safety engineers, and researchers. It also provides valuable input data for computer-based fire modeling.
Read the complete report Thermal Properties Database (pdf, 25 pages)
Use of Scanning Electron Microscopy/Energy Dispersive Spectroscopy (SEM/EDS) Methods for the Analysis of Small Particles Adhering
to Carpet Fiber Surfaces as a Means to Test Associations of Trace Evidence in a Way That is Independent of Manufactured Characteristics
Authors: David A. Stoney and Paul L. Stoney
This project developed and tested an innovative instrumental trace-evidence approach for the recovery and quantitative analysis of very small particles (VSP) that adhere to carpet fibers. The project developed methods for quantitatively removing VSP from carpet fibers and preparing them for analysis by SEM/EDS (scanning electron microscopy/energy dispersive spectroscopy), and it used an existing computer-assisted SEM/EDS method for testing whether the resulting VSP profiles are useful in quantitatively associating shed fibers with a source carpet. The study demonstrated the regular occurrence of hundreds to thousands of VSP on individual carpet fibers. The quantity and character of VSP were sufficient to associate fibers with their carpet area of origin. The findings led to a rejection of the hypothesis that there is a strictly quantitative relationship among VSP, as measured using environmental particle profiles. Environmental particle profiles were found to be unsuitable for assessing VSP variability. An alternative method was developed based on target particle types (TPTs) defined by their elemental profiles, which were measured by computer-assisted SEM/EDS. Within-carpet and between-carpet variations showed an approximately even distribution for most TPTs, and between-carpet variations showed a wide range in types and quantities of VSP. The usefulness of VSP in linking carpet fiber evidence has been established. There is now an achievable potential to use VSP for independent, quantitative testing of the common origin of carpet fibers.
Law Enforcement
Automated License Plate Recognition Systems: Policy and Operational Guidance for Law Enforcement
Authors: David J. Roberts and Meghann Casanova
This project was designed to assess automated license plate recognition (ALPR) implementation among law enforcement agencies in the United States, and to identify emerging implementation practices to provide policy guidance to the field. Law enforcement agencies are adopting ALPR to improve capabilities, expand their collection of relevant data and expedite the process of comparing license plates with lists of vehicles of interest. Within seconds, ALPR systems automatically capture images of license plates, transform those images into data, compare the data to databases and alert officers when there is a match. ALPR must be carefully managed to ensure quality of data, security of the system, compliance with laws and regulations, and privacy of information. A random sample of 500 state, local and tribal law enforcement agencies was surveyed, and 305 agencies responded to the survey. Three-quarters of respondents (235 agencies) indicated they were not using ALPR and 70 agencies said they were using ALPR. A longer, more detailed follow-up survey was sent to the 70 agencies that were using ALPR; 40 agencies responded. Respondents typically implemented mobile ALPR systems and were using ALPR for auto theft, vehicle and traffic enforcement, and investigations. Agencies reported increases in stolen vehicle recoveries, arrests and productivity. More than half of the agencies updated their ALPR hot lists wirelessly, and nearly half updated these lists daily. Forty percent of respondents retain ALPR data for 6 months or less, five respondents retain ALPR data indefinitely, and two indicated that retention is based on the storage capacity of the equipment installed.
Read the complete report Automated License Plate Recognition Systems: Policy and Operational Guidance for Law Enforcement (pdf, 128 pages)
Determining the Relative Impact of PSAs and Brochures upon General Public Drivers Interfacing With Emergency Service Vehicles
Authors: William F. Jenaway, Steven Austin, William Troup, and Darren Basch
This project examines the impact of public service announcements and brochures on drivers. Results show that viewing information suggesting that a driver move over and slow down when meeting an emergency vehicle on the highway will result in actions that are more responsive. This report supports four conclusions. First, “Slow Down and Move Over” is an effective and easy-to-remember safety phrase. Second, the use of PSAs in communicating slow down and move over practices should be continued and expanded. Third, the results of accidents and PSA impacts should be monitored to decide message variations in the future. Fourth, research on how distractions affect drivers when they see emergency responders and their vehicles provides valuable driver safety data. The data collected included a literature review and related field testing findings using driving simulation.
Read the complete report Determining the Relative Impact of PSAs and Brochures upon General Public Drivers Interfacing With Emergency Service Vehicles (pdf, 48 pages)
