Is Sherlock Holmes' "reasoning backwards" a reliable method for discovering truth?

Analyses of four medicolegal cases.

Note: The figures referenced in the article and described at the conclusion of the article were not published on the website due to their graphic nature. They are available upon request. Simply contact me with your request.

The method of “reasoning backwards” expounded by mystery author Sir Arthur Conan Doyle through his famous fictional detective, Sherlock Holmes, begins with observing items of physical evidence and from them surmising the past events that led to the evidence. Although used with seeming great success by the great detective and many other fictional detectives, scientists have not to any significant extent examined this technique to see if it works in reality. The four medicolegal cases presented in this study not only reveal apparent problems and limitations with reasoning backwards but also demonstrate the robust approach of “reasoning forwards” in most cases — first by allowing the hypothesis for past events to be generated from the observations of eyewitnesses and then by comparing the eyewitness-generated hypothesis to the physical evidence for consistency or inconsistency.
Reasoning backwards; reasoning forwards; hypothesis; physical evidence; eyewitness; Sherlock Holmes


“In solving a problem of this sort, the grand thing is to be able to reason backwards. That is a very useful accomplishment, and a very easy one, but people do not practise it much. In the every-day affairs of life it is more useful to reason forwards, and so the other comes to be neglected. There are fifty who can reason synthetically for one who can reason analytically.”

“I confess,” said I, “that I do not quite follow you.”

“I hardly expected that you would. Let me see if I can make it clearer. Most people, if you describe a train of events to them, will tell you what the result would be. They can put those events together in their minds, and argue from them that something will come to pass. There are few people, however, who, if you told them a result, would be able to evolve from their own inner consciousness what the steps were which led up to that result. This power is what I mean when I talk of reasoning backwards, or analytically.”

“I understand,” said I1.

Famous mystery author Sir Arthur Conan Doyle explained the method behind Sherlock Holmes’ brilliant crime solving skills through a conversation of his fictional hero with Dr. Watson, Holmes’ fictional colleague. Unlike ordinary detectives and police officers, Conan Doyle’s hero combined intuition with careful observation to solve seemingly unsolvable crimes. “Reasoning backwards” from clues at the crime scene successfully allowed Holmes to discern the past events that brought about the clues. The creator of the Sherlock Holmes character made sure that the detective almost always inferred the correct answer with this method.

This plot device of reasoning backwards has been employed by writers of detective fiction over many years and persists in many of the forensic and crime solving novels and television shows of today. On its face, this device possesses an aura of plausibility that appeals to many readers and viewers of mystery fiction.

That same aura of plausibility has also appealed to forensic scientists and physicians. Over many years since the inception of the forensic sciences to this present day, doctors have also heavily utilized backward reasoning. Conan Doyle believed he offered through Sherlock Holmes a legitimate method of forensic reasoning inspired by his mentor and medical professor, Dr. Joseph Bell2. Dr. Bell not only presented demonstrations during his medical lectures of the kind of backward reasoning that inspired Conan Doyle’s Sherlock Holmes but Bell also utilized what he called “the method” in service for Scotland Yard3. Scientific philosophers and advocates during the Victorian era such as Thomas H. Huxley advocated this form of “retrospective prophecy” for the historical sciences of geology and paleontology4. This technique remains heavily in use to this present day in forensic medicine and science just as fingerprinting is still used for forensic detection and identification. It is puzzling, however, that the National Academy of Sciences in the United States of America recently criticized the scientific basis of fingerprinting5 but never once addressed the suitability and limitations of backward reasoning6. Only two articles in the forensic pathology literature, to the author’s knowledge, address the limitations of backward reasoning but only as small portions of longer, more general treatises78.

This paper presents analyses of four medicolegal cases. The first three cases demonstrate apparent problems with reasoning backwards from physical evidence to the past events that were surmised to have led to that evidence. One of the cases is famous and thoroughly described in a book9. The names of the individuals from this famous case from Australia are identified in the case presentation. Two other cases are from the consultation file of the author of this article. The final case exhibits not only a limited and legitimate role for reasoning backwards but also a useful, robust approach of “reasoning forwards” utilized by the author of this article. Prior to the presentation of the cases and their analyses, a general explanation of reasoning forwards and backwards as applied to forensic analysis is provided.

General explanation of reasoning forwards and backwards

The terms, “forwards” and “backwards” are in reference to time or, more specifically, a timeline. Movement from left to right along the timeline is considered forward and movement from right to left is backward.

Most cases evaluated by our courts of law involve past events. These events no longer exist in the present; they are now abstractions that exist only in the form of memory or record. Without the memories of those who witnessed the events and without records, these events would be unknown. The court system by necessity relies heavily on the memories of witnesses and the recordings of past events, frequently in written or audiovisual form.

If a past event involves a possible crime or tort, investigators and scientists are summoned to collect and evaluate evidence. The evidence may be in the form of witness evidence derived from memory or record (anamnestic evidence) or tangible evidence providing potential information about the critical past events (physical evidence). The physical evidence lies to the right of the past events along the timeline. The findings from a medicolegal autopsy, radiographs, and DNA testing are examples of physical evidence.

Scientists, including physicians, utilize the scientific method. This time-honored technique involves first the observation of phenomena. To explain the phenomena, a hypothesis is generated. This explanation in the form of a hypothesis allows the prediction of certain events if the hypothesis is true. This prediction is evaluated by testing. In medical practice, the testing involves the ordering of tests from the laboratory, the radiology department, or from other physicians who are trained to perform specialized tests. In the natural and physical sciences, such testing is in the form of an experiment where the scientist directly studies the factors to be tested and controlled under well-defined circumstances. If numerous scientists predict phenomena accurately on numerous occasions and if a hypothesis resists numerous attempts at falsification10, the hypothesis may eventually become a theory in the basic science context or a syndrome in the medical context.

Reasoning forwards and reasoning backwards differ from one another in hypothesis formation. With reasoning forwards, the hypothesis is generated from the anamnestic evidence, or essentially the observations of others who witnessed what happened. This hypothesis then allows testing in the form of comparison to physical evidence. If the physical evidence agrees with what would be predicted by the hypothesis, the physical evidence is then considered to be consistent with the anamnestic evidence or witness accounts. If the physical evidence does not agree with what would be predicted by the hypothesis, the physical evidence is then considered to be not consistent with the anamnestic evidence or witness accounts. This is forward reasoning because the scientist looks at the physical evidence from the perspective of the witness accounts, starting from the left on the timeline and looking to the right.

Sherlock Holmes claims that, “Most people, if you describe a train of events to them, will tell you what the result would be.” Note that the word, result, is singular. This is because the precise order and timing of events lead to one result and one result only. That one result is also readily and accurately predicted, so much so that “most people” — even people without special skill or practice — can successfully do it. In the scientific and medical context, however, one may need knowledge of anatomy, physiology, pathology, and basic physics to make predictions in a forward fashion; nevertheless, most who possess such knowledge may be able to compare accurately and unambiguously whether the anamnestic evidence is consistent or not consistent with the physical evidence.

For example, if a witness gave an account of someone holding a revolver, pressing the muzzle of it against his forehead, and firing it, a forensic pathologist would readily predict a contact gunshot wound in the forehead. Making such a finding at autopsy would make the witness account consistent with the physical evidence. On the other hand, finding an indeterminate or distant range gunshot wound in the right side of the jaw at autopsy would be inconsistent with the witness account. Knowledgeable forensic pathologists make such comparisons easily and without ambiguity.

On the other hand, with backward reasoning, the hypothesis is generated from the physical evidence. This is the way most scientists are accustomed to making hypotheses, but unfortunately, many of them seem unaware that this method in most cases does not seem to work reliably for past events. Past events no longer exist and cannot be observed by scientists. Furthermore, scientists cannot reliably make predictions of past events from hypotheses generated from physical evidence because for any set of physical evidence items, more than one possible past event scenario can be generated. Rather than the scientist being allowed the luxury to consider a consistent/not consistent dyad, the scientist now has to consider numerous possibilities, the total number of which is unknown.

Sherlock Holmes claims that “there are few people, however, who, if you told them a result, would be able to evolve from their own inner consciousness what the steps were which led up to that result.” Rather than one result, the scientist would then have to be among a few special individuals who through intuition could correctly surmise the multiple steps leading to the result. With reasoning backwards, deriving the correct answer becomes problematic because of the numerous possibilities presented by the evidence. The scientist has to rely on a “sense” of probability, selecting the answer that “seems” right using intuition guided by limited experience. What was hoped to be an analysis with a modicum of scientific method now looks more like a subjective determination, potentially heavily influenced by one’s biases.

Consider the previous example of the contact gunshot wound. A contact gunshot wound in the head could be self-inflicted or could be inflicted by someone else. It could be inflicted at any time day or night by anyone under a wide variety of potential circumstances. With such numerous possibilities, the investigator or scientist may then be left to make subjective assessments as to what seems likely or not likely. He or she might even conclude because of prejudice that the contact gunshot wound represents an execution-style murder, even in the face of discrepant anamnestic evidence.

Another backward reasoning example of current medical interest involves retinal hemorrhages in children. There are many who support the use of the examination of a child’s retinas to determine whether or not abusive head injury is “likely” to have occurred, believing that severe retinal hemorrhages are “more common” with child abuse11. Determinations of this nature however become more problematic and unreliable as scientists discover more and more potential causes of retinal hemorrhages besides child abuse1213.

In spite of these apparent limitations, there appear to be three situations where backward reasoning is not only legitimate and useful but also necessary.

1. Backward reasoning applied as a heuristic during an investigation. At the outset of an investigation, little may be known about a case. Both anamnestic and physical evidence must be developed in order to find answers to potential questions that may be asked during a court proceeding or other kind of formal systematic evaluation. In order to generate leads and to anticipate future questions that may be asked, an investigator may use his or her experience and common sense to surmise what may be likely or unlikely in a given case. As such, backward reasoning employed as a heuristic — a loosely defined, common sense, trial and error method — allows investigators to investigate thoroughly and to find the answers to potential questions. It allows them to generate lists of witnesses and to consider other forensic tests in order to rule out or rule in foul play.

Physicians pursuing a diagnosis appropriately apply such backward reasoning to “rule out” critical conditions generated by a differential diagnosis. Diagnostic medicine employs both forward and backward reasoning. A patient history, a past medical history, and a review of systems are important elements of forward reasoning and may lead to the diagnosis of most maladies; however, for the sake of thoroughness and for ruling out other critical and life- threatening conditions implied by the patient’s signs and symptoms, the physician will also generate a differential diagnosis to guide him or her at the outset for ordering important studies.

Backward reasoning during an investigation helps generate leads but it should not substitute for the forward reasoning that should prevail once all the leads are explored and developed.

2. Backward reasoning applied with reasonable certainty when only one plausible explanation for the physical evidence exists. Once an investigation is complete, the witness evidence may be insufficient or completely absent. A case may hinge on circumstantial evidence when there are critical gaps in the timeline of witnessed events. In such situations, a forensic scientist or physician may be able to reason backwards and be reasonably certain when there is only one plausible explanation.

Assigning a manner of death as homicide to a child found dead in the woods from numerous stab wounds and massive blunt force head injury is an example of this form of backward reasoning. Other manners of death — suicide, accident or natural — do not provide plausible explanations for the physical findings, and it is not reasonable to declare a death under such circumstances as “undetermined” for manner.

Reasoning backwards in this way often requires a finite set of categories to assign a death or an injury. Manner of death is an example of such finite categorization because there are only five categories. Even determining if a child death or injury is “accidental” or “non-accidental” — a categorization often applied by pediatric health care providers — is another example. Regardless of the simplicity of the categorization, a scientist or forensic physician can only be reasonably certain if there is only one plausible explanation. More than one plausible explanation then requires the application of probability, making the explanation less than certain.

This limitation of backward reasoning can be stated succinctly in the following way:

One can be reasonably certain if witness accounts of the past are consistent or not consistent with physical evidence in the present, but one cannot reliably surmise past events from physical evidence unless there is only one plausible explanation for that evidence14.

The first part of the statement summarizes forward reasoning. This approach allows a scientist or physician to be reasonably certain because a single set of past events will have only one physical evidence outcome and not multiple outcomes. The latter part of the statement explains the only reliable form of backward reasoning – reliable because it offers only one explanation and not multiple potential explanations. If there is more than one potential explanation for the evidence, the scientist or physician must couch these explanations in terms of uncertainty – even employing terms such as “possible” or “probable.” The expert however should never express that he or she is certain if there is more than one plausible explanation.

3. Backward reasoning applied as a falsification strategy in court. This approach is also used in circumstantial evidence cases where there is no witness account or if witness accounts are insufficient. Backward reasoning in this way may be employed by prosecuting attorneys to evaluate the merits of a case for trial or by defense attorneys to defend a client in trial. These attorneys simply provide other plausible explanations for the physical evidence in circumstantial evidence cases; however, if witness accounts exist, it is more persuasive for a defense attorney to use a forwardly reasoned hypothesis from witnesses to counter the prosecutor’s backwardly reasoned hypothesis from physical evidence. Juries tend to believe eyewitness accounts if those accounts are entirely consistent with physical evidence.

Case reports and analyses

Case #1

While vacationing near Ayers Rock (now Uluru/Ayers Rock) in Australia on August 17, 1980, the family of Michael and Lindy Chamberlain spent an evening at a campsite with Greg and Sally Lowe, a couple they had only recently met. During that evening, the couple witnessed Lindy Chamberlain rocking her 10-week-old baby, Azaria, to sleep. Sally Lowe saw the baby’s legs kicking while Sally talked to Lindy. When the child fell asleep, Lindy Chamberlain and her oldest child, six-year-old Aidan, walked to the family tent. After placing the infant in a bassinet, Lindy and Aidan returned to family and friends. Four-year-old Reagan Chamberlain was already asleep in the tent.

Later in the evening, Aidan and Michael Chamberlain and Sally Lowe heard a baby’s cry. Lindy returned to the tent and found her baby missing. Upon arriving at the tent, Lindy saw a dingo, a wild Australian canid, exiting the tent, but she did not see the baby with the dingo.

Reagan remained asleep on the other side of the tent. A smear of blood stained the mattress in the bassinet. Several people noted paw prints by the tent flap. The family and others at the campsite initiated a frantic but unproductive search. A week later, portions of plastic diaper, an infant jump suit, a singlet, and infant boots were found 6 km west of the camp. The jump suit and singlet were bloodstained. A matinee jacket with pale lemon yellow edging that the child was allegedly wearing was not found.

During a subsequent investigation, law enforcement officers surmised that the infant was not taken by a dingo but was instead murdered by Lindy. The alternate scenario offered by law enforcement personnel was that the infant was killed by her mother days before August 17 and that Azaria’s parents concocted an elaborate ruse to conceal the crime. Multiple forensic science tests performed on several items of evidence seemed to support the alternate scenario. There were no dingo hairs, dingo saliva, or teeth marks in any of the clothing items. Experiments performed with dingoes confined in the zoo using a kid goat dressed in baby clothing fueled skepticism of the dingo account because these captive dingoes did not leave findings similar to the questioned case. A professor of forensic medicine in London, England examined the clothing evidence in the case and concluded that 1) there was a bloody handprint from an adult female, and 2) that the child’s neck had been cut with a sharp instrument, consistent with scissors or a knife. Also, the family car below the dashboard disclosed a spray of material testing positive for fetal hemoglobin.

Lindy Chamberlain was tried and found guilty of murder. She was released 5 years later when the matinee jacket was found in a dingo lair on the opposite side of Ayers Rock9.

Analysis of Case #1

The tragic events of that evening were all witnessed events, many of them involving several witnesses. Members of the Chamberlain family and a couple only recently acquainted with the Chamberlains noted the infant to be alive and in her usual state of health prior to the mother placing the infant in the tent. Individuals other than the mother later heard the child cry. Also, multiple other individuals at the campsite participated in the search and noted the items of physical evidence, including the bloodstains in the tent and the paw prints by the tent flap.

Utilizing forward reasoning, the hypothesis derived from the witness accounts, including the account of the defendant herself, is that a dingo had taken and had eaten the baby. The physical evidence from these events was as one would predict from this hypothesis, even if one has little knowledge of the nature and habits of dingoes. A powerful wild canid predator would be expected to move quietly and quickly, thereby not disturbing Reagan Chamberlain from his sleep. It would be expected to grasp the child’s head with his powerful jaws, leaving bloodstains in the neck portions of the garments, no tooth marks in the garments, and no evidence of saliva. The dingo would be expected to consume its prey totally and leave the garments and the diaper behind at sites far from the campground. Since the physical evidence was as one would anticipate and since there were testimonies of multiple witnesses to the events of that evening — even from people who barely knew the family — there was no reason logically why the authorities should not have believed the Chamberlains.

Unfortunately, the authorities and scientists decided to reason this case backwards, Sherlock-Holmes-style. To do so, they started with the physical evidence and forensic scientific testing, developed a hypothesis or scenario of foul play based on the testing of the evidence, and strived to confirm that hypothesis without giving proper attention to the testimonies of witnesses. The absence of saliva and tooth marks on the garments then became positive evidence that a dingo did not carry away the baby. The bloodstains in the neck portion of the garments then became evidence that the infant’s throat was slashed. The perceived female handprint in blood and the positive testing for fetal hemoglobin in the car then supported a foul play scenario imagined by police authorities and not supported by the testimonies of multiple witnesses. This left the family and their attorneys needing to provide explanations for the two falsely positive tests. The results of the fetal hemoglobin test and the evaluation of the garment by the London forensic pathologist would never have been sought, much less reported and presented as evidence had the investigators and scientists allowed the witnesses to provide the hypothesis to be tested. Furthermore, the experiments performed on dingoes in captivity may have provided interesting data about the behavior of dingoes in captivity, but no experiment performed in the present will answer satisfactorily any issue occurring in the past. Experiments could not determine what happened on the evening of August 17, 1980. On the other hand, the observations of multiple witnesses who viewed the events of that evening provide better answers.

With forward reasoning, there is minimal bias introduced into the analysis. This is because the hypothesis is generated from the eyewitness accounts, and the scientist is not given the opportunity to introduce bias from his or her own internally generated hypothesis. Certainly, witnesses may be biased, but that bias can be discovered and tested through a careful comparison with the physical evidence. Also, simply determining whether or not the evidence is “consistent” or “not consistent” with the witness accounts does not allow room for rampant speculation.

On the other hand, backward reasoning is particularly susceptible to bias. When the scientist is allowed to generate the hypothesis from physical evidence and to make the kinds of probability assessments inherent with backward reasoning, the influence of the scientist’s experiences and his or her relationships with others will color the hypotheses he or she may generate15. In this case, media misunderstanding of the family’s religious practices and the general public perception of the mother as being cold and unfriendly appeared to fuel the bias of law enforcement officers and forensic scientists against Lindy Chamberlain. Only the discovery of the matinee jacket five years later in a distant dingo lair far from the activities of the Chamberlain family was sufficient to reverse these negative perceptions.

Case #2

While on patrol in the town of Blagoevgrad in Bulgaria, five police officers in a police vehicle detained a known drug dealer in that town who was driving another vehicle. The police officers spotted the drug dealer at 2050 hours, and attempts to detain him began at 2053 hours on the evening of November 10, 2005. Four of the police officers exited their vehicle and chased the subject on foot as he ran and attempted to hide among nearby apartment buildings. Upon reaching him, two of the officers, according to testimonies from the officers translated into English, grabbed his arms but were successful in placing a handcuff around only one of his wrists. During the struggle, the subject lunged forward to the ground along with the three officers restraining him. Over an area of about 5 to 6 meters in diameter, the officers struggled with him in the dirt for a time interval variably reported by the police officers from 2 to 5 minutes. Throughout the struggle, the detained person loudly cursed and yelled. This was heard by several nearby who heard the commotion but were unable to see clearly in the darkness of the evening. Eventually, the subject stopped struggling and cursing, and the police officers noted him at that point to be unresponsive. The police officer remaining in the vehicle summoned an ambulance at 2109 hours. The other police officers did not perform cardiopulmonary resuscitation. A doctor and an ambulance crew arrived at 2111 hours. Using a small flashlight, the doctor noted the pupils to be fixed and dilated, and he was unable to palpate a radial pulse. The ambulance transported the subject to a medical facility where an electrocardiogram at 2121 hours demonstrated asystole.

Three pathologists performing the autopsy the following day noted cyanosis of the head and upper body. Hemorrhages, including petechial hemorrhages, involved the conjunctivae of both eyes. Multiple abrasions involved the face, but there were no facial fractures. A small laceration involved the lower lip. Multiple abrasions were scattered throughout the body. Soft tissue hemorrhages corresponded to some of the abrasions, and there was also soft tissue hemorrhage in the back adjacent to the third and fourth thoracic vertebrae. Also internally, hemorrhages in deeper scalp tissues lay in the right forehead and temple areas, in the left temple, and over the left parietal bone. There were no skull fractures. The brain revealed some broadening and flattening of the gyri with narrowing of the sulci, and there were grooved impressions in the cerebellar tonsils. Although two small hemorrhages were described in the cerebrum initially, a subsequent examination did not note any evidence of hemorrhage in the brain. Internal neck structures were intact and without evidence of trauma. Numerous firm adhesions lay in both pleural spaces. The heart was structurally normal. A 5 mm defect lay in the arch of the aorta. Fluid blood stained the tissues around this defect, and 150 ml of blood lay in the left chest cavity. The pathologists found no other visceral injuries. Toxicologic testing later disclosed small amounts of cocaine metabolite in the urine only and no other drugs.

Following the autopsy, the body was released that afternoon to the funeral home, and the pathologists stated the cause of death as “cardiogenic shock and acute heart failure.” The following day, the body was removed from the funeral home, and five other pathologists performed a second autopsy. The gross findings from this autopsy were essentially the same as the first autopsy, but their microscopic examination indicated several additional microscopic hemorrhages lying in the pons and medulla. Portions of the brain were reported to show severe edema. The five pathologists concluded that the subject died as a result of brain trauma that put him into a coma prior to his death. Furthermore, blows to the spine, they opined, resulted in a tear in the aorta, and the precipitous death from the head trauma allowed only 150 ml of blood to leak into the chest cavity.

The five police officers were sentenced to a total of 82 years in prison. An appeal of that conviction was unsuccessful.

Analysis of Case #2

As with Case #1, this case also had multiple witnesses. The police officers offered detailed accounts of what they observed from several different points of observation. When the physician and the ambulance personnel arrived at the scene, they found the subject in cardiac arrest and not in a coma. Forward reasoning allows a cardiac arrest following a violent struggle to be readily inferred from these accounts. The phenomenon of sudden cardiac death during violent activity and restraint, even in individuals without significant drug levels or structural heart disease, is common and amply described in the medical literature under a variety of settings involving both hospital personnel and law enforcement officers1617. This inference is consistent with the autopsy evidence. The abrasions and contusions in many areas of the body are consistent with the account of a struggle, but these indications of struggle in and of themselves were not fatal injuries. Conjunctival petechial hemorrhages are consistent with a mechanism of death either from sudden heart stoppage or asphyxia, but the constant yelling and cursing throughout the struggle indicate that the subject was at least able to breathe sufficiently to phonate loudly. Even the 150 ml of blood in the left chest cavity indicate a lack of circulation from sudden heart stoppage because a large volume of blood in the chest would have been anticipated if the heart had been functioning. One of the pathologists performing the initial autopsy conceded during his testimony in court that they could have cut the aorta during the removal of the organs.

On the other hand, findings from the second autopsy of a dubious character, namely the microscopic hemorrhages later discovered in the brainstem and the discovery of severe cerebral edema also through the microscope, supported only an imagined scenario and were not consistent with the detailed accounts of five witnesses. Just as with the bloody female handprint in Case #1, the findings of forensic scientists became misleading and falsely accusatory when used to support a scenario provided by the scientists themselves rather than the witness accounts.

As with the first case, bias appeared to have played a role in the backward reasoning of the pathologists performing the second autopsy. Throughout the world, there is a general perception that police officers are brutal and that they conspire together to cover up bad behavior. Unfortunately, these assumptions are not scientific and not fair to individuals who may spend major portions of their lives incarcerated. Also, it is not plausible to conclude that five individuals — each interested in his own fate and not necessarily the fate of the others — could quickly invent stories that are consistent in every detail with the physical evidence, particularly when they possessed little knowledge of anatomy, physiology and pathology. The consistency of the accounts with the physical evidence seems to support the truthfulness of the accounts.

Case #3

Two biological parents not married to each other shared custody of an 18-month-old male child. According to their arrangement, each parent would take care of the child for one week and then transfer the child to the other parent. The female parent lived in Chillicothe, Missouri, in the United States of America, and the male parent lived in Lenexa, Kansas.

The maternal grandparents of the child transferred the child to the biological father at a point between Chillicothe and Lenexa. The father took the child to the home of his current girlfriend’s parents in Lenexa at noon on Saturday, March 12, 2005.

On the following Monday, March 14, 2005, the father delivered the child to a childcare facility at 1453 hours. Personnel at the facility did not report any unusual events, and they did not report any injuries in the child during the time they took care of him.

The girlfriend of the father of the child picked up the child from the facility at 1700 hours. She brought the child to the house in Lenexa, KS. A route check performed by the police indicated that the drive time from the daycare to the house would be almost 16 minutes for that hour of the day. A neighborhood canvass performed by the police disclosed one witness who stated that she saw the girlfriend walking to the front of the Lenexa residence holding the hand of the child.

According to an interview of the girlfriend conducted shortly after the events of the afternoon, she heard a “hard knock” in the living room while she was in the kitchen warming up a dinner to deliver to the child’s father who was working that afternoon. She went into the living room and found the child lying face down on the carpet next to a coffee table. The child first screamed, then became quiet. She picked the child up, and as she held him in her arms, she noted that he was “limp” and his head was drooping. The child vomited after being given water to drink.

The girlfriend telephoned the father of the child, and upon his instruction, she drove to his workplace with the child. The father clocked out at 1751 hours and drove the child to the hospital with the girlfriend. The child was conscious, according to the father, but the child was drifting off to sleep. The girlfriend stimulated the child to try to keep him awake during the drive to the hospital.

At 1800 hours, a nurse in the triage area noted the child to be pale and lethargic. When asked why, the girlfriend reported that the child “fell and hit his head.” The nurse and other health care providers noted that the child’s right pupil was larger than his left. In addition to the dilated pupil, the emergency room physician palpated obvious deep scalp swellings on both sides of his head. The child had Glasgow Coma Scores varying from 6 to 3, with the worst score notated at the end of his emergency room stay. A hemoglobin level drawn at 1850 hours was 9.9 g/dl. A helicopter delivered the child to the children’s hospital in Kansas City, Missouri, arriving there at 1923 hours.

The first computed tomography (CT) scan reportedly performed at 1946 hours disclosed a right frontoparietal convexity subdural hematoma and a 9 mm right-to-left midline shift of the cerebrum. The hematoma was of mixed density and interpreted to be consistent with a hyperacute subdural hematoma. A large cranial fracture revealed fracture diastasis of 3 to 4 mm. After comparison with a follow-up CT scan taken the following morning, the radiologist and the child abuse pediatrician further opined that the injury occurred 2 to 3 hours before the initial CT scan.

The child was pronounced dead at 1617 hours on March 15, the day following admission to the children’s hospital. An autopsy performed on the day following the death disclosed a wrap-around fracture with gaping fracture diastasis and associated cephalhematomas (Figures 1 and 2) and a right frontoparietal subdural hematoma (Figure 3). Also seen at the autopsy was a purple mark in the posterior scalp about 3.5 cm in diameter (Figure 4), described by the autopsy pathologist as a “strawberry hemangioma.” Histologically, the lesion was reported as showing “multiple small sclerotic blood vessels and no hemorrhage consistent with a hemangioma.” The review of the histological section by the author of this paper revealed only normal scalp tissue. Lying beneath the posterior scalp lesion was a branching focal calvarial fracture from which emerged the wraparound fracture (Figure 1). The autopsy pathologist also found “infrequent retraction bulbs in the brainstem consistent with diffuse axonal injury,” but the author of this paper found no retraction balls on subsequent review of the sections. Microscopic examination of the dura revealed hemorrhage but no evidence of hemosiderin or fibroblast proliferation. The eyes revealed retinal hemorrhages.

Multiple pediatricians and the autopsy pathologist expressed opinions that the severe head injury was “non-accidental,” that a simple fall in the living room onto a coffee table would not cause sufficient force to explain the injury, that it would take a head first fall from a three or four story building to cause a comparable injury, that the child would have been immediately unconscious and not able to take a single step, and that the incident appeared to occur while the child was under the care of the father’s girlfriend.

The girlfriend eventually pled “no contest” to charges of murder after being unable to find a pathologist supportive of her account of the events. Although the author of this article became involved late in the course of this case after the no-contest plea and testified at a hearing where there was an attempt to remove the plea, the trial judge still declined to allow a trial by jury and sentenced the girlfriend to prison where she remains to this day.

Analysis of Case #3

The defendant herself surmised that the child had injured himself from a fall — a fall she did not see but only heard in the form of a “hard knock.” The severe nature of the injury — the gaping wraparound fracture, the subdural hematoma, and the cerebral edema — is not consistent with the child falling either against a coffee table or the floor. Such a simple fall as described by the defendant lacks the energetic elements that characterize such a remarkably severe injury. This apparent lack of consistency between the defendant’s statement and the clinical/autopsy findings seemed to provide evidence to the law enforcement officers and the physicians that she was not speaking truthfully about the event and that she intended to cover up abusive behavior. Is it possible, however, that these conclusions by law enforcement officers, pediatricians, and the autopsy pathologist were mistaken?

Hymel et al proposes a paradigm for the analysis of pediatric head injury18. Although the paradigm is based on backward reasoning (“in short,” they state in their article, “our paradigm guides pediatric head trauma case analysis backwards — from injury to history.”), it is helpful to consider it initially in the analysis of this case.

Pediatric cranial injuries have both primary and secondary characteristics. The contusion in the left posterior scalp, the focal branching fracture underlying the contusion, the wrap-around fracture extending from the branching fracture, and the right frontoparietal convexity subdural hematoma underlying the wrap-around fracture — all illustrated in Figures 1, 2 and 3 — are the primary focal injuries formed from the impact itself. The secondary injuries, namely the focal cerebral edema leading to a right to left midline shift, the fracture diastasis, and the extensively spread cephalhematomas arising from the fractures resulted from the primary focal injuries and developed over time.

Further following the paradigm proposed by Hymel et al, the constellation of injuries in this case falls under their rubric of “contact injury.” A contact injury is brought about by the forceful impact of a blunt object or the forceful contact with a surface, as opposed to an “acceleration (or deceleration) injury” which is brought about by severe motion of the brain inside the cranium. The injuries do not fit under their rubric of “acceleration (or decleration) injury” because the subdural hemorrhage lies beneath a fracture instead of away from a fracture. A fracture crossing and potentially injuring the superior sagittal sinus and associated bridging veins provides the likely source for the subdural bleeding — not the severe oscillation of the brain within the cranium. Furthermore, the statements from multiple witnesses and the absence of retraction balls due to axonal damage indicate no immediate loss of consciousness from a concussion or no immediate and prolonged traumatic coma from diffuse axonal injury — elements characteristic of an acceleration/deceleration injury. The small diameter of the scalp contusion (3.5 cm) also decreases the likelihood of cranial acceleration/deceleration because significant cranial acceleration or deceleration typically requires a large surface area of impact18.

Accordingly, even further analysis along the paradigm indicates that “the head was struck by a moving impactor” rather than a “fixed surface.” Such impacts from a moving impactor (blunt object) would be more severe if a) the “deformability of the impactor was decreased,” and/or b) the “surface area of contact was decreased (i.e., nondistributed loading),” and/or c) the “impactor mass, velocity and kinetic energy were increased.” The remarkably large wraparound fracture implicates an energetic blow to the cranium by a hard blunt object with a small surface area of impact. The absence of evidence for significant cranial acceleration/deceleration would allow the child to remain conscious for a period of time after receiving the blow but later succumb to unconsciousness from increasing intracranial pressure due to focal cerebral edema and a subdural hematoma.

Such an analysis utilizing this paradigm — a paradigm based on backward reasoning — is useful in a limited sense, but it fails to provide answers to important questions. Could the defendant have landed a severe blow to the backside of the child’s head during the 45 minutes that she was alone with the child, or could that blow have occurred at a previous time — even as far back as the child’s time in Chillicothe? Could the secondary effects of that blow — the extensively spread cephalhematomas associated with the fracture, the diastasis of the fracture line without tearing of the underlying dura mater, the development of the focal cerebral edema leading to a midline shift — become manifest in less than an hour? Would it take less than an hour for the child to become anemic from intracranial and deep scalp hemorrhage, or was the child anemic prior to the blow from another cause? Is the accuracy of the timing of injuries using a head CT scan sufficiently established to allow a radiologist or pediatrician to be reasonably certain that an injury occurred only 2 to 3 hours before the scan was taken19?

Scientists may speculate and reason backwards, yet never come to a satisfactory conclusion suitable for being reasonably certain. This is because the scientific data adequate to answer the questions above in a clear, unambiguous way do not exist.

On the other hand, reasoning forwards may provide answers in a remarkably simple way.

With forward reasoning, the hypothesis is derived from witness accounts but not witness explanations. A witness may provide explanations for events — such as a child developing a head injury from a fall as in this case — but those explanations may be from ignorance of the underlying biomechanical realities. On the other hand, the observations of the witness without her interpretations may prove to be remarkably accurate.

When the girlfriend first observed the child, the child was conscious, but there was something wrong with the child’s neurologic function. The child was limp. He vomited the fluid he had been given to drink. The child had to be stimulated to remain awake. All of these are signs of increased intracranial pressure. The observations of signs of increasing intracranial pressure made by medical personnel — specifically the dilated right pupil, the shifting consciousness in the emergency department, the midline shift of the cerebrum, and the fracture diastasis — are entirely consistent with what the defendant observed at the outset of the child’s abnormal behavior. This is remarkable because the defendant would have no knowledge at the time of her statement how increasing intracranial pressure manifests itself in the form of signs and symptoms.

The defendant does not describe an event capable of causing the injuries noted in the child’s head. A hypothesis derived from all witness statements accordingly would be that the underlying reason for the increasing intracranial pressure — the initial primary injury — occurred at a time prior to the involvement of the girlfriend. The physical evidence in this case is consistent with this hypothesis, thereby supporting the defendant’s observations as truthful.

How the injury occurred and when it occurred is not known because there is no witness account to describe it. The injury may have occurred at the daycare outside of the notice of the daycare providers. The injury may have occurred prior to the time in the daycare facility. In either case, one cannot know for certain without a witness account because more than one plausible scenario can be imagined.

Injuries such as the one in this child are not necessarily abusive. One cannot surmise child abuse from an injury when plausible explanations not involving abuse can be imagined. A blow from a golf ball, a baseball or the end of a stick may provide potential accidental reasons for the injury. In any of these cases, the child may suffer immediate pain, but a loss of consciousness might not occur until after a lucid interval.

The pediatricians and the autopsy pathologist, however, felt that they could be certain of child abuse in this case without realizing that backward reasoning might not provide answers to questions with reasonable certainty. It is interesting to note that with their backward reasoning, they could only state that there was abuse but they could not provide in any detail how that abuse occurred. They could surmise that a three to four-story fall could cause injuries of comparable energy, but they could not provide an explanation as to how the defendant — a relatively small woman — inflicted such an injury inside a house to a 13 kg child. Reasoning backwards appears to oversimplify the explanations and to provide more confusion than clarity.

The hypothesis of abuse developed from backward reasoning also caused misunderstanding and misinterpretation of some of the findings. The autopsy pathologist implicated a strawberry hemangioma — a “birth mark” — as the reason for the purple mark in the scalp, yet no one in the child’s family ever described the child as having a birthmark in that location. Also, the presence of retinal hemorrhages — a finding frequently associated with abusive head injury — may also be associated with intracranial hemorrhages in general and sudden increases in intracranial pressure12.

Case #4

On January 17, 1994 at 0253 hours, police and paramedics responded to an apartment in Atlanta, Georgia, United States of America following a 911 distress call. The mother and grandmother of a two-year-old male child allegedly discovered him unresponsive in bed. When ambulance, fire and police personnel arrived, the grandmother — a registered nurse — was performing cardiopulmonary resuscitation (CPR). The child was transported to a local children’s hospital and pronounced dead in the emergency department. Medical personnel found no evidence of injury, but they did suction a small amount of bloody fluid from the stomach.

An autopsy later that morning disclosed patterned abrasions in the mucosa of the upper and lower lips that matched adjacent teeth (Figures 5, 6). No other evidence of injury or disease was noted in the body. A follow-up investigation of the bed and bedroom where the child was found disclosed no significant additional findings.

The pathologist reported to police personnel his concern that the death may have been due to homicidal smothering and that further police investigation was warranted. The police refused to investigate further until the pathologist rendered an opinion that the manner of death was homicide.

The pathologist aided by investigative personnel at the medical examiner office subsequently performed interviews of multiple witnesses. Ambulance personnel described the manipulations performed on the mouth of the child during the resuscitation, including endotracheal intubation. They claimed that not only did they not cause the injuries in the upper and lower lips but they also saw the injuries during the resuscitation. The mother and grandmother of the child were interviewed separately on January 28. A transcriptionist for the medical examiner office recorded the verbatim responses to interview questions in both audio and written forms.

According to the mother, the child was in his usual state of health on Saturday, January 15. The mother said she went to a party with some of her friends on Saturday night. She stayed awake most of the night and did not return to the apartment until nearly noon on Sunday. The grandmother watched the child that morning. He was reportedly active but well behaved throughout the day, and there was no report of any unusual occurrence. The child was with the mother or the grandmother the entire day. He ate popcorn and potato chips with no evident discomfort. The mother put the child in bed at 2000 hours. At approximately 2300 to 2330 hours, the child got out of bed. The mother said she placed the child back in the bed and told him that he was a big boy and needed to sleep. The grandmother was in another bedroom and the mother heard her toss and turn at that time. The following morning, about 0200 to 0230 hours, the mother arose to use the bathroom. She sensed that something might be wrong, so she checked on the child and discovered him in bed not breathing. She awakened the grandmother who performed CPR. The mother then called the 911 telephone operator.

The grandmother provided a very similar account but with one exception. Between 2300 and 2330 hours, the child suddenly awakened the grandmother by screaming, “I want my mommy! I want my mommy!” She arose from her bed, looked into the child’s room, and saw the mother sitting on the bed next to the child, talking to him. The grandmother went back to sleep.

At the end of each interview, the pathologist revealed to the mother and grandmother the images in Figures 5 and 6, asking each if they knew how the child received these injuries. Both were asked if either could have placed their hand or other object over the child’s mouth and forcefully pressed. Neither one could account for the injuries. The pathologist asked the mother and grandmother to contact him if either subsequently recalled how and when the child might have received the injuries. Neither one contacted the pathologist after the interviews.

On February 2, five days later, the pathologist completed the death certificate, stating that the cause of death was smothering and the manner of death was homicide.

Homicide investigators from the Atlanta Police Department interviewed the mother and grandmother at length. Both were administered polygraph testing on February 10, and each test disclosed no evidence of deception.

On February 15, the mother returned to the homicide office for another interview. She admitted at that time that she was very tired, that the child was crying and very loud, and that she put her hand over his mouth to quiet him down. She recalled pressing firmly. The child reportedly tried to push up out of his bed, but she forced him back down into bed with her hand. When asked why she had not disclosed this information, she stated “she felt it was not important at the time.”

The mother was charged with murder. The case was settled and never went to trial.

Analysis of Case #4

Unlike the previous three cases, this case demonstrates the use of mostly forward reasoning and an appropriate but limited form of backward reasoning. The pathologist in this case is the author of this paper.

The finding of patterned abrasions in the mucosal linings of the upper and lower lips is suspicious for foul play, but one cannot be reasonably certain how these abrasions were made without witness accounts.

The police initially did not believe that such witness evidence was needed. They believed the medical examiner could provide the cause and manner of death from the autopsy alone without witness accounts. Abrasions in the lips, however, are not fatal injuries. The abrasions must be placed in the context of a historical account to determine their significance.

The pathologist during the interviews selected January 15 to begin a timeline of witnessed events. He selected that date because the injuries at autopsy did not reveal the kind of healing that usually takes place in two days. The small amount of blood in the gastric fluid also indicated a recent event.

There were no gaps in this timeline. Someone observed the child during the entire time in question. The accounts gave no information that the child might have injured himself outside of the presence of his caretakers. None of the witnesses, including the mother of the child, the grandmother, the fire and ambulance personnel, and the emergency department physicians provided any explanation for the abrasions inside the lips. Although several of the individuals responding to the emergency call believed that they saw the abrasions, none of them described any procedure or event to explain the abrasions, including their methods to establish an airway. Neither the mother nor the grandmother initially described any event consistent with the abrasions nor did they note any apparent discomfort in the child when he ate even salty foods. The hypothesis was essentially that the child died suddenly and unexpectedly without injury, yet the patterned injury was not consistent with that hypothesis.

A discrepancy in the witness accounts was also noted. Why did the grandmother describe the child screaming and the mother not make any mention of the screaming?

Although the mother and the grandmother were given ample opportunity to provide an explanation for the abrasions, neither one did. Abrasions, however, do not appear inside the lips of young children without cause.

With reasoning backwards, one can be reasonably certain if there is only one plausible explanation for the physical evidence. The patterned injuries were from the lips pressed tightly against the teeth, and smothering was the only plausible explanation for that pattern after a careful consideration of all the evidence. Homicide for manner of death was also the only plausible manner.

When the mother provided a form of confession to the smothering act, that statement also needed to be compared to the physical evidence for consistency or inconsistency. This was necessary because even confessions to crimes can be false under a wide variety of circumstances20. In this case, her statement explained the lesions entirely, and there were no inconsistencies.


Despite advances in science and technology, many are dissatisfied with the quality of forensic science worldwide and its application to matters of justice. The National Academy of Sciences in the United States expressed this dissatisfaction recently in a lengthy report21. Although the report heralds some advances, such as with DNA technology, it also notes that innocent people continue to be convicted and imprisoned on the basis of faulty forensic analyses. The report stops short of proposing a solution or solutions; instead, it makes recommendations for further study of this problem.

Also recently, a commission in the province of Ontario in Canada investigated numerous false convictions and imprisonments brought about by the faulty methods of a Canadian pediatric forensic pathologist22. The report makes several recommendations but falls short of pointing out the basic fallacy behind this pediatric forensic pathologist’s analyses. Scientists realize that there is a problem with forensic scientific evidence, but no one seems to understand what the problem entails or how to solve it.

On the other hand, the philosophy of the world’s most famous fictional detective — a philosophy seemingly embraced by forensic professionals in several different disciplines — remains unquestioned and unexamined. Could an unquestioning reliance on backward reasoning have led even in part to the many injustices decried in these major reports?

Currently, it is not known nor discussed to what degree reasoning backwards may be a cause or contributing factor to forensic science mistakes. At present, discussions of backward or forward reasoning do not appear in peer-reviewed medical or forensic science literature. The author is aware of only two instances where the topic was only obliquely addressed in past literature. One is from a published speech delivered to the American Society of Clinical Pathologists in 19567. In it, Moritz criticized a “Sherlock Holmes type reasoning” that he described tersely as “categorical intuitive deduction.” Although Moritz did not explain what he meant, the term is explanatory in the context of the backward reasoning described in this article: after studying the physical evidence in a case, the scientist surmises the events that caused the evidence (“intuitive”) and from that surmising develops a chain of seemingly logical reasons (“deduction”) to place the events into a category (“categorical”). Medical categories to explain past events often have the word, syndrome, associated with them. Shaken Baby Syndrome, Sudden Infant Death Syndrome, and Excited Delirium Syndrome23 are three such categories used by scientists to explain what they believe happened in the past. As explained previously in the analysis of Case #3, such categorizations based on backward reasoning tend to oversimplify the complex and unique chain of past events in any particular case.

Another article by Wright and Tate from 1980 — a general description of the discipline of forensic pathology and the autopsy — compares and contrasts how an academic hospital pathologist and a forensic pathologist determine cause and manner of death in any given case8. The description in the text of the article and the accompanying flow diagrams both state that the hospital pathologist frequently determines the cause and manner of death from the findings of an autopsy while the forensic pathologist instead derives hypotheses from the terminal event, the medical history, and the environment. The forensic pathologist then tests the hypotheses by correlation with the autopsy findings. The terms, reasoning forwards and reasoning backwards, are not used in the article but the concepts are similar.

Other than these limited mentions in the forensic pathology literature, there is no recognition of the limitations of backward reasoning in the medical or forensic literature at all. Much of the literature on child abuse, for example, solely utilizes backward reasoning. Consequently, many of the hypotheses derived by physicians and scientists studying child abuse have caused the formation of simplistic categories bereft of the unique context provided by primary witness accounts for each individual case. Over the years, scientific studies purportedly demonstrated that severe abusive head injuries without evidence of impact were due to shaking2425, that short falls less than 1.5 m in vertical height in infants and children did not cause skull fractures or subdural hematomas26, that abusive shaking of a child was the most likely cause of severe retinal hemorrhages when associated with subdural hematomas27, that children with abusive head injuries experienced an immediate decrease in their level of consciousness at the time of injury28, that certain types of fractures in infants (spiral fractures of long bones, classic metaphyseal lesions, posterior rib fractures, and multiple fractures at varying stages of healing) are “highly specific” for child abuse2930, and that child abuse is a diagnosis of exclusion to be rendered once other natural disease explanations were ruled out31. These general and simplistic statements supported solely by backward reasoning have been offered with scientific authority in courtrooms throughout the world, yet no one seems to question the legitimacy of the reasoning that brought these ideas into existence. Even the critics either attempt to falsify existing backwardly reasoned concepts32 or propose alternate unified hypotheses33, but none of the critics ever questions if reasoning backwards from physical evidence to past events is valid. They also employ the same backward reasoning in many of their assertions.

Questioning the legitimacy of widespread backward reasoning is important because backward reasoning involves issues that go far beyond the hospital or morgue setting. Law enforcement officers rely on the scientific assertions of medical and forensic experts when they perform interrogations. When police investigators receive information from scientists offered with reasonable certainty, the interrogations then no longer focus exclusively on learning witness information. The law enforcement officers then attempt to evoke a confession from a witness through the use of psychological techniques. Once a witness becomes aware that arrest is imminent, he or she may alter his or her witness account to obtain a personally favorable outcome. The witness may also back down and agree with his or her interrogators after experiencing the strain of long hours of questioning. Such “confessions” lead to an unfortunate confirmation bias in the medical literature by providing flawed bases for backwardly reasoned hypotheses3435.

The four cases presented in this article and their accompanying analyses represent the first attempt known by this author to study formally the reliability and limitations of forward and backward reasoning. This subject requires further study by many others. Scientists and physicians should learn the limitations of both forward and backward reasoning and should perform case analyses with forward reasoning, not only as a means to falsify predetermined categorizations but also for learning about child injury and other forensic issues in general. Much of what scientists assume to know has been on the basis of backward reasoning. If further studies of pediatric injury and other topics utilize forward reasoning, the scientific and medical communities may learn that concepts assumed to be true for many years might not be true at all.


The author would like to thank Drs. Lindsey Thomas of Hastings, MN, USA and Daniel W. Young of Birmingham, AL, USA for reviewing the manuscript and offering constructive criticism.


1 Doyle AC. A study in scarlet. In: Klinger LS, editor. The new annotated Sherlock Holmes (Vol. III). New York: W. W. Norton & Company; 2006, pp. 197-8.

2 Snyder LJ. Sherlock Holmes: scientific detective. Endeavour 2004;28(3):104-8.

3 Liebow EM. Dr. Joe Bell: model for Sherlock Holmes. Bowling Green, OH: Bowling Green University Popular Press; 1982.

4 Huxley TH. On the method of Zadig. Project Gutenberg Literary Archive Foundation. 2001. Accessed 8 Nov 2009.

5 National Research Council. Improving methods, practice, and performance in forensic science. In: Strengthening forensic science in the United States: a path forward. Washington, DC: The National Academies Press; 2009. p. 184.

6 National Research Council. The principles of science and interpreting scientific data. In: Strengthening forensic science in the United States: a path forward. Washington, DC: The National Academies Press; 2009. pp. 111-26.

7 Moritz AR. Mistake of substituting intuition for scientifically defensible interpretation. In: Classical mistakes in forensic pathology (American Journal of Clinical Pathology, 1956). Am J Forensic Med Pathol 1981;2(4):303.

8 Wright RK, Tate LG. Forensic pathology: last stronghold of the autopsy. Am J Forensic Med Pathol 1980;1:57-60.

9 Bryson J. Evil angels. New York: Summit Books; 1985.

10 Popper KR. The logic of scientific discovery, 2nd ed. New York: Harper and Row; 1968.

11 Togioka BM, Arnold MA, Bathurst MA, Ziegfeld SM, Nabaweesi R, Colombani PM, Chang DC, Abdullah F. Retinal hemorrhages and shaken baby syndrome: an evidence-based review. J Emerg Med 2009;20(10):300-8.

12 Aryan HE, Ghosheh FR, Jandlal R, Levy ML. Retinal hemorrhage and pediatric brain injury: etiology and review of the literature. J Clin Neurosci 2005;12:824-31.

13 Lantz PE, Stanton CA. Postmortem detection and evaluation of retinal hemorrhages. Proceedings of the American Academy of Forensic Sciences; 2006 Feb; Seattle, WA. p. 271.

14 Young TW. Young’s Postulate. Heartland Forensic Pathology, LLC. 2009. Accessed 22 Nov 2009.

15 Heuer RJ. Chapter 12: Biases in estimating probabilities. In: Psychology of intelligence analysis. Washington DC: Center for the Study of Intelligence, Central Intelligence Agency; 1999. pp. 147-60.

16 Brice JH, Pirrallo RG, Racht E, Zachariah BS, Krohmer J. Management of the violent patient. Prehosp Emerg Care 2003;7(1):48-55.

17 Pollanen MS, Chiasson DA, Cairns JT, Young JG. Unexpected death related to restraint for excited delirium: a retrospective study of deaths in police custody and in the community. CMAJ 1998 Jun 16;158(12):1603-7.

18 Hymel KP, Bandak FA, Partington MD, Winston KR. Abusive head trauma? A biomechanics-based approach. Child Maltreat 1998;3(2):116-28.

19 Dias MS, Backstrom J, Falk M, Li V. Serial radiography in the Infant Shaken Impact Syndrome. Pediatr Neurosurg 1998;29:77-85.

20 Conti RP. The psychology of false confessions. J Credibility Assessment Witness Psychol 1999;2:14-36.

21 National Research Council. Summary. In: Strengthening forensic science in the United States: a path forward. Washington, DC: The National Academies Press; 2009. pp. 1-33.

22 Goudge ST, commissioner. Inquiry into pediatric forensic pathology in Ontario: report. Ontario, Canada: Ontario Ministry of the Attorney General; 2008.

23 Di Maio TG, Di Maio VJM. Excited delirium syndrome: cause of death and prevention. Boca Raton, FL: CRC Press; 2006.

24 Caffey J. The whiplash shaken infant syndrome: manual shaking by the extremities with whiplash-induced intracranial and intraocular bleedings, linked with residual permanent brain damage and mental retardation. Pediatrics 1974;54:396-403.

25 Guthkelch AN. Infantile subdural hematoma and its relationship to whiplash injuries. BMJ 1971;2:430-1.

26 Chadwick DL, Bertocci G, Castillo E, Frasier L, Guenther E, Hansen K, Herman B, Krous HF. Annual risk of death resulting fro short falls among young children: less thatn 1 in 1 million. Pediatrics 2008;121:1213-24.

27 Pierre-Kahn V, Roche O, Dureau P, Uteza Y, Renier D, Pierre-Kahn A, Dufier JL. Ophthalmologic findings in suspected child abuse victims with subdural hematomas. Ophthalmol 2003;110(9):1718-23.

28 Case ME, Graham MA, Handy TC, Jentzen JM, Monteleone JA. Position paper on fatal abusive head injury in infants and young children. Am J Forensic Med Pathol 2001;23:112-22.

29 Offiah A, van Rijn RR, Perez-Rossello JM, Kleinman PK. Skeletal imaging of child abuse (non-accidental injury). Pediatr Radiol 2009;39(5):461-70.

30 Kleinman PK. Diagnostic imaging of child abuse, 2nd ed. St Louis: Mosby; 1998.

31 Tongue AC. The ophthalmologist’s role in diagnosing child abuse. Opthalmology 1991;98(7):1009-10.

32 Plunkett J. Fatal pediatric head injuries caused by short-distance falls. Am J Forensic Med Pathol 2001;22(1):1-12.

33 Geddes JF, Tasker RC, Hackshaw AK, Nickols CD, Adams GGW, Whitwell HL, Scheimberg I. Dural haemorrhage in non-traumatic infant deaths: does it explain the bleeding in ‘shaken baby syndrome’? Neuropathol Appl Neurobio 2003;29:14-22.

34 Biron D, Shelton D. Perpetrator accounts in infant abusive head trauma brought about by a shaking event. Child Abuse Negl 2005;29:1347-58.

35 Starling SP, Patel S, Burke BL, Sirotnak AP, Stronks S, Rosquist P. Analysis of perpetrator admissions to inflicted traumatic brain injury in children. Arch Pediatr Adolesc Med 2005;159(2):195.

Legends to Figures

Note: The figures referenced in the article were not published on the website due to their graphic nature. They are available upon request. Simply contact me with your request.

Fig. 1 Wrap-around fracture from Case #3. Fracture lines with both branching and tight curves lie in the left posterior calvarium. From this, a wrap-around fracture extends anteriorly along the left side of the calvarium (seen in the top of the image).

Fig. 2 Wrap-around fracture from Case #3. The fracture line in the left side of the calvarium then crosses from left to right across the near front of the calvarium, then passes posteriorly along the right side of the calvarium, ending as a single downward curve in the right posterior calvarium. Copious hemorrhage lies in reflected portions of the scalp. The fracture line spread apart from increasing intracranial pressure on both sides and in the front reveals hemorrhage in the epidural space.

Fig. 3 Top portion of calvarium removed at autopsy revealing a right-sided subdural hematoma. Although a portion of the dura is artifactually separated from the inner table of the calvarium in the front, the dura is otherwise intact and not lacerated by the fracture overlying it.

Fig. 4 The autopsy pathologist pushing aside the hair in the back of the head reveals a contusion in the posterior scalp lying near the origin of the fracture in the left posterior calvarium (compare with Fig. 1). The contusion indicates that an impact from a blunt object with a relatively small surface area of impact initiated the wrap-around fracture.

Fig. 5 The mucosa of the upper lip from Case #4 demonstrates abrasions conforming to the cusps of the maxillary teeth. The two rows of abrasions are consistent with the lip shifting slightly in position during the struggle while the lip was pressed against the teeth.

Fig. 6 The mucosa of the lower lip from Case #4 also demonstrates abrasions conforming to the cusps of the mandibular teeth.