Using network analysis to visualize the role of emotion in murder mystery novels

Heightened emotion is a key element at the root of the crimes in most mystery novels. Despite this, not much literature exists examining the part emotion plays in the structure of classic detective fiction. Network analysis provides a new avenue for exploring the role of emotion between characters in murder mystery novels. Software, like the program Gephi, takes advantage of graph theory to visualize the connections between individual points of data in a network. By graphing the hundreds of small interactions in a novel, larger patterns emerge, allowing interpretations about the broader role of emotion through the novel. In this paper I aim to track character interactions based on emotion to draw wider conclusions about the structure of the novels and writing styles of their authors. I primarily examine five books written by Agatha Christie, the "Queen of Crime." I selected books by publication date to provide a survey of her 50+ year writing career. To provide contrast, as well as insight into how the genre has since changed, books from Louise Penny's Armand Gamache series were also analyzed. In the course of my analysis, I show that Christie and Penny use emotion for fundamentally different purposes: Christie uses emotion as a tool to advance her plot, and Penny uses her plot as a means for exploring emotion. This paper opens an avenue for future use of network theory in the analysis of detective fiction.

ABSTRACT Heightened emotion is a key element at the root of the crimes in most mystery novels. Despite this, not much literature exists examining the part emotion plays in the structure of classic detective fiction. Network analysis provides a new avenue for exploring the role of emotion between characters in murder mystery novels. Software, like the program Gephi, takes advantage of graph theory to visualize the connections between individual points of data in a network. By graphing the hundreds of small interactions in a novel, larger patterns emerge, allowing interpretations about the broader role of emotion through the novel. In this paper I aim to track character interactions based on emotion to draw wider conclusions about the structure of the novels and writing styles of their authors. I primarily examine five books written by Agatha Christie, the “Queen of Crime.” I selected books by publication date to provide a survey of her 50+ year writing career. To provide contrast, as well as insight into how the genre has since changed, books from Louise Penny’s Armand Gamache series were also analyzed. In the course of my analysis, I show that Christie and Penny use emotion for fundamentally different purposes: Christie uses emotion as a tool to advance her plot, and Penny uses her plot as a means for exploring emotion. This paper opens an avenue for future use of network theory in the analysis of detective fiction. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 LITERATURE REVIEW 3 METHODS 8 RESULTS 14 DISCUSSION 36 REFERENCES 41 APPENDICES A: Methodology Exceptions and Examples 46 B: Additional Graphs 50 iii 1 INTRODUCTION “Motives for murder are sometimes very trivial, Madame.” “What are the most usual motives, Monsieur Poirot?” “Most frequent—money. That is to say, gain in its various ramifications. Then there is revenge—and love, and fear, and pure hate, and beneficence—” (Death on the Nile 87) Greed, revenge, love, fear, hate, beneficence— these are the motives Agatha Christie’s famous detective, Hercule Poirot, gives for murder. His statement points to an important truth: strong emotion lies at the root of the crimes in most mystery novels. While strong emotion might not account for all real-life murders, fictional novels are not nearly as compelling without it. Even though emotion is such a large element of detective fiction, very little literature exists examining its roles, or the roles of interpersonal relationships in the genre. Perhaps the connection appears too obvious to merit much analysis. However, emotion is so integral to the structure of detective fiction that analysis of the genre cannot be fully complete without taking it into consideration. Network analysis offers a way to address this gap. With software such as the program Gephi, it is possible to visually graph the interactions between characters across a novel, and to track the flow of emotion between them. A relatively new technique within literary criticism, network analysis has been used for purposes such as examining the social networks in popular fiction, and predicting future plotlines in the Game of Thrones series (Beveridge and Shan; Waumans et al.). Network analysis is a versatile tool that can be adapted to many different applications. By graphing the hundreds of small interactions in a novel with a focus on emotion, I can see the larger patterns that 2 emerge, and make interpretations about the role of emotion in the novel as a whole. Many of these patterns are subtle enough that they are difficult to see without the assistance of Gephi. For this paper, I am focusing on novels by two authors: Agatha Christie and Louise Penny. Agatha Christie is lauded as the “Queen of Crime,” and her prolific corpus of novels and short stories has set the standard for writers in her genre. While her works are now considered archetypal in detective fiction, at the time she wrote them she was famous for playing with form and structure, and for breaking the established rules of the genre. Louise Penny published the first book in her Chief Inspector Armand Gamache series in 2005, and the most recent book came out in August 2021 (“Louise Penny”). The five-time winner of the Agatha Award, her books follow after Christie’s tradition as a part of the subgenre that has come to be called “cozy mysteries.” Malice Domestic, the mystery convention that presents the Agatha Award defines cozy mysteries as “mysteries that contain no explicit sex, excessive gore, or gratuitous violence, and would not be classified as ‘hard-boiled.’” (“The Agatha Awards”). In analyzing Christie’s books, I hope to get a sense of how emotion influences and supports the structure of cozy mysteries, and what role it plays. Penny’s books will help provide context for Christie’s, and will give me a glimpse into how the genre has evolved over time. In my analysis of these books, I will look specifically at what I call “directed emotion.” Murders are emotional by nature, and in each book there are a lot of emotions floating around. I am not interested in the ambient emotion in the book, but in the heightened emotion directed specifically at or between characters. Emotion can be onesided, such as one character expressing love for another, or mutual, such as a shouting 3 match. This emotion can be anything — hate, love, lust, vengeance, spite, etc. The patterns that emerge will show a lot about the structure of the novels, and the role that emotion and relationships play within them. In the first part of this paper, I will examine the flow of emotion in several of Christie’s works and show how they point directly to the murderer. I will then examine a contemporary author to examine what elements of Christie’s legacy have been passed on, and how the genre has since shifted. By comparing Penny and Christie’s works, I will show that Christie makes heavy use of what Bayard calls “exhibition,” and Singer calls “the block element.” Both phrases refer to the use of obscuring details to hide the truth in plain sight. I will also show how Christie and Penny use emotion for completely different purposes. These differences will provide insight into larger structural differences between their works, and provide a basis for future analysis of emotion in the detective fiction genre. LITERATURE REVIEW Classic detective fiction is a genre defined by its rules and conventions. Earl F. Bargainner notes that, “As early as 1913 Carolyn Wells found an equivalent to detective fiction in the fixed forms of verse: the sonnet and the French forms, such as the triolet, the sestina and the rondeaux. Others have agreed, among them Sayers, C. Day Lewis, Richard Lockridge, Michael Gilbert, Hillary Waugh and John Cawelti, particularly as to the sonnet” (5). While some have criticized the set structures of the form, others have suggested that it is precisely that structure of “a well-ordered world, one which shows clearly the work of a creator who is above all else logical” that draws readers (Ramsey 7). 4 What are the key rules in detective fiction? First and foremost, there must be a crime (typically a murder), a criminal, and a detective. “The role of the detective is to find out whodunit, whatever crime the ‘it’ may be” (Ramsey 5). The criminal and detective are joined by a host of characters who each play a role in providing information or obscuring the truth (Bargainnier 131). With the identity of the murderer concealed, every character “is to be considered a suspect and is to be highly mistrusted” (Ramsey 2). By the end of the story, all extraneous details must be swept aside, and the true solution to the crime revealed. It is not simply enough, however, for the detective alone to have the ability to solve the crime. Singer models the detective novel as a riddle and explains the inherent contract between the reader and the writer: “The poser of the enigma is omnipotent at the whim of the posee, and that whim lasts only so long as the solutions are satisfying” (Singer 158). In other words, the detective novel is only successful so long as the reader is satisfied with the solution. What is necessary to produce a satisfying solution? First, “The reader must be given all the clues necessary to solve the crime himself, but they must be cunningly presented so that he sees them without realizing their significance” (Ramsey 5). If the detective produces the solution to the puzzle using clues kept hidden from the reader, the solution feels like a cheap trick rather than a true act of deduction. The reader must have all the same information as the detective to be satisfied. Second, “Every detail must be explained by or in the last chapter, and every one of the characters accounted for in such a way that the bad are punished and the good rewarded – no small feat when you stop to consider that all through the book we have been unsure which were 5 which” (Ramsey 7). If loose ends are left untied, or characters do not receive their proper reward, the book will end with an unfinished air. Sigmund Freud’s theories also had an impact on detective fiction, adding complexity and depth. With the increasing popularity of Freud’s theories, Amy Yang explains, “No longer was it enough to simply describe the events that unfolded in a crime incident, nor was it satisfactory to end the story at the discovery of the murderer. The readers wanted more than just plot – the notion of motive and intent had become an important element of the storytelling” (597). Detective stories became increasingly about personality, psychology, relationships, and motivation. Agatha Christie’s books fall into this more psychological trend. Ramsey attributes much of her continued success to this focus, saying, “when an author eschews gadgetry, and concentrates on the never-changing mental and thought processes of his characters, his books are much less likely to fall by the wayside” (9). Christie’s focus also draws the reader’s attention from the gore and unpleasantness of the murder committed. “The significance of the puzzle element is that it places the emphasis upon the contest between the detective and the murderer, rather than upon the actual deed and the victim” (Bargainnier 7). Bargainner calls Christie “the doyenne of the Cosy School,” a subgenre of detective fiction that focuses on the puzzle of the crime rather than the shock value (7). Much research has been done on the various methods Christie employs to construct her puzzles. She is famous for playing with the genre and breaking the rules. Bargainner notes, “S.S. Van Dine’s ‘Twenty Rules for Writing Detective Stories’ and Ronald A. Knox’s ‘A Detective Story Decalogue’ are perhaps the two most famous such lists of the ‘twenties, and though most of their rules at first glance may seem 6 commonsensical, they have all been successfully broken. Christie alone broke fifteen of Van Dine’s twenty and seven of Knox’s ten” (Bargainnier 6). Despite this willingness to play with convention, there are multiple techniques Christie reliably uses to distract the readers’ attention from the solution to her mysteries. Bayard identifies three techniques: distraction, disguise, and exhibition (Bayard 25). Disguise and distraction, where “the truth is made unrecognizable” and “the false, we might say, is dressed up to draw attention to itself,” are common across most if not all detective fiction (Bayard 25). What sets Christie apart is her frequent use of exhibition. Bayard says, “This technique consists of making the truth invisible by inscribing it in every letter. The murderer is, shall we say, hidden behind the murderer” (25). Christie frequently employs this technique by making the murderer obvious, then confusing the reader with abundant evidence showing that character couldn’t possibly be the murderer. These three techniques are not mutually exclusive, but are instead frequently used together throughout her stories (Bayard 29). Bayard argues that Christie’s overuse of these techniques can lead to a multiplicity of possible correct answers, and that this confusion reduces the readability of her books. Singer, however, offers a different reading. He introduces the concept of a block element from the theory of riddles, saying, “A riddle is enigmatic because there is an obstruction between the image it presents and the referent the riddlee is supposed to guess. In riddling scholarship, this obstruction, following Petsch, is usually called the block element” (Singer 160). Singer argues that Christie employs the use of block elements to obscure the truth in her stories, much like a riddler would to hide the true answer to their riddle. He states, “What Dame Agatha consciously and insidiously does is close the reader’s mind. The clues themselves, then become insignificant, and the 7 solution lies not in untangling their pattern, but in discovering the mechanism by which the reader’s mind is closed” (Singer 160). This is an important distinction because riddles are solved differently from puzzles. “Puzzles are really solved through the accumulation of clues. Riddles can almost never be solved deductively; the key to their hermeneutic structure is the block element” (Singer 170). Singer argues that readers will be able to solve Christie’s mysteries when they stop treating them as puzzles, and instead look for the riddling elements she uses to block them from seeing the clear truth. These various analyses of Christie’s work are very thorough in cataloguing the plot devices she uses to construct her mysteries. However, none of them take into account the interpersonal element necessary to a murder mystery. Kellog notes, “The books of Agatha Christie typically reveal a deep understanding of human nature,” but very little literature exists that examines the various interpersonal relationships or the flow of emotion in her novels (Kellog 46). This is a surprising oversight. No matter how clever Christie’s plots are, they would not be fulfilling without an emotional, interpersonal element. Emotion figures heavily into the structure of the plot, and adds important complications. Because of this, analysis that ignores emotion can only produce a halfformed picture of Christie’s techniques. The software Gephi provides a new methodology for studying these elements more in depth. For this project, I will examine the role of emotion in Christie’s books, and how that role supports Singer and Bayard’s theories. I will also explore books by Louise Penny, a contemporary author, to further understand Christie’s structure, as well as how the genre has shifted over time. 8 METHODS Book Selection I decided to focus on Agatha Christie and Louise Penny for their many similarities, and for some key differences. Both authors are female and write “cozy mysteries” set in the time period in which they are writing. All of Penny’s books center on her male detective, Armand Gamache. In the interest of continuity, I also focused solely on books starring Christie’s male detective, Hercule Poirot. The books follow the same format: the series centers on the detective, with each book presenting a new mystery that is solved by the end of the book. There are other recurring characters throughout the series, but the heart of each book is the challenge the detective faces in solving the case. The focus is less on the graphic details of the murder, and more on the unspooling of events and clues, and the puzzle that they pose. Beyond these similarities, there are several differences between Penny and Christie’s works that are important in the scope of my project. Penny focuses even more on interpersonal relationships and character growth than Christie does. There is a much larger supporting cast that recurs throughout Penny’s books. Where Hercule Poirot is a private detective, Gamache works for the Sûreté, the local police force in Quebec. This means he has a whole team of officers that he works with. These officers are not static characters either, instead they grow and change as they reappear across the series. The bulk of the series also takes place in the tiny town of Three Pines, with Gamache and his wife eventually retiring to move there. While new characters are introduced in each book, 9 this means that the supporting cast remains constant over many of the books. This emphasis on recurring characters shifts the focus of the story even further from the murders themselves to their effects on the town as a whole. The two central detectives are also treated very differently. Hercule Poirot is static and detached from the crimes he investigates. Though his books span over 50 years, he does not change or age. Bargainnier notes, “By strict chronology… he would be over 135 by the time of Curtain, the last novel” (45). The closest things he has to family are his friend Hastings, and his secretary Miss Lemon. Outside of the mysteries he investigates, we see only glimpses of his personal life. By contrast, Armand Gamache is a much more fleshed out, dynamic character. Over the course of the series, we get to know each member of his family intimately. He ages and changes, and his relationship to the cases he is investigating also shifts. In addition to the single mystery that dominates each book, there are larger multi-book arcs following Gamache’s career, and his struggle to fight corruption in the force. In some of the books, this larger arc overshadows the murder and shifts it to a secondary position. All of these differences alter the focus of Penny’s books in ways that are clearly reflected in the graphs. Because of the time-consuming nature of the analysis I am conducting, and the limited scope of this project, I decided to limit myself to eight books. Since Agatha Christie is the primary focus of my project, I chose five of her books, and three of Penny’s. I used publication date as the main criteria for selecting the books I would analyze. I wanted to get a good sample of each author’s writing over time, so I selected the first and last book in each series, as well as a couple spaced evenly through the middle. I tried to pick books I was less familiar with to begin with, so that my analysis 10 was not colored by my foreknowledge of the solution to the mysteries. The one exception to this was The Murder of Roger Ackroyd. This is one of Agatha Christie’s most controversial novels, and I was curious to see how it would look in Gephi. Once I selected my books, I used the same process with each one to catalogue the character interactions and input them into Gephi. What is Gephi? The software I am using for this project is a program called Gephi. Gephi graphs networks visually, exposing relationships and allowing users to identify patterns. A Gephi graph consists of individual points called nodes. In the case of my analysis, each node represents a character in the book. If any two characters interact with each other, their nodes are connected by a line called an edge. I assign these edges different weights depending on how much the characters interact, and what emotions are present in those interactions. The weight is represented by the thickness of the edge — the thicker the edge, the more closely connected the characters. Once I input the nodes, edges, and weights, Gephi plots the graph. Nodes that are more heavily connected are plotted closer to each other. Nodes that have a lot of edges connected to them are larger and situated more towards the center of the graph, whereas nodes with only one or two connections are smaller and sit along the periphery. Finally, Gephi can divide the nodes into groups based on who interacts with each other more. These groups are represented by different colors and help show the different social circles within a book. With so many hard numbers, it would be easy to view the Gephi graphs created from these books as completely objective and incontrovertible. This is not the case. The 11 graph is instead a visual representation of many small interpretive decisions made in collecting the data. The final shape of the graph depends entirely on the system I set up for cataloguing and weighting interactions, and how well I stick to that system. A different system, or even a different researcher using the same system might well yield very different results. Project Procedure In developing the procedure for analyzing these books, the first question dealt with how to catalogue interactions. What counts as an interaction between two characters? I define an interaction as the exchange of words, information, or objects between two or more characters. Most interactions were clear and easy to identify: conversations, phone calls, letters, etc. Since all of these books deal a lot with death, I also decided to include will bequests as an interaction. There were places, however, where identification became more difficult. All participants in an interaction must be identified in order for the interaction to be input into Gephi. In some cases, however, the identity of one or more characters was concealed from the reader. In these instances, I noted the interaction, but did not record it officially until all parties were identified. In addition to these overarching challenges, some of the books offered specific challenges. Elephants Can Remember deals with a crime that happened years in the past, so many interactions are told repeatedly as memories and rumors. The Beautiful Mystery is set in a monastery full of monks accustomed to a vow of silence, so many interactions are entirely nonverbal. I will address these challenges in more depth when I discuss those books individually. 12 Once I decided what counted as an interaction, my next task was to develop a weighting system for rating each interaction. I am focusing specifically on directed emotion, so I settled on a simple five-point scale. The rating scale is as follows: Table 1 Gephi Weighting Scale 1 Basic, informational interaction with little to no directed emotion. 2 Negative emotion expressed towards someone not present in the interaction 3 Positive emotion 4 Negative emotion expressed towards or between people in the interaction 5 Physical violence This weighting scale for interactions is very simple. This is intentional, but it also leads to limitations. In order to keep my weights standardized across the different books, I wanted to have a clear system with very little ambiguity. Each type of emotion is assigned a specific number. As long as the emotions are clear, the number they should be assigned is equally clear. There is relatively little room in this scale for disagreement about weights for a given interaction. What this scale gains in clarity, however, it loses in specificity. A snide, angry comment is rated the same as a shouting match. A slap is rated the same as outright murder. Intensity of emotion is not recorded. I chose to rate negative emotion higher than positive emotion because the goal of these graphs is to find the murderer, but that doesn’t mean each instance of negative emotion is more intense than those of positive emotion. I also decided deliberately to include negative emotion directed outside of the interaction because I felt like it needed to be recorded, even if it wasn’t directed between the people 13 present. I considered recording it as an interaction that included the person being talked about, but ultimately decided not to because that person was not truly involved. It is also important to note that this weighting scale does not take duration into account. A lengthy conversation that is only factual will be rated lower than a brief, emotionally charged interaction. Despite these limitations, I think that my scale accomplishes its main goal. It provides an accurate, if simplistic, depiction of the flow of emotion over the many individual interactions in the book. I also decided to make multiple graphs for each book to show how character relationships change as more information is introduced. Since I wanted to see if the graphs could be used to solve the mystery, I made at least two graphs for each book: one with interactions only up to the reveal, and one with all interactions in the book. For books in which the murder took place partway through, I also created a separate graph that only showed interactions up to the murder. I wanted to see how early the evidence for the murder was established. With multiple graphs establishing a timeline for the books, I decided to record interactions at the point at which the reader becomes aware of them, not at the point at which they occur. For instance, if a conversation takes place between characters, and one or more is unidentified, I did not officially record the interaction until all characters are identified. This only makes a difference in the graphs if information is revealed in a different section of the book than it occurs, but it was important to me to portray the information that the reader has at each stage, rather than simply recording events in timeline order. 14 RESULTS Agatha Christie Books Located chronologically at the center of my selection of books by Agatha Christie, Death on the Nile is perhaps the book most suited to this methodology. After a couple of chapters introducing the main characters, the entire book takes place on a Nile cruise on the S.S. Karnak. This keeps the mystery very self-contained, with only one or two characters introduced after the ship sets sail. It also provides ample opportunity for the different characters to interact with each other, and for relationships to form, twist, and break. Figures 1, 2, and 3 show the evolution of the character relationships as the book progresses. Fig. 1. Death on the Nile pre-murder character interactions. 16 Each of the graphs above are cumulative from the beginning of the book. Figure 1 shows all character interactions prior to the discovery of the murder. Figure 2 shows all character interactions prior to the reveal of the murderer. Figure 3 shows all character interactions in the whole book. A basic understanding of the key characters is important for reading these graphs. The story centers around the murder victim, Linnet Ridgeway. A young, beautiful heiress, Linnet is honeymooning with her new husband, Simon Doyle. Their enjoyment is hampered however, because they are being stalked by Jacqueline de Bellefort. Jacqueline is Linnet’s former best friend, and Simon’s former fiancé. She is angry at Linnet for stealing Simon from her, and has decided to get her revenge by following them from place to place as they vacation. The beleaguered couple finally flees to Egypt, where they meet Poirot. They make secret plans to cruise on the S.S. Karnak in hopes of escaping Jacqueline, but their hopes are dashed when they find her on board. They are joined by a cast of other characters, many of whom turn out to have some connection to Linnet. Partway through the cruise, Poirot’s friend, Colonel Race, also boards the ship in search of a dangerous murderer hidden among the guests. In addition to his own investigation, he helps Poirot investigate Linnet’s murder. Looking at the graphs, it’s easy to see how the character relationships develop over the course of the book. In Figure 1, the characters are still fairly separated into the groups they are travelling with. Since most of the groups were strangers to each other only a few days before, most of the edges connecting characters from different groups are thin and tenuous. The graph is fairly spread out. Over the course of the next two graphs, the relationships between all the passengers on the boat grow stronger. Poirot and Race 17 take on a central position as they interact with every passenger of the boat through their investigation. The lavender portion of the graph centers on the victim, Linnet, and those she interacts with. Easily the strongest emotional connections on all three graphs are those connecting her, Jacqueline, and Simon with each other. This does not come as a surprise; emotions are heightened between all three of them from the beginning. Linnet is incandescently in love with her new husband, and frustrated and a little frightened by Jacqueline. For her part, Jacqueline is furious, and takes pleasure in the discomfort she is causing the new couple. She even reveals to Poirot that she carries a gun and has considered violence. Jacqueline would be the obvious suspect if she didn’t have an ironclad alibi. She had a breakdown the evening before the murder, shot Simon in the leg, and spent the night sedated and under the watchful eye of a nurse. Simon was likewise incapacitated by the bullet in his leg. With the two obvious suspects cleared of suspicion, the confused reader is left to wonder which of the other boat passengers could have a strong enough motive to kill Linnet. In the end, however, the graphs reveal the truth. Christie has used the seemingly airtight alibis of Jacqueline and Simon as the block element that keeps the reader from the solution. The murderer is not Jacqueline or Simon, but both of them working together. They have been in love the whole time, and planned for Simon to marry Linnet for her money. While the plot is full of twists, confusion, and red herrings, the graphs of the flow of emotion bypass the block element to show the truth of the murder clearly. In addition to the solution to the murder, the graphs also show the progression of other relationships over the course of the novel. In particular, there are two romances that 18 develop. When Rosalie’s mother is killed, Mrs. Allerton takes Rosalie under her wing. While Mrs. Allerton’s son, Tim, is initially dismissive of her, they eventually grow close, finally admitting their feelings for each other. Cornelia is the poor relation and traveling companion of the elderly Miss Van Schuyler. She attracts the attention of both Dr. Bessner and Mr. Ferguson. Both vie for her attention, with many emotional outbursts from Mr. Ferguson. Eventually, she chooses the patient, intelligent Dr. Bessner over the abrasive Mr. Ferguson. Both of these relationships are evident in the progression of the graphs. Rosalie’s connection to Tim and Mrs. Allerton, and Cornelia’s connection to Mr. Ferguson and Dr. Bessner grows stronger in each successive graph. I analyzed four other books by Agatha Christie. Of those four, the graphs of The Mysterious Affair at Styles and Elephants Can Remember produce similarly clear pictures of the murder. Figure 4 shows the graph for all character interactions in The Mysterious Affair at Styles: 19 Fig. 4. The Mysterious Affair at Styles full book character interactions. The Mysterious Affair at Styles is the first published novel by Agatha Christie. As is clear in the graph, the story is narrated by Hastings, so much of the emotion centers on himself, and his friendships with John, Mary, and Poirot. The murder in this novel follows a very similar structure to Death on the Nile. The victim is the wealthy Mrs. Inglethorp, who is found poisoned early one morning. Suspicion immediately falls on her strange new husband, Alfred. He does nothing to discourage this suspicion, but there are key elements in the murder for which he has a solid alibi. Poirot and Hastings are baffled until they learn that he has been working with Mrs. Inglethorp’s good friend Evie. Evie 20 has claimed to hate Alfred the entire time, and has been absent for the majority of the book, but she secretly helped him procure the poison to kill Mrs. Inglethorp. The solution to the murder is not revealed until the connection between Evie and Alfred is discovered. As with Death on the Nile, Alfred, Evie, and Mrs. Inglethorp form a clear, separate group in the graph. It’s interesting to note that the connection between Alfred and Evie is formed of mainly negative interactions, as she continually proclaims her hatred of him and accuses him of the murder. Mrs. Inglethorp’s relationships with Alfred and Evie, on the other hand, are made up of primarily positive interactions. These initial interactions are completely counter to the truth: that Alfred and Evie are in love and plotting Mrs. Inglethorp’s murder. This shows that it’s frequently the height of the emotion, rather than the type of emotion, that is most telling in Christie’s books. Heightened emotion, not negative emotion, leads to the killer. This makes sense, given that murder mysteries are full of lies, and people often try to hide their true emotions. While they may succeed in masking their emotions, it seems that the intensity of emotion still remains in this book, clearly identifying the killers. 21 Fig. 5. Elephants Can Remember full book character interactions. Elephants Can Remember presents another seemingly impossible mystery. General Ravenscroft and his wife, Molly, are found dead on the cliffs by their seaside house, a single gun between them. Twelve years later, their daughter Celia seeks answers to what happened. Was it a suicide pact? A murder-suicide? If an outside person killed them, the obvious suspect would be Molly’s twin sister Dolly, who has an emotional past with the couple, and a history of mental illness and violence. The only catch is, Dolly died six weeks before the death of the couple. Once again, however, the graph shows the truth hidden behind the block element: Dolly killed her sister. In order to fulfil Molly’s dying 22 wish that Dolly not go to prison for her crime, General Ravenscroft claims it was Dolly who died, and disguises the real Dolly as her twin sister. Eventually, however, he decides she is too dangerous to be allowed to live, and shoots her and then himself. This book was interesting to analyze because the actual murders happened twelve years in the past. The entirety of the investigation is conducted through interviews with various people who knew Dolly, Molly, and General Ravenscroft. This is why the graph has so many minor characters who are only connected to the two investigators, Poirot and Mrs. Oliver. None of the facts of the murder were witnessed by the reader firsthand. They were all given through the foggy, frequently incorrect lens of the interviewee’s memories. This led to a unique challenge in cataloguing the interactions surrounding the actual murder. Many scenes were repeated in the memories of different people. Details changed in retellings, and many stories were missing key details, such as who was actually involved in the interaction. I tried to only record concrete observations in which both characters in the interaction were clearly identified. I also made sure not to repeat interactions. Still, with all of this ambiguity I am surprised that the graph was so clear. In each of the graphs above, the shape of the murder is very clear, with the murderer(s) and victim(s) clearly in a separate group from the rest of the characters. The other two books that I analyzed did not produce as clear of results. The first was The Murder of Roger Ackroyd. This is one of Agatha Christie’s most well-known and controversial books. The controversy comes because she breaks one of the leastquestioned rules of detective fiction: she makes the narrator, who plays the assistant role to Poirot, the murderer. This would be like Sherlock investigating a crime, only to find 23 out that Watson is the killer. Many fans were upset with Christie because they felt that she didn’t “play fair” with them — by violating the accepted convention of who is allowed to be considered a suspect, she didn’t allow them a fair chance to solve the mystery themselves (Davis 29). Despite the trickery, Christie leaves many clues through the novel that the narrator, Dr. Sheppard, may not be all that he seems. He is a brand-new character who has not appeared in any other Poirot books, so the reader has no attachment to him. There are several points throughout the novel in which it is revealed by other characters that he has not been completely honest to the reader about his actions. These small omissions point to the larger secret he is likewise concealing. With this unusual set up, I was very curious to see how this novel would look when put into Gephi. The investigators are usually the most prominent characters on the graph, so I expected that Dr. Sheppard’s position in the graph would look similar to that of Hastings or Race in A Mysterious Affair at Styles and Death on the Nile. I predicted that if the pattern of connecting the murderer and victim continued, Dr. Sheppard would be closely connected to Roger Ackroyd, and that no other connections to the victim would be as prominent. 24 Fig. 6. The Murder of Roger Ackroyd full book character interactions. As Figure 6 shows, my predictions were correct. Roger Ackroyd does not show up very prominently on the graph because he is killed very early in the book. Still, his strongest connection is with Dr. Sheppard. While it’s not as immediately obvious, this graph still supports my thesis. It’s also interesting to note that this isn’t a very emotional book. Most of the weights for the cumulative interactions between characters are very 25 low. The strongest connection is between Poirot and Dr. Sheppard, and that is based almost entirely on frequency of interaction, not on emotion. The other book that wasn’t as clear was Hickory Dickory Dock. Figure 7 shows the graph of all the interactions. This book takes place in a boarding house for university students. While the main murderer, Nigel, is very prominent, he is only strongly connected with his third victim, Patricia. His connections to his other two victims, Celia and Mrs. Nicoletis, and to his partner in crime, Valerie, are not at all evident. Fig. 7. Hickory Dickory Dock full book character interactions. 26 Mary S. Wagoner points out that Hickory Dickory Dock departs from Christie’s standard structure for her Poirot novels. She says, “Poirot uncovers clues more in the style of Miss Marple than in his own. He senses hidden relationships and motives instead of deducing them from clues available to the reader” (Wagoner 83). Throughout the book, Poirot makes some large leaps that are not readily obvious to the reader, such as deducing that the false bottoms of cheap rucksacks are being used to transport drugs into the country. I find it interesting that this muddiness surrounding the physical clues is also apparent in the relationships present in the novel. The relationships between the boarders are all jumbled, and very little information is gained from the graph. One element that added to the confusion of the character interactions is that many of the interactions took place in group settings in which all or most of the characters were present. It was very difficult to sort out who was interacting with whom. I discuss my approach to group conversations in more depth in Appendix 1. With another book, this confusion in the graphs might suggest that Christie is doing something new with structure or devices. With this book, however, that doesn’t seem to be the case. While some of the elements in the book are a departure from her normal writing style, the choice doesn’t feel deliberate. Christie still builds elaborate clues, and multi-layered stories, but, as Wagoner points out, “she fails to create a sense of logical necessity between events” (Wagoner 83). Many of her twists feel contrived, so that the final reveal doesn’t seem like the only logical solution. With so many leaps in logic, the puzzle element is removed for the reader, and the reading experience is significantly less satisfying. 27 Overall, the graphs for the Agatha Christie books were surprisingly successful at solving the mystery. Other than the expected complications in The Murder of Roger Ackroyd, and the confusion of Hickory Dickory Dock, the graphs clearly highlighted the strong emotional relationships between the victim and the killer(s). The murderer and the victim are isolated in their own groups, separate from the rest of the characters. This supports the theory that Christie frequently makes the most obvious suspect the murderer, and then employs the block element to confuse the reader. When ignoring plot devices and only looking at emotion, the murderer is the person with the closest connection to the victim on each of the graphs. Even in Hickory Dickory Dock, the murderer is clearly highlighted even when his victims are not. These graphs show Christie’s style clearly, and highlight the commonalities between seemingly disparate plots. Similar analysis with books by Louise Penny provides more insight into how this process works. While Christie uses emotion in pursuit of plot, Penny has a different goal in her use of emotion. Gephi provides interesting insight into the structural differences between these two authors. Louise Penny Books Each of the Louise Penny books has unique elements that are important to this project, so I will consider them individually before examining the patterns that emerge between all three. The first book, Still Life, is important because it introduces Inspector Gamache and sets up the village of Three Pines, where many of the following books take place. The mystery centers around the death of Jane Neal, a retired school teacher who is a pillar in the Three Pines community. Jane is shot by a hunting arrow shortly before she is to participate in her very first art show. At first, it seems like it may have been a 28 hunting accident. However, it is eventually revealed that she was shot by one of the central characters, Ben, who feared that her art would reveal incriminating details connecting him to the sudden death of his elderly, sick mother. Fig. 8. Still Life full book character interactions. As figure 8 shows, the graph for Still Life is very confused. There are many more characters than in any of the Agatha Christie novels. Almost all of them are tightly interconnected with each other. Most of the characters in the green section are police officers, or otherwise associated only with Inspector Gamache. In contrast with Christie’s 29 books, which usually only have one or two people investigating the crime, this book features an entire investigative team. The purple and orange sections are the residents of Three Pines. The victim, Jane, sits at the center of the two sections, with connections to almost everyone in the village. Her connection with the murderer, Ben, is not even close to her strongest connection. It looks about the same as most of her other connections. Her strongest connection is with Ruth, as the result of a decades-old hurt between the two of them, along with years of friendship. Positive and familial relationships make up the majority of the strong connections on the graph. Compared with the very clear Agatha Christie graphs, this graph is completely confusing. This suggests that the role of emotion in this book is more generalized than in the Christie books. This book deals heavily with the effect a violent death has on a small community. Penny is very interested in the emotions of all of the characters, and the way their relationships stretch and change. This is a very different approach from Agatha Christie, where the puzzle of the mystery is the central focus of the book. The second book I analyzed, The Beautiful Mystery, is the eighth book in the series. It takes Gamache and his second in command, Beauvoir, out of their normal setting of Three Pines and sends them to investigate a murder in a monastery. The monastery houses a long-forgotten order of monks who have recently become famous worldwide for their beautiful chants. While preparing to record a second album of chants, their choir director, Mathieu, is found dead. The murderer is eventually revealed to be Luc, a young monk who is horrified by Mathieu’s attempts to compose his own chants. However, the central mystery is completely overshadowed by personal drama in Gamache’s life. There is corruption in the Sûreté, and top leaders are not happy with 30 Gamache’s attempt to put an end to it. One of these corrupt leaders, Francoeur, arrives at the monastery to undermine Gamache and to attempt to separate him from Beauvoir, whom he loves like a son. Francoeur succeeds in getting Beauvoir addicted to pain medication and disillusioned with Gamache’s outlook on life. Figure 9 shows that this secondary drama is actually the central element of the story. Fig. 9. The Beautiful Mystery full book interactions. 31 Because it’s set in a monastery with a limited number of monks and no one else allowed in, there are far fewer characters in this book than in Still Life. Since the main murder is so overshadowed, I was curious what the graph would look like if I isolated the crime. Figure 10 shows the graph of the murder with everyone but the monks removed. Fig. 10. The Beautiful Mystery monks only interactions. This graph is much more simple, with the Abbot, Don Philippe, at the center. In this graph, Luc and Mathieu are very clearly highlighted in their own little green group. Most of the other monks are barely connected to each other. This makes sense, since the novel is narrated from the viewpoint of the investigators, so most of the interactions occurred with them, and not between the monks themselves. The weights for the character interactions in this graph are not very high. Most of the connections between 32 the monks have a weight between one and five, and even the connection between Mathieu and Luc only has a weight of 15 (for contrast, the connection between Gamache and Beauvoir has a weight of 58). The connection between Mattheiu and Luc includes five points for the actual murder itself. Without that one act, the graph looks very different. Fig. 11. The Beautiful Mystery monks only interactions without the murder. Figure 11 shows the graph of the monks with the weights adjusted to exclude the murder. Matthieu and Luc are no longer in a group of their own, and the most prominent 33 connection to Mathieu is Don Philippe. This raises an important question about whether or not the actual murder should be included in the graph attempting to solve the murder. In the other books I analyzed, including the murder as one of the interactions didn’t greatly change the graph — it simply strengthened already obvious connections. In this case however, it means the difference between a clear solution and none at all. I would argue that this indicates a failure of the graph to truly predict the murderer. If the one thing indicating the murderer is his act of murder, that’s not much of a solution at all. The third book, number eighteen in the series, is titled All the Devils Are Here. It once again takes Gamache to a new setting. In the novel, Gamache and his wife, ReineMarie, are in Paris visiting their children. Beauvoir is now married to Gamache’s daughter, Annie, and they are expecting their second child. The family is happily vacationing until Gamache’s godfather, the billionaire Stephen Horowitz, is the victim of what looks like an intentional hit and run. As the family tries to find answers to who would want to kill Stephen, they uncover a vast system of corruption that stretches from the police force to the private sector, and is supported by some of the richest people in the world. All the Devils Are Here included some unique challenges not present in the other books. First, the mystery centers around Gamache and his family. This breaks the tradition that the detective is always separate from the murder. Unlike the other Louise Penny books, where Gamache and his team are clearly separate from the mystery, here Gamache and his family take on the roles of both investigators and, in some cases, suspects. The inability to separate out the detective makes the analysis of relationships in this book much more complicated. 34 The second challenge unique to this novel lies in the identity of the murderer. There is no single murderer in this book. Rather, the bad guys are the leaders of a powerful and corrupt corporation that Stephen has been trying to bring down. Some of the key antagonists are not even introduced until the final chapters of the book. This breaks another key rule of detective fiction: that all the suspects must be introduced at the beginning so that the reader has a fair chance of solving the mystery. Fig. 11. All the Devils Are Here full book character interactions. 35 These elements contribute to making the final graph for All the Devils Are Here complicated and confusing. The graph centers entirely on Gamache and his family. The only other prominent character is Dussault, the Prefect of Police for Paris and Gamache’s old friend. For a while, he seems to be the main antagonist, but it turns out he has pretended to join with the evil corporation in order to infiltrate them. The true antagonists, Roquebrune, Girard, and Pinot, are so minor they barely show up on the graph. Since they contracted a person to attack Stephen, they have little to no connection with him personally. Overall, the graph highlights just how far the novel departs from the classic detective fiction format. This decentralization in the graphs provides a more modern commentary on evil. Rather than a single, easily defeated antagonist, villainy in this novel is faceless, systemic, and ubiquitous. Even at the end of the novel, it is unclear how many people were actively aware of the corporation’s evil planning. Without a single antagonist to imprison, the focus instead shifts to revealing the secret workings of the organization to people powerful enough to stop it. The difficulty lies in identifying which of these people are unaware of the crimes being committed, and which are in on the secret. In this novel, Penny shows clearly how our perceptions of evil have changed over time from an easily resolved hero-and-villain narrative, to a decentralized, systemic structure. From these three examples, we can see how Penny is constantly varying her structure. Unlike the graphs of the Christie books, which all have similar characteristics, these three graphs look entirely different from one another. The Beautiful Mystery follows the pattern of the traditional detective story most closely, whereas Still Life focuses more on relationships outside the scope of the murder, and All the Devils Are 36 Here departs from the classic structure entirely. None of the graphs are entirely successful in portraying the solution to the murder, though The Beautiful Mystery comes closest. The victim and murderer are not strongly connected by emotion in any of the books. This is surprising, given that Penny places so much emphasis on emotion. I would suggest that this failure of the graphs to solve the murder indicates that Penny is using emotion for a much different purpose than Christie is. DISCUSSION Using Gephi to plot directed emotion within character relationships results in a very clear depiction of the mystery in Christie’s books. Following the interactions in the book cuts through all the red herrings and block elements to the truth they are trying to obscure. At first, this might suggest an interpretation counter to Singer’s: that the key to solving the mystery lies not in the riddling elements, but in the psychology and relationships of the characters. An examination of Louise Penny’s works provides a different interpretation, however. Louise Penny’s books follow in the tradition of Agatha Christie’s in many ways, most especially in their focus on character relationships and psychology. Her books center on emotion even more than Christie’s do. Because of this, if emotion really is the key to solving this type of mystery, one would expect Gephi graphs of her books to show the picture of the murder even more clearly than those for Christie’s books. In fact, the opposite is true. In Penny’s graphs, the focus is largely pulled from the actual crime to the myriad other relationships surrounding the case. 37 This contrast suggests that Penny utilizes emotion in a much different way than Christie does. Christie uses emotion as a tool to build her mystery, and to set reader’s expectations. As Bayard showed, Christie usually makes the most obvious suspect the murderer, then convolutes the plot until the solution is no longer evident. In most of the books I analyzed, the murderer was the person closest to the victim, who had the biggest motive. Different plot devices make it seem like it couldn’t possibly be that person. This technique leads to an immensely satisfying twist when it is finally revealed that the murderer is exactly who it should be. Within this structure, emotion plays an important role. It is a tool designed to either strengthen the connection to the victim, or to obscure it in some way. As a reader, we are interested in emotion only as a clue to help us solve the mystery. When a character expresses sadness over a murder, both the detective and the reader’s interest is not in the shape and unfolding of the character’s grief, but rather in whether or not the grief is real, and what it reveals about their relationship with the victim. Emotion is useful as a clue, as a motive, as a red herring, and as a way to manipulate people into action. With the exception of a few romantic relationships that Christie loves to sprinkle in, not much emotion exists that is unconnected from the murder. This is why graphing the emotion works so well for Christie’s books. All of the emotion she includes serves a purpose and works with the physical clues to reveal the shape of the murder. These patterns show up clearly when put into Gephi. By contrast, Louise Penny is interested in emotion for its own sake, and instead uses murder as an excuse to explore the complexities of human emotion and relationships. She fully examines the different forms grief takes, and the ways in which it 38 evolves. Her use of many recurring characters allows her to develop emotional arcs across multiple books. This allows her to include more subtlety and nuance in her characters’ development. It also means that many of the emotional arcs have nothing to do with the central murder. We see this especially in The Beautiful Mystery, where much of the strong emotion is between the police officers on the case, and is the fallout from events in the previous novel. Thus, in these books the role of emotion is flipped. The murder is used to examine emotion, rather than emotion being used to solve the murder. This is why the graphs for Louise Penny’s books are so confusing. This difference in the role of emotion between Christie and Penny’s books points to larger revelations about the structure of the two authors’ respective writing. While Agatha Christie was famous in her time for breaking the ‘rules’ of detective fiction, she still sticks to the basic form. At the heart of classic detective fiction is the perceived contract between author and reader: that the reader must also be given the clues, and that their role is to try to solve the crime themselves. An author is successful if the reader is unable to solve the case on their own, but the solution is clear once it is revealed. With her utilization of the block element, Agatha Christie is the master of this form. While other elements of her structure may change, this contract remains at the core of all her books. As discussed above, Louise Penny uses much more varied structures in her books. She is less concerned with setting up a puzzle for her readers to solve, and more concerned with how the story unfolds. While she still includes enough clues to make the solution believable once it is revealed, she also frequently withholds information from the reader. In some cases, such as All the Devils Are Here, the mystery gets expanded so 39 much it is hardly about the murder by the end. In a sense, Penny’s books are much more like a police procedural than a whodunnit, with the focus being on the steps taken to solve the crime rather than on the puzzle itself. These differences also suggest a way in which the cozy mystery genre has evolved over time. In Christie’s books, the cozy element lies in the focus on the puzzle, and the “sanitizing of the ghastliness of murder, as in such euphemisms as ‘a neat bullethole,’ ‘a pool of blood,’ or ‘a dark stain’” (Bargainnier 7). Through much of the time Christie was writing, England was either in the middle of or recovering from a World War. Ramsey suggests of her contemporary readers, “people in times of crisis like to be able to live briefly in a well-ordered world, one which shows clearly the work of a creator who is above all else logical” (Ramsey 7). This logic, neatness, and distance from the murder is what makes Agatha Christie’s books ‘cozy.’ Penny’s books include some of these elements, but they also take a slightly different approach to the cozy mystery genre. While Penny does not include excessive or gratuitous violence and gore, she does lean into the darker elements of the crimes she depicts. She does not shy away from fully exploring the darker side of human nature. Greed, jealousy, anger, and hate are all examined in depth in her books. It’s not just her murderers that fall prey to these emotions, either. Some of her most beloved returning characters exhibit these traits: Olivier’s greed for nice antiques gets the better of him; Daniel allows resentment towards his father to fester for years; Beauvoir spirals down a road of PTSD and substance abuse. Rather than keep a safe distance from the uglier side of her mysteries, like Christie does, Penny dives right in. 40 In the midst of this onslaught of negative emotion, Inspector Gamache stands out as a bright light. He is a reliable, steady symbol of good who is bruised, but not destroyed by the evil around him. He protects his family and band of police officers from the forces of chaos. He’s very good at managing his own emotions, and is not afraid to confront unsavory emotions in others. Penny relies on Gamache to rein in the negativity, and to set everything to rights by the end of the story. This provides insight into how cozy mysteries have changed. In Christie’s day, a cozy mystery was a clever puzzle to work out before the detective solved it. In Penny’s more modern writing, the cozy element is that the reader explores the dark sides of human nature with the understanding that the world will end right side up. Network analysis provides an interesting new way to examine these books, and to find patterns that might otherwise be much harder to identify. It shows very clearly the different roles emotion plays in Penny’s and Christie’s books. It also provides insight into how the genre of detective fiction has changed in the hundred years since Christie began writing. While the scope of this project was very limited, there are many ways in which this research might be expanded. Christie’s work has long been shown to be archetypal for the genre, but it is unclear if Penny’s books are similarly representative of today’s detective fiction, or if they are an outlier. Analysis of more authors and books would provide a clearer picture into the conventions of the genre and how it has changed over time. It would be especially interesting to compare male authors with Christie and Penny, to see if emotion plays as prominent a role in their books. It would also be interesting to compare the success of a graph in predicting the murderer to the critical reception it received. Do more successful graphs predict more popular books? I look forward to 41 seeing how network analysis continues to be used to analyze literature, especially in the genre of detective fiction. 42 REFERENCES Bargainnier, Earl F. The Gentle Art of Murder: The Detective Fiction of Agatha Christie. Bowling Green University Popular Press, 1980. Bayard, Pierre. Who Killed Roger Ackroyd? The Mystery behind the Agatha Christie Mystery. Translated by Carol Cosman, New Press, 2000. Beveridge, Andrew, and Jie Shan. “Network of Thrones.” Math Horizons, vol. 23, no. 4, Taylor & Francis, Apr. 2016, pp. 18–22. Taylor and Francis+NEJM, https://doi.org/10.4169/mathhorizons.23.4.18. Christie, Agatha. Death on the Nile: A Hercule Poirot Mystery. William Morrow, 2011. ---. Elephants Can Remember. Harper, 1972. ---. Hickory Dickory Dock. Harper, 1955. ---. The Murder of Roger Ackroyd: A Hercule Poirot Mystery. Harper, 2011. ---. The Mysterious Affair at Styles: The First Hercule Poirot Mystery. First Vintage Books edition, Vintage Books, a division of Penguin Random House LLC, 2019. Davis, J. Madison. “Playing by the Rules.” World Literature Today, vol. 89, no. 3–4, Aug. 2015, pp. 29–31. JSTOR. Kellog, Richard L. “The Psychology of Agatha Christie.” Teaching of Psychology, vol. 10, no. 1, Sage Publications Inc., Feb. 1983, p. 46. EBSCOhost, https://doi.org/10.1207/s15328023top1001_13. “Louise Penny.” Fantastic Fiction, https://www.fantasticfiction.com/p/louise-penny/. Accessed 10 Nov. 2021. Penny, Louise. All the Devils Are Here. Minotaur Books, 2021. 43 ---. Still Life. Minotaur Books, 2005. ---. The Beautiful Mystery. Minotaur Books, 2013. Ramsey, G. C. Agatha Christie: Mistress of Mystery. Dodd, Mead & Company, 1967. Singer, Eliot A. “The Whodunit as Riddle: Block Elements in Agatha Christie.” Western Folklore, vol. 43, no. 3, Western States Folklore Society, 1984, pp. 157–71. JSTOR, https://doi.org/10.2307/1499897. “The Agatha Awards.” MALICE DOMESTIC, https://www.malicedomestic.org/agathaawards.html. Accessed 15 Nov. 2021. Wagoner, Mary S. Agatha Christie. Twayne, 1986. Waumans, Michaël C., et al. “Topology Analysis of Social Networks Extracted from Literature.” PLOS ONE, edited by Irene Sendina-Nadal, vol. 10, no. 6, June 2015. DOI.org (Crossref), https://doi.org/10.1371/journal.pone.0126470. Yang, Amy. “Psychoanalysis and Detective Fiction: A Tale of Freud and Criminal Storytelling.” Perspectives in Biology and Medicine, vol. 53, no. 4, Johns Hopkins University Press, 2010, pp. 596–604. Project MUSE, https://doi.org/10.1353/pbm.2010.0006. 44 APPENDIX A: METHODOLOGY EXCEPTIONS AND EXAMPLES Group conversations posed a challenge when it came to cataloguing interactions. When the group only consists of a few characters, it is easy to record it as individual interactions between all members of the group. In larger groups, or settings where multiple conversations are occurring, this method becomes less practical. I settled on two different methods. For groups in which all members seemed equally involved in a single conversation, I still recorded it as individual interactions between each member. I adjusted the weighting of the individual interactions to show any particular flow of emotion between two or more characters that went beyond the normal level of the group. For exceptionally large or fractured groups, I only recorded interactions between people that directly addressed each other. This led to some instances in which two characters were in the same room, and possibly even the same large conversation, but no interaction was recorded between them. Despite this, I felt that this method was more accurate than it would have been if I had recorded an interaction between every person in the room, regardless of whether they addressed each other or not. Below is a sample passage of a diffused group conversation, and how I scored it. In this scene from Still Life, Clara and Peter Morrow are hosting a Thanksgiving dinner for many of their friends in Three Pines. Most of the guests have already arrived, and are listening to Olivier and Gabri recount a hateful instance of homophobia they experienced early in the day. At the beginning of the section, another neighbor, Myrna, arrives and makes the rounds talking to people: In the kitchen, Clara was greeting Myrna Landers. 45 ‘The table looks wonderful,’ said Myrna, peeling off her coat and revealing a bright purple kaftan. Clara wondered how she squeezed through doorways. Myrna then dragged in her contribution to the evening, a flower arrangement. ‘Where would you like it, child?’ Clara gawked. Like Myrna herself, her bouquets were huge, effusive and unexpected. This one contained oak and maple branches, bulrushes from the Rivière Bella Bella which ran behind Myrna’s bookshop, apple branches with a couple of McIntoshes still on them, and great armfuls of herbs. ‘What’s this?’ ‘Where?’ ‘Here, in the middle of the arrangement.’ ‘A kielbassa.’ ‘A sausage?’ ‘Hummuh, and look in there,’ Myrna pointed into the tangle. ‘The Collected Works of W. H. Auden,’ Clara read. ‘You’re kidding.’ ‘It’s for the boys.’ ‘What else is in there?’ Clara scanned the immense arrangement. ‘Denzel Washington. But don’t tell Gabri.’ … ‘How are you?’ Myrna came in from the kitchen, followed by Clara, and hugged Gabri and Olivier while Peter poured her Scotch. ‘I think we’re all right,’ Olivier kissed Myrna on both cheeks. ‘It’s probably surprising this didn’t happen sooner. We’ve been here for what? Twelve years?’ 46 Gabri nodded, his mouth full of Camembert. ‘And this is the first time we’ve been bashed. I was gay bashed in Montreal when I was a kid, by a group of grown men. That was terrifying.’ They’d grown silent, and there was just the crackling and muttering of the fire in the background as Olivier spoke. ‘They hit me with sticks. It’s funny, but when I think back that’s the most painful part. Not the scrapes and bruises, but before they hit me they kind of poked, you know?’ He jabbed with one arm to mimic their movements. ‘It was as though I wasn’t human.’ ‘That’s the necessary first step,’ said Myrna. ‘They dehumanise their victim. You’ve put it well.’ She spoke from experience. Before coming to Three Pines she’d been a psychologist in Montreal. And, being black, she knew that singular expression when people saw her as furniture. Ruth turned to Olivier, changing the subject. ‘I was in the basement and came across a few things I thought you could sell for me.’ Ruth’s basement was her bank. ‘Great. What?’ ‘There’s some cranberry glass—’ ‘Oh, wonderful.’ Olivier adored colored glass. ‘Hand blown?’ ‘Do you take me for an idiot? Of course they’re hand blown.’ ‘Are you sure you don’t want them?’ he always asked this of his friends. ‘Stop asking me that. Do you think I’d mention them if there was a doubt?’ ‘Bitch.’ 47 ‘Slut.’ ‘OK, tell me more,’ said Olivier … ‘About the mulch, Jane,’ Gabri was saying, his bulk bending over one of Peter’s jigsaw puzzles, ‘I can get it to you tomorrow. Do you need help cutting back your garden?’ ‘No, almost done. But this might be the last year. It’s getting beyond me.’ Gabri was relieved he didn’t have to help. Doing his own garden was work enough. ‘I have a whole lot of hollyhock babies,’ said Jane, fitting in a piece of the sky. ‘How did those single yellows do for you? I didn’t notice them.’ ‘I put them in last fall, but they never called me mother. Can I have some more? I’ll trade you for some monarda.’ (Penny) My ratings for the brief exchange above is as follows: • Myrna brings Clara a flower arrangement [Clara - Myrna 1] • Myrna joins the conversation [Myrna - Olivier 1] [Myrna - Gabri 1] • Ruth asks Olivier for help selling stuff from her basement [Ruth - Olivier 1] • Jane and Gabri talk about gardening [Jane - Gabri 1] Some of the interactions, such as Olivier and Gabri, are not recorded here. This is because they were recorded when they first arrived at Clara and Peter’s house. Throughout the entire scene, I add interactions as new characters associate with each other. For instance, neither Peter nor Ben are present in this segment, even though both are in the room. As they join the conversation, their presence and interactions are recorded. 48 APPENDIX B: ADDITIONAL GRAPHS For full-size graphs, and additional graphs for each book, visit the following link: https://drive.google.com/drive/folders/16E8OqCE79-klXbEx3sIZX7sXFxB8z2L?usp=sharing Because I follow the structure of each novel, the number of graphs for each book varies slightly. My default number of graphs for each book is three: one for the interactions leading up to the murder, one for the interactions leading up to the explanation, and one for all the interactions in the book. This model doesn’t always align with the structure of the book, however. In some of the books, the murder happens right at the start, before the characters have a chance to be introduced, making a pre-murder graph useless. In The Murder of Roger Ackroyd, the narrator clearly divides the investigation into two phases, so I adjusted the graphs accordingly. 49 Name of Candidate: Sofi Hull Date of Submission: 17 December 2021