Threatened suicide and baiting crowd formation: a replication and extension of Mann (1981)

ABSTRACT

Using the lens of Deindividuation Theory, Leon Mann explored the formation of baiting crowds in a small sample of public suicides/suicide attempts. Instead of attempting to prevent the suicide, baiting crowds encourage the victim to jump. Our aim was to replicate Mann's study with a larger more diverse sample of suicide attempts and a broadened theoretical lens (Frustration-Aggression). Using ProQuest, we identified 152 public suicide attempts reported in English language newspapers and failed to replicate Mann’s findings. Rather, baiting crowd formation was associated with variables that increase crowd frustration (e.g., duration of the event, blocked traffic, cordoned off pedestrian walkways). Finally, we failed to find an association between the presence of a baiting crowd and the suicidal individual ultimately jumping.

KEYWORDS:

• Suicide baiting
• crowd behavior
• Deindividuation
• Frustration-Aggression

The term ‘baiting crowd’ was coined by Mann (1981) to describe a crowd of people who assemble near a suicidal individual perched on a tall building or bridge threatening to jump. Instead of attempting to prevent the suicide, baiting crowds encourage the victim to jump. In his now classic and often cited paper, Mann utilized Deindividuation Theory (Zimbardo, 1969) to explain the extreme and inhumane behaviors exhibited by baiting crowds as well as to identify several variables likely associated with the formation of such crowds.

Zimbardo’s Deindividuation theory posits a variety of ‘input variables’ that contribute to a state of reduced self-awareness, thereby causing ‘a lowered threshold of normally restrained behavior’ (1969, p. 251). Several of these input variables include feeling anonymous, experiencing sensory overload, being in a large group, and experiencing altered states of consciousness (e.g., through alcohol or drugs). Archival studies (Mullen, 1986) and multiple laboratory and field experiments support various aspects of Deindividuation Theory (Diener, 1979; Diener, Fraser, Beaman, & Kelem, 1976; Lindskold & Finch, 1982; Zimbardo, 1969). However, it is not the aim of the current paper to carry out a critical review of the theory, but rather to acknowledge it as having guided Mann in his selection of variables to explore in his archival study of baiting crowds.

Mann’s archival study of baiting crowds

Mann searched the New York Times and the Chicago Tribune using the terms ‘suicides and suicide attempts’ for the years 1964–1979 and identified 21 instances where individuals threatened to jump from buildings, cliffs, towers, or bridges in the presence of a crowd. Evidence of baiting behavior was found in ten of the 21 instances. The main variables in Mann’s analyses included approximate crowd size, time of day, distance between the individual threatening to jump and the crowd, the duration of the event, and, finally, whether the event occurred during the heat of summer (i.e., June-September). Several of these variables, most notably crowd size, time of day, and the distance between victim and crowd are clearly inspired by deindividuation theory in that each is assumed to play a role in creating feelings of anonymity. That is, people in large crowds are less identifiable and less likely to hold themselves personally accountable for their actions (Darley & Latané, 1968). Similarly, individuals cloaked in darkness are less easily identified, and as a result, experience lower levels of behavioral inhibition (Gergen, Gergen, & Barton, 1973; Zhong, Bohns, & Gino, 2010). Finally, Mann hypothesized that suicide baiting would occur more frequently under conditions where the victim and crowd were separated by relatively larger distances because each would be less identifiable to the other, thereby increasing the likelihood that crowd members would dehumanize the suicidal individual (Milgram, 1965).

The final two variables in Mann’s investigation, heat of summer and event duration, have less obvious ties to Zimbardo’s Deindividuation Theory and might be better understood as variables capable of increasing the crowd’s capacity to behave aggressively. Both field and laboratory studies clearly demonstrate a link between high temperatures and increased aggression (Anderson, 1989; Bell, 2005). Likewise, the link between frustration and aggression is well-established and most pronounced under conditions where goals are unexpectedly thwarted (Berkowitz, 1989). Mann argued that higher levels of aggression (i.e., suicide baiting) would be observed in events longer in duration because crowd members were likely to become frustrated (e.g., doubting the sincerity of the suicide threat).

As predicted, Mann found a significant association between the time of day and the formation of baiting crowds, with more instances of baiting occurring after 6:00 p.m. than before that time. Contrary to expectations, baiting behavior occurred more frequently when suicidal individuals were in close proximity to the assembled crowd (120 feet or fewer) than when they were further away. Finally, a marginally significant association was found between baiting behavior and the month in which the suicide attempt occurred. That is, the heat of summer was marginally associated with increased baiting crowd formation. The remaining variables, crowd size and the duration of the episode, were not significantly associated with the formation of baiting crowds (p = .16 and .14, respectively).

Extending our understanding of the baiting crowd 40 years later

Mann’s ability to identify instances of suicide baiting was constrained by the search techniques available to him at the time. He acknowledged the limitations of his findings by stating that ‘No claims can be made for the completeness or representativeness of this small sample of 21 cases’ (1981, p. 704). However, his was a novel and groundbreaking study that examined several predictions made by Deindividuation Theory and simultaneously offered explanations for a phenomenon that frequently lead to tragic consequences. Understandably, Mann’s baiting crowd findings are widely disseminated in undergraduate textbooks (e.g., Crisp & Turner, 2014; Levine & Moreland, 2008; Stangor, 2004; Vaughan & Hogg, 2005), in books written for popular consumption (e.g., Hood, 2012; Simonton, 1994; Surowiecki, 2005), and are often mentioned in media coverage of baiting incidents. It is unfortunate, however, that the findings of the study are frequently misrepresented. For example, the study is often erroneously cited to support a positive relationship between crowd size and baiting behavior (e.g., Aubé, Rousseau, & Tremblay, 2011; Forsyth, 1999; Phillips & Mann, 2019; Stangor, 2004; Tsang, 2002).

Today, digitized newspaper archives can be searched electronically with relative ease, thus enabling one to collect a larger and more geographically diverse sample of public suicides/suicide attempts. A larger sample will afford more powerful tests of the variables associated with suicide baiting with increased generalizability. Furthermore, increased statistical power will allow us to more adequately assess the possibility that crowd baiting behavior increases the likelihood that the individual threatening suicide will ultimately jump. Replicating the study also allows for the phenomenon of suicide baiting to be viewed from additional theoretical perspectives. This seems especially beneficial in light of the growing literature documenting anti-social behaviors in environments that vary in the degree to which face-to-face interaction occurs (Phillips & Mann, 2019; Sharkin, 2004). This does not imply that various deindividuation theories are only capable of explaining behaviors that arise from face-to-face interactions, but, rather, that variables beyond those typically considered using the lens of deindividuation may contribute to a more complete understanding of the baiting phenomenon. For example, Phillips and Mann (2019) carried out an archival analysis of live online suicide threats and found that baiting behavior occurred in 11 out of the 26 cases. Audience size was positively associated with suicide baiting even though members were remote and geographically distributed.

Finally, given the aggressive nature of suicide baiting, it seems critical to examine the extent to which heightened frustration on the part of the crowd might increase baiting behavior (Berkowitz, 1989; Marcus-Newhall, Pedersen, Carlson, & Miller, 2000). For example, travel impedance negatively affects mood and is associated with feelings of frustration (Hennessy & Wiesenthal, 1999; Novaco, Stokols, & Milanesi, 1990; Shinar, 1998). Travel impedance interacts with other variables known to increase aggression including heat and anonymity. Specifically, Kenrick and Macfarlane (1986) reported a linear association between temperature and driver aggression -measured by horn latency and duration- when the driver’s travel was impeded. Moreover, Ellison, Govern, Petri, and Figler (1995) reported shorter horn latency and longer horn duration when drivers were impeded by an anonymous other compared to a visible other.

The current study

In addition to examining the variables associated with Deindividuation Theory, we aimed to assess the extent to which variables associated with frustration increased baiting behavior. Thus, when available, we examined each public suicide case in terms of whether traffic was blocked, or a central pedestrian route was made inaccessible during the event. Furthermore, we felt a more rigorous examination of the baiting phenomenon ought to include an exploration of how each of our variables is linked to spontaneous helping behavior from members of the crowd (i.e., dissuading the victim from jumping). That is, a stronger argument might be made regarding a particular variable’s role in creating an environment conducive to suicide baiting behavior if that variable is also negatively associated or unassociated with the occurrence of prosocial behavior within the same environment.

Method

Identification of cases

ProQuest and ProQuest Historical were used to search English language newspaper accounts of threatened suicide during the years ranging from 1848–2016. We carried out the data base search using the terms ‘jumping suicide and crowd.’ The search identified 24,656 news stories in which the words jumping and crowd, suicide and crowd, or jumping suicide and crowd appeared. Each of these stories were read by one of five individuals (the authors and two research assistants) in order to identify cases appropriate for our analysis. The overwhelming majority of the 24,656 news stories were unrelated to public suicide attempts (e.g., Olympic ski jumping that drew a crowd, suicide bombers in crowds). Of those stories related to public suicide, additional incident selection criteria included: 1) The case had to involve a crowd that formed in response to a person who was threatening suicide (i.e., incidents involving crowds that formed following a suicide were not included); and 2) The case had to involve a suicide/suicide attempt that involved jumping from some height (e.g., incidents involving individuals setting themselves on fire or stabbing themselves in public were not included). One hundred and fifty- two cases were identified.

Given the considerable number of news stories identified with our search terms and the fact that each news story was read by only one individual, it was impossible to assess the reliability with which news stories were chosen for inclusion in the final data set. While it is reassuring that our search terms and case identification methods resulted in a data set that included all of Mann’s original cases and all of the cases identified in a well-known suicide baiting prevention Facebook page (Suicide Baiting Crowd Prevention, 2016), it is most probable that our data set includes an excellent representative sample of public suicides covered by English language newspapers rather than the entire population of events.

Coding identified cases

Each news article was coded in terms of the date and location of the instance, whether baiting from the crowd occurred, the gender of the suicidal individual, and whether the suicidal individual ultimately jumped. The classification of each case as having taken place during the warmer months of summer varied as a function of location. Summer months were defined as June-September for cases occurring within the Northern Hemisphere and December-March for cases occurring in the Southern Hemisphere. Additionally, when available, time of day, duration of the event, crowd size, distance between suicidal individual and crowd (determined by the height of structure on which the suicidal individual was perched)¹, the presence of environmental characteristics that could create frustration on the part of the crowd (e.g., cordoned off walkways and traffic jams), whether or not crowd members were intoxicated, and whether or not the suicidal individual reacted to the crowd (i.e., comments made indicating that she or he was jumping/not jumping to please/disappoint the crowd) were coded. Finally, spontaneous behavior that was aimed toward preventing the suicidal individual from jumping was coded. While almost every suicide attempt involved such behaviors, we did not code preventative measures taken by police officers, fire officials, members of the clergy, or victim’s family members. Rather, we were interested in behaviors (both physical and verbal) that were not prescribed by role and were diametrically different from baiting behavior. All newspaper articles were independently content coded by the first two authors. Of the 2,736 discrete pieces of information that were recorded (18 pieces per 152 cases), coders disagreed on only 9 of them (percent agreement 99%). These nine discrepancies were resolved through discussion.

Results

Suicide baiting occurred in 78 (51.3%) of the 152 cases examined. The earliest reported instance appeared in 1886 in the Detroit Free Press and the most recent instance, reported in the Greater Manchester News, occurred in 2016. There has been a significant increase in the reported number of crowd baiting incidents over time, r (N = 152) = .272, p = .001 (see Figure 1). The reported rate of spontaneous helping behavior from the crowd was decidedly lower, occurring in only 34 (22.4%) of the cases examined. There was not a significant change in the reported helping behavior over time, r (N = 152) = .037, p = .65. With respect to location, 57.9% of the reported public suicides occurred within the United States and 42.1% occurred in other geographic locations. There was a significant association between location and baiting, χ² (1, N = 152) = 5.53, p = .02, φ_c = .19. Baiting behavior occurred in 43.2% of the incidents that occurred within the US, and in 62.5% of the incidents that occurred elsewhere. Of the 152 individuals threatening suicide, 77.5% were male and 22.5% were female. There was no association between the gender of the jumper and the likelihood of baiting behavior from the crowd, χ² (1, N = 152) = .830, p = .36, φ_c = .074. There were surprisingly few instances where drunkenness on the part of crowd members was mentioned (N = 3); however, suicide baiting behavior occurred in each of these instances, χ² (1, N = 152) = 2.90, p = .09, φ_c = .138. In the following section, we report the analyses for variables found to be associated with suicide baiting in the original Mann (1981) paper.

Figure 1. Baiting incidents by decade.

Variables associated with baiting in Mann (1981)

A logistic regression tested the effects of distance (in feet) between the individual threatening suicide and the crowd, time of day (before 6pm/6pm or after), and time of year (warm summer months/the remaining months of the year) on the likelihood that baiting behavior within the crowd would occur. The model was not statistically significant, χ² (3, N = 95) = 3.36, p = .33. The model explained 4.7% (Nagelkerke R²) of the variance in baiting behavior occurrence and correctly classified only 61.1% of the cases. Given the differences in the rates at which baiting crowds form inside and outside the US, separate logistic regression analyses were carried out for the two samples. In each instance, the conclusions drawn were identical to those suggested by the pooled analysis. Likewise, because baiting crowd formation has increased over time, we carried out separate logistic regression analyses for public suicide events that occurred before and after the year 2000. Here, again, the conclusions drawn for each subsample were identical to those suggested by the pooled analysis.

Given that cases with missing data on any independent variable were not included in this analysis, we also assessed the association between the occurrence of baiting behavior and the distance between the suicidal individual and the crowd for all cases with distance data but failed to find a significant association, r (N = 120) = .055, p = .552. Finally, we also provide tests of association of each independent variable (coded dichotomously as they were in Mann’s original study) with the dependent variable in Table 1.

Table 1. Distributions and χ² analyses of variables explored by Mann (1981).

	Crowd Type
Variable	Baiting	Non-baiting	χ^2 and φc Values
Distance
<120 Feet 120 Feet +	53.8% (43) 47.5% (19)	46.2% (37) 52.5% (21)	χ^2 (1, N = 120) = .417,p = .52, φc = .059
Time of Day
Before 6pm 6pm or After	40.2% (35) 54.8% (17)	59.8% (52) 45.2% (14)	χ^2 (1, N = 118) = 1.98, p = .16, φc = .130
Month of Event
Summer Non-Summer	50.8% (32) 51.7% (46)	49.2% (31) 48.3% (43)	χ^2 (1, N = 152) = .012, p = .91, φc = .009.
Crowd Size
>300 300 +	36.6% (15) 35.0% (14)	63.4% (26) 65.0% (26)	χ^2 (1, N = 81) = .02, p = .88, φc = .017

Mann found a relationship between crowd proximity and baiting behavior; however, it was in a direction opposite that predicted by Deindividuation Theory. Mann posited that the relationship may in fact be curvilinear in that dehumanizing behaviors from the crowd may be less likely when the victim is relatively close, more likely when the victim is at a moderate distance, and somewhat pointless at very large distances (i.e., the victim would not be able to hear the jeering and taunting). This hypothesis was not tested in the original baiting crowd paper because he did not have the statistical power to assess it. We tested the possibility that the relationship between crowd distance and baiting behavior is curvilinear using multiple regression (R = .14, F (2,117) = 1.12, p = .33). Neither predictor (distance nor distance²) was significant (p = .26 and .17, respectively).

Crowd size

We carried out two tests assessing the relationship between crowd size and the occurrence of suicide baiting. The first was carried out using cases where the size of the crowd was reported in the newspaper article as a specific number r (N = 42) = .187, p = .236. The second test was carried out in a manner closer to that carried out by Mann (1981) in that we also included cases where the crowd size was reported in words (e.g., a small group of passersby, a very large crowd). In order to create a continuous variable, we systematically transformed the words into numbers.² While less precise, this strategy increased the sample upon which our analysis was based. Here too, the relationship between estimated crowd size and suicide baiting was not significant, r (N = 81) = .085, p = .451.

Crowd frustration

There was a significant relationship between the occurrence of baiting behavior within the crowd and the duration of the public suicide attempt, r (N = 97) = .209, p = .04. That is, as the event’s duration increased (measured in minutes), baiting behavior was more likely to occur. There was also a significant association between crowd baiting behavior and blocked traffic/cordoned off pedestrian routes, χ² (1, N = 152) = 4.93, p = .03, φ_c = .180. That is, baiting behavior was more likely to occur under conditions where travel was impeded.

Parallel analyses were carried out to assess the nature of the relationship between spontaneous prosocial behavior and duration/frustration, however, the relatively small number of cases involving spontaneous prosocial behavior from crowd members certainly limits the conclusions that we might draw from these analyses. Spontaneous prosocial behavior from the crowd was not associated with the duration of the event r (N = 97) = .043, p = .676. Likewise, the distribution pattern between crowd helping behavior and blocked traffic/cordoned off pedestrian routes was not statistically significant, χ² (1, N = 152) = 2.27, p = .132, φ_c = .122. Finally, spontaneous prosocial behavior occurred significantly less frequently within the context of baiting crowds, r (N = 152) = −.172, p = .034. Across all cases, baiting and helping behavior occurred together in only 7.9% of the cases. When baiting did not occur, prosocial behavior was observed 14.5% of the time (Neither baiting nor prosocial behavior occurred in 34.2% of the cases and baiting with no prosocial behavior occurred 43.4% of the time).

Tests of association between crowd behavior and jumping

Baiting behavior from the crowd was not associated with the increased likelihood that the individual who threatened suicide ultimately jumped, χ² (1, N = 152) = 1.19, p = .27, φ_c = .088. Of the 59 individuals who jumped, 45.8% were taunted by the crowd and 54.2% were not, whereas of the 93 individuals who didn’t jump, 54.8% were taunted and 45.2% were not. Spontaneous behavior from the crowd, enacted to either discourage jumping or to prevent death in the event that the individual did jump (e.g., shouting life affirming messages or placing soft objects below the jumper to break her/his fall, respectively), was not associated with decreased rates of jumping, χ² (1, N = 152) = .518, p = .47, φ_c = .058. That is, in cases where spontaneous helping behavior from the crowd occurred, 44.1% jumped and 55.9% did not. For those cases where no such helping behavior was reported, 37.3% jumped and 62.7% did not.

Discussion

The primary aim of the current research was to carry out a replication of the archival study of crowd baiting behavior reported by Mann (1981). Using a more powerful and thorough method of identifying instances than was available to Mann, we examined newspaper accounts of public suicides/suicide attempts across a much larger time span and generated a sample of events that was larger and more geographically diverse than the sample upon which the original results were based. With more statistical power and a more representative sample, we failed to replicate any of the original findings. Mann identified cover of darkness, crowd proximity, and warm summer months as variables that significantly increased the likelihood that a baiting crowd would form but we found no such associations.

Like Mann, we failed to find a significant association between increases in crowd size and the occurrence of baiting behavior within crowds. It should be noted, however, that while early theories of deindividuation would predict such a relationship, later theories – most notably the Social Identity Model of Deindividuation Effects (SIDE) – posit a more complex relationship between group size and member behavior (Reicher, Spears, & Postmes, 1995). More specifically, according to SIDE, the identities of the individuals who make up the crowd, the social relationships that exist amongst crowd members, and ingroup norms are stronger determinants of individual behavior than is the overall size of the crowd (Levine, Cassidy, & Jentzsch, 2010; Levine, Taylor, & Best, 2011). Unfortunately, the identities of the individuals who made up the crowds were rarely, if ever, mentioned in the newspaper coverage of the suicides/suicide attempts we studied. Therefore, we could not explore in any systematic manner many of SIDE’s interesting predictions in the current investigation.

The inherently aggressive nature of the behavior under investigation led us to predict that suicide baiting would be more likely to occur in environments where crowd members became angry, frustrated, or had their travel impeded (Berkowitz, 1983; Lupton, 2002). As predicted, baiting behavior was significantly more likely to occur during public suicides/suicide attempts when sidewalks and pedestrian walkways were cordoned off and roadways/bridges were closed or made impassable due to backed up traffic. Moreover, jeering and taunting from crowd members were significantly and positively associated with the duration of the event. Anecdotally, in many instances, evidence of the frustration felt by crowd members can be detected in the content of the words shouted at the suicidal individual. Examples of these comments included complaints about being prevented from traveling to work or some other destination and anger regarding having leisure time/one’s lunch hour cut short. An especially illustrative example of this can be found in the event involving a 23 year old man with a history of mental illness who ultimately took his life after a 21-hour public ordeal. The hostility expressed by crowd members was palpably associated with thwarted goals. While members of the crowd shouted that they were being inconvenienced, an electrician who claimed he had lost 150 dollars because his work site was shut down due to the suicide threat stated, ‘They ought to go ahead and shoot him and get it over with … Any one of these construction guys could get a gun from their truck and take care of the problem right here.’ Several local radio stations received calls from callous individuals complaining about the situation, to which one radio show host responded ‘Our motto around here is: If you’re going to do something stupid, make sure it only affects you’ (Hendrick, 2001).

The patterns of spontaneous helping from crowd members across the two variables designed to capture frustration were not significant; however, spontaneous helping behavior was significantly less likely to occur in contexts where baiting occurred. Our ability to offer an unequivocal interpretation of the spontaneous helping findings is certainly constrained by the small number of instances observed but it might be that situational factors associated with baiting behavior do indeed reduce the likelihood that prosocial behaviors will occur. That is, one might be less inclined to adopt an empathic focus under conditions where important personal goals are thwarted. Alternatively, simply hearing the other crowd members’ taunting and jeering comments may inhibit the expression of caring behavior. That is, baiting behavior may serve to cue feelings of apathy amongst the non-baiting members of the crowd. Interestingly, Phillips and Mann (2019) also found that non-baited individuals were more likely to receive offers of assistance than were baited individuals in their study of online suicide baiting.

While we failed to find a significant relationship between the presence of a baiting crowd and the suicidal individual ultimately jumping, it is important to emphasize, in light of the context, that null results cannot be interpreted to mean that baiting crowds have no impact upon an individual’s decision to jump from their location. Anecdotally, there were numerous instances (although too few to analyze systematically) of suicidal individuals acknowledging the role that the baiting crowd played (both in terms of wanting to jump to please the crowd and not jumping to spite them). It is likely that the methodology used in the current investigation lacks the sensitivity necessary to assess reliably the role that the crowd played in affecting the psychological state of the suicidal individuals involved. More appropriate methods might include conducting idiographic interviews of public suicide survivors in an attempt to assess more systematically the role that bystanders play (Knapp, 2015). However, it should be noted that in the three known studies exploring suicide baiting (Mann, 1981; Phillips & Mann, 2019: the current study) no association between baiting and the victim following through with their threats has been found.

We believe the current investigation makes a valuable contribution to the existing literature on suicide baiting for a variety of reasons. The geographic diversity of our sample demonstrates that baiting crowds are not a distinctly American phenomenon. It is extremely important to note, however, that the geographical differences found along with the finding that the formation of baiting crowds has increased over time should be interpreted with caution, especially in light of the methodology used. While world-wide standards exist regarding how the media should cover suicides and suicide attempts (World Health Organization, 2008), these standards have changed over time and vary in the degree to which they are adhered (Chu, Zhang, Cheng, Schwebel, & Hu, 2018). Also, it makes sense that more reports of suicide baiting were reported in newspapers in the decades following the publication of a scientific study of the phenomenon. That is, increased awareness coupled with the ability to label and discuss the behavior may increase newspaper coverage. Future research regarding the formation of baiting crowds could involve the careful coding of police reports created regarding the events. It seems likely that the content of such reports would be less variable over time and place.

Importantly, the only variables that were significantly associated with baiting behavior indicate that self-awareness is likely very much at play because one’s needs and desires are necessarily acknowledged when feelings of frustration are experienced. The current study, coupled with the growing literature documenting road rage and its various causes, suggests that predictions grounded in frustration-aggression theory ought to be studied more systematically in examining suicide baiting. The failure to find an association between situational variables that increase anonymity and aggressive behavior seems at odds with the growing literature supporting the notion that computer mediated communication is more antisocial under conditions of anonymity, especially when one is exposed to aggressive models (Zimmerman & Ybarra, 2016). However, it is likely that the degree of anonymity one experiences in public spaces is significantly lower than that experienced when cloaked behind the screen of a computer communicating with distant unknown individuals (Phillips & Mann, 2019). Future research should explore the contextual effects on experienced states of anonymity.

Acknowlegments

We would like to thank Andy Bui and Adam Rash for their assistance in locating public suicide cases.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. When well known landmarks were involved (e.g., Golden Gate Bridge), effort was made to ascertain the height variable when it was missing from the newspaper article’s coverage of the event. This is the only variable of those studied for which values other than those explicitly stated in the newspaper article were used.

2. When words were used, Passersby were coded as 15, dozens were coded as 24, small crowds were coded as 50, and large crowds were coded as 300.