A Case Study of Pocahontas Island: Resistance to Post-Impact Evacuation in a Historic Black Community

by Janice H. Cosel and Laura Jo Leffel

Copyright 1999 by Janice H. Cosel and Laura Jo Leffel. All rights reserved. Permission to reproduce material from the JOURNAL is granted for academic research, library or other archives, or classroom instruction provided the source of material is acknowledged by appropriate citation.

Note: The authors of this article are the recipents of a Meritorious Achievement Award from the Disaster Preparedness and Emergency Response Asociation International for the research study documented in the article.


(1) Why does the population of a historic community that has lost the majority of its housing refuse to seek shelter outside its well defined boundaries? On August 6, 1993 this problem confronted emergency responders in Petersburg, Virginia (r1). The normal expectation of responders is that disaster victims will welcome assistance (s4). When they do not, it is important to understand the reasons so that future emergency management efforts can effectively serve community needs, whatever the background situation (Daines 1991).

(2) Pocahontas Island is a small island, located in the Appomattox River, within the City Limits of Petersburg. Shortly after noon the sky over Pocahontas Island turned gray, and a much needed rain started falling. At 1:30 pm, without warning, the community and its 62 homes were devastated by a tornado that had tracked across the Olde Towne area of downtown Petersburg and over the Island. This tornado subsequently caused the collapse of a Wal-Mart store in Colonial Heights, Virginia, with the death of three persons, and thereby attracted national print and electronic media attention (“The Tornado” 1993).

(3) The Petersburg-Colonial Heights tornado was one of nine tornadoes produced by a single storm; eight of these were in the Fujita F-1 to F-2 range, historically predictable for Virginia. Petersburg, Colonial Heights, and Pocahontas Island experienced the single F-4 tornado, which was characterized by a ground track of 12 miles, a path width of 250 yards, and wind speeds of 250 miles per hour. Combined, the nine tornadoes killed four people, injured 246, and caused an estimated $15,000,000 in damage (Virginia State Climatology Office 1993).

(4) The Fujita wind damage scale classifies a F-1 tornado as a weak tornado with winds of 73 to 112 miles per hour accompanied by moderate damage. A F-2 tornado is classified as strong with winds of 113 to 157 miles per hour, causing considerable damage. In contrast, F-4 tornadoes are violent storms, described as devastating with winds of 207 to 260 miles per hour. Such a storm is capable of destroying brick structures, leveling buildings, and ripping buildings from their foundations. Its winds can debark and uproot trees and throw cars for 100 yards (Virginia Department of Health, Office of Emergency Medical Services 1996).

(5) On Pocahontas Island, a large brick winery, which had been converted to a storage facility for lumber products and hazardous materials (Smith 1981), was demolished. The debris from this building blocked the approaches to the bridge from the Island, effectively closing the only ingress for emergency services personnel and egress for the Island’s population. Although damage reports vary, we believe that the best evidence is that all 62 homes were damaged; of these 34 were eventually red- tagged, indicating the building was unsafe and that residents could not reenter it (r1, s3).


(6) Emergency managers and response and recovery agencies have a fundamental expectation that disaster victims will both need and welcome assistance following a disaster (United States, Federal Emergency Management Agency, Emergency Management Institute 1993). Although disaster victims have suffered from the impact of the event, they can be expected to react purposefully and quickly to assess their situation and to work to meet their immediate needs (Drabek 1991). Problems experienced in the mobilization of assistance during previous disasters have highlighted the need for rapid delivery of services to meet both humanitarian and political objectives (Longshore 1998, Cleary 1990). Rapid needs identification and damage assessment sets the stage for response by identifying the types of services needed United States, Federal Emergency Management Agency 1990). A wide range of government and voluntary agency disaster recovery assistance programs, the Disaster Field Office and Disaster Assistance Center outreach organizational structure, and, in Virginia, networks of locally based Disaster Recovery Task Forces are in place to help provide these services (s1, s2, s5).

(7) Among the tactics governments have adopted to prepare for the onset of disasters is evacuation. Although individual evacuees may fear looting or deviant behavior by their fellow citizens on the roadways (Fischer 1994), getting the population out of the impact area and sheltering them in congregate care facilities remains a staple of response to a developing emergency. And when people loose their homes, emergency managers see the local disaster shelter as a key service government can provide until alternate arrangements can be made for the evacuees’ housing (s5, r1).

(8) In order to understand this evacuation process, Perry, Lindell, and Greene (1981) developed a model to explain why and how citizens respond to the need to leave their homes during an emergency. This model requires four key elements: (1) a credible warning delivered by a means the population can access, understand, and believe, (2) activities the citizens undertake to confirm the warning must, in fact, do so, (3) there must be a real threat that can be understood as a danger, and (4) citizens must perceive a high degree of personal risk. It is important to note that this model addresses actions when confronted by an impending disaster, rather than evacuation actions post impact.


(9) On the face of it, having a tornado destroy your community would seem to be a reasonable catalyst for leaving home and asking for immediate assistance. Why then did the residents of Pocahontas Island refuse to seek shelter off the Island?


(10) We approached this research question using a qualitative technique, specifically grounded theory. We selected this method because this incident had not previously been reported in the literature with which we were familiar. In addition, initial discussions with emergency managers yielded conflicting opinions and reports of perceived facts, along with wide disparity in assessments of causation. We concluded that, in these conditions of uncertainty, grounded theory methods afforded the best approach to developing a possible theoretical explanation for the events of August 6, 1993.

(11) We interviewed 14 individuals who either were knowledgeable of the history of Pocahontas Island, were present at the time and place of the tornado, had technical knowledge of the storm’s characteristics, had a role in the emergency response at state level, or were involved in local government at the time. Interviews were open-ended, with questions being generated based on local history and news accounts, video coverage of the response, replies to previous questions, and issues raised in previous interviews. Field notes from the interviews were coded using a simplified coding schema to determine common issues reported by two or more respondents as the basis for theory development.


(12) Pocahontas Island was the first settlement for free Blacks in Virginia. Its history dates to the 1700s when a house on the Island was used as a shelter for escaped slaves (i1, b1). In 1732 commercial development of the Island was started by wealthy white landowners, introducing a Black labor force. Although they lived there, Blacks were not allowed to make decisions regarding the area; governance of the Island was entrusted to a board of nine trustees (Scott and Wyatt, 1960). In 1752 the City of Petersburg built a bridge connecting the Island to the City. The residents of the Island viewed this as “an enemy in disguise for Pocahontas” (Smith et. al. 1981, 6).

(13) From 1771 to 1781 the White population of the Island declined, apparently to a large degree as a result of the American Revolutionary War. This exodus cleared the way for free and emancipated Blacks to rent and purchase homes, further developing their community. In 1787 Pocahontas Island was incorporated into the Town of Petersburg, and “it ceased to be an independent community. From then onward, Pocahontas was a much neglected appendage of Petersburg” (Smith et. al. 1981, 7-8).

(14) After the Civil War, more freed slaves arrived in the Island community creating a framework of sustained social problems based on overcrowding and unemployment with resulting poverty. Although the immediate effects were disease and malnutrition, these abated as Virginia started recovering from the ravages of the War, and by 1875 residents were either running businesses on the Island or working in Petersburg. By 1887, employment figures indicated 153 people were employed on the Island. The subsequent economic history of the community sees a series of boom and bust cycles. For example, in 1893 only 12 people were reported as being employed on the Island, due in part to a severe national downturn in the economy. In 1897 recovery was evident with over 100 employed (Smith et. al. 1981).

(15) In this century two World Wars continued the cyclical nature of Island life. Between 1914 and 1918 another migration began; at the end of 1918, 193 persons were employed on the Island (21 were employed by the armed forces) in 26 different occupations. During both World Wars increased opportunities also encouraged outward movement with younger, educated, and more progressive residents leaving for the North to seek jobs, better living conditions, and education. Islanders today refer to this time as when they lost the “cream of the crop,” with the majority of the remaining population being elderly, retirees who sustain themselves on small fixed incomes (i1).

(16) In 1971 the Petersburg City Council established a light industrial zone on Pocahontas Island, including the homes of 250 residents. This zoning classified the existing homes as “non-conforming users,” a classification that prohibited repairs and additions (Greenfield 1996). As the homes on the Island fell into disrepair they could be condemned without the owners having a chance to rectify the condition. Residents would then be required to move, and their homes would be demolished, with any chance of case-by-case legal recourse. This opened the door for industrial expansion; businesses migrated to the Island without permission from the residents and without monetary recompense to the Islanders for their land. Residents became increasingly fearful, realizing that as land use changed to industrial, they could be forced from their homes with no place to go, and no money to go there (Smith et.al. 1981)

(17) In 1975 the residents secured legal representation in their battle to have Pocahontas Island rezoned as a residential area. Their attorneys argued that the cultural and historic integrity of the Island should be secured and the industrial encroachment should cease. After six months of litigation, Petersburg City Council accepted the residential rezoning, temporarily securing the traditional community (Smith et. al. 1981).


(18) Public safety response to the Island was not significantly delayed following the tornado’s touchdown, given the degree of overall damage to the downtown of the City of Petersburg. Because the water treatment plant for Petersburg and Colonial Heights is located at the eastern end of the Island, personnel, including those of the South Central Wastewater Treatment Authority, responded immediately with heavy equipment to remove the large amount of debris blocking the only entrance to the Island. This made it possible for emergency medical services to respond to the scene and remove the only injured resident approximately one hour after the 1:30 pm touchdown (i1). Considering the blockage of roads by debris, the extent of the damage, and the overloading of the emergency services, this appears to us to have been a reasonably rapid response.

(19) When city officials and local emergency personnel responded and suggested evacuation, based upon the extensive visible damage, they were met with resistance. The degree of this resistance remains an area of controversy, with three distinct groups of perceptions. These include (1) intimations of potential physical resistance by the residents, (2) verbal refusal by the residents, or (3) the authorities benevolently deciding to allow the residents to stay in their homes (i1, r1, s1, s2, s3, s4, s5, s6). As researchers, we were only able to validate one of these three perceptions–verbal refusal–based on the preponderance of interview evidence. It may be that a combination of the passage of time within the event itself and the time elapsed subsequent to the tornado account for this variability.

(20) Given the historical controversy with the City of Petersburg over control of the development of the Island and encroachment by industry, the residents feared that evacuation would mean loosing their homes to the City government without recourse. They had fought too diligently in 1975 to preserve their community to allow a disaster to drive them out (i1). “We don’t want to leave,” chanted a 66 year old resident, “we want to stay right here” (Johnson 1993, A1). Even though there was no gas or electricity service and portable toilets had to be used, city officials could not convince community members to leave their homes. A 76 year old resident was quoted in THE PROGRESS INDEX as proclaiming:

“I’m not going anywhere. This is my home. I stayed here last night, I’ll stay tonight, and I’ll stay the night after that.” (Kirishnamurthy 1993, 1)


(21) Although Perry, Lindell, and Greene’s model of evacuation addresses pre-impact evacuations, we felt that it might be instructive to apply it to the analysis of an evacuation decision situation post-impact. Viewed from this perspective, the four elements (credible warning, confirmation, threat, and risk) were not present on Pocahontas Island. In terms of the disaster cause, the elements of warning, confirmation, and threat were no longer applicable as the tornado impact event was complete. This meant that there was also no longer a risk–nothing more was going to happen to the Islanders as a direct result of the tornado.

(22) However, using the same model, it would seem that the situation may have created a mirror image of the criteria leading to a decision not to evacuate. First, the past history of the Island and its community provided the credible warning. Second, the litigation of 1975 confirmed this warning. Third, residents may have perceived the threat of dispossession as real; their houses were damaged--potentially untenable–and emergency and city personnel were suggesting evacuation. Finally, the citizens perceived a high risk of the threat becoming reality--losing their homes if they left the Island. This suggests that the residents’ hesitancy to evacuate is not surprising and may well have been rationale given their set of perceptions.

(23) Could there be other explanations for the refusal to evacuate? It is possible that older populations are more intensely evacuation resistant than younger ones, or that Black populations are less willing to evacuate that White, Hispanic, or Asian populations. Tornado events may result in lower rates of evacuation compliance and shelter use than other types of events. However, the literature and training courses generally available to emergency managers do not suggest that any of these factors are of a frequency that justifies their inclusion in evacuation planning (see, for example, Auf der Heide 1989, Drabek and Hoetmer 1991, Kramer and Bahme 1992, and Rosenthal, Charles, and Hart 1989).

(24) We believe that the response of the residents of Pocahontas Island to the Petersburg-Colonial Heights tornado of 1993 highlights a key issue for emergency managers. Understanding the difficulties that may be encountered in the response and emergency recovery phases following disaster impact requires a clear understanding of the communities the emergency managers serve. It is not enough to understand governmental perspectives and the perspectives of the dominant social groups. Conflict in perceptions by residents--especially in isolated populations within a broader community--versus expectations of authorities, whether such conflict is real or imagined, impacts the role of each player during the event and sets the stage for future interactions. Our application of Perry, Lindell, and Greene’s model suggests it may be worthwhile to examine other disasters and disaster process models to identify cases which may explain the occurrence of unexpected responses before, during, or after a disaster.

Article Links

Sources and Review Comments
supporting references and peer reviewer comments not included in the text

Journal Links

Electronic Journal of Emergency Management Cover Page
Journal policies and procedures and index to articles

This page has been visited times.