Darrow’s task was diffi cult. He was attempting to represent schemati-
cally a highly mobile population of gay men with a large number of sex-
ual interconnections over several years. Even assigning a city or state
classifi cation was not straightforward, since some cases moved between
cities. Patient 0 was not the only one to travel frequently and switch resi-
dences between 1978 and 1982. The man identifi ed as SF 1, for example,
had moved to San Francisco in 1979, where he would receive a diagno-
sis of KS in March 1982. He had also spent time in New York and Los
128. Luc Pauwels, “A Theoretical Framework for Assessing Visual Representational
Practices in Knowledge Building and Science Communications,” in Pauwels, Visual Cul-
tures of Science, 4– 5.
129. Darrow reenacted his research practice for the documentary fi lm The Zero Factor,
the fi rst in the multipart televised documentary series A Time of AIDS, directed by Anne
Moir (Princeton, NJ: Films for the Humanities, 1992), videocassette (VHS).
Figure 2.5 “Cluster— LA Cases,” “Cluster— Connection of LA and NYC Cases,” and
“Cluster— Relationship of Symptom Onset to Sex Contact,” AIDS Slides for Dr Foege,
AIDS TRACER M 1983 Jan– June (1 of 2), AIDS correspondence (TRACER) archives,
US Department of Health and Human Services, MS C 607, National Library of Medi-
cine, Bethesda. Three slide reproductions on letter- size papers, 21.6 × 27.9 cm. Courtesy
of the National Library of Medicine. Note the simple representational devices of circles
and straight lines used by the CDC sociologist William Darrow, which bear the infl uence
of the sociograms in fi gure 2.9 but also the fi scal constraints of working with the tools at
hand— in this case, pencils, coins, and paper clips. “Out- of- California KS” appears here in
its earliest form, prior to being abbreviated as “Patient O,” and later “Patient 0.” Photo-
copies of these slides, along with one depicting the extended cluster of forty cases, were
shared through the Department of Health and Human Services in early 1983. It is highly
likely that James Curran used these images in his presentation to the National Cancer Ad-
visory Board at the National Cancer Institute in December 1982, as detailed in this book’s
introduction.
126
chapter 2
Figure 2.6 “Patient #O and his Contacts,” pencil, ink, and colored crayon, 27.9 × 21.6
cm, 1982, Professional Papers of William Darrow, Miami. Courtesy of William Darrow.
The original caption accompanying this image has been retained to provide context for the
reader. This fi gure accompanied the original use of the cluster diagram (shown in fi gs. 2.1
and 2.8) in Darrow’s report to his colleagues of his New York research during the sum-
mer of 1982. It offered additional information to contextualize the cluster representation
and formed an important part of the cluster diagram’s initial environment of use. Without
it— as, for example, when the article was published in the American Journal of Medicine in
1984— the cluster diagram is unable to demonstrate the dates and frequency of “sexual ex-
posures” nor the dates of symptom onset. In the title, one can see that the originally writ-
ten “Patient #57” has been overwritten with the letter O, indicating that this particular ver-
sion was destined for readers outside the CDC. Individuals denoted by the 8000 number
series indicate sexual contacts of “Patient #O” who had not yet been diagnosed with an
AIDS- related illness.
Angeles, and “was known by several cases” in those cities; indeed, two
of these sexual connections would be captured in the cluster diagram.130
Despite these challenges, location was one of the features Darrow em-
phasized in the cluster diagram, along with the type of illness each pa-
tient had, and the sequence, though not the date, of the onset of these ill-
nesses. This set of choices made sense, given that the image was intended
130. Darrow, “Trip Report,” pp. 4, 10, Darrow Papers.
Figure 2.7 “Cts of Pt. 57; Members of FFA; Others,” early draft of the cluster diagram in
regular and colored pencil, 21.6 × 27.9 cm, 1982, Professional Papers of William Darrow.
Courtesy of William Darrow. This draft image, with redactions— straight- edged for cases’
and contacts’ initials and last names, round- edged for case numbers— to preserve confi -
dentiality, shows Darrow working to organize the relationships of the men he has linked to
the cluster. The image suggests some of the many challenges facing researchers: the multi-
ple geographic locations of the CDC’s cases (New York, Los Angeles, San Francisco, and
Philadelphia, for example); the existence of sexual partners who linked cases but who were
not themselves reported as sick (the individual above case 226, for example, who had sex-
ual contact with Patient 57 in 1978); and the impossibility of linking those cases, such as
the one at the bottom left of the drawing, who refused to take part in the study. It becomes
clear how important Patient 57’s records of his contacts and his cooperation with the CDC
investigators were to the completion of the cluster study and to the central role he would
acquire as “Patient 0.”
Figure 2.8 “Extension of the Los Angeles Cluster to New York City and Elsewhere.”
Numbered cluster diagram, hand- colored with crayon, 21.6 × 27.9 cm, 1982, Professional
Papers of William Darrow. Courtesy of William Darrow. The original caption accompany-
ing this image has been retained to provide context for the reader. This image, which ac-
companied fi gure 2.6 in Darrow’s September 1982 trip report, shows the cluster diagram
reaching its near- fi nal state. The image was hand- colored to indicate types of illness and
included each individual’s CDC- assigned case number, allocated roughly in order that the
cases were reported to the CDC (with redactions to preserve confi dentiality).
The Cluster Study 129
to demonstrate a sexual network of individuals whose contacts with each
other were deemed unlikely to occur by chance, given their geographic
dispersal.
It appears that an aesthetic decision was made to keep the image
“clean” by retaining a signifi cant amount of white space, which meant
that other pertinent information was left out. Dates and frequency of
sexual contact for each connected pair were excluded from the graphic
(though some of this information would have been available to Darrow’s
colleagues through previous correspondence and additional information
attached to his reports), as were the specifi c dates for the onset of their
illnesses. Again, in the initial circumstances of production, the diagram
that would become the iconic signifi er of the cluster study was inter-
woven in a library of other diagrams and documentary evidence that
gave it support and meaning. As the image moved farther from the ini-
tial workings of the task force and was selected, it seems, as the best rep-
resentation to “illustrate” the study, the accompanying information and
assumed knowledge required to interpret and interrogate the diagram
diminished. It became diffi cult, if not impossible, to determine from
looking
at the image the dates at which patients fell ill and whether there
was any directionality to the spread of infection. Thus, the farther ob-
servers were situated from the inner workings of the task force, the less
information they had to contextualize their interpretations of its repre-
sentation and the greater their chances of interpreting it through other
means and assumptions.
Another feature missing from the image— perhaps not surprisingly,
given the representational challenges it would have brought— was the
number of total partners each patient had reported. Most of the cases in
the cluster had hundreds of contacts that had potentially infected them
with the posited transmissible agent. The diagram’s clean circles, straight
lines, and white space all work to artifi cially fi x these cases and their in-
terconnections within a single frame of space and time, removed from
the messiness— and reality— of their lives. Many viewers would focus on
the connections depicted, rather than the thousands of undepicted con-
nections that had more likely infected these individuals, which had taken
place outside of their memories, or without the necessary confi rmation
by a second party or a friend, and thus outside the allowed representa-
tion of reality framed by the image. In other words, many viewers as-
sumed that the connections depicted in the diagram denoted the spread
of a transmissible agent within the cluster that subsequently caused the
130
chapter 2
patients in question to develop AIDS. In all likelihood, however, these
sexual liaisons occurred after the men had been infected— again, outside
the clean arrangement depicted by the diagram.
For several decades, sociologists had made use of images known as
sociograms to represent the social relationships they studied.131 While
undertaking his PhD in sociology at Emory University, Darrow had
been taught by one of the former research assistants of James Coleman,
a sociologist renowned for his study of the relationships between teenag-
ers in high schools during the 1950s.132 Coleman had made use of num-
bered circles, squares, and triangles linked by lines to represent the al-
liances and groups that made up the student populations in the schools
he studied. The elaborate diagrams he created clearly infl uenced Dar-
row’s representation of the sexual relationships he included in the clus-
ter (see fi g. 2.9).133 Representational conventions had developed over the
years to standardize sociograms and to develop a shared visual knowl-
edge among research professionals. The “most chosen individual” in a
network was typically placed in the center of the image; nodal points (in
this case representing individuals) were to be arranged for maximal clar-
ity and minimal line overlapping; and individuals were to be depicted in
their “‘natural’ groupings of diads, triads, and so forth”— the real- world
relationships being modeled.134
Whether consciously or not, Darrow followed these established visual
conventions by placing “Patient 0” at the center of his cluster diagram.
From a geographic perspective, there was some logic to this positioning—
the placeless “Patient 0” serving as “the missing link” between cases in
Los Angeles on the left and New York on the right (though this geo-
graphic structure disintegrates with cases from other cities). The fl ight
attendant’s superior ability in recalling the names of former contacts led
to his being the “most chosen individual” in the network, an arbitrary
status which was then naturalized and reinforced through visual means.
Later research, which indicated that individuals may mistakenly attri-
bute more importance to a centrally positioned node in a representation
131. Alden S. Klovdahl, “A Note on Images of Networks,” Social Networks 3 (1981):
197– 214.
132. This research assistant was sociologist Martin Levin; William Darrow, e- mail to
author, August 8, 2013.
133. James S. Coleman, The Adolescent Society: The Social Life of the Teenager and its
Impact on Education (New York: Free Press, 1961), 175– 82.
134. Klovdahl, “Note on Images,” 199.
Figure 2.9 Sociogram titled “Network of Reciprocated Friendships among Marketville
Boys,” in James S. Coleman, The Adolescent Society: The Social Life of the Teenager and
its Impact on Education (New York: Free Press, 1961), 175; 13.0 × 18.5 cm. Reprinted with
the permission of Free Press, a Division of Simon & Schuster, Inc., from Adolescent Soci-
ety by James S. Coleman. Copyright © 1961 by The Free Press. Copyright © renewed 1989
by James S. Coleman. All rights reserved. Coleman’s research assistant, Martin Levin,
taught Darrow at Emory University in the 1970s and introduced him to the use of socio-
grams to represent human relationships. In these representations of teenaged boys in a
small, coeducational high school in autumn 1957, the bolded circle represents the boy most
mentioned in interviews as the best athlete in each year group, the square shows the most
scholarly, and the triangle the boy most often mentioned as most attractive to members of
the opposite sex. The larger rounded shapes represent social cliques. The reciprocated re-
lationships between the school’s girls were depicted separately, and attempts to show rela-
tionships between the two sexes were abandoned due to the technical challenges of analyz-
ing such a high quantity of relationship data.
132
chapter 2
of a network, suggests that this combination— of visual conventions and
subsequent viewer interpretation— served to amplify the perceived sig-
nifi cance of “Patient 0.”135
The initial importance of the study— as reported in the June 1982
MMWR article— was the evidence it seemed to offer that AIDS was
caused by a sexually transmissible agent. Later that year, as the inter-
views of the New York cases offered more details, it offered the possibil-
ity of estimating a preliminary incubation period for the condition. Both
of these developments focused on the role of the “Out- of- California
case,” and the excitement he generated is captured by his prominent role
in the cluster diagram. As these conclusions made their way up the chain
of command in the US Department of Health and Human Services, a
fi xity to the results developed. In a message outlining the activities un-
dertaken by the National Institutes of Health in response to AIDS, Ed-
ward N. Brandt Jr., the assistant secretary for Health, described the in-
vestigation as such: “Investigators traced a case from one individual
which was spread to nearly a dozen sexual contacts.”136 This conclusion,
that AIDS had “spread” from the “out- of- California” patient to his sex-
ual partners, had been communicated by James Curran as recently as
December 1982. It was, however, in confl ict with the evolving conclu-
sions of the cluster study. Shortly after Brandt wrote his message, a pre-
publication draft of the extended cluster study was circulated for feed-
back; it urged caution in interpreting any directi
onality in the cluster. “A
cluster of AIDS patients linked one to another by sexual contacts does
not necessarily imply,” the authors noted, “that an infectious agent was
directly transmitted from an infectious host to a susceptible partner.”
They also acknowledged that “most of the patients included in this clus-
ter were exposed to another member of the cluster during a single sexual
encounter or during periods of sexual contact that lasted for several days
or weeks.”137 This fi nding represented another potential weakness in the
135. Cathleen McGrath, Jim Blythe, and David Krackhardt, “The Effect of Spatial Ar-
rangement on Judgments and Errors in Interpreting Graphs,” Social Networks 19 (1997):
223– 42.
136. AIDS events and actions undertaken by the National Institutes of Health, chro-
nology attached to letter from Edward Brandt to Henry Waxman, 4 March 1983, AIDS
TRACER M 1983 Jan– June (1 of 2), AIDS Correspondence (TRACER) archives, 1982–
1990, Offi ce for the Assistant Secretary for Planning and Evaluation, US Department of
Health and Human Services, MS C 607, National Library of Medicine, Bethesda.
137. Auerbach et al, “Cluster of Cases,” 7– 8.
The Cluster Study 133
diagram. The uniform straight lines between circles, which indicated the
sexual connections in the diagram, did not allow readers to distinguish
between the relative “strength” of these sexual connections, which sug-
gested that a single sexual encounter was as likely to transmit the sus-
pected agent as a relationship spanning several months.138
As we shall see in chapters 3 and 4, Randy Shilts was one of the most
infl uential consumers of the cluster study to misconstrue (or at least mis-
represent) its fi ndings, and the diagram featured strongly in his skewed
reading. “At the center of the cluster diagram,” he wrote, “was Gaetan
Dugas, marked on the chart as Patient Zero of the GRID epidemic,”
making use of the acronym for “gay related immune defi ciency,” an early
term coined for the syndrome. Shilts appears to have viewed the dia-
gram as a sort of closed sexual network with infection radiating out from
Patient Zero and the Making of the AIDS Epidemic Page 22
