prospect— unfounded, in his view— that gay men might be singled out
as carriers of a virus. Sonnabend wrote angry letters to David Sencer,
the commissioner of the New York City Department of Health and for-
mer chief of CDC; to the Democratic congressman Henry Waxman; and
to Michael D. Gregg, the editor of MMWR. To Sencer, he declared that
“CDC has shown in this matter an unfortunate lapse in their abilities
to undertake research of this kind, and what is perhaps more serious, a
lapse in appreciation of their responsibilities. It is indeed a very serious
matter to suggest that members of any minority group may be carriers of
what in effect is a cancer virus.”114 To Waxman, he complained, with ref-
erence to the Gay Report, that “the fact that material from a survey con-
ducted in part through pornographic magazines (reference no. 2) is cited
in support of the contention is a further indication of an oversight on
112. Ron Bluestein, “Cries and Whispers of an Epidemic,” Advocate, November 24,
1987, 64.
113. Reminiscences of Neil Schram, 1996, p. 47, Physicians and AIDS Oral History
Project, CCOHA.
114. Joseph Sonnabend to David J. Sencer, 19 July 1982, folder: L.A. Cluster FF! Son-
nabend Papers.
The Cluster Study 119
the part of the editors with respect to the responsibilities associated with
their position.”115 Sonnabend was particularly concerned with the clus-
ter study’s reliance on “generalizations contained in the Kinsey report of
1948,” used to estimate the size of the gay populations in California and
thus the likelihood that the individuals in the AIDS cases included in
the cluster might know each other by chance. He suggested to Gregg that
the Gay Report, upon which some of the study’s evidence relied, “can
hardly be regarded as a valid scientifi c reference,” since it could “support
any point of view.” Sonnabend angrily wrote that “these are the fl imsi-
est of referral sources to support a contention with such far- reaching im-
plications. These derive from the perception of gay men as carriers of a
cancer virus.”116
Andrew Moss, the San Francisco epidemiologist, was also critical
of the CDC’s epidemiologic abilities, as well as the cluster study’s ap-
proach. “I met the CDC people that came out here,” he later recalled in
an oral history interview, “Curran and Jaffe, and I drove them around
in my hideous beat- up Volkswagen. Both of them said, ‘Well, I’m not re-
ally an epidemiologist,’ meaning, ‘I don’t know what I’m doing.’ Which is
true; they had no idea what they were doing.” He continued, dismissively:
They had the CDC kind of three- week course, or whatever it is they get. They
were not formal epidemiologists. They were insecure about their ability as
formal epidemiologists. . . .
The CDC didn’t in those days give them a lot of training before pitch-
ing them in. Infectious disease epidemiology is sort of like, “Get in there and
see what’s going on!” Outbreak investigation— stamp it out. AIDS is a bit
different.
See, the interesting thing about AIDS from a professional point of view is
it’s an infectious disease that looks like a chronic disease. It takes a long time
[to develop]. You don’t go and stamp it out. It’s not like salmonella or some-
thing. You don’t stamp out an outbreak by fi nding the infected chicken. Al-
though that’s what Darrow tried to do. That’s what all that Patient Zero stuff
was about.
It’s like trying to visualize AIDS in the model of an infectious disease out-
115. Joseph Sonnabend to Representative [Henry] Waxman, 15 July 1982, folder: L.A.
Cluster FF! Sonnabend Papers.
116. Joseph Sonnabend to Michael D. Gregg, 14 July 1982, folder: L.A. Cluster FF!
Sonnabend Papers.
120
chapter 2
break. It’s quick; it spreads from person to person; you go in there and you
fi nd the prime cause, and you remove it. This is not the way AIDS works.117
Moss acknowledged that, while more confi dent in his training, he felt
less connected to the CDC than his rival San Francisco epidemiologist,
Selma Dritz of the city’s health department.118 This insecurity in a possi-
ble turf war in San Francisco over access to information about cases may
have colored his frank recollections. Still, by making clear his belief that
the type of work done by the CDC was distinct from that done by “for-
mal epidemiologists,” Moss exhibits views that are suggestive of the divi-
sions rendered by shifts in epidemiological practice in the mid- twentieth
century.
Lengthening Incubation Periods
Following the isolation and identifi cation of a virus in 1983 and 1984 and
the increasingly widespread ability to test for viral antibodies in 1985,
the practical importance of the cluster study waned. Additional research
clarifi ed the virus’s transmission routes, and the key questions shifted
from whether sexual contact spread the virus to the types of activities
that presented the highest risk and, more problematically, to what ex-
tent the disease posed a threat to the “general public.”119 The cluster
study receded from the spotlight, having served its role in helping to re-
direct the attention of researchers and concerned members of the public
to the likelihood of a sexually transmissible agent as a cause for AIDS.
The hypotheses on which the study had been based and which provided
signifi cance to its central visual representation did not receive explicit
challenge— despite evolving knowledge about the time between HIV in-
fection and the appearance of an AIDS- related illness. Early estimates
of an incubation period were hampered by diffi culties in testing for sub-
clinical infections and by a lack of historical data, as well as by an ini-
117. Moss, “AIDS Epidemiology,” 246– 47.
118. Andrew Moss, interview with author, San Francisco, July 24, 2007, recording
C1491/09, tape 1, side A, BLSA.
119. For criticism of the discourse that posited the threat of infection shifting from
“risk groups” to the “general population” or “general public,” see, for example, Grover,
“AIDS: Keywords,” 23– 30; Patton, Inventing AIDS, 25.
The Cluster Study 121
tial hypothesis that AIDS patients were likely to develop complications
at the same speed as patients taking immunosuppressant drugs after or-
gan transplants.
Scientists would soon note that the incubation period seemed to be
lengthening. In 1986, for example, three public health researchers sug-
gested that it was “becoming apparent from several lines of evidence
that the period of latency from exposure to illness may be fi ve years or
longer.”120 And the authors of a paper published in 1988 noted drily, “In
past studies the estimated average incubation period has been disconcert-
ingly close to the time span over which data are available, suggesting that
the average could lengthen as more information accumulates.” The pa-
per also noted that the most up- to- date analysis then available s
uggested
that the average incubation period between HIV infection and the onset
of AIDS was seven to eight years— an estimate which would grow to a
median of just over ten years for adult men in the absence of treatment.121
In his 1988 letter to the New York Review of Books, Andrew Moss
pointed out that this evolving understanding of HIV’s incubation period
made it highly unlikely that the neatly arranged cluster continued to in-
dicate what it had once appeared to do.122 Before, when researchers re-
lied on the comparison to renal transplant recipients, who developed KS
an average 14.9 months after taking immunosuppressive drugs, the cases
making up the cluster could be assumed to link together in a manner
which suggested that the patient who fi rst demonstrated symptoms likely
passed the infection to one displaying symptoms afterward.123 While
James Curran had qualifi ed his statement at the NCAB meeting in De-
cember 1982 with the caution that this was “certainly a very loose fi g-
ure,” he maintained that the “out of California case” showed that the
incubation period between sexual contact and onset of illness was ap-
120. Victor De Gruttola, Kenneth Mayer, and William Bennett, “AIDS: Has the Prob-
lem Been Adequately Assessed?” Reviews of Infectious Diseases 8, no. 2 (1986): 297.
121. Roy M. Anderson and Robert M. May, “Epidemiological Parameters of HIV
Transmission,” Nature 333, no. 6173 (1988): 514; Nancy A. Hessol et al., “Progression of
Human Immunodefi ciency Virus Type 1 (HIV- 1) Infection among Homosexual Men in
Hepatitis B Vaccine Trial Cohorts in Amsterdam, New York City, and San Francisco,
1978– 1991,” American Journal of Epidemiology 139, no. 11 (1994): 1077– 87.
122. Moss, “AIDS without End,” 60.
123. Auerbach, Darrow, Jaffe, and Curran, “A Cluster of AIDS: Patients Linked by
Sexual Contact,” second prepublication draft, May 10, 1983, p. 8, folder: L.A. Cluster FF!
Sonnabend Papers.
122
chapter 2
proximately 14 months, since “he had contact with the people about 9
to 22 months prior to the onset of their symptoms.”124 By the late 1980s,
however, when an average incubation period was understood to be at
least six times that long, some of the supposed transmission events noted
in the cluster and represented in its commonly cited diagram might bet-
ter be interpreted as epidemiological red herrings. The actual exposures
that had infected the cluster’s patients had almost certainly occurred
several months or years before the ones depicted, with different partners
than those denoted by the diagram’s links. These names and faces would
have lain beyond the recall of some of the patients and likely beyond the
historical period examined in the study too. In other words, the cluster
most likely represents a network of gay men who had sex with each other
after they had become HIV- positive, and not a web of transmission of a
causative agent.125
Certainly, a patient’s health and other conditions would affect the
speed at which he or she would experience the onset of AIDS. It is pos-
sible that one or more of the men included in the cluster, many of whom
had extensive histories of sexually transmitted infections, could have ad-
vanced along the HIV continuum to an AIDS diagnosis faster as a result
of their previous or concurrent infections.126 It is highly unlikely, though,
that it would be enough to cause them to advance at a rate suggested by
the cluster study’s calculated incubation period.
124. “NCAB Meeting,” 29– 30.
125. Recent molecular evolutionary analysis of stored serum samples collected from
gay and bisexual men has found that extensive genetic diversity of HIV already existed
in these groups in New York City and San Francisco by 1978– 1979. Furthermore, the re-
search used serum and blood product drawn in 1983 from the CDC’s “Patient 0” to gener-
ate an HIV genomic sequence for this individual and to compare its genetic diversity to se-
quences obtained from eight other samples, as well as sequences from genomes stored in
a national database at Los Alamos. The research found that Patient 0’s HIV sequence ap-
peared “typical” of the strains of HIV circulating undetected in the United States in the
mid- 1970s; in other words, it was not foundational to the North American epidemic. See
Michael Worobey et al., “1970s and ‘Patient 0’ genomes Illuminate Early HIV/AIDS His-
tory in North America,” Nature 539, no. 7627 (2016): 98– 101.
126. Deschamps et al., “HIV Infection in Haiti,” 2515– 21. This study determined that
the median time from infection to AIDS was three to fi ve years in a resource- deprived
country in a population with high levels of respiratory, diarrheal, and skin infections. Such
data offer a tentative point of comparison for the claim that the fi rst homosexual men in-
fected with HIV in the United States, having weathered many STDs already, had compro-
mised immune systems and may have progressed more swiftly to AIDS.
The Cluster Study 123
“A Picture Is Worth a Thousand Words”
Given the centrality of the cluster diagram to later interpretations of the
cluster study and the role of “Patient 0,” it is worth examining the pro-
duction and reception of this illustration in greater detail. This chapter’s
fi nal section draws on insights furnished by historians and sociologists of
science since the 1990s to chart the evolution of this scientifi c image. As
a number of authors have pointed out, scientifi c visual representations
are important, socially produced documents crucial to the development
of ideas and theory. They assist scientists in arranging data, testing hy-
potheses, and convincing colleagues. Although they are often treated as
subordinate to the text, images often play a powerful role in the overall
process of knowledge production and communication. Recent work in
the sociology of science encourages readers to consider the broad trajec-
tory of an image’s life span— its conception, the work it needs to accom-
plish, and the various ways in which it is interpreted, by particular audi-
ences in varying contexts beyond the specifi c working environment in
which it was produced— paying attention at each stage to factors infl u-
encing its development. Visual representations convey information, but
they conceal it as well, and they can occasionally, and unwittingly, com-
municate the worldviews of the working environment in which they are
produced.127
Translation or conversion is involved in every process of scientifi c rep-
resentation, whether taking a photograph, recording an interview, tal-
lying a survey, or drawing an image. A number of actors— including re-
searchers, artists, and technicians— make a series of decisions that will
affect the fi nal appearance of the representation. These choices— some
deliberately made, others outside of conscious thought— will determine
which features of a referent (the object, phenomenon, or concept being
represented) are included and emphasized, which are downplayed or ob-
scured, and
which are left out. Institutional context, disciplinary con-
127. Michael Lynch and Steve Woolgar, eds., Representation in Scientifi c Practice
(Cambridge, MA: MIT Press, 1990); Luc Pauwels, ed., Visual Cultures of Science: Rethink-
ing Representational Practices in Knowledge Building and Science Communication (Leb-
anon, NH: Dartmouth College Press / University Press of New England, 2006); Regula
Valérie Burri and Joseph Dumit, “Social Studies of Scientifi c Imaging and Visualization,”
in The Handbook of Science and Technology Studies, 3rd ed., ed. Edward J. Hackett et al.,
297– 317 (Cambridge, MA: MIT Press, 2008).
124
chapter 2
ventions, cost, and individual aesthetic preferences may all play a role.128
For the representation of conceptual phenomena not visible to the eye—
relationships between individuals, for example— standardization is dif-
fi cult, and the representation will depend a great deal on the individual
producing the original work.
In our example, Darrow faced the challenging task of representing
complex conceptual data— the sexual relationships within a network of
early reported AIDS patients, fi rst in Los Angeles and Orange Counties,
later across the country— in a resource- constrained public health setting.
Darrow used the rudimentary and low- cost tools at hand to make his
fi rst attempts to create visual order from the hundreds of names, places,
and dates of sexual contact gleaned from his interviews and other CDC
data sources. These tools included quarters and paper clips to represent
patients and connections between them, and pencil, pen, and paper to
create fi gures that would accompany the trip reports he produced for
his California and New York research trips, fi rst in May and later in the
summer of 1982.129 The fi rst representations of the Los Angeles cluster
clearly demonstrate this provenance (see fi g. 2.5), as do the subsequent
images that would result in the fi nal cluster diagram in the 1984 article.
Other efforts to organize the additional data Darrow gathered in New
York involved a graph of early known cases (see fi g. 2.6), which drew on
earlier techniques used to chart syphilis cases, and an early pencil sketch
arranging the contacts (see fi g. 2.7), which is a prototypical version of a
color- coded cluster diagram (see fi g. 2.8).
Patient Zero and the Making of the AIDS Epidemic Page 21