One of the strongest arguments against the germ “theory” of disease is the repeated failure to experimentally demonstrate human-to-human transmission. If contagion is a real biological phenomenon, it should be demonstrable under controlled conditions—especially through natural exposure routes. Yet across more than a century of experimentation, this has consistently failed.
Perhaps the most infamous example is the 1918 “Spanish Flu” experiments. Researchers attempted numerous methods to transmit disease from sick patients to healthy volunteers, including close face-to-face contact, exposure to respiratory secretions, and even direct inoculation with bodily fluids. Despite these efforts, not a single volunteer developed influenza. This was remarkable given that the 1918 flu is routinely described as one of the deadliest “viral” outbreaks in history. Even more striking, these failures occurred across multiple experiments conducted on different coasts, eliminating the possibility of a localized anomaly.
These were not isolated incidents. Numerous other failed transmission attempts have been documented throughout the literature, many of which were compiled by Daniel Roytas in his book Can You Catch A Cold?. I have also documented multiple examples in my Infectious Myth Busted series, including failed attempts to demonstrate transmission of HIV, measles, polio, and the common cold.
This absence of experimental evidence has not gone unnoticed by mainstream researchers themselves.
In a 2003 review, Bridges and colleagues openly acknowledged:
“Our review found no human experimental studies published in the English-language literature delineating person-to-person transmission of influenza.”
This was not a fringe critique, nor a historical oversight. It was a review of the available literature concluding that, even after decades of research, no controlled human experiments had ever demonstrated influenza transmission from one person to another.
A decade later, the situation remained unchanged. In 2013, Kingsley and colleagues reaffirmed the same limitation, writing:
“Creation of symptomatic illness in individuals using a GMP challenge virus is well proven, but a successful and robust model of human-to-human transmission with a GMP challenge virus has yet to be established.”
In other words, researchers had learned how to induce symptoms through artificial inoculation, but they still could not demonstrate transmission between humans. “Infection” and transmission were being treated as interchangeable concepts, despite the absence of experimental evidence connecting the two.
Beyond transmission, researchers have repeatedly failed to recreate disease itself—even through deliberate exposure. Throughout medical history, investigators exposed themselves to microbes widely believed to be deadly, often through ingestion, without developing disease. I compiled many such cases in The Germ Duel. Some of the most damning examples come from Drs. Thomas Powell and John Fraser, whose experiments I documented in detail in The Untold Stories of Drs. Powell & Fraser.
Modern human challenge trials were supposed to resolve these uncertainties. Instead, they exposed the same unresolved problem. In these trials, healthy volunteers were deliberately exposed via artificial inoculation to manufactured “viral” samples intended to cause disease. Yet the results have been largely underwhelming. In a 2020 influenza challenge study, only one secondary “infection” was confirmed, and even that relied on “antibody” testing after PCR failed to detect it. A 2022 “SARS-COV-2” challenge trial fared no better: roughly half of participants were labeled “infected,” based solely on PCR detection of genetic material in the very areas where the material had been introduced.
None of these trials demonstrated human-to-human transmission. More telling still, in the case of “SARS-COV-2,” researchers openly acknowledged that attempts to actually cause “Covid” failed.
This was reported in the May 1, 2024 Nature article Scientists tried to give people COVID — and failed. Participants given a dose of “ancestral SARS-COV-2” did not develop sustained “infection,” prompting researchers to repeatedly increase the dose—eventually reaching levels 10,000 times higher than the original amount. Even then, only short-lived “infections” were claimed, and these quickly resolved. Susan Jackson, a study clinician at Oxford and co-author of the study, admitted, “We were quite surprised. Moving forward, if you want a COVID challenge study, you're going to have to find a dose that infects people.” Tom Peacock, a virologist at Imperial College London added, “If you can’t get people infected, then you can’t test those things.”
While these failures are damning, it is not surprising that some challenge studies manage to produce respiratory symptoms. Introducing cell-cultured material directly into the nasal cavity and holding it there with nose clips is an unnatural exposure capable of irritating tissue. At best, such experiments show that foreign substances can provoke symptoms. They do not establish causation, nor do they demonstrate transmissibility under natural conditions.
And when natural conditions are actually observed, transmission often fails to occur.
In a 2021 observational study, three individuals who accompanied pediatric patients in a “Covid” ward remained negative throughout their stay, despite close proximity and shared facilities. Similarly, an April 2020 paper from Singapore reported that 18 patients and multiple healthcare workers were exposed to a confirmed “Covid” case in shared wards and toilets. Despite presumed exposure via air and surfaces, none developed the disease during the 14-day follow-up period.
These examples highlight a fundamental problem. Contagion, as commonly claimed, does not reliably occur in real-world settings—even in environments where it should be most apparent, such as hospital wards filled with allegedly “infectious” patients. The assumption of transmission persists not because it has been demonstrated experimentally, but because it has been repeated often enough to be treated as fact.
This is why studies that attempt to observe natural transmission are the most persuasive, and why their failures are the most damning. Artificial inoculation can always be blamed on dose, technique, or timing. But when sick and healthy individuals are placed together under conditions where transmission is expected to occur, and it does not, the hypothesis itself must be questioned. With that in mind, we can now examine the latest influenza transmission study, an experiment that once again attempted to demonstrate spread under controlled, natural conditions, and once again failed to do so.
A Tale of Two Milton's
The most infamous failed human influenza transmission study—and the latest modern attempt to explain influenza spread—share an unexpected connection: the name Milton.
Milton J. Rosenau
The original Spanish Flu experiments were led by Milton J. Rosenau, long regarded as the foremost authority on preventive medicine, hygiene, and public health. His textbook Preventive Medicine and Hygiene (1913) quickly became the most influential work in the field and was said to have advanced public health worldwide more than any other single factor.
During the Spanish Flu, Rosenau sought to determine the cause of the disease. At the time, only a handful of “viruses” had been conceptualized—yellow fever, rabies, and polio—and because Pfeiffer’s bacillus was routinely cultured from influenza victims, it was widely assumed to be the culprit. Still, Rosenau remained open to the possibility of a “viral” cause and even organized a symposium in 1916 to explore that hypothesis.
In an effort to “determine the mode of spread of influenza,” Rosenau first attempted to “infect” military volunteers using pure cultures of Pfeiffer’s bacillus. When this failed, he escalated the experiment, administering “a very large quantity of a mixture of thirteen different strains” of the bacterium, including strains freshly obtained from the lungs at autopsy. When this too produced no illness, volunteers were inoculated with material taken from the throats and noses of influenza patients, and later with patients’ blood.
Still nothing.
As a final measure, healthy volunteers were instructed to shake hands with, converse with, and be coughed upon by actively ill influenza patients. Yet in every instance, the men remained healthy. Despite his best efforts, Milton Rosenau failed to demonstrate both “infection” and human-to-human transmission.
Donald K. Milton
More than a century later, Donald K. Milton, UMD MPower Professor of Environmental and Occupational Health, would encounter the same problem.
After spending most of his 40-year career studying airborne “pathogens,” Milton played a pivotal role during the “Covid” era in overturning longstanding opposition among national and global health authorities regarding airborne transmission. A controversial proponent of the idea that aerosols—fine particles capable of remaining suspended in air—could carry enough exhaled “virus” to cause “infection,” Milton was previously dismissed by critics as a “quack” and a “charlatan.”
In response, he developed a device in 2013 called Gesundheit II, designed to capture exhaled breath and “extract viruses” from volunteers. In March 2020, Milton and colleagues published the brief communication Respiratory virus shedding in exhaled breath and efficacy of face masksin Nature Medicine using this device to claim that surgical masks could block the release of “coronaviruses” and influenza from “infected” individuals. Despite decades of research failing to establish the effectiveness of masks experimentally, this study is widely credited with transforming perceptions of mask “effectiveness” and cementing masks as a preventive intervention.
In 2021, Milton received approximately $15 million in NIH funding as part of a broader, long-standing research program focused on respiratory aerosols and influenza transmission dynamics. This funding supported efforts to further investigate how influenza—already presumed to be caused by a “viral” agent—might spread between humans under controlled conditions.
As co-principal investigator Jelena Srebric admitted:
“Of course, you can’t actually see individual viruses causing someone to be infected, but the design of this study will give more understanding of every step along the way. We’ll know what made you sick and what didn’t make you sick.”
Yet, just as with Rosenau’s experiments more than a century earlier, human-to-human transmission once again proved elusive.
Different era. Different Milton. Same unanswered question.
Failure to Transmit
On January 7, 2026, Milton's human challenge study was published in PLOS Pathogens detailing his failure to demonstrate human-to-human transmission of disease. As reported by Science Daily in the article A room full of flu patients and no one got sick, the study was described as “a striking real-world experiment” in which “flu patients spent days indoors with healthy volunteers, but the virus never spread.” Notably, this is the first controlled clinical trial specifically designed to examine airborne influenza transmission using individuals who were “naturally infected,” rather than intentionally inoculated in a laboratory. In other words, there were no artificial injections of lab-created cell cultured soups into volunteers.
The aim of the study was to better understand how influenza moves from person to person, and the research was conducted throughout 2023 and 2024 on a quarantined floor of a Baltimore-area hotel. Five participants with “confirmed” influenza symptoms and eleven healthy volunteers lived together on the isolated floor for two weeks, following daily routines designed to mimic ordinary social interaction. These included casual conversation, shared physical activities such as yoga, stretching, and dancing, and the handling of shared objects including pens, tablet computers, and a microphone, which were passed among participants.
The failure to transmit influenza could not be attributed to vaccination status. Only two of the healthy volunteers had received a flu vaccine, and “antibody” levels to the detected strains were reported to be low across participants. By the researchers’ own criteria, the volunteers were therefore in an optimal position to become “infected.”
Following the null result, Milton and his co-authors offered several post hoc explanations for why no transmission occurred. These included limited coughing among the symptomatic participants, effective indoor air circulation, and the middle-aged status of the healthy volunteers. None of these factors, however, were exclusion criteria prior to the experiment, nor were they presented as conditions under which transmission should be considered unlikely. Examining the study clarifies exactly why these explanations are merely hypothesis-saving rescue devices rather than legitimate factors in stopping the spread of the “virus.”
Enjoy the Stay
According to the introduction of the study, Milton and colleagues expected an average of 1.3 secondary cases from an “infected” person based on an estimated reproductive number. This is supported by the Cleveland Clinic who state that, “for every person infected, they spread the flu to one to two more people.” Therefore, under the study’s own assumptions, placing 5 sick individuals in close contact with 11 healthy individuals should have resulted in multiple “infections.” It would have been expected that approximately 6–7 of the 11 healthy volunteers would become “infected.” Not zero.
However, what is important to note is that an effective reproductive number of ~1.3 for seasonal influenza does not describe robust person-to-person transmission. In fact, it quietly admits the opposite. An average of 1.3 secondary cases means that most “infected” individuals are expected to “infect” no one at all, with the mean sustained only by a small minority of cases under favorable conditions.
For example, in a hypothetical group of 10 “infected” people, 6–7 may “infect” no one, while 2–3 “infect” one or more people, producing roughly 13 new cases total—giving the population-level average of 1.3. Because R is an inferred population statistic rather than a directly observed effect and explicitly depends on environmental conditions and contact rates, failed transmission can always be attributed to context rather than the agent itself, creating an unfalsifiable scenario. Such fragile, slow growth collapses with minor changes in assumptions and requires statistical smoothing, inferred transmission chains, and population-level modeling rather than direct observation. This aligns with the experimental and household literature, where consistent human-to-human transmission cannot be reliably demonstrated and many index cases produce zero secondary “infections.”
Reading further, it becomes clear that the study repeatedly admits uncertainty at the very core of influenza transmission. While Milton and colleagues assert that “humans shed infectious virus into fine particle aerosols,” this claim relies primarily on indirect evidence—post-hoc household analyses and model-based inferences—rather than direct observation. They concede that “the relative importance of these transmission modes remains poorly understood” and that “the lack of in-depth understanding…impedes development” of effective control strategies. Their prior experiment, EMIT-1, is openly described as a failure: it did not meet the expected secondary attack rate, producing only a single asymptomatic, PCR-negative “infection” (oddly contradicting the actual study’s report that the participant was symptomatic). Rather than questioning the assumption that influenza spreads efficiently between humans, Milton attributed this failure to limitations of the model, specifically the use of an egg-adapted challenge “virus,” and proposed a redesigned study using community-acquired cases.
From the outset, EMIT-2 is framed defensively, emphasizing expected “challenges, limitations, and insights gained,” language more indicative of a salvage operation than a genuine test of hypothesis. Stripped of its public-health framing, the introduction reads as an admission that, despite decades of study, human-to-human influenza transmission remains experimentally unproven. In other words, the study is structured not to test the assumption, but to preserve it.
Evaluating modes of influenza transmission (EMIT-2): Insights from lack of transmission in a controlled transmission trial with naturally infected donors
In a typical year seasonal influenza is associated with hundreds of thousands of hospitalizations and tens of thousands of deaths in the US [1]. Although the 2009 pandemic influenza virus had a broadly similar severity profile to seasonal influenza [2–4], the three previous pandemics in the 20th century were associated with much greater morbidity and mortality [5]. Influenza is therefore responsible for a considerable burden of disease in the US and globally, and control of influenza is a public health priority. The effective reproductive number of seasonal influenza is estimated to be around 1.3 at the start of an influenza season, meaning that we expect an average of 1.3 secondary cases from an infected person, depending on the environment and number of people whom they come into contact with [6].
It is now well established that humans shed infectious virus into fine particle aerosols both when coughing and also when simply breathing [7]. A post-hoc analysis of household studies suggests that viral aerosols play an important role in transmission, with half of influenza transmissions in households occurring via inhalation [8]. In April 2024, the World Health Organization released a technical consultation report and defined the modes of respiratory pathogen transmission as airborne transmission/inhalation, direct deposition, and contact (direct and indirect) corresponding to a previously proposed transfer-process categorization: inhalation, spray, and touch transmission, respectively [9,10]. Despite recent emphasis on airborne transmission [9,11], the relative importance of these transmission modes remains poorly understood and an important focus of continuing research interest [12,13]. The lack of in-depth understanding of transmission modes and mechanisms impedes development, implementation, and prioritization of effective control strategies.
In 2013, a randomized controlled trial (RCT) of human challenge-transmission, Evaluating Modes of Influenza Transmission (EMIT-1), was conducted to evaluate the relative importance of these transmission modes [14,15]. EMIT-1, using volunteers intranasally inoculated with 105 50% tissue culture infectious doses (TCID50) influenza virus as Donors, failed to meet the expected secondary attack rate (SAR); only one serology-confirmed asymptomatic, PCR-negative infection was identified in a Control Recipient [15]. Post-hoc analyses revealed that Donors intranasally inoculated with influenza virus had mild symptoms and minimal shedding of virus in exhaled breath compared with symptomatic community-acquired cases on a college campus collected in the same year (2013) [14] as well as mildly symptomatic influenza cases from a longitudinal cohort [16]. These findings suggested that if lack of transmission was due to use of an egg-adapted challenge virus, a controlled human influenza virus transmission trial (CHIVITT) with community-acquired cases as Donors could be an improved and more valid approach to studying the dynamics of influenza transmission.
Here, we describe EMIT-2, a first attempt at a CHIVITT incorporating community-acquired cases, and the challenges, limitations, and the insights gained regarding the dynamics of influenza transmission.
Moving on to the Results section, the study cohort and population reveals that, out of hundreds of potential volunteers, only six “naturally infected” Donors were admitted, and the first two quarantine cohorts failed entirely—one because the influenza season ended before enrollment, the other because the supposed Donor was “infected” with a seasonal “coronavirus” rather than influenza. The remaining cohorts relied on just five Donors and 11 Recipients, with individual cohorts including as few as one intervention Recipient, two controls, and four Donors. Even timing requirements were inconsistently applied, with at least one Donor enrolled despite symptom onset more than 48 hours earlier. Far from being an incidental challenge, these practical realities highlight the core problem: in controlled experiments, influenza simply does not transmit reliably between humans. The study’s design and execution thus serve as a modern echo of Milton Rosenau’s century-old experiments—careful planning, motivated participants, and controlled conditions could not produce the phenomenon the study set out to demonstrate.
Results
A schematic of the study design is shown in Fig 1a and the timeline of two cohorts is shown in Fig 1b.
Fig 1. Study design and timeline. (a) Schematic of the study workflow and exposure event room. (b) Timelines for cohorts 24b (started January 24, 2024) and 24c (started February 14, 2024). Shaded blocks indicate phases (Run-in, Donor exposure, Observation, Discharge, Follow-up). Arrows mark donor introductions (D24b-1 on 01/30; D24c-1 on 02/16, D24c-2 on 02/17, D24c-3&D24c-4 on 02/21).
Study cohort and study population
By January 3, 2024, we recruited 80 potential Recipients into a study registry (S1 Fig). A total of 505 potential Donors expressed interest, with six ultimately admitted to the quarantine facility (S2 Fig). Between 2023 and 2024, we ran four quarantine cohorts: one in 2023 (23a) and three in 2024 (24a, 24b, 24c). Recipients were recruited to participate in quarantine cohorts without further selection based on antibody levels or vaccination status, due to the small number available. Recipients were assigned to Intervention Recipient (IR) and Control Recipient (CR) groups using the face shield and hand hygiene intervention previously described for the EMIT-1 study that used nasally inoculated Donors [15].
Despite successful enrollment of Recipients in cohorts 23a and 24a (S1 Table), we were unable to recruit Donors with natural influenza virus infection. For Cohort 23a (February 20 to March 5, 2023), the start of which was delayed due to the coronavirus disease 2019 (COVID-19) pandemic, we could not recruit Donors because the influenza season was unusually early and had ended before the quarantine commenced. For Cohort 24a (January 3–16, 2024), one Donor recruited based on a positive rapid influenza test reported by an outside facility was later confirmed to be infected with a seasonal coronavirus (229E), not influenza virus (S2 Table).
The later two cohorts 24b and 24c included 11 Recipients and 5 naturally infected Donors. We enrolled three IRs, five CRs and one Donor in cohort 24b (Jan 24 to Feb 6, 2024). We enrolled one IR, two CRs, and four Donors in 24c (Feb 14–27, 2024). All Donors reported symptom onset within the last 48 hours. However, a rapid flu test result submitted after enrollment by one of the Donors (D24c-3), on closer examination, demonstrated that they had a positive test more than 48 hours prior to enrollment. All tested positive for influenza A virus by a molecular test (Cepheid GeneXpert Xpress) at recruiting sites on the University of Maryland College Park campus or on arrival at the hotel quarantine. The Donor for 24b (D24b-1) and two of the Donors for 24c (D24c-1 and D24c-2) were infected with H3N2, while the other two Donors (D24c-3 and D24c-4) were infected with H1N1.
For cohorts 24b and 24c, 54% of the Recipients were female and the average age was 36 (range 22–49). Two Recipients (both in Cohort 24b) received an influenza vaccine earlier in the same influenza season before enrolling in a quarantine (Table 1). Among the five Donors, most were female (80%) young adults, with a mean age of 21 years (Table 2).
Table 1. Recipients demographics for cohorts 24b & 24c.
Table 2. Donor demographics for cohorts 24b & 24c.
As detailed earlier, volunteers participated in conversation games, icebreakers, light physical activity, and object-passing exercises—activities intended to maximize contact over 23 events totaling 82 hours. Environmental conditions were carefully controlled, with temperature, humidity, and ventilation set to create a “well-mixed” room. Despite hours of close interaction under engineered conditions, no actual transmission was observed, reinforcing the central point: even under carefully monitored, high-contact circumstances, influenza simply does not spread between humans. It carries on a century-long pattern of failure by echoing Milton Rosenau’s early experiments in which exposure and interaction could not produce the disease.
Exposure events and activities
We conducted 23 exposure events involving influenza Donors, including 6 events in cohort 24b and 17 events in cohort 24c. These events lasted between 111 and 250 minutes, totaling 82.2 cumulative influenza exposure-event hours (19.5 in 24b and 62.7 in 24c). Eleven events involved one Donor and 12 involved two (S3 Table). During the events, volunteers engaged in conversation-focused activities (such as icebreaker questions), interactive games (such as the card game UNO and drawing and guessing game Telestrations), plus some physically active periods (stretching, yoga, or dancing). A designated object was passed following a donor-centered alternating sequence, including a marker in 19, tablet computer in 2, and microphone in 2 events (see S3 Table).
Environmental conditions
Mechanical systems successfully maintained stable environmental conditions in the exposure event room during exposure events through running two fan coil units and two large capacity dehumidifiers with fans constantly on high speed. The indoor air temperature remained within 22-25oC, and relative humidity ranged between 20–45% (Fig 2). The ventilation rate was approximately 0.25-0.5 air changes per hour. Running fan coil units and dehumidifiers at high speed in addition to portable air cleaners without filters served to create a well-mixed environment satisfying assumptions of the standard Wells-Riley model [17]. However, the extensive mixing reduced the extent of Recipient exposure to concentrated plumes of exhaled breath (a detailed computational fluid dynamic analysis will be published separately).
Fig 2. Variations in indoor temperature, relative humidity, and CO2 concentration in the exposure event room and hallway during cohorts. (a) Cohort 24b (Jan 30–Feb 1, 2024); (b) Cohort 24c, first donor block (Feb 16–19, 2024); (c) Cohort 24c, second donor block (Feb 21–23, 2024). The left y-axis shows temperature (°C) and relative humidity (%); the right y-axis shows CO2 concentration (ppm). Line plots show temperature (orange, solid), relative humidity (dark blue, solid), exposure event room CO2 (green, dashed), hallway CO2 (purple, dot-dashed). Yellow bands denote each exposure event windows; vertical ticks mark exposure start (red) and end (black). Temperature and humidity were maintained within stable ranges by the mechanical systems, as these parameters can influence virus stability. CO2 concentration is a proxy for ventilation and a direct indicator of rebreathed air; levels increased during exposure events due to occupancy and slowly returned to background afterward, reflecting the low ventilation condition targeted in these experiments.
CO2 concentrations in the exposure event room increased rapidly during exposure events, as occupants were the sole CO2 source, and returned toward the background level (~420 ppm) after the exposure event. In 24b, with 11 occupants (one Donor, eight Recipients, and two study moderators), CO2 levels rose from 460 to 5300 ppm, while in 24c, with 7 occupants (two Donors, three Recipients, and two study moderators), CO2 levels were lower, rising from 430 to 2400 ppm. Fluctuations in CO2 correlated with variations in temperature and relative humidity. CO2 levels in the corridor remained relatively stable near background concentrations, with minor fluctuations caused by occasional human traffic and the opening of the exposure event room doors.
The “viral” detection data underscore just how unreliable and variable supposed human-to-human influenza transmission is, even among individuals selected to be “infectious.” Across all measures—exhaled breath, nasal swabs, saliva, air sampling, and surface swabs—”viable virus” was detected only in isolated instances. Keep in mind that all results rely on tests detecting “viral” RNA that has never been conclusively shown to belong to a distinct, “infectious virus.” Only one exhaled breath sample, from a Donor on the last day of quarantine while coughing repeatedly, contained “culturable virus.” Mid-turbinate swabs were inconsistently positive, saliva yielded no “culturable virus,” and there was no correlation between nasal, saliva, or breath “viral load.” Ambient air where Recipients were actually present was nearly always negative, and among 23 surface swabs, only one marker pen contained detectable “virus.” These findings reveal extreme intermittency and unpredictability in “viral shedding:” even carefully selected donors, theoretically at peak “infectiousness,” fail to produce consistent evidence of transmissible “virus.” Taken together—and considered alongside the unproven assumption that detected RNA even represents a “virus”—these results reinforce the central reality: despite repeated, carefully controlled exposure events, human-to-human influenza transmission remains elusive, highly variable, and unobservable.
Viral shedding and environmental contamination by donors
Exhaled breath aerosol samples.
Using the G-II, we collected 16 exhaled breath aerosol samples from five influenza Donors. During each 30-minute collection period, Donors coughed and sneezed spontaneously, with cough counts ranging from 0 to 16 (median: 0) and sneeze counts from 0 to 1 (median: 0) (S4 Table).
Among these exhaled breath aerosol samples, we detected viral RNA in seven (44%) fine (<5µm, range: 79 – 8.9 × 103 copies per 30-minute sample) and one (6%) coarse (≥5µm, 53 copies per 30-minute sample) aerosol sample; all from Donors with H3N2 infections. The viral RNA load in breath aerosol samples fluctuated over time and did not follow a clear pattern across Donors (Figs 3 and 4).
Fig 3. Viral RNA and infectious viral load in Donor samples. Violin plots depict the distribution of viral RNA and infectious (fluorescent focus unit, FFU) viral loads (log10 RNA copies and FFU) across four different sample types: breath aerosol <5 µm, breath aerosol ≥5 µm, mid-turbinate swabs (MTS), and saliva. RNA copy and FFU values are expressed as follows: per 30-minute sample for breath aerosols, per swab for MTS, and per mL for saliva. Negative samples were assigned a value of 1.
Fig 4. Donor sample viral RNA, infectious viral load, total symptom score and cough trajectories with timing of positive environmental samples. Each individual plot shows one donor over their days in quarantine. Lines/points show RNA (solid) and FFU (dashed) for: breath aerosol <5 µm (solid circle, orange), breath aerosol ≥5 µm (solid triangle, sky blue), mid-turbinate swab (MTS; solid square, green), and saliva (solid diamond, purple). Cough counts during 30-min breath sampling (dot-dash line, black, open circle) and self-reported total symptom score (dotted line, grey, open square) are overlaid and referenced to the right axis. Shaded bands mark environmental positives centered on the day that positives samples were detected: Ambient+ (light blue) and Ambient + & Surface+ (yellow). Donors’ samples were collected once per day, prior to any exposure events (on the afternoon or evening of admission and otherwise in the morning), ambient samples were collected during afternoon events and surface swabs at the end of each exposure event. RNA copy and FFU values are expressed as follows: per 30-minute sample for breath aerosols, per swab for MTS, and per mL for saliva. Negative samples were assigned a value of 1.
The only culture-positive breath fine aerosol sample contained 147 focus forming units (FFU) per 30-minute sample (Figs 3 and 4). The sample came from the Donor in 24b (D24b-1; H3N2 infection) on the morning of Day 9 (February 1, 2024) their last day in quarantine, when they coughed the most during the 30-minute exhaled breath sampling(six times) and shed their highest breath aerosol viral RNA load (405 copies per 30-minute sample). Eight Recipients were exposed for four hours on this day.
Mid-turbinate swabs and saliva.
We collected 18 mid-turbinate swabs (MTS) samples and 18 saliva samples from Donors. All MTS samples contained detectable viral RNA (range: 5.32 × 104 – 3.91 × 108 genome copies/swab), and 11 (61%) were culture positive with viral loads from 40 to 8.7 × 104 FFU/swab. Among saliva samples, 15 (83%) contained detectable viral RNA (range: 67 – 4.37 × 107 copies/mL); however, none were culture positive (Figs 3 and 4). We did not observe significant correlations between exhaled breath aerosol viral RNA and that measured in MTS or saliva samples in this small data set (S3 Fig).
Ambient and personal bioaerosol sampling.
We collected two ambient air samples and two personal breathing zone air samples during each of 12 afternoon exposure events (one per day) when Donors were present, resulting in a total of 24 ambient and 24 personal samples.
Among the ambient samples, we detected viral RNA (75 genome copies/m3 air) by dPCR in one sample from cohort 24b, at sitting height (129 cm) in the 1–4 µm size fraction, on January 31, 2024 – the afternoon prior to the culture positive exhaled breath sample from Donor D24b-1. One other ambient sample, collected from cohort 24c (Donors D24c-1 & D24c-2 present; February 19, 2024) at standing breathing level height (150 cm), contained detectable viral RNA (25 RNA copies/m3 air) in the > 4 µm size fraction (S4 Fig). None of the personal bioaerosol samples contained detectable viral RNA by dPCR.
Surface swab viral load.
We collected 23 surface swabs from fomites used in exposure events with a Donor, including two from a microphone, two from a tablet computer, and 19 from a marker pen, depending on the activities performed (S3 Table and S4 Fig). Among these, one sample, collected from a marker pen during the third exposure event of the Donor’s second day in quarantine (24b), contained virus detected by dPCR; it was also positive for culture (total viral load of 454 RNA copies and 20 FFU).
The symptom and serology data make the study’s outcome unambiguous: by their own measures, human-to-human influenza transmission did not occur. Recipients reported only sporadic, mild symptoms, often coinciding with exposure event days and not persisting afterward, suggesting these were unrelated to influenza. Post-exposure nasal and saliva samples from all Recipients were uniformly negative by PCR and culture, despite “confirmed infections” in Donors. Serological analyses reinforced this absence of transmission as none of the Recipients exhibited a four-fold rise in HAI titers, and ELISA results confirmed no “antibody” response to the Donor strains. Many Recipients were serologically susceptible at baseline, and some even had higher baseline ELISA values than Donors, yet exposure produced no detectable “infection.” Even one Recipient who received a flu vaccine after quarantine showed no response attributable to exposure itself. Taken together, these findings confirm that, under repeated, carefully monitored exposure to what are claimed to be clearly “infected” Donors, influenza failed to transmit, and the mild, inconsistent symptoms reported cannot be attributed to “viral infection.”
Symptoms reported by Donors and Recipients
Donors reported varying levels of symptom severity across a range of individual symptoms (Fig 5a). Throughout the quarantine period, total symptom scores fluctuated for the Donor in 24b (D24b-1), but generally declined for Donors in cohort 24c, though the rate of decline differed among Donors (Fig 5b).
Fig 5. Symptom severity and progression in Donors during quarantine. (a) Heatmap shows the maximum individual symptom scores reported by individual Donors. The color intensity represents the symptom severity, with scores ranging from 0 to 3. Symptoms are listed along the x-axis, and Donors are on the y-axis. Columns are grouped by domains: Upper respiratory (running nose, stuffy nose, sneeze, sore throat, earache), Lower respiratory (chest tightness, shortness of breath, cough), and Systemic (malaise, headache, muscle & joint ache, chills, feverish). (b) Line plots present total symptom scores and symptom group scores (upper respiratory, lower respiratory, and systematic symptoms) over time for each individual Donor. The x-axis represents cohort timelines, and the y-axis indicates the sum of each symptom group scores.
Recipients sporadically reported individual symptoms, with overall symptom severity lower than that of Donors. Malaise was the most frequently reported symptom (Fig 6a). Total symptom scores fluctuated over time, with most Recipients reporting more symptoms after exposure to a Donor (Fig 6b). Three Recipients reported symptoms only on days with exposure events but not afterward. R24b-Ctl4 in cohort 24b recorded the highest total symptom score (10) three days after exposure to the Donor (Fig 6b).
Fig 6. Symptom severity and progression in Recipients during quarantine. (a) Heatmap shows the maximum individual symptom scores reported by individual Recipients. The color intensity represents the symptom severity, with scores ranging from 0 to 3. Symptoms are listed along the x-axis, and Recipients are on the y-axis. (b, c) Line plots present total symptom scores over time for Recipients. (b) shows data for the Recipients in cohort 24b, while (c) shows data for those in cohort 24c. The x-axis represents the timeline, with the yellow-shaded areas indicating periods of Donor quarantine. Each subplot corresponds to an individual Recipient, with blue lines (circle) representing the control group and green lines (triangle) representing the intervention group.
Recipient post-exposure PCR and culture
We collected a total of 89 post-exposure MTS and saliva samples from Recipients. In cohort 24b, each Recipient provided daily MTS and saliva samples on 7 days (quarantine cohort Days 8 through 14) following their first exposure event on Day 7. In cohort 24c, each provided 11 samples (Days 4 – 14) following their first exposure event on Day 3. None of the samples tested positive by dPCR.
Serology from Donors and Recipients
Consistent with qRT-PCR test results, hemagglutination inhibition (HAI) titers for Donors measured at admission and follow-up confirmed three H3N2 (D24b-1 from cohort 24b; D24c-1 and D24c-2 from cohort 24c) and two H1N1 (D24c-3 and D24c-4 from cohort 24c) infections (Figs 7 and 8).
Fig 7. HAI titers over time for Donors and Recipients in Cohort 24b. HAI titers are shown for Donors (solid lines, star symbols), Vaccinated Recipients (dot-dashed lines, filled triangles), and Unvaccinated Recipients (dot-dashed lines, open circles) at four time points: T0 (Recipient screening), T1 (Admission), T2 (Discharge), and T3 (Follow-up). Virus targets are color-coded: A/Victoria/4897/22 (H1N1, blue), A/Darwin/6/21 (H3N2, green), and B/Austria/1359417/21 (red). Each facet represents an individual study participant.
Fig 8. HAI titers over time for Donors and Recipients in Cohort 24c. HAI titers are shown for Donors (solid lines, star symbols) and Unvaccinated Recipients (dot-dashed lines, open circles) at four time points: T0 (Recipient screening), T1 (Admission), T2 (Discharge), and T3 (Follow-up). Virus targets are color-coded: A/Victoria/4897/22 (H1N1, blue), A/Darwin/6/21 (H3N2, green), and B/Austria/1359417/21 (red). Each facet represents an individual study participant.
Among the exposed Recipients, 8 out of 11 had low HAI titers (≤10) at admission against the vaccine strains corresponding to the Donor strains, including 6 out of 8 in cohort 24b, 2 out of 3 in cohort 24c (Figs 7 and 8), and one of the two vaccinated for the current season. However, none of the Recipients showed a four-fold increase in HAI titers against the vaccine strains from admission to follow-up, except for one Recipient in cohort 24b (R24b-Ctl3), who responded to all vaccine strains and reported having received an influenza vaccine two days after discharge from quarantine.
The enzyme-linked immunosorbent assay (ELISA) results were generally consistent with the HAI titer findings, as none of the Recipients showed a four-fold increase in area under the curve (AUC) values against vaccine strains HAs corresponding to the Donor infections, including the Recipient who received an influenza vaccine after quarantine (S5 and S6 Figs). However, ELISA results indicated higher AUC values in Recipients than in Donors at baseline. In cohort 24b, all eight Recipients had higher AUC values at admission compared to the Donor. Similarly, in cohort 24c, one of three Recipients had a much higher AUC value than their Donor counterparts at admission.
Evident from the explanations provided by Milton and colleagues in the Discussion, the absence of transmission in EMIT-2 is not explained by direct empirical findings but by a sequence of post hoc rationalizations introduced only after transmission failed to occur. Rather than testing predefined transmission conditions against falsifiable criteria, the study retrofits explanations—low donor “source strength,” recipient “immunity,” aerosol heterogeneity, “infectious” dose thresholds, plume dispersion, and air mixing—to preserve the assumption that influenza must be transmissible by aerosol, even when this is not observed.
Critically, the experimental conditions themselves were deliberately established a priori to favor transmission. Low ventilation (~0.25–0.5 ACH), prolonged close contact, crowding, stable indoor temperatures (22–25 °C), and relative humidity (20–45%) were explicitly justified as conducive to influenza survival and spread. Yet when transmission did not occur under these deliberately permissive conditions, the same parameters were retrospectively reframed as inhibitory—particularly air mixing and plume dilution—despite being intrinsic features of the study design. Conditions selected to enable transmission were thus reinterpreted as reasons it failed, revealing a methodological asymmetry in which negative outcomes are never allowed to challenge the underlying hypothesis.
Each proposed explanation depends on unmeasured or inferred variables (e.g., “infectious” quanta, short-range plumes, “supershedding” distributions) whose values are defined retrospectively to fit the null result. When “viral” RNA is present but transmission does not occur, the dose is declared insufficient; when dose appears high, “immunity” is invoked; when “immunity” is low, airflow becomes the limiting factor. None of these factors were independently manipulated or prospectively specified as failure conditions, rendering them non-predictive and non-falsifying.
Of importance to note, the same environmental parameters (low ventilation, crowding, prolonged exposure) are routinely cited in the literature as supportive of influenza transmission, yet here they are reframed as inhibitory only after transmission fails. This reversal underscores that the explanations function as theoretical escape hatches—mechanisms invoked to prevent the null result from challenging the underlying contagion model.
After two controlled EMIT studies, it is clear from the explanations provided by Milton and colleagues that transmission remains hypothetical, while their explanatory framework grows increasingly complex and conditional. The accumulating need for rare behaviors (coughing at the right moment), exceptional hosts (“supershedders”), narrowly defined exposure geometries, and precise airflow conditions presents a model that is insulated from falsification. In this context, the study’s conclusions illustrate how a failed prediction is repeatedly rescued rather than reconsidered.
Discussion
We developed a novel framework for investigating human influenza virus transmission using Donors with community-acquired infections under controlled conditions. However, despite enrolling five naturally infected Donors within 48 hours of reported symptom onset, promoting prolonged close contact under low ventilation conditions with 11 Recipients, many with low HAI titers, no secondary infections were observed. Three possible explanations related to Donor source strength, Recipient susceptibility, and rapid air mixing may provide insight into influenza transmission dynamics and guide future study designs.
Donors in our present study, EMIT-2, shed detectable viral RNA somewhat more frequently, but not significantly more frequently than had been observed in EMIT-1 Donors infected with an egg-adapted laboratory virus: 44% (7/16) versus 22% (14/64) of fine and 6% (1/16) versus 9% (6/64) of coarse aerosol samples in EMIT-2 and EMIT-1, respectively [16]. EMIT-2 Donor aerosol shedding frequency was on the low end of previous observations of natural influenza infections, which ranged from 38% to 86% for fine aerosols and 29% to 86% for coarse aerosols [7,16,18]. The rate of viral RNA shedding into fine aerosols (≤5 µm) by EMIT-2 Donors was also similar to rate from the small fraction of experimentally infected Donors in EMIT-1 who had detectable aerosol shedding (range 1.6 × 102 to 1.8 × 104 copies per hour in EMIT-2 v. 2.6 × 102 to 1.6 × 105 copies per hour in EMIT-1) [15,19].
Looking specifically at symptomatic ILI cases on our campus infected with H3N2 from the 2012–13 and 2018–19 seasons, we previously noted that cases recruited during milder influenza seasons may, on average, shed lower quantities of viral RNA into exhaled breath [16]. We also noted that cases recruited by intensive case finding had milder symptoms and shed less frequently into aerosols. The results here are consistent with those earlier observations in that 2023–24 was a relatively mild season, most Donors were recruited by offering free tests at a kiosk on campus, and Donor D24b-1, with the strongest evidence for potential infectiousness, was the only donor recruited from the University Health Center.
Based on evidence for transmission to one Recipient in EMIT-1, Bueno de Mesquita et al. estimated a lower bound for an infectious quantum (i.e., the ID 63% via inhalation [20]) of approximately 1.4 × 105 RNA copies [19]. Recent aerosol inoculation experiments suggest that, with an egg-adapted laboratory strain, the infectious dose in approximately 104 TCID50. During EMIT-2, using an average minute-ventilation rate of 16 L/min based on Recipient activities and the two positive NIOSH bioaerosol samples, we estimate that on January 31, Recipients in Cohort 24b were exposed to approximately 750 RNA copies each. On February 19, each Recipient in Cohort 24c received an exposure of approximately 290 RNA copies (S4 Text). These exposures fell well below the infectious dose estimated from EMIT-1 and recent aerosol inoculation experiments. Therefore, low overall contagiousness of our Donors may be a key reason why no transmission was observed in this study.
This occurred despite two Donors (D24b-1 and D24c-2) having low Ct values from mid-turbinate swabs tested by Cepheid at admission to the hotel quarantine: 12.1 and 16.1 for D24b-1, who was recruited from the University Health Center, and 12.2 and 14.6 for D24c-2. We previously reported that viral load in the nose did not reliably predict exhaled virus concentrations [7]. Donor D24b-1 did not report cough and only coughed during the last of three exhaled breath collections – when we detected infectious virus in their breath. Similarly, Donor D24C-2 only coughed during the last exhaled breath collection. Cough observed during sample collection is a major predictor of viral shedding into aerosols. For future controlled studies of transmission from natural infections, it may be necessary to focus on recruiting Donors who cough.
In addition, we previously observed substantial heterogeneity in viral aerosol shedding in studies of community-acquired influenza and SARS-CoV-2 infections [7,16,21]. Recent human challenge studies have reported similar variability. For example, in a SARS‑CoV‑2 challenge, two participants accounted for 86% of airborne virus released, despite many being infected [22]. Similarly, a recent influenza challenge study found wide variation in the aerosol emissions of infected donors [23]. These consistent findings across both naturally acquired and controlled infections underscore the importance of inter-individual variability in aerosol shedding. Such heterogeneity likely contributes to the overdispersion of transmission seen in real-world outbreaks and may help explain the generally low aerosol shedding observed in our Donors.
The absence of transmission may also be attributable to Recipient pre-existing immunity. Although most of the Recipients had low HAI titers (≤10) against the vaccine strains corresponding to Donor viruses, their ELISA AUC values against these strains were higher at baseline than those of Donors, suggesting potential for some degree of pre-existing immunity from prior infections or vaccinations. The EMIT-1 challenge study began with a registry of 496 volunteers with HAI titers ≤10 against the challenge virus, who reported not having had a seasonal influenza vaccine in the last 3 years [15]. In contrast, our current study had a much smaller registry pool (80 volunteers). Over one-fourth of Recipients whom we later exposed to an influenza Donor had baseline HAI titers >10 and nearly one-third of registry members received an influenza vaccine during the same influenza season as the quarantine, including two who were later exposed to an influenza Donor.Given that our Recipients were middle-aged adults, they would have had greater cumulative exposure to influenza virus infections and vaccinations than our Donors, further contributing to broader immunity. Older adults are more likely than children to develop enhanced pre-existing cross-reactive antibodies due to prior infections and vaccinations [24], and middle-aged adults are less likely than young adults to show evidence of infection in large population studies [25]. Therefore, future transmission trials should focus on recruiting younger adults into the pool of volunteer recipients.
Finally, temperature and humidity during exposure events were within the range likely to promote transmission and therefore do not explain the lack of observed transmission in this study. Laboratory and animal studies reviewed by Arundel et al. [26] suggest that influenza virus survival and transmission are enhanced at low relative humidity (<40%) and may increase again at high humidity (>70%), with a mid-range of ~40–70% minimizing combined infectivity and survival. According to standards from the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), thermal comfort for most individuals lies between 67–82 °F (19.4–27.8°C) [27]. In our study, temperatures were maintained at 22–25°C and relative humidity ranged from 20% to 45%, a range that supports participant comfort while enabling influenza survival and transmission.
Although the relationship between ventilation and airborne transmission is complex, low air change rates have been linked to increased risk in healthcare and public settings [28]. Menzies et al. [29], for example, observed higher tuberculosis infection risk in hospital rooms ventilated at <2 air changes per hour (ACH). While no universal minimum threshold has been validated, we maintained ventilation at ~0.25–0.5 ACH to support transmission. To achieve temperature and relative humidity control in the hotel room used for exposure events, in the presence of low ventilation and seven or more occupants, we ran the room’s two fan coil units at maximum speed and added two large capacity portable dehumidifiers. Operation of these fans produced high rates of air mixing.
Controlled test chamber and modeling studies show that high rates of air mixing reduce short-range aerosol exposure [30,31] by rapidly breaking up concentrated plumes of exhaled breath. Exposure to concentrated plumes may be a critical factor in transmission via aerosol inhalation, especially from hosts with low to moderate viral aerosol shedding rates. Given the low viral aerosol shedding rates of our Donors, rapid air mixing with reduced short-range aerosol exposure may have contributed to lack of transmission in this study. Future studies using an indoor climate facility designed for indoor air quality and climate studies and capable of controlling temperature and humidity with controlled, minimal air movement and ventilation could avoid this problem.
Together, the three possible explanations for lack of transmission in this study (source strength, susceptibility, and air mixing) may provide insight into the transmission of seasonal influenza. For seasonal influenza, previous studies suggest that 4–7 out of 10 cases shed some virus into aerosols. But the geometric mean shedding rate per hour observed was at most one tenth of the infectious dose estimated from EMIT-1 for a selected population with minimal immunity. Although the average infectious dose for a population with some cross-reactive immunity may be higher, susceptibility to circulating human virus without laboratory adaptation is likely lower. Therefore, an infectious dose associated with exhaled breath aerosols in the rage of 104 to 105 viral RNA copies seems reasonable. Shedding rates are lognormally distributed and the upper tail extended to 107 to 108 RNA copies/hour, suggesting that some cases rare cases may shed as many as 100–1000 infectious quanta per hour [7]. However, reported geometric mean aerosol shedding rates suggest that only more symptomatic cases with cough would be expected to shed one or more infectious doses per hour and that mild cases would rarely be contagious [16]. This evidence for influenza virus supershedding suggests that while the average symptomatic case may be able to transmit to someone through prolonged exposure to concentrated breath plumes, a few cases may be able to transmit relatively rapidly with less intense and prolonged exposures. An additional implication may be that prolonged face-to-face contact and a moderate degree of superspreading combine to generate seasonal influenza epidemics.
These considerations suggest that future controlled experiments aiming to observe transmission from naturally infected cases focus on recruiting Donors with cough and young adult volunteers with low rates of prior immunity as Recipients. It also points to advantages of a controlled environment with limited air mixing that allows exhaled plumes to linger, such as certain displacement ventilation arrangements [32].
The unpredictable timing and duration of influenza seasons pose a challenge for conducting a CHIVITT using Donors with community-acquired infection. The inability to recruit an influenza virus-positive Donor in 2023, due to an early and short influenza season, highlighted the need to prepare for potential quarantines from November through early March to capture both early and late influenza seasons, as we did for the 2023–24 influenza season. The resulting prolonged hotel rental including months without influenza activity, added considerable expense. New more flexible designs that can respond to timing of seasonal influenza activity without tying up resources for prolonged periods are needed. An approach that can leverage existing campus-based indoor environment research infrastructure and also study other respiratory viruses when influenza is not actively circulating could be cost effective. Another limitation was that Control Recipients may have reduced behaviors such as touching their face or eyes, which are relevant to touch-contact transmission. To better ensure potential for occurrence of touch-contact transmission additional behavioral requirements, such as instructing Control Recipients to touch their face at regular intervals as previously described for rhinovirus studies [33], could be incorporated and create a clear contrast with the hand hygiene intervention. Finally, we note that mucosal immunity, both innate and acquired, play an important role in susceptibility to initial infection and were not examined in this study.
The conclusion illustrates that this was a study designed to produce transmission that, when it did not occur, was explained away rather than considered evidence against the hypothesis. Transmission is portrayed as dependent on donor cough, recipient susceptibility, and concentrated breath plumes. Failure is attributed to mild symptoms in donors, middle-aged recipients’ “immunity,” or plume dispersion. Future studies are suggested to adjust these variables, insulating the model from falsification: the hypothesis of aerosol transmission is never directly challenged. Across the study, EMIT-2 demonstrates how a null result is repeatedly rescued via post hoc rationalizations, producing increasingly complex, conditional explanations to save a falsified hypothesis.
Conclusion
Lack of transmission from mild influenza cases, who rarely coughed, to healthy middle-aged volunteers in a room with rapid air mixing may provide some insights into influenza transmission dynamics. Transmission from healthy young adults may be dominated by the symptomatic cases with cough. Consistent with population data, middle-aged adults may be relatively immune to circulating influenza viruses. Finally, concentrated breath plumes may play an important role in transmission of seasonal influenza viruses. Future studies addressing these three factors are needed to advance our understanding of the role of inhalation exposure in transmission of seasonal influenza viruses.
One final point worth noting, drawn directly from the Methods section, concerns the study design itself. EMIT-2 was explicitly constructed to maximize the opportunity for influenza transmission. Recipients were randomly divided into two groups—intervention Recipients (IR) and control Recipients (CR)—with both groups exposed to similar risk from inhalation of purportedly “infectious” respiratory aerosols. The key difference was that IRs wore face shields to interrupt direct deposition and sanitized their hands every 15 minutes to reduce direct and indirect contact transmission, whereas CRs were deliberately exposed to all hypothesized transmission routes simultaneously: inhalation of aerosols in a low-ventilation, well-mixed room; direct deposition through unshielded face-to-face interaction; and direct and indirect contact via shared objects without hand hygiene. These conditions were specified a priori and repeatedly justified as favorable for influenza transmission. Yet despite prolonged, repeated exposure to donors classified as actively “infected,” no secondary “infections” occurred in either group by PCR, culture, symptom monitoring, or serology. Because the control arm was exposed through every pathway the contagion model permits, there is no untested route to invoke. The explanations later offered—insufficient donor shedding, recipient “immunity,” aerosol heterogeneity, plume dilution, or air mixing—are therefore not independent experimental findings but post hoc rationalizations introduced only after transmission failed to occur. Taken together, EMIT-2 does not merely fail to demonstrate transmission; it demonstrates that under conditions deliberately engineered to permit it, human-to-human influenza transmission did not occur, leaving the underlying assumption intact only through explanatory rescue rather than empirical support.
The Mystified Miltons
After numerous exposure experiments on Gallops Island failed to produce a single case of influenza in healthy volunteers in 1918, Milton Rosenau was forced to concede that the mechanism of influenza transmission was unknown. He concluded his report with an admission that remains striking today:
“As a matter of fact, we entered the outbreak with a notion that we knew the cause of the disease, and were quite sure we knew how it was transmitted from person to person. Perhaps, if we have learned anything, it is that we are not quite sure what we know about the disease.”
More than a century later, Donald Milton arrived at a remarkably similar impasse. Reflecting on EMIT-2—despite prolonged close contact, low ventilation, and deliberate exposure—he remarked:
“At this time of year, it seems like everyone is catching the flu virus. And yet our study showed no transmission—what does this say about how flu spreads and how to stop outbreaks?”
The persistence of this confusion suggests a deeper problem. Rather than reconsidering the underlying assumption, influenza research has spent a century attempting to rescue it. Despite repeated experimental failures to demonstrate human-to-human transmission, the contagion hypothesis is preserved through increasingly abstract constructs—“infectious quanta,” “supershedders,” invisible aerosol plumes, inferred doses, and digitally mediated laboratory effects—none of which are directly observed to produce disease in humans under controlled conditions. Negative results are not treated as falsifications but as prompts to introduce new, conditional explanations that render the hypothesis unfalsifiable.
In this light, EMIT-2 does not represent progress toward understanding transmission but a reiteration of Rosenau’s dilemma. What changes is not the result, but the explanation. Each experimental failure is met with new contingencies—insufficient coughing, inadequate “source strength,” recipient “immunity,” plume geometry, air mixing, dose thresholds—none of which are prospectively specified, independently manipulated, or capable of falsifying the underlying assumption. Transmission is never allowed to be wrong; it is merely declared absent for reasons proposed after the fact.
This is the defining characteristic of an unfalsifiable model. When transmission occurs, it is credited to the “virus.” When it does not, the environment, the host, the airflow, or the timing is blamed. The hypothesis is insulated from failure by design. Despite technological advances, the fundamental experimental outcome remains unchanged: there are still no human experimental studies that demonstrate reproducible human-to-human transmission of influenza. The century separating the two Miltons has not resolved the question—it has merely refined the language used to avoid answering it.
Dr Sam Bailey examined a “virus” I’m often asked about—the so-called “parvovirus”—and, as usual, the Baileys provide an excellent analysis and reliable resource.
When i bring up this repeated failure of researchers to prove transmission, be it the flu and colds, "COVID," "HIV,"..., defenders of official narratives, even those who are critical of countermeasures such as masks and vaccines, almost always attribute this to the alleged "incompetence" of the researchers. Never mind that they believe the medical profession from which these researchers come out of regarding the validity of the germ/contagion theory. :-) I've heard regarding Rosenau that "science has developed a lot since 1918," as if this was a matter of treating the disease rather than transmission (e.g. an Irish doctor named Gerry, at the 2023 gathering of Doctors For Covid Ethics.) And after all else fails, they appeal to "common sense," saying "i was around sick people, and i got sick."
Transmission is never allowed to be wrong; it is merely declared absent for reasons proposed after the fact.”
A multi-billion dollar allopathic/pharma industry is in large part built upon assumptions of contagion therefore the reluctance to question transmission is a result of the financial interests of those interlocking cartels. Full disclosure would challenge those interests.
When i bring up this repeated failure of researchers to prove transmission, be it the flu and colds, "COVID," "HIV,"..., defenders of official narratives, even those who are critical of countermeasures such as masks and vaccines, almost always attribute this to the alleged "incompetence" of the researchers. Never mind that they believe the medical profession from which these researchers come out of regarding the validity of the germ/contagion theory. :-) I've heard regarding Rosenau that "science has developed a lot since 1918," as if this was a matter of treating the disease rather than transmission (e.g. an Irish doctor named Gerry, at the 2023 gathering of Doctors For Covid Ethics.) And after all else fails, they appeal to "common sense," saying "i was around sick people, and i got sick."
Transmission is never allowed to be wrong; it is merely declared absent for reasons proposed after the fact.”
A multi-billion dollar allopathic/pharma industry is in large part built upon assumptions of contagion therefore the reluctance to question transmission is a result of the financial interests of those interlocking cartels. Full disclosure would challenge those interests.