An Analysis of Data and Hypotheses
Related to the Embassy Incidents
Contact:
JSR-21-01
Version 1.1
February 10, 2022
DISTRIBUTION F. Further dissemination only as directed by U.S. Department of State.
Reason: Sensitive information for the short term. Requests for this document shall be re-
ferred to U.S. Department of State, 1800 Wilson Blvd, Arlington, VA, 22209;
JASON
The MITRE Corporation
7515 Colshire Drive
McLean, Virginia 22102
JSR-21-01 DoS Embassy Incidents February 11, 2022
Revision History
Version Date Description
Original 10/16/2021 Original Version
Corrected typographical errors, resolved missing
1.1 2/10/2022 references, and minor wording changes for clarity.
Appendix H changed to a separate document.
i
JSR-21-01 DoS Embassy Incidents February 11, 2022
Contents
1 EXECUTIVE SUMMARY 1
2 INTRODUCTION 11
2.1 Some Definitions and Terminology . . . . . . . . . . . . . . . . . . . . . . . 15
3 ANALYSIS OF INCIDENT AND SIGNALS DATA 17
3.1 Assessment of Case Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2 Signals Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4 MEDICAL DATA AND THEIR INTERPRETATION 23
4.1 NIH Traumatic Brain Injury Study . . . . . . . . . . . . . . . . . . . . . . . 24
4.1.1 Hypothesis-testing on combined datasets . . . . . . . . . . . . . . . . 27
4.1.2 Example of vestibular migraine with auditory symptoms . . . . . . . 28
4.1.3 The look-elsewhere effect . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2 Differentiating Organic, Functional, and Psychogenic Conditions . . . . . . 31
4.2.1 Functional Disorders: Example of persistent postural-perceptual dizzi-
ness (PPPD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.2.2 Psychogenic illness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3 Potential Underlying Causes . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.4 Clinical Context of the Affected Individuals . . . . . . . . . . . . . . . . . . 41
4.5 Understanding and Managing the Role of Social Interactions . . . . . . . . 43
5 ACOUSTIC ATTACK HYPOTHESES 45
5.1 Introduction and Background . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Infrasound and Reported Effects . . . . . . . . . . . . . . . . . . . . . . . . 47
5.3 Ultrasound and Reported Effects . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
ii
JSR-21-01 DoS Embassy Incidents February 11, 2022
6 ELECTROMAGNETIC ATTACK HYPOTHESES 57
6.1 Introduction and Background . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Ionizing Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.2.1 Generation, focusing, absorption, and interference . . . . . . . . . . 60
6.2.2 Effects on the human body . . . . . . . . . . . . . . . . . . . . . . . 62
6.3 Bounds on RF Wave Properties . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.3.1 Penetration Depth, Skin Heating, and the Carrier Frequency Range 63
6.3.2 Peak and average flux . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.3.3 Pulse duration and repetition frequency . . . . . . . . . . . . . . . . 67
6.4 RF Radiation Beaming, Propagation, and Absorption . . . . . . . . . . . . 68
6.4.1 RF beaming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.4.2 RF absorption and interference . . . . . . . . . . . . . . . . . . . . . 69
6.5 RF Effects on Human Health . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.5.1 The f < 500 MHz Range . . . . . . . . . . . . . . . . . . . . . . . . 72
6.5.2 The 500 MHz < f < 30 GHz range . . . . . . . . . . . . . . . . . . . 74
6.5.3 Requirement for RF-induced Traumatic Brain Injury . . . . . . . . . 79
6.5.4 Thermoacoustic phenomena and the “Frey Effect” . . . . . . . . . . 80
6.5.5 Are there non-thermal effects of microwaves on tissue? . . . . . . . . 84
6.6 Examples of RF Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.6.1 The Moscow Embassy Lilienfeld Study . . . . . . . . . . . . . . . . . 87
6.6.2 The Norway Ship Radar Exposure . . . . . . . . . . . . . . . . . . . 88
6.7 Conclusions on the RF hypotheses . . . . . . . . . . . . . . . . . . . . . . . 91
7 AN INSTRUCTIVE ANALOG 96
8 COMBINED ASSESSMENT AND CONCLUDING REMARKS 98
A APPENDIX: Sample Incident Intake Questionnaire 102
iii
JSR-21-01 DoS Embassy Incidents February 11, 2022
B APPENDIX: Acoustic Waves, Fundamentals, and Detailed Calculations 108
B.1 Sound Speed and Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . 108
B.2 Acoustic reflection and transmission . . . . . . . . . . . . . . . . . . . . . . 109
B.3 Sound attenuation and localization . . . . . . . . . . . . . . . . . . . . . . . 112
B.4 Sound Transmission through a Slab . . . . . . . . . . . . . . . . . . . . . . . 116
C APPENDIX: Mathematical Details of the Frey Effect 120
D APPENDIX: BlueJay 2nd-gen RF Detector 124
E APPENDIX: Microwave Pulser 126
E.1 Commercially Available Components . . . . . . . . . . . . . . . . . . . . . . 126
E.1.1 The low-level pulsed RF . . . . . . . . . . . . . . . . . . . . . . . . . 126
E.1.2 The power amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
E.1.3 Getting to Megawatts . . . . . . . . . . . . . . . . . . . . . . . . . . 128
E.2 Portable Power Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
F APPENDIX: Classified (S) 132
G APPENDIX: Classified (TS//SCI) 134
H APPENDIX: Briefings 136
iv
JSR-21-01 DoS Embassy Incidents February 11, 2022
1 EXECUTIVE SUMMARY
As of the time of our study, nearly two hundred federal employees and their family members
assigned to posts in embassies and consulates around the world have reported adverse
health symptoms perceived to have been precipitated by incidents experienced in their work
places or residences. The reported incidents range from unfamiliar sounds to a sensation
of head pressure, while the associated symptoms can include vertigo, nausea, headache,
and tinnitus. In some cases, the severity and duration of the symptoms have been intense,
leading to lasting cognitive and vestibular deficits. Identifying the causes of these incidents
and providing mitigation and prevention strategies constitutes a high priority for the United
States government. As part of this effort, JASON has been tasked by the Department of
State to consider all available data and evaluate potential mechanisms with regard to their
ability to produce effects that are consistent with those reported.
Making progress on identifying the causes of the health incidents (often referred to
as Anomalous Health Incidents or AHIs) requires two parallel but related methodologies.
The first one requires a well-defined medical case statement that is based on identifying a
unique cluster of symptoms and establishing that such symptoms exceed their prevalence in
a comparable background population. The second approach relies on identifying a cause or
causes that can be physiologically linked to the reported symptoms and physically linked to
the anecdotal
1
evidence in the incident reports, under the assumption that these symptoms
are related to a single underlying cause and are not manifestations of a collection of medical
conditions and functional disorders. We refer to these ‘first’ and ‘second’ approaches and
discuss them in more detail below.
The Centers for Disease Control and Prevention (CDC) currently adopts a work-
ing case definition that requires the onset of symptoms in two phases: a first phase that
includes at least one of head pressure, disorientation, nausea, headache, vestibular distur-
bances, auditory symptoms, or vision changes, followed by a second phase that includes
either vestibular disturbances or cognitive deficits, with no readily recognizable alternate
1
We use anecdotal in the following sense, in accordance with the Oxford English Dictionary: ”consisting
of, or based on, reports of individual cases rather than systematic research or analysis.”
1
JSR-21-01 DoS Embassy Incidents February 11, 2022
explanation. This case definition is based on the incidents self-reported by the affected
personnel and on subsequent medical evaluations of the associated symptoms.
The Department of State provided to JASON incident data files that included, to
varying degrees of detail, the location and narrative of a reported incident, the immediate
sensations described by the individual, medical symptoms that developed over time, and a
description of environmental factors that may have been pertinent to the incident or the
associated symptoms. The accounts are highly varied in many regards: the number and
specificity of the queries and responses naturally evolved substantially with time; the timing
of the reports ranged from minutes to years after the incident, introducing different degrees
of recall bias; the number of reported symptoms accompanying an incident varied from none
to the full set included in CDC’s criteria.
The narrative nature of the reports precluded a quantitative analysis. Upon performing
a qualitative systematic review, JASON finds that the reported incidents themselves did not
share any uniquely identifiable common set of characteristics. Further, JASON finds natural
and credible explanations for these incidents in all but 20-30 out of the approximately 200
cases because either the incidents and symptoms fall within the realm of everyday common
occurrences or the symptoms can be explained in ways that are not related to the perceived
incidents.
With regard to the medical data, investigators at the National Institutes of Health
(NIH) presented to JASON preliminary population-aggregated data of clinical test results
for 65 individuals enrolled in an ongoing 5-year study on traumatic brain injury (TBI)
of overseas U.S. personnel Shahim et al. (2021). The study began in May 2018 and was
designed to evaluate whether signs of brain injury were present in any of the personnel in
the days to years following the incidents. The data included outcomes from a multitude of
diagnostic tests, such as MRI imaging, audiology exam, serum biomarkers, and assessment
of vestibular function (inner ear and balance) tests including eye movement, gait, and
Vestibular Evoked Myogenic Potential (VEMP). JASON finds no compelling evidence of
TBI in the data obtained to date. We note that the NIH study has not been completed and
that with additional data, the NIH may come to conclusions different from JASON’s.
2
JSR-21-01 DoS Embassy Incidents February 11, 2022
Having carefully evaluated all available incident and medical data, we find that sta-
tistical conclusions from the first, data-driven approach cannot be drawn with even a low
degree of confidence because:
• Appropriate control groups for the personnel in question need to be lifestyle matched
and would, therefore, ideally be limited to embassy personnel on active duty. Because
this requirement poses significant logistical challenges, the control groups enrolled in
current studies are small, heterogeneous, or not appropriately matched.
• The number of cases showing a substantial range of signs and symptoms is small,
leading to significant statistical fluctuations in the sample and uncertainties in the
results.
• Administering many successive independent tests to the affected individuals, without
a clear hypothesis, increases the risk of encountering a spurious correlation.
• Self-reporting, varied initial evaluation criteria, and additional criteria for admission
into the NIH clinical study each contribute to selection effects that are difficult to
quantify and disentangle.
The net result of these complications is an inability to provide a rigorous case definition
and, therefore, a corresponding lack of guidance toward an underlying cause based on this
data-driven approach at this time.
Turning to the second parallel methodology, JASON evaluated multiple mechanisms
as potential causes that could establish a physical and physiological connection between
the reported symptoms and the information in the incident reports. In particular, the
anecdotes provided in the incident data would require mechanisms of damage that can be
directed covertly and over short periods of time at individuals who are indoors (as reported
in more than 90% of the nearly 200 cases) and localized to a space of less than a few meters
(as reported in approximately 75% of the cases). The proposed mechanisms would also
have to induce recognizable and lasting health effects on humans through some identifiable
physiological pathway. In our evaluation of potential mechanisms, we assessed the physics
3
JSR-21-01 DoS Embassy Incidents February 11, 2022
of delivering directed energy to an individual consistent with the incident reports and we
evaluated the potential of such mechanisms to cause sensations and symptoms consistent
with the medical evaluations. We refer to these as our incident consistency criteria. In
devising these criteria, we considered more heavily the smaller subset (approximately 15%)
of cases that did not have immediate common or environmental explanations, rather than
include the remainder that appeared to be more consistent with everyday occurrences of
events and symptoms. The incident consistency criteria, discussed further in Section 2,
contain the following: energy must be able to penetrate into buildings through walls and
windows from a distance of several tens of meters, be localized to areas the size of a room
or smaller, be generated with equipment that is mobile and covert, lead to audible and/or
recordable sounds, cause immediate symptoms of pressure, vertigo, headache, or nausea,
and cause long-term vestibular and/or cognitive dysfunction.
With regard to direct physical attacks, the scope of this JASON study was focused on
mechanisms that can deliver energy from a distance, owing to the thorough work presented
in previous investigations on mechanisms involving biological agents and toxins (such as the
2018 JASON study and the 2020 study by the National Academy of Sciences). We inves-
tigated electromagnetic energy delivery in the form of radio-frequency (RF) and ionizing
radiation, and acoustic energy delivery in the form of infrasound, audible, and ultrasound
waves. Present data allow us to rule out some of these mechanisms with a high degree of
confidence. We propose methods for collecting data to assess the remaining possibilities
more fully.
We also evaluated the potential role of functional disorders, i.e., conditions involving
altered biological function without known structural change, in producing the reported
cluster of symptoms (in contrast to physical attacks that can directly cause physiological
damage). We considered functional disorders with different types of precipitating events,
including purposeful attacks, inadvertent physical insults, events that can be considered
ordinary but have an amplified effect due to prevailing circumstances (akin to post-traumatic
stress disorder), medical conditions unrelated to any external insult, or unidentified events
that precipitate a functional condition. We found that these disorders potentially explain at
least a subset of the long-term symptoms reported by the affected personnel, even though
4
JSR-21-01 DoS Embassy Incidents February 11, 2022
it is not possible to identify a common initial stimulus consistent with all the events and
criteria.
Our analysis provides a way to focus future data collection on a hypothesis-based
definition of Anomalous Health Incidents that combines both methodologies employed here.
Having reviewed all incident data to select the cases that are potentially consistent with AHI
and having explored a range of physically plausible mechanisms of attack or harassment, we
identify the most relevant criteria on which a future incident questionnaire can be based.
Furthermore, assigning numerical weights to the responses in the questionnaire can facilitate
a rapid and uniform initial assessment of reported incidents across all embassies, allow
for an initial triage, and help guide appropriate medical and monitoring response. We
recommend that only incidents that meet a threshold be labeled as AHI and selected into a
different database to increase the statistical robustness in subsequent hypothesis evaluations.
Naturally, medical care and support of the affected individuals should be a high priority
independent of the AHI-consistency decisions. In other words, we emphasize that judging a
case to be inconsistent with an AHI carries no implications about its importance or severity
from a medical standpoint.
Regarding adversarial intent, it is not possible to conclude at this time that the events
reviewed by JASON are the result of intentional attacks that cause physical harm. However,
it is not possible, either, to rule out mechanisms that do not cause any physical harm but
which might constitute harassment and lead to health conditions and functional disorders,
for example through unpleasant sounds or pressure sensations. Given this, and in the
interest of protecting embassy personnel and their families, it would be prudent to be
vigilant against tactics intended to produce anxiety and trauma, with an intent to either
disrupt operations and/or cause long-term harm. The US government could minimize the
effects of such tactics, if present, through open communication, education, and appropriate
rapid medical response to any conditions that develop.
5
JSR-21-01 DoS Embassy Incidents February 11, 2022
We provide our full findings and recommendations below.
Findings
F1. 85-90% of the reported incidents are consistent with common and known symp-
toms of pathophysiological or environmental origins; the remainder appear more complex
and defy a straightforward explanation at present.
F2. The use of uniform, structured incident intake questionnaires with quantifiable
responses can help homogenize collected data, be used for preliminary assessment (e.g.,
through assigning numerical values to responses) and appropriate rapid response, and may
help illuminate patterns and potential causes.
F3. The medical data collected through the NIH study are not, on their own, suffi-
cient to explain the reported abnormal signs and symptoms. Furthermore, the aggregated
medical data are not sufficient to infer the existence of a novel clinical syndrome
2
.
F4. Under the CDC’s working definition of a case, the reported collection of symp-
toms could derive from a structural cause, functional disorder, psychogenic response, or
some combination of the three. The government’s response to the reported incidents may
be causing additional stress to the affected individuals and may exacerbate normal psycho-
logical responses that are almost certainly present.
F5. On evaluation of specific energy delivery hypotheses on the basis of the combina-
tion of medical and situational data, and physical considerations, we find that:
• No single hypothesized mechanism can explain all of the incidents.
2
Used here in accordance with the medical definition of a syndrome: a group of signs and symptoms that
occur together and characterize a particular abnormality or condition, for which a direct cause may not be
understood.
6
JSR-21-01 DoS Embassy Incidents February 11, 2022
• Ionizing radiation, electromagnetic energy below 500 MHz and above 30 GHz, and
sonic (infrasound/audible/ultrasound) energy delivered from a distance can, with high
confidence, be eliminated as potential causes.
• Short-pulse radio frequency (RF) radiation primarily in the range of 500 MHz to 30
GHz cannot be ruled out conclusively as a modality for a subset of the incidents at
this time. However, it is unlikely given that there is no well-documented or broadly
accepted mechanism to induce lasting neurological or other damage with focusable
RF energy that would not produce a sensation of heating on the skin.
• Sounds associated with cranial acoustic sensations due to pulsed RF waves (the “Frey
effect”) cannot produce brain damage by mechanical means at tolerable sound levels.
Further, such sensations cannot be recorded by electronic microphones and, therefore,
cannot be associated with any recorded acoustic phenomena.
F6. The absence of observed disruption of electronics (e.g., cellphones, televisions,
computers) reduces the allowed parameter space of any potential pulsed RF signals but
does not rule them out, given that contemporary consumer electronics may not be affected
by short pulses at peak flux levels below approximately 25 W/cm
2
.
F7. Persistent monitoring with appropriate sensors at select locations is needed for
further evaluation of RF hypotheses and could help exclude this remaining potential mech-
anism.
F8. It is JASON’s judgement that, on the basis of available reports of events, related
data, and health evaluations, it is not possible to conclude at this time that these events
are the result of intentional attacks that cause physical harm.
7
JSR-21-01 DoS Embassy Incidents February 11, 2022
Recommendations
R1. DOS should modify its questionnaire for incident reporting to enable a uniform,
structured, and quantitative dataset, focusing on surveying specific physical sensations that
would accompany remote energy attacks. Such a questionnaire can be used to minimize
the variance and noise in data collection, to provide an initial triage, and to initiate rapid
deployment of follow-up environmental monitoring. JASON provides an example of such a
questionnaire in the report.
R2. DOS should categorize only the incidents that meet a threshold as AHI and select
these into a different database. The threshold should be based on the intake questionnaire
to provide clarity as quickly as possible to the affected personnel. DOS should guide individ-
uals to appropriate medical care and support independent of the AHI-consistency decisions.
R3. DOS should develop criteria (e.g., linear classifiers) to rank the relevance of inci-
dent reports in the context of hypothesis testing. The focus should be on hypotheses that
have not been ruled out by existing analyses.
R4. In conjunction with other agencies, DOS should consider using the uniform triage
process to select affected individuals for inclusion in future medical studies, carrying out a
specific series of tests all done in the same order and within a specified time frame, striving
to make appropriately selected control groups larger, and providing incentives for individ-
uals to participate in the control groups.
R5. Informed by related efforts by other agencies, DOS should analyze existing and
future patient-level health data in conjunction with reported sensory symptoms to identify
possible clusters, which may help define a potentially novel syndrome.
R6. DOS and collaborating agencies should adapt their baseline medical evaluation
8
JSR-21-01 DoS Embassy Incidents February 11, 2022
program so that it is more comprehensive and responsive to a variety of hypothesized harm
mechanisms, to be administered prior to deployment overseas. These evaluations should
be used on an individual basis to provide context for any future health events and in an
aggregated manner to provide an appropriate control group for the reported symptoms. It
may be useful to consider offering incentives for individuals to participate in this program.
R7. DOS should deploy broadband RF sensors to locations of interest. The sensors
should be capable of detecting RF waves in the 0.5–30 GHz range, from continuous waves
to nanosecond pulses, sensitive to average flux levels down to 20 mW/cm
2
and peak flux
levels down to 1 W/cm
2
, and with sufficient sampling rates to detect a broad range of
pulse repetition rates from 1 to 10,000 pulses per second. A two-tier monitoring program
could be employed, where some sensors sensitive to a subset of the frequencies are deployed
more broadly and sensors sensitive to the whole frequency range are deployed at select
locations. Sensors for continuous data recording should be deployed in various, randomly
selected locations. In addition, sensors should be deployed speedily at any location where
a documented incident has occurred.
R8. DOS should carry out experiments using RF on relevant and commonly used
electronics to establish the frequency, power, and pulse duration ranges that might disrupt
their operation and track specific models of electronic devices that might be disrupted by
future events. The data from such experiments, together with the absence of disruptions
during reported events can be used to set parameter limits on directed RF power that might
be of concern.
R9. Given that fear and stress can cause or exacerbate real harm to DOS employees,
DOS should consult with relevant experts concerning messaging, training, guidance, and
other measures to reduce anxiety associated with the reported incidents. These could
include building awareness of the very limited mechanisms capable of producing physical
harm via energy delivery at a distance. DOS should recognize psychological phenomena as
real afflictions that cause pain and treat those cases the same way as other types of injuries.
9
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
10
JSR-21-01 DoS Embassy Incidents February 11, 2022
2 INTRODUCTION
The possibility that US personnel have been subjected to intentional attacks leading to
lasting adverse health effects is, appropriately, an ongoing source of concern to the US
government. This issue impacts the individuals involved as well as the federal agencies that
employ them, and has serious implications for US foreign policy and for the conduct of our
diplomatic operations. This document presents the results of the second JASON study of
this issue. The previous JASON report, JSR-18-017 “Acoustic Signals and Physiological
Effects on U.S. Diplomats in Cuba,” contains deliberations, findings, and recommendations
for the situation that obtained in 2018. In the intervening time, a number of other studies
(by e.g., the National Institutes of Health, the National Academy of Sciences, etc.) were
conducted. Other investigations are being undertaken in parallel with this 2021 JASON
study.
We were charged by the Department of State to:
• examine all available studies and documentation related to the constellation of symp-
toms commonly referred to as “Havana Syndrome”.
• examine data and attempt to identify any stimuli that may have caused the effects
and quantify how those stimuli could have been produced.
• analyze current monitoring protocols and make recommendations to enhance preven-
tative measures where possible.
Further, we were asked to include three components in the study:
• Analysis: JASON will define a credibility threshold for provided technical hypothe-
ses based on retrospective analysis of data, and then list which theories exceed said
threshold. In the process, JASON will suggest new tests for ranking hypotheses.
• Sensing: JASON will recommend what monitoring should be carried out in the mid-
term and in the long-term to validate or eliminate probable causes.
11
JSR-21-01 DoS Embassy Incidents February 11, 2022
• Mitigations: JASON will suggest measures to mitigate probable causes.
Reports of incidents continue to accumulate, particularly after personnel were asked
to come forward and report incidents they deemed unusual or that elicited concern. The
challenge at this stage is to assess and categorize the candidate incidents, on the basis of
available data (medical and physical) from both the reporting population and appropriate
“control” groups, to facilitate an evaluation of various hypotheses. Reports need to be
assessed against a range of hypotheses, including:
• Actions undertaken by one or more adversaries with the intent of inflicting physical
harm through the transmission of harmful amounts of energy,
• Actions undertaken by one or more adversaries where physical harm is an unintended
consequence,
• Actions undertaken by one or more adversaries where real harm, for example via the
causation of functional disorders, arises as a consequence of an attack which does not
cause identifiable physiological damage,
• Instances of medical conditions that are of “normal” origin, consistent with the rates
seen in an appropriate control group,
• Instances where individuals respond to an environment of high tension and aware-
ness, where otherwise unremarkable sensations are interpreted as initiating events or
symptoms (potentially in conjunction with functional disorders),
• Cases where genuine and lasting adverse health effects are brought on by a climate of
heightened apprehension and anxiety (psychogenic illness),
• Environmental sources or naturally occurring pathogens (viral or bacterial),
• An increase in reported events due to reinforcement on social media or other interac-
tions, whether spontaneous or incited.
Given the range of phenomena and circumstances described in the reports that were shared
with us, we consider it unlikely that a single explanation applies to all of the candidate
12
JSR-21-01 DoS Embassy Incidents February 11, 2022
incidents. In addition, several mechanisms may be at play at the same time. For example,
an adversary may take advantage of heightened tension and awareness and attempt to
inflict real harm using methods which do not transmit harmful amounts of energy but do
trigger sensory responses. The findings of this report contain JASON’s assessment of the
information currently on hand for the candidate incidents. The recommendations contain
JASON suggestions on how to improve the scope and utility of information gathered for
future candidate events.
Our approach entailed (1) reviewing the information obtained from interviews with
individuals, (2) reviewing the diagnostic health data that were shared with us, and prior
reports and meta-analyses of this medical information, (3) reviewing the known mechanisms
by which standoff devices could impart lasting adverse health effects, and (4) a systematic
evaluation of possible means for transmitting effective amounts of energy in ways that
are consistent with the reporting and that could produce lasting adverse health effects.
We also took into consideration other proposed, but not well-accepted by the health science
community, mechanisms for inflicting harm from afar (such as the “grey literature” discussed
in Section 6.5.5).
Our study focused on evaluating scenarios where damaging amounts of energy could
be broadcast from tens of meters of separation, at levels that could produce direct physical
harm, as well as on assessing plausible scenarios where temporary physical stimuli trig-
gered from a distance could lead to short-term and long-term symptoms (e.g., through
precipitating a functional disorder). The conclusions from this study are intended to al-
low the government to rule out attack mechanisms that cause direct physical harm and
provide guidance on how to sense for signals that can cause harm directly and indirectly.
We do not address viral and bacterial infections, the possibility of environmental toxins or
contaminants, or food or water-borne vectors, as these were studied in detail in the 2018 JA-
SON report and in the 2020 report from the National Academies of Sciences, Engineering,
Medicine (National Academies of Sciences, Engineering, and Medicine, 2020).
We think it is informative to briefly review the history and evolution of what started off
as the “Havana Syndrome” or, more formally, as Anomalous Health Incidents. The initial
13
JSR-21-01 DoS Embassy Incidents February 11, 2022
reports of anomalous experiences were tightly linked at the time to loud audible sounds,
some of which were recorded. The previous JASON study provided an in-depth analysis of
these sounds and (we believe) identified a natural and possibly benign source for the noises.
The personnel involved in those initial candidate incidents were subjected to a variety of
medical evaluations, including neurological functional tests and medical imaging. At that
time the medical evidence for cerebral anomalies was presented as clear and compelling,
a finding that was later called into doubt by subsequent analyses of those data (Muth &
Lewis, 2018; Della Sala & Cubelli, 2018) as well as by the NIH Traumatic Brain Injury Study
(Shahim et al., 2021). Less clear at the time was whether the incidence of anomalies exceeded
that to be expected from an age, gender, and life-experience–matched control group. It was,
and remains, a challenge to crisply identify a set of signs and symptoms that would comprise
a clear case definition. The paucity of physical data associated with candidate events posed
a challenge to making a definitive association with causal mechanisms.
In the time between the 2018 JASON study and this writing (Aug 2021), the situation
has evolved. The audible-acoustic association with candidate events has declined dramati-
cally. The interpretation of the earlier medical diagnostic information has been called into
question, and the geographical distribution of candidate events is much broader. Given the
range of experiences that appear in the reports on the various candidate incidents, it is not
surprising that the medical signs and symptoms, when they exist, are not entirely consistent
between cases and are complicated by confounding factors. The statistical interpretation
of the medical information is impeded by a lack of data from appropriate control groups
of sufficient population size and by the extreme nonuniformity of the data collection. We
comment on physical data collection in the Appendices of this report.
Another consideration to which we draw attention is the time that has elapsed since
the initial reports, and the substantial press coverage that they continue to receive. Even
if no intentional adversarial actions occurred early on, at this point we need to anticipate
adversaries exploiting the current situation and perhaps taking actions to confuse, confound,
and impede our resolving the issue. We have further comments on this in the Appendices
as well.
14
JSR-21-01 DoS Embassy Incidents February 11, 2022
This report begins with our assessment of the case reports that were provided to us,
for the candidate incidents. We highlight the benefits of implementing the collection of
structured data that is amendable to automated analysis, and the merits of taking prompt
action (including the installation of in situ RF monitoring systems) when certain conditions
are satisfied. We provide a summary appraisal of the existing medical data. We then
evaluate all available data in the context of known mechanisms that can produce lasting
adverse health effects, including through pathways that may trigger functional disorders,
including ionizing radiation, sound (ranging from low frequency infrasound to inaudible
ultrasound), and electromagnetic waves. We assess these possible intrusion mechanisms in
the context of stand-off means of producing long-lasting health effects that are consistent
with the medical data presented to date.
We are grateful to the many individuals who provided briefings, data, and background
material for this study. A list of briefings delivered to JASON can be found in Appendix
H. We also take this opportunity to extend our gratitude to the many US personnel who
undertake the nation’s business overseas.
2.1 Some Definitions and Terminology
We provide the following terms and their definitions as used throughout this report.
• Organic Disease: Any health condition in which there is an observable and mea-
surable disease process, such as inflammation or tissue damage. An organic disease is
one that can be validated and quantified through the standardized biological measures
known as biomarkers.
• Non-organic Disease: Any health condition which manifests with symptoms but
whose disease process is either unknown or unable to be measured by current scientific
means.
• Syndrome: A recognizable complex of symptoms and physical findings that occur
together and characterize a particular abnormality or condition, for which a direct
cause may not be understood.
15
JSR-21-01 DoS Embassy Incidents February 11, 2022
• Anecdotal: Consisting of, or based on, reports of individual cases rather than sys-
tematic research or analysis.
• Traumatic Brain Injury: An injury that affects how the brain works.
• Functional Disorder: A medical condition involving altered biological function
without known structural change.
• Psychogenic Illness: A medical condition involving loss or alteration of function
that does not have a precipitating physical cause.
• Situational: Relating to location and surroundings of a place.
• Nociceptor: A sensory receptor for painful stimuli. Nociception is the perception of
a painful or injurious stimulus.
• Etiology: The cause, set of causes, or manner of causation of a disease or condition.
• Afferent: Conducting or conducted inward or toward something (for nerves, the
central nervous system; for blood vessels, the organ supplied).
16
JSR-21-01 DoS Embassy Incidents February 11, 2022
3 ANALYSIS OF INCIDENT AND SIGNALS DATA
3.1 Assessment of Case Data
JASON was provided with a set of incident data from personnel assigned to posts in US
embassies and consulates who reported experiencing sensory and medical incidents over the
course of the last four years. The dataset spanned multiple countries and was complete
up to May 2021. We provide a general assessment of these data here and include some
additional details in an appendix.
The data provided to us are in the form of spreadsheets that included various details
about the incidents in multiple columns. They are all in narrative form, providing an ac-
count of the perceived incident, sensations experienced by the affected personnel, symptoms
that developed immediately or after a period of time, the location of the incident, and other
environmental details that may be pertinent to the experience (e.g., if it took place indoors
or outdoors, whether other people were present and if they also experienced any sensations,
whether the affected individual changed locations and if this alleviated the sensations).
Each incident was self-reported by the affected personnel. A small subset (10-15%) of the
reports were prompted by embassy-wide meetings or agency-wide requests that asked all
personnel to report “anything unusual” experienced during assignments.
The level of detail and the content of the incident reports is highly heterogeneous. The
files show that the incident reports have evolved significantly over time. Early data from
Havana include fewer pieces of information, can be captured in a handful of columns, and do
not appear to be responses to a uniform set of questions. Later reports include many more
questions and responses and fill several tens of columns. We took this evident evolution
in the incident files as a testament to the significant efforts by the Department of State
and other government agencies towards gathering relevant data, identifying a cause, and
providing a solution to these incidents. The level of detail captured also varies by location.
In addition, because the reports contain answers in narrative form, the details provided are
not consistent across reports, even when reports are guided by a uniform set of questions.
Finally, the delay between the time of the incident and the time of the report varies from
17
JSR-21-01 DoS Embassy Incidents February 11, 2022
hours (typically for more recent reports) to years (for some of the earlier reports.) In general,
reports prompted by requests occurred after longer delays than unprompted ones. In our
opinion, this range of delays introduces different levels of recall bias to different reports.
Analyzing the incidents as a whole, we find that the data show a multitude of differ-
ences across cases.
• The incidents described do not follow one single or clear pattern but show significant
variation across events, sensations, and symptoms. The severity likewise spans a wide
range: some affected individuals only report an unfamiliar sound, others communicate
a multitude of sensations and near-term symptoms, including sensations of pressure
and sound accompanied by vertigo, nausea, tinnitus, and/or headache. Less than 20%
of cases appear to report all of the sensations and symptoms in CDC’s current case
definition.
• The nature of the incidents show significant evolution over time, from reports of loud,
external sounds in earlier cases in Havana to a sensation of pressure or headache at
the onset for later ones. Overall, they do not appear to describe the same phenomena.
• In later incidents, the majority of the affected personnel express awareness of the prior
incidents in their reports and appear to verbalize a stress response to the possibility
that they have been targeted.
• The reports are divided between descriptions of one-time, sudden events and those
that continue for hours, days, or weeks, or occur repeatedly.
We emphasize that even among the subset of incidents with similarities (e.g., a report
of a sound), the specifics described vary a great deal. In the case of sound, descriptions range
from high pitched sounds of short duration to low-frequency bass beats to constant noises
that resemble running water or rolling marbles. In some cases, the sounds last for days or
more; in others, they represent a singular incident. These descriptions do not immediately
point to a single origin for the sounds or a single underlying cause for the events.
18
JSR-21-01 DoS Embassy Incidents February 11, 2022
All of these characteristics make it difficult to provide a quantitative assessment of
the incidents in order to look for patterns and correlations. With that caveat in mind, we
nevertheless attempt to quantify a few key features that will be important for connecting
the events to an underlying cause (or causes).
In 85-90% of the incidents (all but 20-30 cases out of nearly 200), it was our judge-
ment that the reported sensations could be explained by normal, benign, environmental or
natural causes. These include sounds that seemed to be originating from appliances, noisy
hallways or natural sounds; ear or head pressure accompanying reported colds; and pressure
sensations that accompany the onset of headaches. Eliminating such incidents left 10-15%
of incidents that evaded an immediate alternative explanation. (Finding 1)
The accounts of the remaining cases have similarities with one another and likely
serve as the basis for the current case definition. This pattern of sensations and symptoms
includes a sudden onset of a sound or a sensation of pressure, followed by a subset of
the following symptoms: headache, vertigo, nausea, tinnitus, gait problems, confusion, or
other cognitive problems. The large overlap between these non-specific symptoms and other
common medical conditions makes it difficult to identify a link between these reports and a
physical cause. Nevertheless, we identified a set of frequently reported characteristics and
distilled these into ”incident consistency criteria” we will use to evaluate various hypotheses.
Below, we highlight these criteria and our reasoning for inclusion in our assessment.
Incident Consistency Criteria:
More than 90% of the cases were reported to take place indoors. At least 75% of the
cases took place in personnel residences or hotels. The remainder took place either within
embassy buildings or occurred in other indoor or outdoor public locations (the latter making
up less than a total of 10 cases). This gives rise to the first evaluation criterion: in the
majority of the cases, an external source must be able to penetrate through the walls
or windows of a building and still be able to have adverse health effects on people.
19
JSR-21-01 DoS Embassy Incidents February 11, 2022
In the majority of cases, personnel report being a significant distance away from other
people or residences. This includes being inside an embassy building with controlled outside
access, the interior room of a house, or in an outside facing room that is not adjacent to a
neighbor. We assess from these descriptions that if harm is inflicted by an external energy
source, that the source should be able to operate from a distance of tens of meters
from its target, at least in a significant fraction of the cases.
More than 50% of cases explicitly refer to localization and/or directionality. Examples
include affected personnel reporting that the sensations subside after changing positions in
a room, other individual(s) in close proximity not reporting signs or symptoms, and the
sound being described as coming from a particular direction or location. This leads us to
our criterion that if an external source of energy is responsible for the events, it should be
focusable to an area of a few square meters, at least in a subset of the incidents.
There were no reports of disruptions in the functionality of personal electronic devices.
Moreover, any reported interruptions around the time of an incident could be traced to
normal events such as electrical maintenance. As a result, we require that an external
energy source used to cause harm to people should not cause discernible interruption
in electronic devices or communications.
Future Data Collection:
Going forward, incident assessments could be facilitated with a questionnaire with
multiple choice answers that aim to gather only pertinent data. The questions should be
structured around factors described in Sections 5 and 6 and address specific physical sensa-
tions that are expected to accompany remote energy attacks. A quantifiable questionnaire
can help reduce the level of heterogeneity present in the data and potentially illuminate
patterns, if present. JASON provides an example of such a questionnaire in Appendix A.
(Recommendation 1)
This questionnaire could be used for a uniform initial assessment and triage process
both to guide rapid medical care as well as to trigger physical/environmental monitoring
for any signals of interest (we discuss this further in Section 6). It would be useful to
20
JSR-21-01 DoS Embassy Incidents February 11, 2022
assign numerical scores to responses to answers in the questionnaire, based on severity and
relevance. A total numerical score can form the basis of the initial assessment and triage:
certain sets of medical and surveillance actions can be taken when threshold conditions are
met. (Recommendation 2)
Finally, it would be useful to develop classifiers to rank the relevance of incident reports
in the context of hypothesis testing. For example, linear classifiers sort data into categories
based on a linear combination of input features. Different hypotheses would form different
categories based on the different expected set of sensations, symptoms, and situational
information. (Recommendation 3)
3.2 Signals Data
JASON has been provided with documentation and briefings on a variety of environmental
data that has been collected at incident locations. We discuss our assessment of these data
in an appendix.
We briefly note here that
• Some recordings of the sounds have been made by personnel during the incidents they
reported.
• Other sensors that are sensitive to various signals and hazards have been deployed to
some locations.
• No significant detections have been reported from such sensors.
• Some sensors were active during incidents while others were used for post facto in-
spections.
We factor in the recorded sounds and the negative result from sensors in our assess-
ment of potential energy delivery mechanisms that may cause adverse health effects
(Chapters 5 and 6.)
21
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
22
JSR-21-01 DoS Embassy Incidents February 11, 2022
4 MEDICAL DATA AND THEIR INTERPRETATION
The investigators at the National Institutes of Health (NIH) presented to JASON prelim-
inary population-aggregated data of clinical test results for 65 individuals enrolled in an
ongoing 5-year study on traumatic brain injury (TBI) of overseas U.S. personnel (Shahim
et al., 2021). We base our analysis of the medical findings in this report on these data, not-
ing that NIH may arrive at different conclusions at the end of their study with additional
data. However, we first briefly review earlier medical studies performed on the affected
personnel to identify underlying causes of their symptoms in order to provide some context
for the variety and evolution of the medical tests and evaluations that were performed.
Because the initial incident reports described loud sounds accompanying the symp-
toms, the first evaluations for the early cohort from Havana were performed at the Uni-
versity of Miami, with a focus on audiological and otolaryngological tests (i.e., related
to hearing, ear, nose, and throat). Between February and April 2017, the medical team
in Miami evaluated 80 embassy staff and family members, of which 16 were judged to
have clinical symptoms consistent with mild traumatic brain injury or concussion, although
other causes including stress and underlying medical conditions were not ruled out (Baloh &
Bartholomew, 2020). These 16 individuals, along with 8 others identified in the intervening
time, were then evaluated at the University of Pennsylvania Center for Brain Injury and
Repair. Out of the 24, 21 individuals completed the set of evaluations that included cogni-
tive, vestibular, oculomotor, audiometric, and imaging tests. This study, published in the
Journal of the American Medical Association (JAMA; Swanson et al. 2018), concluded that
all 21 individuals exhibited acute and persistent symptoms consistent with concussions.
The JAMA study was later criticized for making claims that were not well supported
by the data and for attempting to interpret the data without appropriate comparisons to
control groups (see, e.g., Muth & Lewis 2018; Della Sala & Cubelli 2018). Subsequent
analyses of these data concluded that most medical findings were within normal ranges
and that there were no structural changes in the brains of the individuals. For example,
follow-up studies argued that imaging revealed “nonspecific white matter changes” in only
three individuals that could have resulted from a wide range of conditions, including aging
23
JSR-21-01 DoS Embassy Incidents February 11, 2022
and depression. In addition, these critiques pointed out that many of the symptoms overlap
with numerous other medical conditions and did not provide clear evidence for brain injury.
In our view, these examples serve to illustrate some obvious but important aspects of
these medical evaluations. First, it is important to enroll appropriate control groups. Be-
cause the non-specific symptoms reported by the affected personnel can result from a range
of normal situations and medical conditions, defining appropriate age-, sex-, and lifestyle-
matched control groups and enrolling statistically relevant numbers of individuals in the
studies is crucial for obtaining robust conclusions. Second, administering many successive
independent tests to the affected individuals, without a clear hypothesis, increases the risk
of encountering spurious correlations (we revisit this in section 4.1.3). Therefore, JASON
advises formulating specific hypotheses – such as expected physical damage associated with
particular attack mechanisms – and performing tests specifically designed to look for those
effects (as opposed to a broad range of more general medical tests).
4.1 NIH Traumatic Brain Injury Study
JASON reviewed preliminary aggregated medical data from the NIH’s ongoing 5-year study
on TBI of overseas U.S. personnel, which began in May 2018. At the time of this review
(August 2021), the NIH investigators presented JASON with population-aggregated data
of clinical test results for 65 cases: 26 from Cuba, 6 from China, 9 from Vienna, and 24
from other locations. As of June 2021, this sample represents about 0.6% of an estimated
11,000 overseas non-military personnel who are candidates for conjectured harm.
1
The NIH study evaluated patients at different stages in their clinical progression,
ranging from days to years after the initial onset of symptoms. The selection of patients for
inclusion in the study was subject to several filters, including self-reporting of a perceived
anomalous health incident, triage based on credible medical findings that led to an NIH
referral, a willingness to participate in the study, and a determination by NIH investigators
that there was no obvious alternative cause for those findings. Each of these selection criteria
1
There are ˜9,000 State Department employees overseas. The study sponsor estimates that ˜20% of non-
military overseas employees are from agencies other than the State Department, giving a combined total of
˜11,000 non-military overseas U.S. personnel.
24
JSR-21-01 DoS Embassy Incidents February 11, 2022
has a subjective component and shaped the selected cohort in ways that may influence the
prevalence of specific medical findings, potentially making it more difficult to identify a
causal mechanism underlying the reported incidents.
NIH enlisted a variety of control groups in an attempt to reach different comparison
populations. Foreign service members posted in Cuba during the first set of events (n=12)
were the most lifestyle-matched, followed by foreign service members who did not serve
in Cuba (n=15). Because enrolling active-duty embassy personnel in a long-term medical
study poses significant logistical challenges, these control groups are post-duty and thus
understandably small. NIH also enrolled other, larger control populations in their study
to increase the size of the control groups, albeit not lifestyle-matched, consisting of local
NIH volunteers (n=40) and TBI patients in a different longitudinal study (n=109). These
choices necessarily resulted in heterogeneous control groups. NIH readily acknowledges this
and other limitations in their study, such as the prior treatment some affected personnel
received, recall bias, enrollment delays, and confounding illnesses. Going forward, it would
be beneficial to use the uniform triage process to select affected individuals for inclusion
in future medical studies, to carry out a specific series of tests all done in the same order
and within a specified time frame, and to put even more effort into assembling larger and
appropriately selected control groups. (Recommendation 4)
Based on the aggregated medical data provided, JASON does not find any evidence
of a novel medical syndrome that would go beyond the expectation of intercurrent illnesses
in the general population. (Finding 3) In the future, the researchers may be able to
take advantage of the individual incident reports and patient medical records to design
improved selection criteria that provide evidence for such a novel syndrome, for example,
based on the detection of rare clusters of symptoms not normally found to occur in the
general population. (Recommendation 5)
Reviewing the entire cohort of affected individuals, JASON finds no strong evidence
of TBI in the data obtained. JASON was not given access to the results of the patients’
MRI studies, either conventional or diffusion tensor, but was told that up to that date, the
studies showed no abnormal findings that would support the hypothesis of TBI. Data on
25
JSR-21-01 DoS Embassy Incidents February 11, 2022
two serum biomarkers, neurofilament light chain (NF-L) and glial fibrillary acidic protein
(GFAP), showed mean elevated levels in the affected individuals, but these values were not
significantly different from those of the tested control individuals. There are (at least) two
credible explanations: either there was no adverse exposure causing TBI, or the signal of
TBI is drowned out by a significant background of unharmed individuals.
This situation emphasizes the importance of how the selection criteria are applied in
defining the “treatment” group. We further note that the measured serum biomarkers are
not unique to TBI. In particular, elevated NF-L could result from damage to peripheral
nerve axons as a consequence of various causes, including inflammatory or demyelinating
neuropathies, autoimmune axonopathies, chemotherapy-induced toxicity, and crush injury
(Mariotto et al., 2018; Meregalli et al., 2020). Because peripheral nerve damage is a com-
ponent of some of the hypothesized mechanisms of harm described in this report, further
analysis should be done to differentiate between central and peripheral sources of elevated
serum NF-L. We raise this concern because the clinical focus on TBI may be misdirected.
As explained in section 6.5, it is very difficult to deliver harmful quantities of energy to the
central nervous system at a distance, making TBI implausible.
Just before completion of the JASON study, the NIH provided us with aggregated
medical data for a subgroup of 6 individuals from their study, together with 12 control
cases selected by the Department of State that were matched for age and gender. The 6
members of the subgroup were selected from a larger group of 29 individuals that JASON
identified as having had similar anomalous experiences, based on our review of the nearly
200 case reports and without reference to the medical data. The medical data were provided
for only 6 of the 29 requested cases because only those individuals had given prior consent
for data sharing. JASON has asked that efforts be made to obtain consent from the other
23 identified cases.
Given the small number of cases in the subgroup, it is difficult to draw firm conclu-
sions. In summary, however, there were no significant differences for any of the measures of
vestibular function in the subgroup compared to the control group (Shahim et al., 2021).
These studies included both cervical and optical vestibular myogenic evoked potentials,
26
JSR-21-01 DoS Embassy Incidents February 11, 2022
which are objective measures of neuronal function of the saccule and utricle, respectively
(the two otolith organs of the inner ear). There also were no significant differences between
the two groups with regard to rotational measures (gain, phase, and symmetry), and to
more subjective tests of oculomotor function, including: optokinetic nystagmus response,
smooth pursuit, vergence pursuit, fixation and spontaneous square waves jerks, horizontal
saccades, anti-saccades, self-paced saccades, predictive saccades, motor reaction saccades,
audio-visual reaction times, and pupillary reflex.
There were statistically significant differences between the subgroup and controls with
regard to some aspects of their general neurological exam, with the selected subgroup of 6
individuals exhibiting higher depression scores and higher scores on both an inventory of
medical symptoms and a general health questionnaire. With regard to serum biomarkers,
the mean level of both NF-L and GFAP was higher in the subgroup compared to the
controls, although not reaching overall statistical significance. This is because there was
much more scatter in the subgroup, including one outlier who scored markedly above the
mean on both measures.
4.1.1 Hypothesis-testing on combined datasets
By combining the medical data with the incident reports pertaining to circumstances at the
time of onset of symptoms, there is the potential to improve significantly the assessment of
alternative causal hypotheses. A selection criterion recommended by the Centers for Disease
Control and Prevention, which is already in use by some agencies (although perhaps not
applied uniformly) requires a biphasic onset of symptoms. Specifically, the CDC’s criteria
are:
1. An initial phase that includes at least one of the following:
• Head pressure
• Disorientation
• Nausea
• Headache
27
JSR-21-01 DoS Embassy Incidents February 11, 2022
• Vestibular disturbances
• Auditory symptoms
• Vision changes
2. A secondary phase that includes at least one of the following, with no alternative
explanation:
• Vestibular disturbances
• Cognitive deficits
These criteria, subsets of these criteria, and extrapolations involving additional situa-
tional data could be used to test alternative hypotheses. The NIH study found evidence of
vestibular dysfunction among a significant fraction of the study cohort.
2
However, incidents
of vestibular dysfunction are very common in the general population, affecting 15–20% of
adults on an annual basis, with vestibular vertigo accounting for about a quarter of these
cases (Neuhauser, 2016). Thus the finding of vestibular dysfunction alone could result in
individuals meeting the study selection criteria who have a condition that might be better
attributed to other causes, such as benign paroxysmal positional vertigo, vestibular neuritis,
or vestibular migraine, thus diluting the ability to identify a novel syndrome. However, if
paired with other selection criteria, for example, auditory symptoms or a sensation of ear
pressure, the selection process might identify a cohort of individuals who have a similar
underlying cause for their symptoms. (Recommendation 5)
4.1.2 Example of vestibular migraine with auditory symptoms
To illustrate the usefulness of applying more refined study selection criteria, consider vestibu-
lar migraine as a potential explanation for the symptoms experienced by the affected indi-
viduals. Vestibular migraine is a recurring type of migraine that produces dizziness, nausea,
and vestibular disturbances (Lempert et al., 2012). Clinical studies have shown that the risk
of cognitive dysfunction is increased in vestibular migraine (Rizk et al., 2020). It is notable
2
Based on abnormal test scores for self-paced saccades, motor reaction saccades, and audio-visual reaction
times, and an attenuated amplitude of vestibular evoked myogenic potentials.
28
JSR-21-01 DoS Embassy Incidents February 11, 2022
that about half of vestibular migraine patients report abnormal sounds and/or a sensation
of ear pressure. This condition would be a potential explanation for the biphasic pattern
of sound or pressure, followed by vestibular disturbances, in the reported health incidents.
One can then ask, what is the probability that the selection criteria have simply picked out
naturally-occurring cases of vestibular migraine with auditory symptoms within the pop-
ulation of overseas U.S. personnel? Could this affliction account for the approximately 25
affected personnel who, in our estimation, are showing the broader and more severe range
of symptoms that evade an immediate alternate explanation based on the incidents?
The lifetime prevalence of all types of vestibular migraine is estimated to be 0.01 (95%
CI, 0.0070–0.0137) and of those, the fraction producing auditory symptoms or ear pressure
is estimated to be 0.45 ± 0.16 (Neff et al., 2012). The onset of vestibular migraine can occur
at any age, with a mean age of onset around 40 years (37 years for females and 42 years
for males; Hilton & Shermetaro 2021). Migraines typically predate the onset of vertiginous
symptoms by an average of 8 years and are more common among women by a ratio of
approximately 5:1.
One might suspect that a vestibular migraine subject, having not yet experienced such
a migraine, would not understand what is happening to them and report their first episode
as a health incident. Age-stratified data for first onset are not available in the literature, but
we can still make a useful estimate of the probability of this scenario under the assumption
that the onset is uniformly distributed between ages 20 and 50. If the lifetime prevalence of
any vestibular migraine is 0.01, then the probability of developing the migraine at any given
year between the ages of 20 and 50 would be 0.01/30=0.00032. About half of those migraines
will have an auditory component, roughly consistent with the prevalence of sound in the
reports. For the entire population of n = 11, 000 non-military overseas U.S. personnel,
we then expect 3.5 such events per year. While this appears insufficient to explain all
of the approximately 200 reported events, it could be plausible for the selected group of
approximately 25 cases since 2016, for which the rate would be 25/5 = 5/year.
29
JSR-21-01 DoS Embassy Incidents February 11, 2022
This is a significant number that, on the surface, could make up a large fraction of the
health incidents. However, the above calculation does not account for the clustering of events
in geographical location (several appearing in one post) and in time (several appearing
within a year instead of over several years). For a 100-person post (such as Havana),
the expected number of first-time vestibular migraines would instead be 0.00032 × 100 =
0.03/year, which could not explain the appearance of several similar health incidents within
one year.
We can ask a similar question using different assumptions about human behavior.
Instead of requiring that vestibular migraine be a first-time event, we can relax that re-
quirement and consider that the fear of an attack by an adversary may cause any migraine
subject to report such a migraine as a health incident, even if they had experienced one
before. In this case, we need to know that the 1-year prevalence of vestibular migraine is
0.009 (95% CI, 0.0062–0.0127), because those having vestibular migraines tend to experi-
ence episodes about once a year (Neuhauser, 2016). The expected number of vestibular
migraines for the total embassy population is then 0.009 × 11, 000 = 99/year, whereas for a
single 100-person post would be 0.009 × 100 ≈ 1/year.
From the foregoing, we can conclude that the occurrence of vestibular migraine in the
general population does not, by itself, explain the cluster of symptoms reported by U.S.
overseas personnel. One must carry out a similar analysis for other potential explanations
for various constellations of signs and symptoms among the affected individuals to determine
if their occurrence deviates significantly from what might be expected for the age-matched
general population.
4.1.3 The look-elsewhere effect
It is important to keep in mind that there may be several different clinical conditions that
underlie the reported symptoms, each of which is far too uncommon to provide a unifying
explanation, but that collectively account for a significant fraction of the cases. For any
given uncommon condition, one might conclude from the small probability of occurrence
that it cannot possibly explain the observed cluster of reported symptoms. However, such
30
JSR-21-01 DoS Embassy Incidents February 11, 2022
conclusions fail to take into account what is known as the “look-elsewhere effect” (Lyons,
2008), also known in statistics as the multiple comparisons problem or the multiple hypoth-
esis testing problem. The look-elsewhere effect refers to the overestimation of significance of
an observation due to failure to search the entire relevant space of parameters. Even if the
probability of occurrence of a particular medical condition among the U.S. overseas popu-
lation is very small, the probability of a few people simultaneously experiencing symptoms
due to some type of unusual condition may not be rare at all.
As discussed in section 4.1.1, most of the symptoms experienced by the affected per-
sonnel are very common among the general population across many illnesses. However, the
notion that these symptoms may have a more sinister origin can be socially propagated,
especially in a stressful environment. Hence, a very small cluster of individuals who expe-
rience these symptoms due to an uncommon medical condition could form the seed of a
socially-amplified cluster of patients who report similar symptoms. Thus, to determine the
probability of a true syndromic cluster occurring at a particular post, one must integrate
over random coincidences of all possible conditions, however rare, that might provide the
seed for social propagation of similar reported symptoms.
With regard to the many reported anomalous health incidents, it is difficult to deter-
mine the appropriate search space that must be considered when attempting to carry out
a quantitative evaluation of the look-elsewhere effect. It should nevertheless be understood
that the probability of occurrence of an odd symptom cluster of some kind greatly ex-
ceeds the probability computed for the incidence of a cluster of any specific illness that has
been identified after the fact. An alternative possibility is that there are one or more com-
mon conditions that explain the majority of cases, together with a collection of uncommon
conditions that account for the remainder.
4.2 Differentiating Organic, Functional, and Psychogenic Conditions
It is important to consider and evaluate multiple types and causes of illness with regard to
the anomalous health incidents. In our evaluation, JASON considered the potential role of
the full range of illnesses including organic disease, functional disorders, and psychogenic
31
JSR-21-01 DoS Embassy Incidents February 11, 2022
illness. Organic diseases involve structural, biological changes. Functional disorders have
physical origins, but involve altered biological function without known structural changes.
Psychogenic conditions do not have a precipitating physical cause. It is helpful to identify
the contributions of these different components in an attempt to differentiate between the
following possibilities:
• Direct physical and physiological harm caused by energy delivered as a result of a
physical attack;
• Functional disorders that arise as a consequence of physical attack;
• Functional disorders that have other precipitating events, including inadvertent phys-
ical insults, medical conditions unrelated to any external insult, and events that can
be considered ordinary but have an amplified effect due to prevailing circumstances;
• Psychogenic disorders that do not have a precipitating physical cause.
All medical conditions have a real, psychological component. Despite this, functional
disorders and psychogenic illness have different etiologies. In persistent postural-perceptual
dizziness (PPPD) and other functional disorders, there is a strong psychological component,
but this is not meant to imply that these individuals have psychopathological abnormalities.
Rather, the psychological (and accompanying psychosocial) factors should be regarded as
amplifiers that cause an acute precipitating event to evolve into a functional disorder. Such
a dynamic process is thought to underlie PPPD (Figure 4.1). Individuals who by nature
or circumstances are in a state of anxiety and heightened body vigilance are more likely to
convert an acute precipitating event to a functional condition (Staab & Ruckenstein, 2007).
This is because such individuals have heightened sympathetic activation, which provides
a physiologic basis for their stress and anxiety. We note that such a condition could be
triggered by adversarial action without the use of attacks that produce direct physical
harm.
32
JSR-21-01 DoS Embassy Incidents February 11, 2022
Psychological
Functional
(PPPD)
Structural
Precipitants
• vestibular
• other medical
Predisposition
• anxiety
• personality traits
Precipitants
• psychological distress
Comorbidity
• depression
• phobia
Figure 4.1: Putative mechanism of PPPD, involving a dynamic process that begins with
a precipitating structural condition that is amplified by psychological factors to result in a
functional disorder (based on Dieterich et al. (2016).)
4.2.1 Functional Disorders: Example of persistent postural-perceptual dizzi-
ness (PPPD)
One of the most common vestibular disorders is persistent postural-perceptual dizziness
(PPPD), which occurs in 15–20% of patients who present for evaluation of vestibular symp-
toms (Dieterich et al., 2016; Staab et al., 2017). It is difficult to estimate the prevalence of
PPPD in the general population, but the annual incidence of dizziness (including vertigo)
in the U.S. adult population is 15–20%, with vestibular vertigo accounting for more than
one-third of these cases (Neuhauser, 2016). This disorder is characterized by symptoms of
dizziness, unsteadiness, and non-spinning vertigo that are exacerbated by motion or expo-
sure to complex or moving visual patterns (Staab et al., 2017). Patients often describe a
feeling of fuzziness or lightness in the head, disorientation, or lack of visual focus. PPPD
is triggered by a variety of conditions that cause dizziness, unsteadiness, or problems with
balance, include acute or episodic vestibular syndromes, especially when coupled with psy-
chological distress. The symptoms of PPPD may fluctuate, but are present on most days
and can persist for months or longer.
Several features of PPPD match the secondary phase of the CDC criteria for Anoma-
lous Health Incidents, but this disorder does not explain the triggering condition(s) that
would have given rise to symptoms in the initial phase of the syndrome. One of the precip-
33
JSR-21-01 DoS Embassy Incidents February 11, 2022
itating causes of PPPD is vestibular migraine, but PPPD occurs more commonly following
other acute conditions, such as benign paroxysmal positional vertigo or vestibular neuri-
tis, and in some patients the precipitating cause is never identified. Given the common
occurrence of these precipitating conditions and the high prevalence of PPPD in the gen-
eral population, it would be difficult to rule out this condition as an explanation for the
secondary phase of symptoms among at least some of the affected U.S. overseas personnel.
PPPD is regarded as a chronic functional disorder, which refers to altered biologi-
cal function in the absence of known structural changes at the level of cells or tissues.
“Functional” is not a synonym for “psychogenic” or “psychosomatic”, but rather indicates
a pathophysiological state that has no discernable molecular or anatomical correlate. With
the progression of medical knowledge, some disorders that previously were described as func-
tional have come to be recognized as having a structural basis (e.g., fibromyalgia). Other
disorders may have a strong functional component that overlays an organic component (e.g.,
chronic fatigue syndrome). PPPD is considered to be a pure functional disorder. It is not
a diagnosis that is made following the exclusion of all others, but it does require one to rule
out other disorders that might better account for the symptoms, and does not preclude the
existence of an underlying trigger that has a more clearly defined pathophysiological basis
(Staab et al., 2017).
There is a subset of affected U.S. overseas personnel who might be described as having
PPPD, a condition that would be expected to occur in a few percent of the overall popu-
lation, irrespective of any link to some postulated “attack”. Clearly, not all of the affected
individuals meet the diagnostic criteria for PPPD as being the cause of the secondary phase
of their condition. Yet, it is important to recognize that individuals who are under stressful
conditions are especially sensitive to environmental sensory triggers, which can help to elicit
PPPD and other functional disorders. For these individuals, an initial noxious stimulus may
instigate a feedback loop that involves activation of the stress response pathways, which in
turn heighten anxiety and sensitivity to external stimuli. In addition, persons with certain
underlying medical conditions, especially those that might give rise to the initial phase of
the reported symptoms, are more susceptible to exacerbation of those conditions during
times of stress. (Finding 4)
34
JSR-21-01 DoS Embassy Incidents February 11, 2022
4.2.2 Psychogenic illness
Distinct from a functional disorder, there also are psychogenic conditions in which the
medical symptoms do not derive from a precipitating structural event. Psychogenic illness
may occur at the level of an individual, a local cluster of individuals, or within a broader
population. It has been suggested, based solely on publicly accessible information, that the
embassy health incidents are the result of a mass psychogenic illness (Baloh & Bartholomew,
2020). JASON reviewed the literature on psychogenic illnesses, which have been reported
to occur in diverse settings and with diverse associated symptoms. It is difficult to make
quantitative estimates of the prevalence of psychogenic illness, but attempts to do so suggest
that the phenomenon is not rare (Page et al., 2010). One study, for example, estimated the
prevalence of psychogenic, non-epileptic seizures in the general population to be between
1/50,000 and 1/3,000 (Benbadis & Hauser, 2000).
Psychogenic illness occurs when a patient’s psychological concern about a perceived
physical threat or condition produces actual bodily symptoms. There is broad agreement
in the medical literature on the following:
(i) The experience of physical symptoms in patients with psychogenic illness is real, not
faked. The illness can entail objectively observable physiological changes (Jones, 2000;
Bartholomew, 2000).
(ii) Psychogenic illness can occur in any healthy population, without underlying psychi-
atric conditions (Broderick et al., 2011; Sapkota et al., 2020).
(iii) Some stress situations increase the risk of psychogenic illness (Weir, 2005), and the
symptoms may be specific to fears and preoccupations of a given culture (Bartholomew,
2000).
Statements (i) and (ii) are not well understood by the general public or even by many
medical professionals, resulting in a stigma being associated with psychogenic illness. In
the face of such stigma, a somatogenic explanation may seem preferable over a psychogenic
one, even if the latter is better supported by the available data.
35
JSR-21-01 DoS Embassy Incidents February 11, 2022
A psychogenic etiology is a hypothesis to either confirm or rule out based on empirical
data, whether in an individual or a group experiencing similar symptoms. The diagnosis of
psychogenic illness is obtained by exclusion, when no satisfactory alternative structural or
functional explanation can be found, especially when there is a pre-existing social connection
among the affected individuals and there are more cases than would be expected for the
corresponding population in the same period of time. Indicators of a psychogenic illness
include:
• Attribution of symptoms by the affected individuals to a threatening external agent;
• Symptoms that are incompatible with what the individuals believe to be the causal
event;
• Lack of a plausible connection between what is believed to be the causal agent and the
circumstances under which exposure to such an agent might reasonably occur (Page
et al., 2010).
The medical literature tends to focus on local, temporal clusters of mass psychogenic
illness. However, there is also the possibility of more isolated or sporadic cases of psychogenic
illness that are distributed across a broader population, especially if there is some form of
shared knowledge among the affected individuals. In particular, social media has become
a vehicle for amplifying the connection between an individual experience and a collective
syndrome of psychogenic illness (Bartholomew et al., 2012). Furthermore, somatic and
psychogenic illness are not mutually exclusive, and could give rise to the same constellation
of symptoms, amplified at the level of the individual through psychological factors and at
the level of a group of individuals through both psychological and psychosocial factors.
One implication of a hybrid model is that a plausible physical hypothesis need not account
for every symptom experienced across the set of affected individuals. Rather, competing
explanations for the reported symptoms may include physical causes, functional syndromes
derived from physical causes, and purely psychogenic effects.
Given the complexity and diversity of symptoms experienced by the affected U.S.
personnel over the past 5 years, as well as the diverse circumstances associated with the
36
JSR-21-01 DoS Embassy Incidents February 11, 2022
onset of those symptoms, it would be unreasonable to expect all cases to have a common
etiology. The CDC’s diagnostic criteria, which emphasize the biphasic nature of symptoms,
may help to stratify patients who have experienced an acute adverse event from those
who developed a chronic disorder as a consequence of such events. In some cases, the
chronic symptoms of vestibular disturbances and cognitive deficits may derive from the
acute event, in other cases these late symptoms may reflect the conversion of a structural
disorder to a functional disorder, and in still other cases the syndrome may be psychogenic
without any adverse triggering event. For the latter group, it is important to emphasize
that psychogenic illness is common and does not imply that the affected individuals have
a psychiatric disorder. These individuals should be reassured that their symptoms are
real, even though they may derive from acute stress response or misinformation. It also
is important to remind these individuals that information they may have heard regarding
“suspected” causes is not equivalent to a validated explanation for the reported cases (Jones,
2000).
We end this section by noting that in all conditions, whether organic, functional, or
psychogenic disorder, anxiety can worsen and prolong the symptoms in affected individ-
uals. Some level of stress and anxiety is evident from the incident reports that describe
sensations and symptoms when the affected personnel experience an event. Inconsistent,
uncommunicative, or delayed responses in the aftermath of the incident can significantly
worsen the situation. The U.S. Government could aim to reduce such stress by a variety of
means including messaging, education, and rapid follow-up health and environmental mon-
itoring. It could also help ease anxiety to use the rapid questionnaire-based triage process
to let the affected personnel know if and when their incidents or health symptoms are not
consistent with AHI, but still refer them as quickly as possible to appropriate medical care
for whatever other condition they do have. It would be valuable to build awareness that
mechanisms that can deliver energy at a distance and cause brain injury, without accom-
panying damage and sensations on other parts of the body, are extremely limited and to
reassure overseas personnel that there is ongoing monitoring for signals associated with any
such mechanism. (Finding 4, Recommendation 9)
37
JSR-21-01 DoS Embassy Incidents February 11, 2022
4.3 Potential Underlying Causes
Whether both phases of symptoms derive from a structural cause or only the initial phase
does and then gives rise to a functional disorder, it is important to consider potential mech-
anisms that might underlie the initial phase. There are several medical conditions that can
give rise to a syndrome of vestibular disturbances with auditory symptoms, variably accom-
panied by headache, head pressure, nausea, visual disturbances, or disorientation. These
conditions include: benign paroxysmal positional vertigo, vestibular neuritis, labyrinthitis,
vestibular migraine, mal de d´ebarquement syndrome, and M´eni´ere’s disease.
There also are conditions of “central vertigo” that derive from pathological brain pro-
cesses, such as stroke, cerebral tumor, or multiple sclerosis (Vanni et al., 2017). In addition,
there are many conditions that can cause “dizziness”, which lacks the sensation of self-
motion of true vertigo, but may also be associated with headache, nausea, lightheadedness,
and disorientation, symptoms that have commonly been reported by the affected U.S. per-
sonnel. Dizziness can result from many causes, including anemia, hypoglycemia, electrolyte
imbalance, cardiac arrythmia, and orthostatic hypotension, all of which are common med-
ical conditions. Collectively, the conditions that give rise to peripheral vertigo, central
vertigo, and dizziness could explain the initial phase of symptoms in many, if not most, of
the affected personnel. Clearly, however, no single condition can explain all of the reported
cases. Conversely, there may be a subset of cases that cannot be explained by any of the
currently recognized medical conditions.
In principle, there are several afferent neural pathways that could result in acute
vestibular disturbances that may be accompanied by disruption of auditory, oculomotor,
cerebellar, or cognitive function. In considering the possibility of extrinsic causes, we will
focus on pathways that might be activated by external stimuli, excluding injury due to
trauma or direct physical contact. We also exclude obvious external stimuli that do not
match the case reports, such as damaging sounds, intense IR, visible, or UV light, or
exposure to ionizing radiation. Within the electromagnetic spectrum, frequencies above
10–20 GHz cannot penetrate human skin by more than 1 mm (see section 6.3.1), and thus
could not affect afferent nerves that lie below the dermal layer, which has a thickness of 1–4
38
JSR-21-01 DoS Embassy Incidents February 11, 2022
semi-circular
canals
vestibular
nerve
cerebellum
oculomotor
nuclei
sensory nerves of
head, neck, trunk
motor nerves of
spinal cord
vestibular
nuclei
otolith
organs
Figure 4.2: Neural pathways of the vestibular system. Sensory input from the otolith
organs (saccule and utricle) and the semi-circular canals is carried by the vestibular nerve
to the vestibular nuclei, where it is integrated with positional sensory input from the body,
coordinated by the cerebellum. Output from the vestibular nuclei to the cerebellum, motor
nerves of the spinal cord, and oculomotor nuclei of the brainstem help to control balance,
posture, and eye movement, respectively.
mm. Frequencies below 1–2 GHz would require a very large emitter and could not be tightly
focused to provide the type of localized effects that have been reported (see section 6.4).
However, frequencies in the range of 1–20 GHz might conceivably stimulate or modulate
the firing of peripheral nerves, which in turn might excite central pathways that result in
acute vestibular disturbances with auditory symptoms.
The vestibulocochlear nerve (cranial nerve VIII), mediates sensations of movement
(vestibular portion) and sound (cochlear portion). The vestibular nerve carries sensory
information to the vestibular nuclei of the brainstem, with most of that input coming from
the otolith organs and semicircular canals of the inner ear, which sense linear and angular
acceleration, respectively (Figure 4.2). The vestibular nuclei also receive input from the
cerebellum, which helps to coordinate movement and maintain balance, and from sensory
pathways in the head, neck, and trunk that provide information about spatial positioning.
The primary outputs from the vestibular nuclei are to the cerebellum (feedback circuits for
movement and balance), motor pathways of the spinal cord (control circuits for postural
equilibrium), and oculomotor nuclei of the brainstem (control circuits for eye movement).
39
JSR-21-01 DoS Embassy Incidents February 11, 2022
There are pathways that link stimuli near the skin directly or indirectly to the vestibu-
lar circuits. For example, the trigeminal nerve (cranial nerve V) has sensory branches orig-
inating in the upper (ophthalmic) portion of the face that project to the vestibular organs.
Stimulation of these trigeminal branches can result in increased blood flow to the vestibular
organs, with both short-term effects on vestibular function and potential longer-term effects
of vestibular hypersensitivity. Vestibular migraine, for example, is thought to be mediated
through these pathways (Baloh, 1997; Vass et al., 2001).
Direct electrical stimulation of the skin of the forehead has been shown to elicit spon-
taneous nystagmus (repetitive involuntary eye movements), especially in patients who have
a history of migraine (Marano et al., 2005). In contrast, no nystagmus is seen following
stimulation of other nerves, such as the medial nerve at the wrist. The character of the
nystagmus indicates that the effect of trigeminal nerve stimulation is stimulatory to the
vestibular organs on the same side of the body. Of course, direct electrical stimulation of
the trigeminal nerve is not being considered with regard to the affected U.S. personnel.
However, it has also been shown that nystagmus can be elicited by gentle vibratory stim-
ulation to the forehead (Sheth et al., 1995), suggesting that there may be more indirect
means of stimulating the trigeminal nerve that give rise to vestibular disturbances.
Another afferent pathway that links to the vestibular circuits involves sensory inputs
carried by the vagus nerve (cranial nerve X), with some of those inputs originating within
a centimeter of the skin surface. Apropos of its name, which means “wandering”, the
vagus carries sensory input from many locations of the body, including the skin, heart,
respiratory tract, digestive tract, and other internal organs. It also controls motor function
of the mouth and throat, as well as involuntary (parasympathetic autonomic) function of
the heart and digestive tract. Of particular interest here is the input the vagus receives
from the skin behind the ear, the external portion of the ear canal, and the meningeal
surface of the brain. The vagus can also be stimulated indirectly by massaging the carotid
sinus (located just below the jawline), splashing cold water on the face (the dive reflex), or
applying modest pressure to the orbits of the eyes (the oculocardiac reflex) (Ceylan et al.,
2019). These maneuvers are sometimes used to slow abnormal heart rhythms, and non-
invasive stimulation of the vagus nerve has been used to treat both PPPD and vestibular
40
JSR-21-01 DoS Embassy Incidents February 11, 2022
migraine (Eren et al., 2018; Beh & Friedman, 2019). As with the trigeminal system, surficial
stimulation of the vagus nerve might alter vestibular function, either increasing or decreasing
the symptoms of vertigo. For the vagus, the situation is more complicated because its
stimulation also results in reduced heart rate and blood pressure, which can cause symptoms
of dizziness and lightheadedness that are often reported as vertigo.
The above discussion notwithstanding, it remains difficult to link a plausible mode
of external nerve stimulation to the constellation of symptoms seen in the initial phase of
the syndrome experienced by affected U.S. overseas personnel. Stimulation by electromag-
netic radiation sufficient to cause structural damage would require delivering such a large
amount of energy that it would be accompanied by a sensation of heat and/or pain. It is
possible, as discussed above, that many of the affected individuals are experiencing dizzi-
ness rather than true vertigo during the initial phase of symptoms. It would be easier to
bring about dizziness, rather than vertigo, through external stimuli. Finally, there may be
some unknown neural pathway that links external stimuli to the reported symptoms that
lies beyond the current body of medical knowledge. Human neurophysiology is remarkably
complicated and is intertwined with other physiological, cognitive, and emotional factors,
compounded by the variability of the individual human experience.
4.4 Clinical Context of the Affected Individuals
The variability of human experience may provide an opportunity to identify causal relation-
ships between an individual’s reported symptoms and an underlying medical condition. As
noted in the earlier subsections, there are many potential causes of symptoms of headache,
nausea, dizziness/vertigo, lightheadedness, disorientation, strange sounds, and visual dis-
turbances. An individual who reports these symptoms may be given medication to try to
alleviate them, or may already be taking medication for a different condition. In either case,
the response to those medications could shed light on the underlying cause of symptoms
that were associated with an anomalous health incident.
For example, individuals who suffer migraines, which have a prevalence of approxi-
mately 12% in the general population (Yeh et al., 2018), may be treated with analgesics,
41
JSR-21-01 DoS Embassy Incidents February 11, 2022
ergotamines, triptans, beta blockers, tricyclic antidepressants, or calcitonin gene-related
peptide (CGRP) antagonists. In addition to their potential benefit for migraine, these med-
ications have relevance to some of the causal hypotheses discussed above. A beta blocker
decreases sympathetic tone and may reduce the anxiety that contributes to a functional dis-
order such as PPPD. Conversely, ergotamine increases sympathetic tone and can increase
anxiety, at least in the short term. A CGRP antagonist directly affects the physiological
pathways that underlie vestibular migraine.
A similar case can be made for many other medications, including: gabapentin, which
is used to treat neuropathic pain, but can precipitate depression and other mood disorders;
beta adrenergic agonists, which are used to treat asthma, but increase sympathetic tone
and can increase anxiety; benzodiazepines, which reduce anxiety and therefore may ben-
efit someone with a functional disorder; and carbamazepine, an anti-convulsant that also
modulates conduction in the trigeminal nerve. This list goes on.
Every individual has a story to tell and in effect constitutes an “n-of-1” clinical trial
that relates their medication history to their past and present medical condition. It appears
highly unlikely that all of the nearly 200 reported cases of anomalous health incidents
derive from a single, or even a small number, of medical conditions. By individuating
rather than aggregating the data associated with those cases, it may be possible to identify
subgroups that meet the diagnostic criteria for a known medical condition. Collectively,
these subgroups may account for the majority of cases, without the need to invoke a new
syndrome of unknown etiology.
To aid with future efforts in this direction, JASON recommends adapting the baseline
medical program to capture the full health history of each individual prior to their assign-
ments to posts around the globe. Such evaluations may be beneficial in uncovering the
particular medical conditions that either already affect the individual or that may play a
role in determining their medical response to environmental factors. It would also provide
a more rigorous basis for evaluating whether particular medical conditions can account for
some or all of the reported health incidents. (Recommendation 6)
42
JSR-21-01 DoS Embassy Incidents February 11, 2022
4.5 Understanding and Managing the Role of Social Interactions
The role of social interaction in reinforcing and exacerbating some of the conditions de-
scribed above cannot be overstated, especially in an age of social media and ubiquitous
connectivity. Society’s understanding of social dynamics in the context of social networks
continues to evolve. This topic has a voluminous literature that is well outside the scope
of this report. However, the study of the social dynamics and best leadership practices of
groups of people operating with a common mission in stressful situations has a history that
long predates social media, especially in the U.S. Navy (Allingham, 1964; McGlynn, 2005;
Roos, 1943). We cite two illustrative examples:
• A 1956 study mapped and examined the social networks aboard a Navy ship and
their role in the spread of rumors aboard ship. Examining the failure of artificially
seeded rumors to take hold during the experiments, the author concluded: “Here, it
now seems, lies the only suitable explanation of the test rumor’s limited ‘success’. . .
something in the makeup of the community basically prevented its taking hold. . . Of
major importance would seem the active and consistent program of information and
education pursued by the commanding officer. The captain meticulously informs the
crew of all matters, which affect their personal planning, as they arise.” (Allingham,
1964)
• A more recent dissertation in 2005 looked at best practices when decommissioning
Navy ships and found the existence of rumors and the effect of their spread to be
one of the major themes presented as leadership challenges and discussed at length,
where “...extraordinary efforts to communicate also appear to be important in naval
contexts where rumors can become rampant.” (McGlynn, 2005)
A thorough study of the best practices concerning messaging, training, guidance, and other
measures, with an eye towards reducing anxiety and building resilience in an open and
transparent way, appears warranted, especially given that adversarial attacks via social
networks are well documented (Prier, 2017; West, 2016; Singer & Brooking, 2018). (Rec-
ommendation 9)
43
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
44
JSR-21-01 DoS Embassy Incidents February 11, 2022
5 ACOUSTIC ATTACK HYPOTHESES
In this section, we consider acoustic-energy projection as a possible covert mechanism for
causing harm to individuals. Acoustic-energy projection includes high-energy directed infra-
sound, audible sound waves, and ultrasound, which are sound waves in different frequency
ranges and, as a result, have different interactions with the human body and varying pene-
tration properties into buildings.
In our assessment of sound waves of these different frequency ranges, we follow the
criteria we devised from the incident and health reports, as discussed in Section 2: energy
must be able to penetrate into buildings through walls and windows from a distance of
several tens of meters, be localized to areas that the size of a room or smaller, be gener-
ated with equipment that is mobile and covert, lead to audible and/or recordable sounds,
cause immediate symptoms of pressure, vertigo, headache, or nausea, and cause long-term
vestibular and/or cognitive dysfunction. To that end, we identify the intensity levels, in
decibels (dB), that are reported to be linked to health effects and compare these intensities
with what may be achieved with delivery mechanisms in realistic operational settings.
5.1 Introduction and Background
Infrasound (IS), is generally described as acoustic waves in the frequency range
5 Hz . ν
is
. 20 Hz , (5-1)
sometimes regarded as extending somewhat below 5 Hz or somewhat above 20 Hz, i.e., a
range of acoustic frequencies below the audio spectrum typically perceived by the human
ear. In particular and in the present context, infrasound is reported to display, “a special
capacity to affect human health and adaptation because its frequencies converge with those
generated by the human body” (Persinger, 2014).
Acoustic excitation in the audio range, i.e.,
20 Hz . ν
au
. 20 kHz , (5-2)
45
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 5.1: Sound-frequency regimes.
brackets human audio perception, with actual low- and high-frequency limits and minimum
intensity levels (dB) that can be consciously perceived that vary across individuals and, for
any one individual, will also depend on age.
Ultrasound typically refers to acoustic excitation overlapping and extending beyond
the high-frequency end of the audible spectrum. Very high frequency sound (VHFS) con-
ventionally refers to the range of
11.2 kHz ≤ ν
vhfs
≤ 17.8 kHz , (5-3a)
and ultrasound (US) to
ν
us
> 17.8 kHz , (5-3b)
as noted by Fletcher et al. (2018), even though “ultrasound” sometimes refers to the two
concatenated frequency regimes in equations (5-3).
Even higher sound frequencies are used for diagnostic and other imaging applications,
as indicated in Fig. 5.1.
We emphasize that humans may be aware of exposure to sound, i.e., acoustic excita-
tion, both below and above audio/audible frequencies, especially of intense sound. People
may sense the presence of infrasound even though they may not be able to attribute such a
sensation to sound, sometimes perceived and described as a whole-body experience. They
may also be able to perceive the presence of ultrasound, as employed as an insect or rodent
repellent, for example, even though they may not be able to hear those frequencies, as we’ll
discuss below.
46
JSR-21-01 DoS Embassy Incidents February 11, 2022
Land animals may be able to hear both infrasound, especially large animals such as
elephants, or ultrasound, such as rodents, dogs, and cats. Dogs are reported to be sensitive
to frequencies as high as 45 kHz, while cats are sensitive to frequencies as high as 65 kHz,
and can serve as sentinels alerting humans to the presence of such frequencies. Ultrasonic
dog whistles (23–45 kHz) that humans may perceive as quiet hissing, exploit dogs’ sensitivity
to such high-frequency sounds.
Appendix B provides a review of basic concepts as well as estimates of sound atten-
uation and the limits of focusing sound waves. These calculations support the discussion
below.
5.2 Infrasound and Reported Effects
Sound waves interact with the human body in several ways. We first note that there is
a large mismatch of acoustic-impedance, Z
ac
, between air and water, and hence between
air and the human tissue/body as a whole. Nevertheless, special transduction by human
ears allow very low pressure oscillations to be detected. Further, the air-body interface
cannot be considered to be rigid at lower frequencies, e.g., for ν < 1 kHz, where tissue
elasticity and viscosity becomes important (e.g., Fry, 1952) and leads to a non-negligible
absorption of acoustic energy by the human body. In very intense sound fields, wide-spread
stimulation of somatic mechanoreceptors can occur (e.g., Stephens & Bate, 1966). Below
100 Hz, corresponding to sound wavelengths in air, λ
air
> 3.4 m, the body responds as a
whole (e.g., Jauchem & Cook, 2007).
Humans routinely interact with a variety of infrasound sources in their environment.
These waves are generated by means ranging from natural sources, such as wind turbulence,
to man-made sources, such as helicopters and ceiling fans at their blade-passing frequency,
road noise, buffeting in cars with only one window (partially) open, Helmholtz resonances
in building HVAC-systems, and more recently, from wind turbines.
47
JSR-21-01 DoS Embassy Incidents February 11, 2022
Owing to the long wavelengths of infrasound waves, it is not possible to focus infrasonic
acoustic energy (and the resultant pressure excitations) in free space at length scales relevant
to this report (i.e., meters or tens of meters). However, human sensation, perception,
and detection are not entirely linear processes. For example, it is possible to generate a
localized sensation at a beat frequency ν between two frequencies ν
1
and ν
2
= ν
1
+ ν,
where the frequency ν
1
might be short-wavelength ultrasound (which can be localized) and
frequency ν may be in the infrasound range (which cannot be localized). Similarly, non-
linear transduction of an ultrasonic acoustic wave at a carrier frequency of ν
1
, modulated at
an infrasound frequency ν, will generate harmonics at ν
1
±ν, with beating that will generate
excitation and a possible sensation at both ν and 2ν. In both cases, the sensation of the
infrasound frequency at a frequency ν, can be localized as dictated by the generation and
transmission of sound at an ultrasonic frequency, ν
1
. Infrasound can also be delivered in
interior spaces from/via HVAC furnace combustion chambers, ducts, and plena, localizing
excitation, to some extent, creating standing-wave resonance modes in sufficiently large
rooms.
Below audible frequencies, the sensation of sound from vibratory stimulation within
the ears or by the whole body can appear similar. There is a large variance in sensation,
however, attributable to individual differences in perception by humans in response to the
“same” physical stimulus (Persinger, 2014).
A number of studies report variable and mostly mild effects of infrasound waves on
human health.
Bolin et al. (2011) report a statistically significant association between noise levels
and self-reported sleep disturbance in two of three studies. They note that low-frequency
noise (LFN) from wind turbines may cause other health problems, but acknowledge that
empirical support for these claims is lacking.
Kasprzak (2013) reports on influences by infrasound on a cohort of 17 men and 18
women, aged 19-23, who declared they had no prior medical conditions and were not under
the influence of medicines. In the research, the cohort was exposed to infrasound at a
frequency of ν = 13 Hz and a sound pressure level of SPL = 109.1 dB(LIN), 62.1 dB(A),
48
JSR-21-01 DoS Embassy Incidents February 11, 2022
112.9 dB(G),
1
for an experiment duration of 35 min, with an exposure lasting 20 min, while
recording EEG signals. This study reported statistically significant (p < 0.01) effects on
the sensory motor rhythm (SMR) brain wave rhythm in the 12 − 15 Hz frequency range,
but statistically insignificant effects on other brain-wave rhythms.
Modern wind turbines can generate significant infrasound intensities and harmonics.
A three blade wind turbine at 4 rpm will generate a frequency ν = 0.2 Hz and a relative
pressure level at about 100 m distance about 50 dB outside a brick house and only slightly
less inside the house. The frequency range between 1 and 4 Hz, the delta range for brain
waves (EEG measures), is the one involved with slow wave (deep sleep). Disruptions to
this sleep can affect the synchronized release of hormones and proteins that facilitate tissue
repair and normal homeostasis (Persinger, 2014).
Considering wind-turbine infrasound noise, Berger et al. (2015) find no evidence that
exposure to such G-weighted infrasound levels can directly impact human health, attributing
self-reporting effects to subjective variables. Baliatsas et al. (2016) in their review paper on
low-frequency noise related to wind farms, reported “some associations . . . between expo-
sure to LFN and annoyance, sleep-related problems, concentration difficulties and headache
in the adult population living in the vicinity of a range of LFN sources. However, evidence,
especially in relation to chronic medical conditions, was very limited. The estimated pooled
prevalence of high subjective annoyance attributed to LFN was about 10%.”
Tonin et al. (2016) also considered low-frequency noise related to wind farms and report
that, “at least for the short-term exposure times conducted here-in, that the simulated
infrasound has no statistically significant effect on the symptoms reported by volunteers, but
the prior concern volunteers had about the effect of infrasound has a statistically significant
influence on the symptoms reported. This supports the nocebo effect hypothesis.”
A longitudinal, randomized experimental pilot study by Ascone et al. (2021) to inves-
tigate the effects of infrasound on “human mental health, cognition, and brain structure”
concluded that infrasound (6 Hz, 80–90 dB) applied in the bedroom for 28 consecutive
1
Units of dB(A)/dBA, dBB, dBG, etc., refer to different spectral-weight filters applied to acoustic spectra
to approximate human ears’ response to sound. See discussion, for example, in https://en.wikipedia.org/
wiki/Decibel.
49
JSR-21-01 DoS Embassy Incidents February 11, 2022
nights produced no effect on healthy individuals. They proposed further, fine-grained stud-
ies focused on bodily sensations and perception.
Summarizing the review of infrasound and its effects on humans, available reports
acknowledge complaints of discomfort with some affecting sleep patterns, but do not docu-
ment significant evidence of medical/physiological consequences, especially persistent ones,
often ascribing causes and complaints to subjective annoyance and the like.
5.3 Ultrasound and Reported Effects
As discussed earlier in this section, ultrasound (US) acoustic excitation is at frequencies
above human normal auditory sensory means. Nevertheless, in that frequency regime, non-
linear response of biological tissues can provide a sensation of perception unrelated to hear-
ing and alert to intense ultrasound exposure. In addition, with respect to the interactions
with the human body, inertia by human tissues that cannot react elastically dominates, and
acoustic impedance mismatch considerations apply.
Humans exposed to very-high-frequency sound (VHFS) or US acoustic excitation re-
port a number of effects, including pain in the ear, discomfort, and irritation (Ueda et al.,
2014). There are similar reports by other researchers, with typically insufficient exper-
imental controls. In a more controlled study on the effects of ultrasound on humans,
Fletcher et al. (2018) selected frequencies where individual participants in their cohort
had a SPL = 63 dB hearing threshold level and then set the VHFS/US tone level 25 dB
higher to avoid exceeding the maximum allowable SPL = 92 dB, set to meet allowable daily
noise exposure. The test frequency started at 12 kHz and subsequently increased by a set
pattern to 20 kHz, adjusted for individual hearing thresholds. Tests and effects included
a Conjunctive Continuous Performance Task (CCPT), Galvanic Skin Responses (GSR),
subjective ratings of symptoms. Results are summarized in Fig. 5.2 below.
While Fletcher et al. (2018) report discomfort rating differences between ultrasound
and the 1 kHz reference sound, as summarized above, they do not report lasting effects.
They note, however, that “the sound pressure levels and exposure durations were limited
50
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 5.2: Mean difference in rating between VHFS/US and 1 kHz reference conditions
for all ratings, averaged over four time periods and all participants (32 asymptomatic, 10
symptomatic). Statistically significant (p < 0.05) differences marked with asterisks. From
(Fletcher et al., 2018, Fig. 3).
in this study by the application of the precautionary principle, and the researcher’s duty of
care to participants,” and that, “These findings cannot be used to predict outcomes from
exposures at higher sound pressure levels or longer durations.”
Macc`a et al. (2015) report on ”significant correlations between workers exposed to
audio-frequency noise and tinnitus, sensation of fullness in the ears, and hearing loss [when
exposed to sound with intensities] > 25 dB [referenced to p
0
ref
= 20 µPa (B-5)] at 4 kHz.”
Macc`a et al. (2015) also note adverse effects on hearing and other health indicators among
people who deal with ultrasonic devices in the workplace, such as for cleaning, drilling,
homogenizing, etc., in agreement with earlier studies.
In seeking feasible mechanisms for reported adverse effects of ultrasound, Leighton
(2016) notes that the “subjective symptoms reported (migraine, nausea, tinnitus, headaches,
fatigue, dizziness, feelings of ‘pressure’) are so common and so non-specific that most clini-
cians are unlikely to attribute them to a physical cause.” Leighton (2016) also reviews and
reanalyzes the Macc´a et al. data, agreeing with Macc´a et al. that the data show significant
increases in asthenia (loss or lack of bodily strength) and vertigo. Case histories show that
51
JSR-21-01 DoS Embassy Incidents February 11, 2022
symptoms persist for a few days after exposure and cease subsequently, with no further oc-
currence, in the absence of exposure. The tabulated results are summarized and reproduced
in Figs. 5.3 and 5.4. The overlap with complaints, symptoms, and syndromes reported in
the present context is noteworthy.
Figure 5.3: Occurrence of a range of symptoms related to ultrasound and industrial noise
based on the Macc`a et al. (2015) data, from (Leighton, 2016, Fig. 7).
Conflicting reports and etiologies abound, however, and (Leighton, 2016) notes that
any author raising issues of possible health hazards related to ultrasound “comes at repu-
tational risk,” as many initial reports have later been attributable to psychosomatic causes.
Nevertheless, the evidence suggests that VHF/US exposures elicit greater adverse effects in
some individuals than others, possibly even without the need to hear the signal, or without
the degree of suffering related to the degree of audibility of the VHF/US signal by an indi-
vidual. As a consequence, the usable/acceptable dynamic range between what is perceptible
and what
52
JSR-21-01 DoS Embassy Incidents February 11, 2022
may cause adverse effects may be unexpectedly small (Leighton et al., 2020). It is worth
reiterating that even among individuals who appear to show symptoms related to ultrasound
noise, these symptoms are transient and do not continue well beyond exposure.
Figure 5.4: The p-values from the data of Macc`a et al. (2015) based on cohort sizes of 24
and 56 for the number of subjects exposed to ultrasound and industrial noise, respectively.
Entries with p-values below 5% are in bold and enclosed in parenthesis if they do not have
at least 5 instances for the corresponding symptom. Numbers in braces denote the number
of occurrences for which p < 5% (Leighton, 2016, Table 2).
Closing this part of the discussion, we estimate the characteristics of a device that
could be used to cause adverse health effects. An ultrasonic beam in the tens of kHz
range would experience tolerably low atmospheric attenuation, a beam directivity that can
be achieved at the requisite level, and resonant emitters readily available. We will adopt
40 kHz as a nominal frequency below, but the results also hold for frequencies in the range
of 20 kHz (below which the signal would be in the audible range) to 100 kHz (above which
atmospheric attenuation becomes too large.)
An emitter with a characteristic dimension of D ' 1 m would provide a diffraction-
limited beam-divergence at 40 kHz of ∆θ ' 8.5 mrad. For a standoff distance of x ' 30 m,
53
JSR-21-01 DoS Embassy Incidents February 11, 2022
such an acoustic beam would expand by 0.25 m, which is small enough compared to the
D ' 1 m beam width to treat the beam as focused.
A requisite power level for harm could be estimated based on the occupational exposure
limit set by OSHA for ultrasound at 40 kHz,
2
and is SPL = 115 dB (relative to 20 µPa). For
the sake of this exercise, we will presume that lasting physiological damage from a one-time
exposure requires ultrasound levels akin to that needed at audible frequencies, i.e., 150 dB.
Conversely, we will treat this level as providing an important exposure threshold that, if not
exceeded, effectively rules out ultrasound acoustic-wave energy as an admissible modality
in many, if not most, cases.
Transmitting a 40 kHz beam over a distance of x = 30 m and then through a window
pane introduces 20 dB of attenuation along the path, and another 60 dB through the window
pane, as discussed in Appendix B. Harmful levels would then require a source that can emit
at a Sound Pressure Level of SPL = 150 + 80 = 230 dB. While this may not require a large
amount of propagating acoustic power, only of order 100 mW, coupling power from an
acoustic emitter to air is inefficient. By way of example, the LRAD 500X-RE is a high-
power audio-frequency device with a peak power of 154 dB and marketed to armed forces
for long range transmission of audible sound as a deterrent or weapon.
3
Transferring relevant ultrasonic power levels over a distance of 30 meters and through
a closed window would require a device to broadcast 150 times as much power as is available
from high-end commercial devices. Moreover, the audio power levels outside the building
(including reflections) would be 1000 times higher than inside, hardly covert, and leading
to acute distress on bystanders and operators.
We conclude that remote ultrasonic attacks are, therefore, highly unlikely, on technical
power grounds.
2
https://www.osha.gov/otm/section-3-health-hazards/chapter-5\#appendixc
3
https://genasys.com/lrad_products/lrad-500x/
54
JSR-21-01 DoS Embassy Incidents February 11, 2022
5.4 Concluding Remarks
We considered acoustic-energy projection as a possible covert attack modality. High inten-
sity infrasound and ultrasound emerged as possibilities on the basis of the health effects
reported to be associated with exposure to these sound waves. Among the two, effects in
response to ultrasound were more significant and had a larger overlap with reported syn-
dromes. However, such effects are not reported to persist, at least following exposure at
intensity levels that have been investigated.
In terms of health effects, the challenge of etiology and attribution of biomedical effects
remains, however, complicated by the likely admixture of psychogenic effects, the large
variance in susceptibility between individuals to infrasound and ultrasound, and possible
differences in the mechanisms responsible for exposure. Further and importantly, careful
controlled experiments with human cohorts are understandably limited to exposures deemed
safe for human subjects and, as a consequence, have not probed to explore permanent effects
extending to damage resulting from levels that are, or could be, considered unsafe. In the
case of ultrasound, these are long-established and accepted based on experience with, e.g.,
medical ultrasound imaging, extending to fetuses.
However, that the investigations of medical effects of exposure to infrasound and ultra-
sound have been conducted with due regard to safe exposure intensity levels when applied
to human subjects is important. In particular, while some human subjects registered and
reported discomfort and irritation, at such exposure levels no persistent effects have
been reported.
Such reports, coupled with our theory, analysis, and measurements suggest, however,
that it is difficult, if not infeasible, to project such acoustic-wave energies/fluxes into the in-
terior of a building, where many occurrences were reported to have taken place. Specifically,
infrasound cannot be delivered from a distance of tens of meters with the level of focusing
consistent with our incident criteria, without equipment that is tens of meters in size. We
regard that as implausible, and can rule out infrasound as a candidate modality. For ul-
trasound, atmospheric attenuation with typical relative humidity (RH%) levels, geometric
55
JSR-21-01 DoS Embassy Incidents February 11, 2022
attenuation if impinging from the outside as spherical waves from some distance, and the
small acoustic transmission through a glass window pane or wall, combine to limit possible
exposure in the interior of buildings to Sound-Pressure Level (SPL) values not reported to
produce lasting effects, for covert acoustic sources. As a result, acoustic energy impinging
from a building exterior with no open windows does not emerge as a candidate modality
inside buildings. Ultrasonic acoustic power could be beamed/focused in open spaces at SPL
values that may be injurious, but not easily by covert means.
Figure 5.5: Combined assessment of acoustic waves as a potential cause for embassy inci-
dents on all relevant criteria. An admissible modality, including the requirement of remote
delivery, would require green along a vertical swath in this graph.
We summarize our assessment of sonic hypotheses in Figure 5.5. The color coding in
the figure corresponds to the following: red indicates that the mechanism is inconsistent
with the requirements or effects considered, yellow indicates some level of inconsistency, and
green represents consistency. The fact that no frequency is green across a swath crossing the
multiple rows shows that acoustic modalities can be ruled out as a cause for the embassy
incidents (Finding 5).
56
JSR-21-01 DoS Embassy Incidents February 11, 2022
6 ELECTROMAGNETIC ATTACK HYPOTHESES
6.1 Introduction and Background
Electromagnetic (EM) radiation refers to oscillating coupled electric and magnetic fields
that propagate in space as traveling waves. The wavelength λ and the frequency f of the
wave have a fixed relation to one another given by λ = v/f, where v is the speed of the
wave. In vacuum, EM waves travel with the speed of light c = 2.998 × 10
8
m/s. As with
acoustic waves, the energy flux (energy per unit area per unit time, or power per unit area)
carried by the wave is proportional to the square of the wave’s amplitude (here, the peak
electric or magnetic field strength).
The properties of EM radiation vary with wavelength λ over the electromagnetic spec-
trum, ranging from radio frequency (wavelengths greater than about λ ∼ 1 m, corresponding
to frequencies less than about 300 MHz), to microwaves (wavelengths from about 1 m to
1 mm, and frequencies from 300 MHz to 300 GHz), to optical photons (400 to 700 nm wave-
lengths) to X-rays and gamma rays (wavelengths shorter than 10 nm). In the discussion
that follows, we will specifically use the term “radio frequency (RF) waves” and “microwave
frequencies” to mean electromagnetic radiation with frequencies in the 300 MHz - 300 GHz
range (wavelengths ranging from 1 meter to 1 millimeter.) Further, within this range, the
following distinctions are commonly made: the UHF band spans 300 MHz - 1 GHz (1 - 0.1
m); the SHF band spans 3 to 30 GHz (10 - 1 cm); the EHF band spans 30 to 300 GHz (10
- 1 mm).
EM radiation is generated by oscillators: charged particles accelerated back and forth
by either natural or artificial means. An antenna typically creates electromagnetic radiation
by driving an oscillating current of electrons in a conductor. EM radiation can be generated
as a continuous wave, with a single carrier frequency, or as pulsed radiation that consist
of packets of duration τ
p
and pulse repetition frequency (PRF) that measures the number of
57
JSR-21-01 DoS Embassy Incidents February 11, 2022
pulses per unit time, i.e. per second. Generating pulsed radiation creates a finite band-
width around the carrier frequency that increases as the pulse duration decreases. Roughly
speaking, the bandwidth BW ≈ 1/τ
p
.
Electromagnetic radiation incident on matter will interact with charged particles within
the matter, accelerating them back and forth as the wave goes by. Through this interac-
tion, the energy carried by the wave will be transferred to the motion of particles in matter.
From energy conservation, EM energy deposited in matter will be lost from the wave, and
the wave will attenuate as it propagates through matter. If the irradiated matter is a re-
ceiving antenna, the resulting current from the oscillating electrons can be amplified into
a measurable signal. The energy transferred to the material will often end up as heat, but
in general can be converted into any other kind of energy. Energy from an electromagnetic
wave can also be reflected or re-radiated by the material.
The strength of this interaction, and the resulting rate of attenuation of the incident
wave, depends both on the frequency of the wave and on the nature and composition of the
matter. In lossy conductors, attenuation is exponential and can be expressed as
I(x) = I
0
τe
−x/L
, (6-1)
where I
0
is the intensity of radiation incident on the lossy conductor, τ is the fraction
transmitted into the lossy conductor, x is the depth into the lossy conductor and L is the
penetration depth. Penetration depth L refers to the depth where the intensity drops to 1/e
of its original value. The term skin depth, denoted δ
p
, instead refers to the depth where
the field or current density drops to 1/e. Because power is proportional to field squared,
L =
δ
p
2
. Human tissues, including skin, muscle and fat, can be treated as lossy conductors.
EM waves interacting with human tissues experience an exponential attenuation, as do
typical building materials that the waves pass through, analogously to acoustic waves. In
addition, due to the nature of the interaction, shorter wavelengths penetrate to smaller
depths within a conductor. We will return to the concept of penetration depth and its
implications for the types of effects it causes on people, in more detail when we consider
RF waves.
58
JSR-21-01 DoS Embassy Incidents February 11, 2022
In our discussion, we will group ionizing radiation under EM waves, although ionizing
radiation can consist of energetic particles as well as waves. Ionizing radiation is released
by atoms and their nuclei; those in the form of electromagnetic waves include X-rays and
gamma-rays, while particles can be in the form of neutrons, alpha-particles or beta-particles.
Such radiation contains sufficient energy to ionize matter by freeing a bound electron.
The energy carried by an electromagnetic radiation source will spread out in space
as the wave propagates from the source. If the source is isotropic (radiates equally in
all directions), then the power received from the source (energy transferred per time) will
decrease as
P (r) =
P
o
4πr
2
, (6-2)
where P
0
is the emitted power and r is the distance from the source. If the source wave is
beamed, this means that the radiator (e.g. antenna) radiates more strongly in a particular
direction; this is modeled as a gain factor that depends on the direction around the antenna.
Power will still decrease as the inverse distance squared from the source. The properties
of transmitted radiation fields in both frequency and spatial power distribution can be
manipulated by the shapes and sizes of transmitting antennas.
Electric and magnetic fields need not be time-varying; fields which do not change
amplitude with time are called static. Neither static electric fields nor magnetic fields are
relevant to the present study. Static electric fields can certainly cause harm to humans,
usually via the production of large voltages across the human body and resultant damaging
current(s). However, the upper limit to static electric fields in everyday life is set by the
dielectric breakdown of air; that is, sufficiently large fields ionize air molecules, producing
sparks (as in the lightning of thunderstorms or the sparks that result when live wires come
into very close proximity.) The production of large electric fields requires the application of
high voltages across two or more conductors; this is inconsistent with situational reports.
Additionally, although high static electric fields can be harmful, the effects are inconsistent
with those reported.
Magnetic materials can be used to produce static magnetic fields. The SI unit of
the magnetic field is the Tesla (T). For reference, humans are exposed continuously to the
59
JSR-21-01 DoS Embassy Incidents February 11, 2022
Earth’s magnetic field, which has a value on order of 50 µT. Kitchen magnets have strengths
ranging from 1-100 mT; centimeter-scale permanent magnets with strengths on order of
1 T are available commercially (e.g., neodymium magnets). Clinical magnetic resonance
imaging machines (MRI) produce fields on order of 3 − 7 T and research machines can
produce much higher fields, on the order of 30 T, but the bores size is too small for human
bodies. Static fields are generally considered safe up to MRI strengths. Some caveats exist
with respect to exposure time but, in any case, affecting a human body at any significant
distance with >1T strength magnetic fields would be impractical or impossible. A review
of static magnetic field effects, both homogeneous and inhomogeneous (i.e., producing field
gradients) can be found in Zhang et al. (2017).
6.2 Ionizing Radiation
We first consider whether ionizing radiation is a possible attack modality.
6.2.1 Generation, focusing, absorption, and interference
Ionizing radiation is the common term used for particles or electromagnetic waves that have
sufficient energy to ionize atoms by removing electrons from them. The term covers X-rays,
gamma-rays, neutrons, alpha particles, and beta particles. X-rays and gamma-rays are
highly energetic photons, while the rest are energetic particles. This property gives them
the ability to damage cells and the DNA in cell nuclei and makes them harmful to living
organisms.
60
JSR-21-01 DoS Embassy Incidents February 11, 2022
X-rays are a high-energy, short-wavelength form of electromagnetic radiation. The X-
ray band covers the wavelength range approximately between 10 nanometers (10 × 10
−9
m,
corresponding to a photon energy of 0.1 keV) to 12 picometers (12×10
−12
m, corresponding
to a photon energy of 100 keV). They are commonly produced in X-ray tubes by accelerating
electrons through a potential difference, or a voltage drop, and directing them onto a target
material such as tungsten.
Gamma-rays have even higher energies and shorter wavelengths than X-rays, typically
smaller than 12pm (though note that some sources define smaller energy radiation as a
gamma-ray if it originates from a radioactive process). The primary generation mechanisms
are processes that involve the nuclei of atoms, including radioactive decays, nuclear fusion,
and nuclear fission.
Alpha particles consist of two protons and two neutrons bound together (i.e., the
nucleus of a helium atom). They are usually produced as a result of radioactive decay.
Alpha particles are a strongly ionizing form of radiation, but at typical energies up to 8
MeV, their penetration depth is small and they can be stopped by a few to 10 centimeters
of air. As a result, alphas cannot travel long distances from a source and cause harm.
Beta particles are high-energy electrons or positrons. They are more penetrating than
alpha particles but still cannot be used to direct energy from longer distances or penetrate
buildings.
Both X-ray and gamma-rays have high penetrating power. Depending on their energy,
they can pass through several inches of a dense material like lead and up to several feet of
concrete, potentially causing damage even when they travel through a wall.
X-ray beams can be collimated or reduced in size using pinholes or movable slits typ-
ically made out of tungsten or some other material with high atomic charge. A collimator
works by only allowing rays coming from certain angles to pass through and forms a direc-
tional beam that way. To focus them further requires using “grazing incidence” mirrors,
which involves reflecting them at grazing incidence angles. This requires specialized and
often bulky optics. Gamma rays require even smaller grazing incidence angles to focus.
61
JSR-21-01 DoS Embassy Incidents February 11, 2022
However, they can also be collimated. Because collimators typically consist of a honeycomb
of tubes made of absorbing material, they significantly reduce the intensity of the source
by admitting only a fraction of the radiation through.
The combined difficulties of generation, collimation, and penetration of ionizing radi-
ation across the relevant distances renders this attack modality highly improbable with a
portable device (Finding 5).
6.2.2 Effects on the human body
The damaging health effects of short-wavelength or ionizing electromagnetic radiation, such
as X-rays and γ−rays have been known for a long time. The damage due to radiation
depends on the type of radiation emitted, the dose received by a person, and the parts of
the body that are exposed. The radiation dose itself scales with the duration of exposure
and the power of the source, and inversely with the square of the distance from the radiation
source.
When ionizing radiation interacts with tissue, it causes damage to the cells as well as
harmful changes to the genetic material, or the DNA, in the cell nucleus. As a result, it
can cause skin burns, radiation sickness, widespread organ damage, as well as neurological
symptoms, fatigue, and nausea. These symptoms are inconsistent with the medical records
and reports of affected individuals, rendering exposure to ionizing radiation not a viable
explanation for the reported incidents (Finding 5).
6.3 Bounds on RF Wave Properties
We now turn to RF waves and assess whether they can produce effects consistent with the
incidents reported. We place some immediate bounds on the properties of RF waves of
potential interest using some of the Incident Consistency Criteria discussed in section 3. In
addition to those listed there, the lack of reports of a heating or burning sensation during
the incidents provides another bound on the RF parameter space, as we will show below.
62
JSR-21-01 DoS Embassy Incidents February 11, 2022
Regarding the association of sound with RF waves, there is a mechanism, known as
the Frey effect, which has been shown to induce a sensation of sound in people. We will
discuss this in more detail in section 6.5.4. For now, we simply note that this mechanism
requires the use of pulsed RF radiation, a PRF in the audible range, a carrier frequency in
the 300MHz−10GHz range, and a sudden deposition of a small amount of energy into the
ear. As a result, if the sounds reported during the incidents are to be explained through
incident RF radiation, this imposes strong constraints on the type of radiation of interest.
Sound sensations produced by the Frey effect cannot be recorded by external electronic
microphones. This is an important difference between the sensation of sound produced
through the Frey effect and that produced by actual sound waves, and rules out RF waves
as the source of any incident with recorded sounds that match what was heard by the
relevant person(Finding 5).
The multiple attributes of pulsed RF waves form a multi-dimensional parameter space
that is somewhat difficult to navigate. The parameters of interest include carrier frequency
f, peak energy flux F
p
, average energy flux F
avg
, the PRF, and the pulse duration τ
p
. We
will now examine the various constraints on this parameter space and show how we can
narrow it down to a small region that remains interesting.
6.3.1 Penetration Depth, Skin Heating, and the Carrier Frequency Range
We first show that the carrier frequency can be bounded by considering the penetration
depth of RF into the human body. Microwave radiation incident on the human body will
be partially reflected and partially absorbed. A reasonable value for the fraction of power
absorbed, τ, lies between 0.4 and 0.5 (Foster et al., 2016). Because the human body is a
lossy conductor, microwave energy is subject to exponential attenuation as it penetrates
inward, as discussed in Section 6.1.
The penetration depth (and skin depth) of microwave radiation into tissue depends
on the frequency of the radiation and the type of tissue. Gabriel et al. (1996) provides
measured permittivity and conductivity vs. frequency data for various human tissues. This
63
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.1: Penetration depths for various tissue types vs RF frequency as calculated by
JASON, see section 6.3.1. Above ∼ 20 −30 GHz, RF waves penetrate less than 1 millimeter
into tissue. Dotted line is based on reference Blick et al. (1997).
data can be fit to a model based on multiple Cole-Cole dispersions (Eq. 4 in Gabriel et al.
1996) to produce an equation for the relative complex permittivity,
r
. Using
r
calculated
for each tissue, we can calculate the skin depth, δ
p
using
δ
p
=
67.52
f
[
p
02
+
002
−
0
]
−1/2
mm, (6-3)
where f is the frequency of the radiation in GHz and
r
=
0
− i
00
. Note that
0
and
00
are
frequency dependent. Recalling that L =
δ
p
2
, we plot the penetration depth, L, for various
frequencies in Figure 6.1. We also plot the penetration depth estimated using a simplified
approach presented in Blick et al. (1997).
What is apparent is that for frequencies f > 20 − 30 GHz, penetration depths are less
than 1 mm. That is, the incident energy will be deposited at very shallow depths and simply
cause skin heating (and accompanying sensations of warmth and, for sufficiently high pow-
ers, pain). As a result, we will not consider frequencies above 30 GHz further. Conversely,
for frequencies < 500 MHz, penetration depths can be many centimeters, implying that
energy can be deposited into the bulk of an organism. In the 500 MHz − 30 GHz region
64
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.2: The bounds on RF frequencies imposed by consideration of penetration depth
on human body.
(note that these extremes are arbitrary and one could consider somewhat larger or smaller
ranges), there is the possibility of shallow penetration. We will consider these two regimes
in more detail in Section 6.5 but we will focus our discussion of a remotely operated RF
device to the 500 MHz−30 GHz region. We show these regimes in Figure 6.2. We also
denote the frequency range for cell phones on this plot as well as for the Frey effect, which
we will discuss in Section 6.5.4.
6.3.2 Peak and average flux
For pulsed RF waves, the average flux incident on a surface, F
avg
, and the peak flux F
p
incident during the pulse are related by a ’duty cycle’: F
avg
= DF
P
, where the duty cycle D
is just the ratio of pulse duration to the time between pulses. Phenomena such as heating
of tissues are sensitive to the average flux because they are controlled by the total energy
deposited per unit time. Some phenomena, such as the disruption of electronic devices or
the Frey effect, are instead sensitive to the peak flux in the pulse. Using this fact, we can
identify regions in the (F
avg
, F
p
) space that are consistent with the reported phenomena.
In Figure 6.3, we delineate regions where
• F
avg
is large enough to produce perceptible sensations of heating and pain. Average
flux (sometimes called average power density) thresholds are discussed in Section
65
JSR-21-01 DoS Embassy Incidents February 11, 2022
6.5.2. Sensations of heating or skin pain are inconsistent with reports. This bound is
frequency dependent; to be conservative, we choose the highest power density value
in Table 1 which is 63 mW/cm
2
at 2.45 GHz,
• F
p
is large enough to cause disruption of consumer electronics, for pulse widths
τ
p
>∼ns,
• The region occupied by ordinary cell phones and MRI machines, which are deemed
safe for human health.
The regions shown in red can be eliminated from further consideration by these arguments.
Figure 6.3: The bounds on the average flux (F
avg
) vs peak flux (F
p
) parameter space
imposed by various physical and physiological effects and known safe bounds.
Several studies explored the upset and destruction of electronics by nanosecond RF
pulses (Camp et al., 2001; Nitsch et al., 2004; Backstrom & Lovstrand, 2004). These stud-
ies were performed in the early 2000s; therefore, the electronics studied are representative
of that era, with their coarser circuit features and less dense printed circuit boards, and
likely with lower distributed capacitance compared with contemporary electronics. Camp
et al. (2001); Nitsch et al. (2004), in particular, focused on sub-nanosecond irradiation,
66
JSR-21-01 DoS Embassy Incidents February 11, 2022
and reported upsets typically around 10kV/m (corresponding to peak fluxes of 25 W/cm
2
)
and permanent damage at an order of magnitude higher (thus 2.5 kW/cm
2
). We adopt
here the F
p
= 25 W/cm
2
limit, but note that contemporary electronics may have differ-
ent thresholds for disruption (Finding 6). Placing firmer experimental bounds on the
effect of RF waves on contemporary electronics can help narrow down this space further.
(Recommendation 8)
6.3.3 Pulse duration and repetition frequency
We can compare incident data with information in the open literature to estimate the range
of pulse width and pulse repetition frequency (PRF) that may be of interest for sensing. To
begin with, the majority of literature-reported thermoacoustic effects (i.e., sounds produced
by pulsed radiofrequency waves via the Frey effect) use pulse repetition rates of 1000 Hz
(Elder & Chou, 2003). The highest PRF reported for a Frey effect is 20,000 Hz, but the
upper end of this range is not readily reproducible. At fluxes of tens of mW/cm
2
(see Elder
& Chou 2003 and Section 6.5.4) and in the 1 − 10 GHz range, these signals would likely
produce interference in communication systems (e.g., for mobile phone signals as well as 2.4
and 5 GHz WiFi.)
One last concern is the possibility of extremely fast pulses. A pulse of duration δt will
spread over a bandwidth of 1/δt Hz. This means that a very short pulse that encompasses
only one oscillation of the carrier wave will be spread very widely in frequency. In other
words, much of the energy will not be at the carrier energy. A reasonable rule of thumb is
to assume each pulse must contain at least 10 oscillations of the carrier wave. For a 1 GHz
carrier wave, for example, a single oscillation is 1ns and a ten oscillation pulse would have
a bandwidth of 100 MHz. For a 10 GHz carrier, the smallest pulse would then be 10 × 10
ps = 100 ps, corresponding to a bandwidth of 1 GHz. This places a bound on the smallest
pulse to reasonably consider for a given frequency.
67
JSR-21-01 DoS Embassy Incidents February 11, 2022
6.4 RF Radiation Beaming, Propagation, and Absorption
6.4.1 RF beaming
Focusing RF radiation is achieved typically by using parabolic reflectors. The electric field
pattern beamed from a such a reflector (or a circular phased array) is circularly symmetric
and depends on angle θ from on axis. As a rule of thumb, a dish or phased array forms
quite a localized central beam of angular width λ/D radians, where λ is the wavelength of
the radiation, D is the antenna diameter. The ratio of the distance of an antenna of size D
to the localization of focus is roughly equal to size of the antenna measured in wavelengths.
The size of a beam in physical length units, for a target located at a distance L from the
reflector can, therefore, be estimated as X ≈ Lλ/D.
For example, achieving a beam divergence of approximately 1 m at a distance of 50 m,
as indicated by our incident consistency criteria, requires an antenna that is at least 50
wavelengths in size. This results in the following dish sizes for some frequencies of interest:
• 300 MHz ∼ 1 m wavelength: 50 m emitter diameter
• 1 GHz ∼ 30 cm wavelength: 15 m emitter diameter
• 3 GHz ∼ 10 cm wavelength (S band, cell phone): 5 m emitter diameter
• 10 GHz ∼ 3 cm wavelength: 1.5 m emitter diameter
• 30 GHz ∼ 1 cm wavelength (Ka band, 5G): 50 cm emitter diameter
• 100 GHz ∼ 3 mm wavelength (Active Denial System): 15 cm emitter diameter
An antenna required to localize its emission focus to ∼1 m at a distance of 50 m would be
quite noticeable for frequencies lower than ∼5 GHz.
There are also techniques for creating a localized beam using the interference between
two or more antennas (as is done in aperture synthesis interferometry). However, using these
techniques would require tight spatial coordination between the locations of the dishes and
would be logistically very challenging.
68
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.4: Entry loss as a function of frequency through metallized windows (Win) and
fiber backed plasterboard (FBP) of a typical modern house as well as a brick “Mansion”.
6.4.2 RF absorption and interference
Attenuation and entry loss of RF into buildings is described in detail by
https://www.ofcom.org.uk/_data/assets/pdf_file/0016/84022/building_materials_
and_propagation.pdf. Their Figure 1-1, reproduced as Figure 6.4, shows the attenuation
of RF into a building at frequencies between 100 MHz and 6 GHz. In this frequency range,
the attenuation ranges from 5 dB to 30 dB, with ∼ 10 dB variations depending on the con-
struction material and steadily rising toward higher frequencies.
Bas et al. (2009) did a study of the penetration of 28 GHz (∼1 cm wavelength, 5G
cell phone) radiation from outside a building to inside, and found a typical penetration
loss of 23 dB Bas et al. (2019). They used a realistic building with diverse components;
the penetration through window, concrete block, and wood frame and drywall are quite
different.
Albeit significant, the level of attenuation reported in these studies is not prohibitive
69
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.5: Field coverage (received E-field in dB) of building floor with inhomogeneous
periodic walls for v-polarization. In this assessment, the source is inside the building, at
the position of least attenuation (yellow).
at ∼GHz frequencies but penetrating effectively through a building is progressively more
challenging at frequencies higher than a GHz.
RF waves exhibit a high level of interference (producing spatial variations of field
strength) when passing through materials and a high degree of scattering when reflecting
off ordinary surfaces. Thiel & Sarabandi (2009) calculated the effects of attenuation and
scattering of 1 GHz (30 cm wavelength) wave within a typical office layout. Figure 6.5,
reproduced from their Figure 6, demonstrates these effects, including attenuation with
distance, attenuation because of passage through walls, and interference which creates nodes
and antinodes on typical scales of a half wavelength. In particular, this figure illustrates
that the signal remains collimated only when there is a direct line of sight to the source;
70
JSR-21-01 DoS Embassy Incidents February 11, 2022
otherwise, scattering and interference patterns destroy collimation and spread the beam to
scales comparable to the size of the room.
These interference patterns may shift as scatterers such as humans or furniture move
around and, as illustrated, the power flux between a node and antinode can be an order of
magnitude or more, although typically somewhat less. The location of fringes is typically
different for different polarizations. Because of the sensitivity to the exact details of the
layout within a building, without complete knowledge of all materials present it would be
virtually impossible to predict exactly where such interference fringes will occur.
6.5 RF Effects on Human Health
We turn to the effects of RF radiation on human health, focusing in particular on the
question of long-term effects. We also review the literature with an eye toward effects that
can be induced without causing a sensation of pain or warmth, to be consistent with incident
reports.
The study of electromagnetic interaction with biology has a long and storied history.
The academic literature on microwave effects on biology, especially in the context of effects
on humans, is large and spans the better part of a century and many thousands of papers,
books and articles. Two of the most comprehensive and balanced synthesis of microwave
effects on biological systems are Lin & Michaelson (1987) and Barnes & Greenebaum (2007).
As of this report, the prevalent opinion in the literature is that most, if not all, demon-
strated in vivo effects (including thermoacoustic phenomena like the Frey effect) are due to
the deposition of thermal energy by microwaves into tissue.
How thermal energy is deposited into the human body varies by frequency but is
reasonably well understood. There exist detailed analytical (Adair & Black, 2003; Foster
et al., 1998; Foster, 2000; Foster et al., 2010, 2018), lattice (Gowrishankar et al., 2004),
and finite element models on this question, which have been demonstrated and validated
(Hirata et al., 2003; Zasowski et al., 2006).
71
JSR-21-01 DoS Embassy Incidents February 11, 2022
Considering the possibility of direct penetration of energy into the ear canal rather
than through normal pathways of skin and tissue, some recent studies have shown that 60
- 90 GHz RF waves may be able to deliver energy deeper into the ear canal than RF waves
of lower frequency (Vilagosh et al., 2018, 2020). However, this effect is very weak. “Since
the absolute value of the [power flux density] is trivial at 30 GHz, the transmission has no
practical significance. The higher penetration of the 60 GHz and 90 GHz pulses results
in approximately 2.3% of the [power flux density] incident on the outside of the ear canal
being transmitted into the middle ear at 90 GHz.” (Vilagosh et al., 2020)
6.5.1 The f < 500 MHz Range
For frequencies where the penetration depth is on order of a body dimension, microwaves
will penetrate through the entire body. The relevant thermal effect can be well modeled
by a homogeneous thermal input (load) on the human body (Adair & Black, 2003), i.e., by
writing a heat balance equation
A + M ± W = ±R ± C ± E ± S (6-4)
where all terms are in power units (e.g., Watts), and A, M, W, R, C, E and S are the rates
of energy absorption from the RF field into the body, thermal energy production through
metabolic processes, work production by or on the body, heat exchange with the envi-
ronment via radiation, heat exchange with the environment via convection, heat exchange
with the environment via evaporation, and heat storage in the body, respectively (Adair &
Black, 2003). Additionally, microwave resonance effects are possible at these scales if the
wavelength of the radiation, λ, is also on order of a relevant body dimension (Foster, 2000).
Thus, to first order, incident microwaves introduce a bulk thermal energy input, which must
be dealt with by the organism.
At sufficiently high RF intensities, enough thermal energy is deposited into the system
to produce a measurable increase in temperature. The thermal gradients produced by
such energy deposition differ from those of ambient heating. Above certain intensities, RF
energy can produce morbidity, and, after thermoregulatory mechanisms are overwhelmed,
mortality (Barnes & Greenebaum, 2007).
72
JSR-21-01 DoS Embassy Incidents February 11, 2022
Can exogenous “whole body” inputs of thermal energy via microwaves below
thresholds of perception produce long-term effects in humans?
There remain open questions about the limits of human thermoregulation (i.e., what
does the human body do when provided with this “extra” thermal energy) and thermoper-
ception (i.e., at what threshold and how does the human body perceive this “extra” thermal
energy), especially for very gradual heat inputs over very long periods of time (months) and
during periods where conscious detection of thermal excursions may be impaired (e.g., dur-
ing sleep).
Adair et al. (2005) (and references therein) carried out a series of trials in which humans
were bulk irradiated at frequencies of 100, 200, 450, and 2450 MHz at specific fluxes. At
these frequencies, subjects were subjected to three fluxes (9, 12, and 15 mW/cm
2
) at a
whole body average normalized specific absorption rate of 0.045 W/kg at each of three
ambient temperatures, T
a
= 24, 28, and 31
◦
C. We note, as a metric, that typical human
metabolism generates heat at a rate of ∼ 0.5 − 1 W/kg. The study included control groups
that had no RF exposure. Measured responses included esophageal and skin temperatures,
metabolic rate, local sweat rate, and local skin blood flow. Derived measures included heart
rate, respiration rate, and total evaporative water loss.
Several conclusions from this study are noteworthy. First, subjects thermoregulated
efficiently through increased heat-loss responses, particularly sweating. Second, they con-
firmed that RF energy at 100 and 220 MHz was absorbed into deep tissues of the body
rather than on the body surface. As a result, there was an absence of the thermal sensa-
tion that occurs at frequencies at and above 2 GHz, which are derived from stimulation of
temperature-sensitive neurons in the skin. Nevertheless, thermoregulatory responses were
generated owing to the presence of similar temperature-sensitive neurons in the brainstem,
spinal cord, and elsewhere in the central nervous system. The authors concluded that “The
rapid mobilization of sweating observed during the first minute of a 220 MHz RF exposure
at 15 mW/cm
2
in all tests ... can only be understood in terms of the thermal stimula-
tion of temperature sensitive neurons residing in critical regions of the brainstem and spinal
cord.” They further noted that “Calculations of localized RF energy deposition in the body,
73
JSR-21-01 DoS Embassy Incidents February 11, 2022
through FDTD [Finite Difference Time Domain] modeling, indicate that there is especially
high power absorption (SAR > 1.0 W/kg) in the third and fourth ventricles, the medial
preoptic/anterior hypothalamic nuclei, and the cerebrospinal fluid bathing these regions.”
The observation that bulk heating, well below thresholds required for tissue damage (see
below), can trigger hypothalamic nuclei is intriguing.
We conclude from these studies that, at flux levels of interest to the incidents, the main
effect in this frequency range (< 500 MHz) is a trigger of the thermoregulation response in
humans, inconsistent with direct causality for the reported incidents (Finding 5).
6.5.2 The 500 MHz < f < 30 GHz range
This regime has been modeled and studied extensively (see, e.g., Foster et al. 1978, 1998;
Foster 2000; Foster et al. 2010, 2018; Gowrishankar et al. 2004 and references therein). We
find the set of experiments on human skin by Blick et al. (1997) and a subsequent modeling
effort on those data by Riu et al. (1997) particularly useful.
At these frequencies, most of the energy is deposited within the first few millimeters of
skin. Between 500 MHz and ∼ 3 GHz, the time constant for thermal conduction in tissue is
greater than that for thermal convection of heat from blood flow, so convection dominates.
The convection time constant is frequency-independent and of order 150 s (Foster et al.,
1998). Above 2-3 GHz, the dominant transport mechanism for acute effects is thermal
conduction of tissue, with a time constant ranging from 15 s at 10 GHz to 0.6 s at 35 GHz
(Foster et al., 1998). Above 30 GHz, energy penetration is so shallow that it is effectively
equivalent to pure surface heating.
The acute perception of warmth is produced by nerve endings within the first 0.6
mm of the skin and are sensitive to changes in temperature as small as 0.06 − 0.08
◦
C (Riu
et al., 1997; Barnes & Greenebaum, 2007). For acute application of microwave energy,
there is consensus in the literature on sensation and pain thresholds. Table 1, reproduced
from Blick et al. (1997), presents experimentally determined human skin warmth sensation
thresholds as well as the tissue penetration depth for different carrier frequencies. These
74
JSR-21-01 DoS Embassy Incidents February 11, 2022
levels represent conservative values; they were determined by subjects who were aware
that there was incident radiation in a controlled environment and were asked to detect
sensations. In other words, higher power levels might be required for thermal perception in
real scenarios where subjects are not aware energy is incident on their skin. The threshold
for pain is much higher than for thermal sensation, occurring at about 46
◦
C (Barnes &
Greenebaum, 2007). Given the sensitivity of human skin to temperature changes, dermal
warmth sensation thresholds provide an upper bound for the average flux of any incident
signal for which skin warmth was not detected.
Can microwave energy modulate the activity of peripheral nerves, espe-
cially those that innervate the head, without producing sensations of warmth
or pain?
Given the relative absence of reports of warmth or heat-induced pain and the fact
that for frequencies above 3 GHz, the penetration depth is 1 cm, it is natural to ask
whether microwave energy can stimulate or modulate the activity of peripheral nerves,
especially those that innervate the head (and potentially could lead to vestibular symptoms),
without producing sensations of warmth or pain. We judge that this is improbable because
modulating peripheral nerve activity thermally requires local temperature excursions (at
the nerve depth) of 6-10
◦
C (see below) and because microwave energy deposition at the
surface will be at least the same as (for penetrant frequencies) or much higher than (for low
penetration depth frequencies) at nerve depth. This would inevitably cause temperature
excursions at the skin surface that would be perceived.
The fact that microwave radiation could produce responses in peripheral nerves in
vivo was demonstrated 60 years ago in a seminal paper by McAfee (1962). In experiments
on cats, dogs, rabbits and rats, that work demonstrated that “heating afferent peripheral
fibers within exposed peripheral nerve trunks or heating cutaneous nerves lying within the
tela subcutania by 3-cm microwave irradiation, infrared radiation, or by a warm-water
thermode (a type of temperature sensor) results in a nociceptive response at corresponding
temperatures, indicating that the response derives from thermal stimulation of these nerves
75
JSR-21-01 DoS Embassy Incidents February 11, 2022
Table 1: Dermal Warmth Detection Thresholds as a Function of Frequency
Frequency (GHz) Threshold (mW/cm
2
) Tissue Penetration Depth (mm)
2.45 63.1 ± 6.7 32
7.5 19.5 ± 2.9 6.3
10 19.6 ± 2.9 3.9
35 8.8 ± 1.3 0.8
94 4.5 ± 0.6 0.4
1.3 × 10
5
5.34 ± 1.1 < 0.1
.... The response occurs abruptly when a temperature averaging 46
◦
C is reached at the
nerve” (McAfee, 1962). The nerves irradiated in these experiments include radial, sciatic
or trigeminal nerves. The microwave power levels incident on the animals were adjusted
such that they would produce a subcutaneous temperature rise to about 45
◦
C within 5 min.
This elicited a wide range of responses consistent with pain response, including respiratory
hyperpnea, pupil dilation, changes in blood pressure and pulse rate, as well as skin and
visceral vasoconstriction.
The authors opine that, “...because the response to thermal stimulation may be main-
tained for some time..., adrenal medullary secretions may elicit any one or more of many
possible secondary effects. Secretion of adrenocorticotropin (ACTH) may occur, resulting
in increased elaboration of adrenal cortical hormones. Changes in the composition of the
blood which characterize the stress response are expected to follow.” Lastly, they observed
that “...thermode heating and microwave irradiation does not elicit the spontaneous high-
voltage activity associated with myelenated fibers” consistent with earlier work that “only
the unmyelenated or thinly myelenated afferent sensory fibers exhibit spontaneous activity
at these temperatures.”
One interpretation of these results is that microwave heating produced increased firing
rates in peripheral nerves, specifically those involved in nociception. That is, the microwave
energy produced unpleasant sensations, ostensibly by triggering some sense of pain.
These and subsequent studies also concluded that, again, due to the low penetration
depth in tissue for carrier frequencies above 3 GHz, very little energy would penetrate into
76
JSR-21-01 DoS Embassy Incidents February 11, 2022
the central nervous system and, thus, the cause of the effects seen on the animals would
need be restricted to peripheral structures; only minute fluxes of radiation could possibly
reach the deeper structures of the central nervous system. McAffee (1971) also noted that
it is the neurons themselves that respond to the specific temperature and produce reflex
behavior, as opposed to the terminal end organs. The resulting reflexes and responses
produced by thermal stimulation at 45
◦
C of peripheral nerves is called nociceptive because
of the unpleasant stimulations it produces. It is important to note that, for short durations,
reaching a temperature of 45
◦
C at the nerve does not cause tissue damage. McAffee (1971)
estimates the flux at the skin to be of order 20 − 40 mW/cm
2
, albeit with a great degree
of uncertainty. We note that these fluxes applied over 5 mins, required to reach the effect
threshold of 45
◦
C in the experiments, is equivalent to a total dose of 6 − 12 J/cm
2
.
Taken together, then, there is evidence of increased firing rates in peripheral nerves due
to continuous wave microwave radiation. The literature cited would suggest that humans
subjected to this energy would not show lesions but would experience sensations of heat
and/or pain.
Moving from continuous waves to pulsed radiation, we review findings from a recent
study that provides the energy doses and temperature excursions required to fire a peripheral
nerve thermally. Using optical lasers at different frequencies, Wells et al. (2007) stimulated
rat sciatic nerves in vivo and showed that direct neural activation with pulsed laser light is
induced by a thermal transient. They further quantified the spatial and temporal nature of
this required temperature rise, including a measured surface temperature change required
for stimulation of the peripheral nerve (6
◦
C–10
◦
C). The temperature rise results in direct
or indirect activation of transmembrane ion channels causing action potential generation
and occurs at dosages of 0.3-0.4 J/cm
2
for 5 µs, 350 µs, 1 ms, 3 ms and 5 ms pulses,
corresponding to fluxes during a pulse of 60 kW/cm
2
, 857 W/cm
2
, 300 W/cm
2
, 100 W/cm
2
,
and 60 W/cm
2
, respectively, at 0.3 J/cm
2
.
77
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.6: From Foster et al. (2018). “Skin surface temperature increase in 1D baseline
model to pulse train at varying carrier frequencies in GHz and for the surface heating model
for comparison. Pulses are 1 s duration, repeated at 1/10 Hz, peak pulse intensity 1000
W/m2. Also shown (dotted lines) are the increases in surface temperature from exposure
to continuous-wave radiation at the same time-averaged power density at 3 and 100 GHz.
The energy transmission coefficient was assumed to be 1 at all frequencies to facilitate
comparisons of responses.”
The average power is lower depending on the pulse repetition frequency. We note that
these doses are about 10,000 times those reported to produce thermoacoustic phenomena
(the Frey effect), which occur at ≈ 40 µJ/cm
2
(see Section 6.5.4).
1
What we can conclude by linking this with the earlier work by McAffee (1971) and
others is that by modulating a microwave carrier wave (preferably at or above 1 GHz, so
that the energy is only absorbed just under the skin in the peripheral nerves), it is feasible to
produce a pulse that could thermally fire an action potential in a peripheral nerve. Although
Wells et al. (2007) argue that thermally-induced firing (with a laser) can be repeated up to
about 5 Hz PRF, this is misleading. As stated earlier, there are two thermal time constants
1
Some care must be taken when comparing the optical and microwave powers, given the different absorp-
tion and scattering phenomena at work. The authors of the study estimate that, for 2.12 µm wavelengths,
about 68% of the light incident on the peripheral nerve penetrated, entered the axon, and was converted to
heat. Given this, order of magnitude comparisons appear justifiable.
78
JSR-21-01 DoS Embassy Incidents February 11, 2022
for tissue near the skin: a frequency-dependent conduction time constant and a convection
time constant. The convection time constant is on order of 150 s and the conduction time
constant ranges from 2000 − 0.6 s for 1 − 35 GHz, respectively (Foster et al., 1998). For
energy above 2–3 GHz in frequency, the conduction time constant dominates. Although a
pulse would produce a very rapid temperature rise, the temperature decay (i.e., the time
it takes for a section of tissue to cool down to the local steady-state temperature) would
be set by a multiple of the conduction time constant. Thus, the fastest one could repeat
a thermal firing event without slowly raising the steady-state temperature of the tissue is
likely about 5τ, or one firing event every 3 seconds. This is illustrated in Figure 6.6 (Foster
et al., 2018).
This time interval between pulses is much slower than the endogenous firing rates of
most peripheral nerves. Just as importantly, because microwave energy deposition at the
surface will be at least the same (for penetrant frequencies) or exponentially higher (for
low penetration depth frequencies), pulsing in this way would inevitably cause temperature
excursions at the skin surface that would be perceived. We thus conclude that pulsed-
microwave induced thermal firing of peripheral nerves at fast enough rates to produce
significant biological effects and explain the direct sensations and symptoms associated
with anomalous health incidents is unlikely, especially in light of the absence of reports of
thermal sensations (Finding 5).
6.5.3 Requirement for RF-induced Traumatic Brain Injury
We conclude that radio-frequency energy is unlikely to produce Traumatic Brain Injury
(TBI) without concomitant sensations of heat and or pain. In addition, we see that RF
energy is unlikely to produce lesions in tissue unless power levels are sufficient to raise
the temperature of tissue to > 45
◦
C (Barnes & Greenebaum, 2007). For RF energy with
penetration depths on order of the centimeters to meters, such power levels would lead to
general hyperthermia, because, in order to raise the temperature of a putative target such as
a location in the central nervous system, the energy would increase the temperature of most
of the body or head by similar amounts. For RF energy with penetration depth less than
79
JSR-21-01 DoS Embassy Incidents February 11, 2022
1 centimeter, the elevation of an internal tissue location to 45
◦
C would necessitate a much
larger increase in skin surface temperature, given the exponential decay in power density
as a function of tissue depth. This would produce sensations of warmth, at minimum, and
pain, in all likelihood (Finding 5).
6.5.4 Thermoacoustic phenomena and the “Frey Effect”
Acute auditory effects of microwave exposure have also been documented, and are often
referred to as the “Frey effect.” Essentially, an abrupt (typically t < 100µsec) but very
low level heating of the tissue, with ∆T 1
◦
C, by the absorbed RF raises the pressure in
the tissue, and this launches a sound wave within parts of the auditory system inside the
human head. The effect is called “thermoelastic” or “thermoacoustic”. At RF frequencies
f & 1 GHz, the absorption length of the electromagnetic energy of the order of a centimeter,
as shown in Figure 6.1, whereas that of the sound is typically much longer. As a result,
this effect can be used to image the brain or other tissues acoustically (Jiang, 2020). This
auditory phenomenon, characterized by the sound of clicking, buzzing, knocking, or chirping
in the head, has no known adverse health consequences in and of itself (Frey & Etchert,
1985; Barnes & Greenebaum, 2007; Elder & Chou, 2003).
The Frey effect has been reported at frequencies ranging from 2.4 MHz to 10 GHz,
although the upper end of this range is not readily reproducible. For example, 8,900 MHz
was not effective at an average flux of 25 mW/cm
2
and peak flux of 25,000 mW/cm
2
(Frey,
1962). At 216 MHz, the threshold in average flux was found to be 4 mW/cm
2
and the peak
flux to be 670 mW/cm
2
(Frey, 1963). At the lowest effective frequencies (2.4–170 MHz)
reported in the literature, the peak flux thresholds were up to 9,000 mW/cm
2
(R¨oschmann,
1991). The lowest threshold value expressed in units of average incident flux is 0.001
mW/cm
2
(Cain & Rissmann, 1978); this value was due to the low PRF of only 0.5/s.
Overall, the hearing phenomenon has been shown to depend on the energy in a single
pulse and not on average power density. Guy et al. (1975) found that the threshold for RF
hearing of pulsed 2450 MHz fields was related to an energy density of 40 µJ/cm
2
per pulse,
or energy absorption per pulse of 16 uJ/g, regardless of the peak flux, for pulses less than
80
JSR-21-01 DoS Embassy Incidents February 11, 2022
32 ms. That amount of incident energy is expected to increase the tissue temperature by
5 × 10
−6 ◦
C (Elder & Chou, 2003).
One other important characteristic of the auditory response to microwaves is that
the waves must be modulated, generally in the 7 to 10 kHz range (Chou et al. 1977,
1982b; Watanabe et al. 2000), or 8 to 15 kHz (Lin 1977a,c). If the sounds reported
in the incidents arise from the Frey effect, this limits the PRF that is required in the
incident radiation. However, we repeat here that any directly recorded audio by an external
electronic microphone cannot be explained by the thermoacoustic effect.
The thermoacoustic effect has been calculated theoretically (e.g., Gournay 1966) for
fluids such as water by starting from the isobaric thermal expansion coefficient
β ≡
1
V
Å
∂V
∂T
ã
p
≈ 3.6 × 10
−4
K
−1
at 37
◦
C. (6-5)
For a sufficiently short pulse ∆t < h/c
s
, however, where h is the RF absorption length and
c
s
is the speed of sound in the fluid, it is not directly the expansion coefficient (6-5) at
constant pressure that is relevant, but rather
Å
∂P
∂T
ã
V
=
α
κ
≈ 8 × 10
5
Pa K
−1
at 37
◦
C (6-6)
because the affected layer is heated at constant volume, and only later expands. (Here
κ = −(∂ ln V/∂P )
T
≈ 4 × 10
−10
Pa
−1
is the isothermal compressibility). For example, from
Fig. 6.1, h ≈ 2 cm at f
RF
= 2.45 GHz, and so h/c
s
≈ 14 µs in water (c
s
≈ 1520 m s
−1
).
Indeed, it has been verified experimentally that the pressure amplitude of sound produced
depends only on the energy per pulse, not the peak RF intensity (power/area), in the limit
of short pulses (Foster & Finch, 1974).
Since the reference pressure level for sound is 2 × 10
−5
Pa, and ordinary conversation
is typically a thousand times this (60 dB), the acoustic pressure is exquisitely sensitive
to small temperature changes
2
. Thus for example, a 60 dB pressure level (0.02 Pa) would
require a temperature rise of no more than 3 × 10
−8 ◦
C. Of course, a train of N pulses
would heat the tissue N times as much if the length of the pulse train were short compared
to the thermal diffusion time, which is ≈ 11(h/cm)
2
minutes in water.
2
This is the reference level for sound in air; in water, 1 µPa is often used instead.
81
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure 6.7: (U) The human auditory system. Credit: Wikipedia Commons.
Based on these calculations, we find that at tolerable levels of audibility, the thermoe-
lastic excitation of sound waves in the brain, due to small acoustic pressure established for
the Frey effect, cannot produce brain damage by mechanical means (Finding 5).
The ability of a mechanical disturbance to cause damage to the brain is often quantified
by the stress or strain that it produces (Wright & Ramesh, 2012). For example, according
to this reference, a 50% probability of mild traumatic brain injury (TBI) is induced by a
shear stress of 7.8 kPa; and a 50% probability of mild diffuse axonal injury (DAI) with a
strain 0.2-0.3, corresponding to a shear stress of 26 kPa. Now the units of shear stress are
the same as those of pressure, whereas even at the threshold of pain (140 dB), the pressure
of audible sound is only 0.2 kPa. The strain associated with a sound wave of amplitude δp
is δp/ρc
2
s
(which is dimensionless), and ρc
2
s
≈ 2.3 GPa in water at 37
◦
C. In other words, at
levels that are audible but not painful, the stresses and strains associated with sound waves
produced in the brain by the Frey effect are orders of magnitude lower than those that have
been associated with brain injury. This leads to our finding above.
A few caveats are in order here:
1. According to Howard (2021, http://www.tulane.edu/
~
h0Ward/BrLg/AuditoryTransduction.
html) the mechanical advantage of the connection from the tympanum (eardrum) to
82
JSR-21-01 DoS Embassy Incidents February 11, 2022
the cochlea via the ossicles (malieus, incus, & stapes in Fig. 6.7), is such that a given
sound pressure level (SPL) in the air is amplified by a factor ≈ 22 in the inner ear.
We have not been able to find a similar factor for the acoustic coupling from the other
side, i.e., from the brain to the cochlea, which presumably would go primarily via
the internal auditory meatus, the channel through which the cochlear and vestibu-
lar nerves connect the inner ear to the brain. As a result, when comparing sound
waves excited in the brain to those from external sources, one should subtract at least
20 log
10
(22) ≈ 27 dB to account for the transduction of pressure by the auditory sys-
tem. Comparing the power levels and pulse durations of their experiments with water
(Foster & Finch, 1974) to those of Frey & Messenger (1973), Foster & Finch (1974)
estimate that the threshold of audibility corresponds to an SPL ≈ 90 dB (0.65 Pa) in
the fluid. They further estimate that this corresponds (at least at f
RF
= 2.45 GHz)
to an incident energy in microwaves ≈ 50 µJ per pulse, of which 60% would be re-
flected by the skull, and the remainder absorbed in the brain. The corresponding
peak temperature increase per pulse would thus be
2 × 0.65 Pa/(∂P/∂T )
V
≈ 1.7 µK ,
in which the factor of 2 has to do with the acoustic waveform (see below). This is far
too small to cause direct thermal damage.
2. Most of the acoustic power produced is at frequencies f
s
. c
s
/2πh, where, again,
h refers to the RF absorption length. At f
RF
= 2.45 GHz, where h ≈ 2 cm, this is
about 12 kHz. That is to say, a single short RF pulse should produce a very broad-
band acoustic signal with cutoff ∼ f
s
. If much higher RF frequencies were used,
the characteristic acoustic frequency would lie outside the range of human hearing.
Therefore, a given perceived sound level would correspond to a larger peak pressure in
the brain for the same incident microwave power (or energy per pulse). For example, at
f
RF
= 100 GHz, h ≈ 0.3 mm and f
s
≈ 1 MHz. Furthermore, the ratio of the perceived
SPL to the actual SPL in the brain is then sensitive to boundary conditions: linearly
with f
audible
/f
s
for a rigid boundary condition (as may be appropriate at a few GHz
where h is larger than the thickness of the skull), but as (f
audible
/f
s
)
2
if the boundary
is free (as may be appropriate at 100 GHz, where h is less than the thickness of the skin
83
JSR-21-01 DoS Embassy Incidents February 11, 2022
covering the skull). This has to do with the symmetries of the acoustic waveforms
produced in these two cases (see Appendix C). Thus, at f
RF
& 100 GHz, audible
clicks might correspond to acoustic stress levels in the brain of a few kPa, which
gets somewhat closer to the thresholds quoted above for mild TBI or DAI. However,
these pulses would be perceived on the skin well before they are perceived as acoustic
sensations.
We provide additional mathematical details in Appendix C.
6.5.5 Are there non-thermal effects of microwaves on tissue?
There is a vast “grey literature” on effects that are a) demonstrated only in vitro or in
small animals, b) inferred from epidemiological studies and reports and/or c) purported to
demonstrate so-called non-thermal effects. The current consensus is that most such findings
are either not reproducible, not easily scalable to humans, and/or not relevant to effects in
humans. Chapter 6 of Lin & Michaelson (1987) provides a valuable resource in this area.
A number of responses have been reported that appear not to be accompanied by
changes in temperature. However, in JASON’s conclusion based on a review of the litera-
ture, these are all most likely caused by the deposition of thermal energy into tissue. The
confusion arises in part because the body thermo-regulates, in part due to heterogeneity
of tissue, in part due to the presence of tissues that have low densities of pain receptors,
and in part due to poor experimental design. This discrepancy in the use of the term
“non-thermal” also arises when reviewing the extensive Cold War-era literature originating
in Soviet bloc countries. An early definition of non-thermal effects from that literature
is based on effects not associated with a measurable local or whole organism temperature
rise from an equilibrated baseline. Most Western investigators use the term thermal in a
somewhat different sense, allowing for the fact that an organism can be affected thermally
without a demonstrable temperature rise. The temperature rise, in fact, is taken to mean
that “the functional reserves of the organism for maintaining homeostasis have been ex-
ceeded in some manner, and thermal regulation has been instituted” (Lin & Michaelson,
1987).
84
JSR-21-01 DoS Embassy Incidents February 11, 2022
It has been proposed that non-thermal effects might be due to RF interaction with cell
membranes, especially those of neurons, perhaps producing depolarization or the opening of
ion channels. This is extremely unlikely. (See Schwan 1971 for a lucid treatment). In brief,
the electric field across a cell membrane arising from the cell’s normal resting potential is
about 500 kV/cm. Based on the electrical characteristics of cell membranes, the electric
fields produced by RF energy above 100 MHz across these membranes are between a factor
of 10
5
and 10
6
smaller than 500 kV/cm. Put bluntly: “According to all modern concepts
of neurophysiology about excitation, this just cannot stimulate nerves” (Schwan, 1971).
The possibility still remains that future studies will reveal reproducible non-linear
effects arising upon excitation of tissue by radio frequency energy in the relevant frequency
range. If these exist, they would occur at very high field strengths. How high these fields
need to be is unknown. We can take as a guess the field strength found across cell membranes
(500 kV/cm or more), although note that such fields are relatively static, while the fields
discussed here are sinusoidal (or at least alternating) at GHz frequencies.
Such field strengths would require very high fluxes. In order to apply such fluxes
without thermal sensation or damage, fast pulses would be required. By way of example,
consider a set of conventional RF pulses studied by Foster et al. (2018), shown in Figure 6.8.
Figure 6.8: (from Foster et al. (2018) )
85
JSR-21-01 DoS Embassy Incidents February 11, 2022
The pulses in Figure 6.8 have a 1.9 GHz carrier frequency, a duration of τ
p
= 0.57 ms,
a flux of of 800 W/m
2
(for a dose of 56 J/m
2
), repeated 217 times per second. Each pulse
produces a temperature rise of less than 5x10
−6 ◦
C.
The radiation flux F , in units of W/m
2
, is related to the electric field, E (units V/m)
as:
F =
E
2
η
, (6-7)
where η is the wave impedance, (in units of Ohms or Ω), of the tissue:
η =
377
1/2
r
Ω ≈ 180 Ω, (6-8)
where we calculated the permittivity using data from Gabriel et al. (1996) and the equation
on page 44 of Hurt (1996) as cited in Table 2 of Blick et al. (1997). Using the flux of
800 W/m
2
, this gives for the electric field,
E = (F η)
1/2
= 380 V/m. (6-9)
Assuming all of the energy in the 800 W/m
2
entered the tissue, the resultant field at the
surface of the skin would be 380 V/m or 3.8 V/cm. A field of 500 kV/cm is roughly 10
5
times larger than this value; achieving the larger field strength would require a 10
10
larger
incident flux. In order to maintain the same total energy dose per pulse (and thus keep the
temperature rise the same), the pulse width would have to decrease by 10
10
, going from 0.57
ms to 57 femtoseconds. For reasons discussed in Section 6.3, such a pulse width is below
the sensation threshold and spreads out the signal over a very large bandwidth. We can
approach this from the opposite direction: if we assume a 1 nanosecond pulse, a 10 GHz
carrier frequency, and ignore penetration depth problems, what is the maximum achievable
field strength for the same energy dose? Such a pulse would be 570,000 times shorter in
duration than the 0.57 ms pulse above, allowing for a 570,000 times higher flux:
E = (F η)
1/2
= (570000 × 800W/m
2
× 180 Ω)
1/2
= 2.8 kV/m. (6-10)
This is still a factor of ≈ 200 below the field strength across cell membranes. From the
above, we conclude that while it may be prudent to provide detection for very short pulses
for completeness (Finding 7, Recommendation 7), JASON finds it highly unlikely that
86
JSR-21-01 DoS Embassy Incidents February 11, 2022
realistic pulses can produce cell membrane-level electric fields within tissue without also
producing thermal sensation at the surface of the skin (Finding 5).
6.6 Examples of RF Exposure
We end this section by considering instructive examples of health effects from documented
and/or accidental exposures to RF waves, such as exposures to radars and directed RF
signals. These incidents provide an experimental bound on the health effects due to RF at
various average and peak powers. They are also illuminating in terms of health problems
that may arise due to the psychological response to these incidents.
6.6.1 The Moscow Embassy Lilienfeld Study
We begin by recalling that between 1953 and 1976, beamed microwave energy was directed
at the United States Embassy in Moscow, perhaps as part of an espionage attempt. Due
to widespread, publicized concern over health effects of the exposure (Brodeur, 1977), re-
searchers at the Department of Epidemiology at Johns Hopkins University, led by Abraham
Lilienfeld, conducted an epidemiological study focused on the staff of the US embassy in
Moscow and their families (Lilienfeld et al., 1978). An excellent and balanced review of this
report and the subsequent, sometimes differing, opinions produced can be found in Elwood
(2012). Because the frequencies and power levels are within ranges considered here, we cite
from Elwood below.
“From 1953 to May 1975, the microwave beam came from a source in a Soviet apart-
ment building about 100 m west of the 10-floor embassy building, affecting the west facade
of the central building, with highest intensities between the third and eighth floors. The
frequency was from 2.5 to 4.0 GHz and maximum exposures are given as up to 5 µW/cm
2
,
9 hours per day” (Lilienfeld et al., 1978). However, Appendix 11 notes that individual
exposures would have in general been much less than the maximums because of movement
of personnel away from windows and to other rooms, as well as the fact that some hours of
signal operation were at night (Elwood, 2012). The average fluxes noted in the Lilienfeld
report are significantly lower than accepted IEEE/ANSI power density standards for cell
87
JSR-21-01 DoS Embassy Incidents February 11, 2022
phones (IEEE Standards, 2019) and the radiation fluxes considered earlier when discussing
thermal effects.
The conclusion of the Lilienfeld report was that, epidemiologically, there were no
demonstrable health effects. Summarized more precisely in Elwood (2012): “literally hun-
dreds of comparisons were made” and only two differences stood out from the medical
record review, the increased rate of protozoal infections in Moscow male employees, and the
slightly higher frequencies of the most common kinds of health conditions in the Moscow
group. There were greater differences in the self-reported data on the questionnaires, but
there were no conditions which were more common in Moscow and showed a relationship to
estimated microwave exposure or length of service within the Moscow group..... The overall
conclusion of the investigators (page 246) was: “To summarize, with very few exceptions, an
exhaustive comparison of the health status of the state and non-state department employ-
ees who had served in Moscow with those who had served in other Eastern European posts
during the same period of time revealed no differences in health status as indicated
by their mortality experience and a variety of morbidity measures. No convincing
evidence was discovered that would directly implicate the exposure to microwave radiation
experienced by the employees at the Moscow embassy in the causation of any adverse health
effects as of the time of this analysis.”
6.6.2 The Norway Ship Radar Exposure
On August 24, 2012 a United States Navy destroyer passed within 70-100 meters of a Nor-
wegian Coast Guard ship. The ships were in close proximity for approximately 7 minutes.
During this period, the Norwegian Coast Guard ship experienced disturbances to instru-
mentation and electronics, including monitors and navigation instruments. Fire alarms were
also set off. The Norwegian crew deduced that the U.S. destroyer had accidentally left its
radar systems on while it passed near them and contacted the U.S. ship. The destroyer
then switched off its radar. Moen et al. (2013) provide a summary of the event, estimates
of exposure and a summary of the medical follow up and findings.
Of the 66 Norwegian crew onboard, 14 people were on the bridge or outside on the
88
JSR-21-01 DoS Embassy Incidents February 11, 2022
deck. From November 2012 to January 2013, 29 people (including the 14 on bridge or on
deck) were referred to the Hospital Department of Occupational Medicine in Bergen, Nor-
way out of concern that the exposure had caused health problems. The cognizant medical
staff utilized a subjective health complaint questionnaire at intake. A battery of clinical
tests were performed, including clinical examination of blood pressure, auscultation of the
chest, observation of the skin, palpation of the abdomen, neurological status and mini men-
tal state exam, eye examinations by an ophthalmologist, as well as blood sample analysis for
haemoglobin level, blood count, red cell volume, liver function (alanine transaminase), renal
function (urea, creatinine), thyroid hormones (thyrotropin, thyroxine and triiodothyronine)
and immunoglobulines (Immunoglobulines A, E, G and M). Among males, luteinizing hor-
mone, follicle-stimulating hormone, testosterone, sex hormone-binding globuline, prolactin
and oestradiol were analyzed. The conclusion of this study was that “the crew was found to
be a healthy group of persons, with no signs of skin affection, no cataract, no general health
problems, no affection of the nervous system or any abnormal blood analyses. The crew
had clearly less subjective health complaints than what we find in a normal population, and
the symptoms had evidently disappeared” (Moen et al., 2013).
This incident is of special interest because a) the exposure was confirmed and, based
on incident reports and fundamental physics, likely of high enough power density to produce
sensations of warmth on the skin; b) it resulted in documented rumor-driven health concerns
and early self-reports of a variety of heterogeneous health problems by the crew (including
crew who were inside the ship and could not have been exposed to radar energy); c) there
was no evidence of lesions or long-term physical damage despite the high intensity exposure,
even to sensitive areas including eyes and genitals; d) clinical evaluation led to a strong
conclusion that the exposed population was healthy 6 months after the event. Additional
detail on the event is summarized below.
Estimates of exposure
Moen et al. (2013) estimated from a review of the open literature that the radar that
produced the exposure was a SPY-1D(V) and operated at a frequency of 3-4 GHz, 6MW
peak power and 58 kW average power (i.e., 1% duty cycle). Additionally, the antenna gain
is estimated to be 9300. The average radiation flux at the exposure site can be obtained
89
JSR-21-01 DoS Embassy Incidents February 11, 2022
from
F = G
P
4πr
2
≈ 6, 000W/m
2
= 600 mW/cm
2
, (6-11)
where G is the antenna gain of 9,300, P is the average power of 58 kW and r is the distance
of the exposure site from the antenna, which we take to be 85 meters as the median of the
range reported (70−100 meters.) The corresponding peak power density is 60 W/cm
2
. The
authors further surmise, and JASON agrees, that the radar was likely performing a search
pattern over the space swept (as opposed to maintaining a steady beam targeted at the Nor-
wegian ship), so that the total dwell time at any spot would likely be of order seconds and
thus the average exposure over several minutes at any one location would be significantly
lower than the value calculated above (perhaps 10–100 times less than the value calculated
above.) At 2.45 GHz, the reported threshold for warmth sensation is 63.1 ± 6.7 mW/cm
2
(see Table 1), which is 1/10 the value calculated above. We would thus expect crew mem-
bers on deck (i.e., not inside the ship) would experience sensations of warmth lasting several
seconds to tens of seconds.
Incident reports
“Five of the crew on the bridge and two of the crew standing outside on the deck on the
same side of the ship where the American ship passed, felt a slight sensation of warmth on
the face during the radar incident. This sensation lasted for a very short period, not more
than seconds or minutes. Two of those on the bridge also felt a mild heat sensation on the
left arm; this side of their body was turned towards the passing ship....Two of the crew
were sitting inside the ship when the fire alarm went off and ran out on the deck to perform
their duties during the alarms. They described a slight warm sensation on the face and on
their arms” (Moen et al., 2013). No burns were reported. A physician on board the ship
reported no signs of skin, eye problems, or health issues of any kind for anyone on board
the ship that day and the crew reported no complaints.
Reported health issues
In the days that followed, crew members grew concerned about health effects of exposure.
90
JSR-21-01 DoS Embassy Incidents February 11, 2022
Some kept health diaries. The ship was eventually diverted to Tromso on September 4,
where crew were examined. No signs of disease or damage were found. At this point, a
meeting was held where the crew was informed of the exposure. Subsequent to this, the
ship set sail with a physician on board. During the journey, the crew began to report a wide
variety of symptoms to the physician on board. “At this time the crew reported problems
with headache (43%), fatigue (26%), sweating (20%), pain/burning sensation on the skin
(15%), and impaired/disturbed vision (7%). Furthermore, they reported strange feelings
in the ear, hearing loss, pain in testicles, nose-bleeding, nausea, and chest discomfort. The
onset of symptoms showed considerable variation. One week later the physician wrote in a
note that: “At the follow-up a week after the original survey, only 1 individual has health
problems. The remaining crew has no problems or complaints.” (Moen et al., 2013)
Subsequently, however, crew members began to report health problems again: “They
all described that they talked a lot about the radar incident on board, and they wanted
to know more about the exposure and the possible health effects. They all claimed that
the information they were given was scarce, and that they became worried. Some of them
described that they thought their symptoms developed because they discussed different
possible diseases that could be caused by radiation, for instance cancer of the brain and
infertility. The weeks passed and members of the crew still complained about the occurrence
of different symptoms. In a note from November, more than 2 months after the incident,
a marine medical officer reported to have a follow-up examination of 33 persons from the
ship.” As a result of this, “the crew were crew was referred for examination to the hospital
in Bergen at the end of November 2012.” (Moen et al., 2013) As summarized above,
the subsequent clinical study found the population affected to be healthy approximately 6
months after the event.
6.7 Conclusions on the RF hypotheses
We considered directed RF energy as a possible mechanism for causing harm to individuals
from a distance. We explored the frequency range, peak and average flux levels, pulse
durations, and pulse repetition frequency that may be utilized toward that end evaluated
91
JSR-21-01 DoS Embassy Incidents February 11, 2022
these characteristics based on multiple physiological and environmental considerations. We
summarize our findings on the RF hypotheses below as well as visually in Figure 6.9.
For the carrier frequency range of interest, the main limits arise from considering the
penetration depth of RF waves through the human body. Frequencies between 500 MHz
and 30 GHz can penetrate below the skin and produce sensations of heat and pain. Lower
frequencies can deliver thermal energy to the entire body.
Log frequency
[Hz]
6
MHz
7 8 9
GHz
10 11 12
THz
Wavelength
(E&M waves)
100m 10m 1m 10cm 1cm 1mm
Localize to one
person
Focus to smaller
than a person
Penetrate a
building
Known to affect a
human
Power 0.1W/cm
2
from distance
Surreptitious 10λ
antenna
Whole body heat Frey Skin pain
Figure 6.9: Combined assessment of RF radiation as a potential cause for embassy incidents
on all relevant criteria.
For the flux range of interest, we eliminated high peak flux levels, F
p
& 30 W/cm
2
,
that would cause a disruption of common consumer electronics devices, and low peak flux
levels, F
p
. 1 W/cm
2
, for which there are no documented health effects. We also narrowed
down the range of average flux levels of interest to 20−100 mW/cm
2
based on common daily
exposures on the low end and avoiding a heating sensation on the high end of this range.
However, we also note that some of these statements are based on “routine” pulse durations
(τ
p
> 10 ns) and, therefore, recommend a conservative approach for monitoring and allowing
higher flux levels to be considered. On the lower end, we caution that routine RF clutter
rises toward lower flux levels and would make individual signal detection challenging.
92
JSR-21-01 DoS Embassy Incidents February 11, 2022
For pulse durations, shorter pulses (e.g., of nanosecond duration) can allow higher
voltages to be induced across the cell membrane but (i) require higher peak fluxes to deliver
the same energy and (ii) spread the signal over a wide bandwidth that scales as the inverse
of the pulse duration. Because of this, practical considerations render pulse durations that
are significantly less than a nanosecond ineffective as a mechanism.
Associating sounds reported during incidents with the “Frey effect” leads to a PRF in
the audible range for human hearing. Auditory response has most commonly been elicited
in the 7-10 kHz range but experiments span the range from 1Hz to (rarely) 20 kHz. Based
on these studies, a PRF in the range of 1-10,000 pps provides an appropriate region of
interest.
It is important to repeat here that no immediate or lasting structural damage or ad-
verse health consequences have been reported in the literature for the signal characteristics
outlined above. Any documented damage mechanism requires high flux levels that can
cause large (> 5
◦
C) temperature changes in tissues and nerves, which would unavoidably
cause accompanying sensations of heat and pain. This is inconsistent with the reported
incidents and symptoms.
We summarize our assessment of RF hypotheses in Figure 6.9. The color coding in
the figure corresponds, as before, to the level of consistency between the mechanism at
the given frequency and the requirements or effects considered: red denotes inconsistency,
yellow denotes marginal consistency, and green denotes possible consistency. The fact that
there is no frequency at which the color is green simultaneously across all rows reflects the
conclusion above that there is no region in the RF spectrum that is consistent with all of
the signs and symptoms reported in the embassy incidents. The region with frequencies
between 500 MHz and 30 GHz cannot be conclusively ruled out but appears highly unlikely
given the arguments presented in this section (Finding 5). Our conclusion is based on
all of the information provided to us in the briefings and the substantial JASON expertise
in electromagnetic phenomenon in this frequency range. JASON will continue to engage
the US national and DOD labs to ensure we have not missed any work that would impact
Finding 5.
93
JSR-21-01 DoS Embassy Incidents February 11, 2022
We conclude by providing thoughts on future RF surveillance. Given that the current
data and analysis significantly limits but cannot conclusively rule out any offensive use of
microwave irradiation, it is important to deploy technology that can continuously monitor
ambient cm-wave and mm-wave microwave RF. The BlueJay detector suite described in
Appendix D includes a first-generation monitor, covering the band 1 MHz to 10 GHz, with
55 dB dynamic range of native sensitivity (1 nW to 0.3 mW, but deliberately attenuated in
practice by some 40 dB, thus 10 µW to 3 W), and with peak-hold functionality for pulsed
radiation down to 100–200 ns. With relatively straightforward circuit and component-level
changes, a second-generation RF monitor could extend coverage by a factor of 7 in frequency
and 100 in pulse-capture speed (thus ∼70 GHz and 1 ns). A suggested path is described in
Appendix D (Finding 7 and Recommendation 7).
94
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
95
JSR-21-01 DoS Embassy Incidents February 11, 2022
7 AN INSTRUCTIVE ANALOG
Before concluding this report with our final remarks, we find it useful to underscore some
of the challenges we face with the methodologies applied to the Anomalous Health Inci-
dents in order to identify an underlying cause and to reach conclusions regarding deliberate
adversarial action.
In assessing and developing methodologies that might be applied to the current sit-
uation with the embassy anomalous health incidents (AHI), we found it informative to
apply our reasoning to a somewhat analogous data set. The goal here is to better under-
stand evaluation criteria for hypothesis testing, and the identification of an appropriate
comparison/control group. Also, we think it useful to have a parallel data set to which a
methodology can be applied, as a way to validate the overall approach.
Of all the U.S. missions overseas, according to Department of State Office of Inspection
reports ISP-I-19-14 and ISP-I-19-18, between 2012 and 2017 Colombia and Haiti reported
anomalously high rates of vehicular accidents. Some of these accidents had the unfortu-
nate consequence of injury and even death. Table 2 compares this accident data to the
AHI situation, in the context of physical injury mechanism, consistent medical effects, and
statistical incidence relative to an appropriate baseline rate.
One could postulate that one or more adversaries are intentionally causing accidents
involving U.S. embassy vehicles on the streets of Bogota and Port-au-Prince, on the basis
of this excess of incidents, and the clear connection of a physical mechanism to a long-
lasting adverse health effects. How might this hypothesis be evaluated? We would look for
a clear and well-defined physical mechanism, and a causally-associated cluster of medical
consequences. Both of those criteria are clearly satisfied in the vehicle accident situation,
much less so for AHI.
Another element of the evaluation of the hypothesis is to look at the rate of events
compared to the background, and see if there is an excess. There were bound to be two
embassies that had accident rates above that of other U.S. missions, and these two were
selected on that basis. As a result, the excess of events above the average for U.S. missions
96
JSR-21-01 DoS Embassy Incidents February 11, 2022
Table 2: (U) A comparison of the reported excess of accidents involving U.S. embassy
vehicles in Colombia and Haiti to AHI. If we require a clear physical mechanism and a clear
set of medical effects connected to it, any methodology that assigns a likelihood to nefarious
intentional action should be applied to both data sets, as a validation mechanism.
Vehicle Accident Analog AHI
Is there a clear physical process that can
lead to the reported lasting health effects? Y ?
Is there a common physiological/medical
outcome consistent with the physical insult? Y ?
Is the frequency of events larger than seen
in other U.S. embassies? Y ?
Is the frequency of events larger than seen
in the domestic US population? Y ?
Can we conclude that this situation is the
consequence of deliberate adversarial action? ? ?
is not significant. A more meaningful comparison might be to compare the accident rate
and medical consequences to that of broader background populations. One indicator is the
incidence of vehicular fatalities in the U.S., in Colombia, and Haiti. The deaths per year
per 100,000 population for those countries are 10.9, 21.3, and 15.5, respectively. But are
any of these the right control group? Are there adequate statistics to assess whether the
embassy cars collide more often than the typical car in the cities in which they drive?
This analog problem has some lessons we can extract: whatever methodology is applied
to the AHI situation to assess the likelihood of intentional adversarial action should also be
applied to analogous situations where we think we know the answer, to validate the logical
process and reasoning. Second, the definition of one or more appropriate control groups
is non-trivial, and requires careful consideration. Finally, beware of selection effects. In
this analog problem, it is easy to recognize that the two embassies were chosen due to high
accident rates. Other selection biases can be much more subtle and lead to the conclusion
of a spurious signal.
97
JSR-21-01 DoS Embassy Incidents February 11, 2022
8 COMBINED ASSESSMENT AND CONCLUDING RE-
MARKS
We conclude by summarizing our analyses and findings on the embassy health incidents
and potential mechanisms that could be responsible for them.
The results drawn from medical and incident data are inconclusive at this time and
do not point clearly to the presence of a new medical syndrome associated with AHI or to
a single cause underlying them. Instead, currently reported collection of symptoms could
result from a combination of structural causes, including those arising from other known
illnesses, functional disorders, as well as psychogenic response to stressful events.
Upon evaluation of incident data, JASON concluded that the incidents are highly het-
erogeneous; approximately 15% of the 200 cases share moderately common characteristics
and could potentially fit the working case definition. We used these incident files to arrive at
a set of “incident consistency criteria” that we used for evaluating potential causes. These
included the physical requirements for operations in the described environments and the
sensations and symptoms that have been reported. We also noted sensations and symptoms
reported to be absent in order to help establish a fully consistent link between potential
mechanisms and outcomes. Similarly, we used the medical data to identify symptoms and
delineate symptoms that should and should not be present in conjunction with a mechanism.
Focusing on mechanisms that could deliver energy from a distance, we evaluated acous-
tic and electromagnetic waves with regard to their potential to produce physiological damage
as well as any transient sensations that could be consistent with those reported and pre-
cipitate functional disorders. We summarize our combined assessment in Table 3. Color
coding indicates the level of consistency between the mechanism and the physical require-
ments, sensations, or symptoms considered: red denotes inconsistent/highly unlikely, yellow
denotes somewhat inconsistent/unlikely, and green denotes possibly consistent/likely.
Based on the fact that no mechanism is green across all columns, JASON judges none
of these mechanisms to be the likely cause underlying the health incidents. In particular,
acoustic waves at all frequencies, ionizing radiation, electromagnetic waves below 500 MHz
98
JSR-21-01 DoS Embassy Incidents February 11, 2022
and above 30 GHz can be ruled out with high confidence because there is at least one aspect
of generating or delivering that energy, or its effects on human health that is highly incon-
sistent with the incidents. Pulsed RF waves with carrier frequencies between 500 MHz and
30 GHz cannot be conclusively ruled out but appear unlikely given the ”yellow” designation
in multiple columns: the radiation flux needed to cause any physiological change in human
tissues, nerves, or organs, even temporarily, would have led to a sensation of skin heating or
pain, in contradiction to the reported sensations during incidents. In addition, the difficulty
of keeping an RF beam at these frequencies focused through structures such as walls and
the unlikely association of sounds produced by pulsed RF (e.g., through Frey effect) with
any damage mechanism render this possibility unconvincing. Finally, for incidents with
matching recorded sounds, RF as a source via the Frey effect can be ruled out conclusively.
Future data, especially if collected in a uniform and quantitative manner, may help
with the resolution of these anomalous incidents. Data collection falls into two categories.
First is searching directly for harmful signals through persistent monitoring. This effort
should be focused on potential signals that have not already been ruled out. Second is a
redesigned uniform incident questionnaire with numerical evaluation criteria that can be
employed at all embassies to provide an initial filtering of events and help guide appro-
priate medical and monitoring response. We recommend that only incidents that meet a
threshold to be included in an AHI database, to provide a robust sample on which studies
and hypothesis testing can be carried out. We also emphasize that judging a case to be
inconsistent with an anomalous health incident carries no implications about its importance
or severity. An inner ear infection, stroke, or onset of PPPD can be serious conditions that
require medical attention even if they are not AHI.
Regarding adversarial intent, it is not possible to conclude at this time that the events
reviewed by JASON are the result of intentional attacks that cause physical harm. However,
it is not possible, either, to rule out mechanisms that do not cause any physical harm but
which might constitute harassment and lead to health conditions and functional disorders,
for example through unpleasant sounds or pressure sensations. Given this, and in the
interest of protecting embassy personnel and their families, it would be prudent to be
vigilant against tactics intended to produce anxiety and trauma, with an intent to either
99
JSR-21-01 DoS Embassy Incidents February 11, 2022
disrupt operations and/or cause long-term harm. The US government could minimize the
effects of such tactics, if present, through open communication, education, and appropriate
rapid medical response to any conditions that develop.
Table 3: An Assessment of Potential Causes
1
Consistent Inconsistent Biological Barrier Focusing Production
Sensations Effects Mechanism Penetration
Ionizing neuro symp. skin burns cell and can can be bulky
Radiation nausea rad sickness DNA penetrate directed equipment
fatigue organ damage damage thick walls
no sounds
100 MHz tingling body heat neuro effects can penetrate beam large
RF muscular eff sensation unestablished building unlikely antenna
1−10 GHz sounds (Frey) skin heat neuro effects medium no beam medium
RF potential nerve electr/comm require penetration through antenna
stimulation disruption high flux capability barriers
30 GHz potential skin pain neuro penetration no beam small
RF thermal electr/comm effects challenging through antenna
effects disruption unestablished barriers
Infrasound headache whole body pressure minor loss no very large
nausea sensation &vibrations through focusing equipment
dizziness on organs walls
Audible hearing loss confusion ear damage significant directed large
Sound headache dizziness at high dB dB loss thru not localized equipment
no sound cases building for high dB
Ultrasound pressure, dizziness long-term hyperthermia significant directed large
tinnitus, headache effects acoustic dB loss thru not localized equipment
nausea, sounds cavitation building for high dB
1. A scoring of potential causes based on several biological and physical criteria in the context of the in-
cident and medical data relevant to the reported events. Color coding indicates the level of consistency
between the mechanism and the physical requirements, sensations, or symptoms considered: red denotes
inconsistent/highly unlikely, yellow denotes somewhat inconsistent/unlikely, and green denotes possibly con-
sistent/likely. “Inconsistent Effects” column refers to signs and symptoms that should have been present for
a given mechanism but are not reported or symptoms that are reported but cannot be produced by that
mechanism. There is an implied power level in this table for each of these remote energy mechanisms, which
is discussed in detail in the report, but assumed here to be the most optimal power level for its relevance
for the reported incidents and symptoms.
100
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
101
JSR-21-01 DoS Embassy Incidents February 11, 2022
A APPENDIX: Sample Incident Intake Questionnaire
This questionnaire is included to serve as an example for the types of quantitative, struc-
tured, and hypothesis-related questions that can be asked during an incident report. It is
not meant to be complete nor optimized; we recommend expanding and modifying it as
needed prior to deploying it widely. For example, some of the situational questions that
exist in the current reports are not repeated in their entirety here.
It is important to administer the same questionnaire to all personnel reporting inci-
dents as soon as possible after an event and equally important to ask them to answer every
question. A question left blank could lead to ambiguities about whether the individual did
not feel that symptom or did not know how to answer that question, defying the attempt
at gathering uniform and structured data.
1. Which of the following symptoms, if any, did you experience during the event?
2 Sound
2 Pressure (specify where on your body)
2 Heat (specify where on your body)
2 Pain (specify where on your body)
2 Smell (describe)
2 Physical sensation such as buffeting, vibration, etc.
2 Ringing in ears
2 Nausea
2 Dizziness
2 Headache
102
JSR-21-01 DoS Embassy Incidents February 11, 2022
2. If you answered yes to Sound in Question 1, check all that apply.
2 high-pitch
2 low pitch
2 continuous
2 intermittent
2 directional (specify how)
2 clicky. Specify if repetitive click, fast clicks, or slow clicks.
2 specify volume. Barely audible, conversation level, loud, or very loud.
2 Did you put your hands over your ears? If yes, did it make a difference?
3. If you answered yes to Heat in Question 1, please answer the following.
Circle the one that applies to the heat level: Slight flush, hot, very hot, painful
Circle the one that applies to the onset: Sudden onset, slow rise, intermittent
4. If you answered yes to Dizziness or Nausea in Question 1, please answer the following.
Circle the one that applies to the severity: General wooziness, sudden sweep through
your head, complete disorientation
Circle the one that applies to the onset: Sudden onset, slow rise, intermittent
Please specify the duration.
5. If you answered yes to headache in question 1, please specify type:
2 whole head
2 localized (specify where in your head)
2 throbbing
103
JSR-21-01 DoS Embassy Incidents February 11, 2022
2 squeezing
2 stabbing
6. Specify the duration of the event.
7. Did the sensations and symptoms change when you moved?
2 Yes 2 No 2 Maybe
If you answered yes or maybe, which ones and how?
8. Was anyone else present?
2 Yes 2 No
If yes, how far away were they from you?
If yes, did they have the same experience? 2 Yes 2 No 2 I don’t know.
9. Check any malfunction of electronics near you that was coincident with the event.
2 Computer reboot or a monitor change
2 Noise out of any device with audio function, such as computer, phone, etc. If yes,
specify.
2 Phone reboot, display changes, or disruption of communication
10. Check all that apply regarding your location during the incident.
2 Indoors
2 Outdoors
2 Single residence home
2 Apartment building
104
JSR-21-01 DoS Embassy Incidents February 11, 2022
2 Embassy or consulate
2 Other (please specify)
11. Please estimate the distance to the following:
Nearest building:
Nearest unit if within the same building:
Nearest street:
If indoors, nearest exterior wall or window:
12. How often do you get headaches?
2 fewer than once per month
2 once per month
2 once per week
2 several times per week
13. Have you ever suffered from or been diagnosed with any of the following? Check all
that apply.
2 Sinus infections
2 Hearing problems
2 Tinnitus
2 Migraines
2 Vertigo
2 Blood pressure problems (specify)
2 None of the above
105
JSR-21-01 DoS Embassy Incidents February 11, 2022
14. How were you feeling otherwise on the day of the incident? Report any illness or other
issue not typical for you on that day or in the days prior.
15. Have these symptoms
2 gotten better
2 stayed the same
2 gotten worse since the event?
If you responded ”stayed the same” or ”gotten worse”, please describe duration and severity.
16. Please describe if anything else out of the ordinary happened at the same time and just
before the event.
106
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
107
JSR-21-01 DoS Embassy Incidents February 11, 2022
B APPENDIX: Acoustic Waves, Fundamentals, and Detailed
Calculations
B.1 Sound Speed and Propagation
Acoustic waves are characterized by a sound speed,
a
s
=
Å
∂ p
∂ρ
ã
s
, (B-1)
the square root of the partial derivative of pressure at constant entropy with respect to
density. Under the assumption of an ideal gas, at ordinary temperatures, and using the
composition for dry air, this leads to
a
s
' 343 m/s = 343 Hz · m = 34.3 kHz · cm . (B-2)
Moist air, at a particular relative humidity (RH%), supports sound speeds that are some-
what higher than that of dry air. Relative humidity plays an important role in the propa-
gation of high-frequency sound, as we’ll discuss below.
The wavelength of the wave is related to its speed and frequency by
λ =
a
s
ν
. (B-3)
Considering a few values at different frequencies, we have λ ' 50 m at an infrasound fre-
quency of ν = 7 Hz, 20 m at 17 Hz, 34 cm ' 1 ft at an audio frequency of 1 kHz, 1.7 cm at
an ultrasonic frequency of 20 kHz, 1 cm at 34 kHz, and sub-millimeter wavelengths at MHz
frequencies often used for medical and material imaging (Fig. 5.1).
Using the standard equations for acoustic energy propagation, we define the acoustic
energy flux as
j
ac
= p
0
u
0
, (B-4)
with units of [ W/m
2
] (e.g., Thompson, 1972), where p
0
is the small acoustic pressure
perturbation in the medium and u
0
is the perturbation in the velocity field introduced by
the acoustic excitation. The (average) threshold of audibility for (healthy) human ears is
used as an acoustic-flux/-power reference and accepted as
j
ac,ref
= 10
−12
W/m
2
. (B-5)
108
JSR-21-01 DoS Embassy Incidents February 11, 2022
For healthy ears, this level is very nearly frequency-independent, especially at lower fre-
quencies within the audible acoustic frequency band.
This level is used to scale acoustic flux on a decibel (dB) sound power level (SPL),
SPL = 10 log
10
Å
j
ac
j
ac,ref
ã
, (B-6)
yielding SPL = 0 dB as the audible acoustic-flux threshold.
Finally, we define the acoustic impedance, which measures the resistance of a medium
to the propagation of acoustic waves, as the ratio
Z
ac
≡
p
0
|u
0
|
. (B-7)
For small perturbations, the impedance is equal to Z
ac
= ρ
0
a
s,0
, where ρ
0
is the density of
the unperturbed medium and a
s,0
is its sound speed.
Because the acoustic flux is quadratic in acoustic pressure, we may also express SPL
in terms of the latter,
1
SPL = 10 log
10
Ç
p
02
p
02
ref
å
= 20 log
10
Å
p
0
p
0
ref
ã
, (B-8)
with (B-5)
p
0
ref
' 20 µPa (rms) , (B-9)
corresponding to j
ac,ref
in (B-5) in air.
2
B.2 Acoustic reflection and transmission
Acoustic reflection at boundaries between two different media occurs for many reasons.
For semi-infinite media on either side of an interface, such phenomena can be ascribed to
differences in the acoustic impedance (B-7), Z
ac
, of the substance comprising the two media.
1
We note that ‘SPL’ is often referred to as the ‘sound pressure level’. We prefer ‘sound power level’ in
this document to avoid the confusion of the factor of 2 in its expression in terms of dB in (B-8).
2
Notable is how small audible rms threshold pressure levels are, as a fraction of ambient pressure, e.g.,
p
0
ref
/p
atm
' 2 × 10
−5
/10
5
' 2 × 10
−10
. Pain threshold is of the order of 140 dB, i.e., human ears serve as
sensors with a dynamic range over 14 orders of magnitude. Some animal ears are known to be better yet.
109
JSR-21-01 DoS Embassy Incidents February 11, 2022
In non-dispersive media, such as dry air, the sound speed (and therefore, its impedance)
has only a weak dependence on the sound wave frequency. The characteristic specific
acoustic impedance of air for audible acoustic waves at T = 25
◦
C is
Z
ac,air,au
≈ 410 kg/(m
2
· s) , (B-10a)
and only slightly higher for ultrasound waves, i.e.,
Z
ac,air,us
≈ 430 kg/(m
2
· s) . (B-10b)
Attenuation and dissipation by various mechanisms are more significant at high fre-
quencies, as described below, especially in dispersive media.
The acoustic impedance of a concrete (wall) for ultrasound waves, for example, is
Z
ac,us,conc
≈ 5 − 8 × 10
6
kg/(m
2
· s) , (B-10c)
measured with a transducer with a central frequency of ν = 0.5 MHz (Lotfi et al., 2013).
For glass, a typical value is,
3
Z
ac,glass
≈ 14 × 10
6
kg/(m
2
· s) . (B-10d)
At boundaries between media with different acoustic impedances, some of the wave
energy is reflected and some is transmitted. The acoustic amplitude reflection coefficient,
C
ac,r
, i.e, the ratio of the reflected pressure p
0
r
to the incident pressure p
0
i
, depends on the
acoustic impedances of the two media on either side of the boundary, Z
ac,1
and Z
ac,2
.
For waves at normal incidence to the boundary surface, the wave-amplitude reflection
coefficient at normal incidence is given by
p
0
r
p
0
i
= C
ac,r
=
Z
ac,2
− Z
ac,1
Z
ac,2
+ Z
ac,1
, (B-11a)
and can be both positive and negative, i.e., inducing a wave phase flip in the latter case.
Reflection of the sound intensity (energy-flux) is given by C
2
ac,r
.
3
https://www.bostonpiezooptics.com/ultrasonic-properties.
110
JSR-21-01 DoS Embassy Incidents February 11, 2022
Absent absorption/attenuation in the media and at their interface, the total energy
flux on the two sides of the interface will be conserved and we’ll have
C
2
ac,r
+ C
2
ac,t
= 1 , (B-11b)
where C
ac,t
is the wave-amplitude transmission coefficient. At a typical interface between
air and a solid wall, for example, this expression would give a near-zero transmitted acoustic
flux, indicating that the vast majority of the acoustic wave flux would be reflected rather
than transmitted into and through a semi-infinite adjoining high-Z
ac
medium.
The relations above are for normal incidence. Acoustic waves incident at an angle will
both reflect and refract with angles and amplitudes reminiscent of Snell’s Law of reflection
and refraction for optics (e.g., Thompson, 1972, Sec. 4.9).
An analysis of transmission of acoustic waves through finite-thickness slabs and bound-
aries with dissimilar materials, e.g., walls, window panes, etc., needs to take both the thick-
ness of the finite-thickness slab, say, d and the acoustic wavelength (B-3), λ, or wavenumber
k = 2π/λ, into account.
Considering a glass slab, as in a window pane, sound speed is much higher in glass than
in air, with a
ac,glass
' 4540 m/s, for example, yielding a sound-speed and a corresponding
wavelength ratio (at the same frequency) with respect to air of
a
ac,glass
a
ac,air
=
λ
ac,glass
λ
ac,air
'
4540
340
' 13.4 . (B-12a)
An ultrasonic acoustic wave with a
λ
ac,air
(ν=20 kHz) ' 1.7 cm (B-12b)
wavelength (B-3) in air, will then have a wavelength in glass given by
λ
ac,glass
(ν=20 kHz) ' 13.4 × 1.7 cm = 22.8 cm . (B-12c)
As a consequence, standing-wave resonances in window panes cannot occur at even ultra-
sonic frequencies of concern. Moreover, a typical window glass-pane thickness, d, will be
much smaller than λ
glass
, even at ultrasonic frequencies, and, as a consequence, will re-
spond as a slab comprised of an incompressible material, impeding transmission by its areal
density [ kg/m
2
], as discussed in App. B.4.
111
JSR-21-01 DoS Embassy Incidents February 11, 2022
Glass window panes are mounted in frames and, therefore, also respond to the equiva-
lent of drum-membrane frequencies that will be much lower, also depending on the damping
offered by the window-pane mount, e.g., low-damping aluminum vs. high-damping wood,
etc. Such frequencies will also depend on the glass thickness, density, elasticity, mounting
details, whether single-pane, double-pane with the latter using purposely different panes to
avoid cross-resonances, etc.
Ultrasonic waves are attenuated when propagating over long distances in air, especially
moist air, as discussed below, further impeding the ability to project them at a distance at
high frequency.
B.3 Sound attenuation and localization
Sound propagates effectively in air with an intensity (e.g., acoustic flux) that can decrease
either geometrically, by spreading, as from a point source at distances large compared to
its wavelength, or attenuated by dissipation through the action of bulk viscosity and heat
conductivity.
Dissipative (as opposed to geometric) attenuation of acoustic intensity/flux per unit
propagation length, x, i.e.,
I(x) = I
0
e
−αx
. (B-13a)
The attenuation coefficient, α, has units of reciprocal length and may be approximated with
a nominally quadratic scaling with frequency in dry air, i.e.,
α(ν) = α
0
(ν/ν
0
)
2
, (B-13b)
where α
0
is the attenuation specified at the reference frequency ν
0
.
Attenuation generally depends on the gas-mixture bulk viscosity and heat conductivity
with a frequency dependence that is more complicated in mixtures with molecules that
have energy modes with characteristic frequencies in the vicinity of sound frequencies. An
important example in our context is moist air where sound attenuation depends on relative
humidity (Fig. B.1), and other air-mixture constituents (e.g., Harris, 1966; Bass et al.,
1995).
112
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure B.1: Sound attenuation in dry and moist air, also illustrating the quadratic de-
pendence of the attenuation coefficient for dry air. Data from Sivian (1947). Figure from
Thompson (1972, Fig. 4.38).
Dissipative attenuation begins to become significant at frequencies of tens of kHz and
above, as can be seen in Fig. B.1, with attenuation of 20 kHz sound in dry air of order
α
dry air
' 6 × 10
−3
/m and α
air, 37%RH
' 8 × 10
−2
/m. For a collimated acoustic beam
traversing a distance x = 30 m, this gives
I(ν=20 kHz, x=30 m)
I
0
' e
−8×10
−2
×30
= e
−2.4
' − 10 dB , (B-14a)
for RH=37% air, and α
air, 37%RH
(40 kHz) ' 0.15/m
I(ν=40 kHz, x=30 m)
I
0
' e
−0.15×30
= e
−4.5
' − 20 dB , (B-14b)
illustrating the importance and role of water vapor content in sound attenuation in air.
Ultrasound can be projected and localized/focused in air in open spaces from such
distances, and perhaps from beyond, if generated with a large D/λ transmitter of adequate
power. Such a beam, projected from a distance into the interior of a building, might rely
on transmission through a glass window pane, and would undergo a total attenuation that
113
JSR-21-01 DoS Embassy Incidents February 11, 2022
is at least as high as the sum of the attenuation in air and that through the window:
−10 dB
| {z }
20 kHz, air
+ −40 dB
| {z }
window
' −50 dB
| {z }
20 kHz, total
, (B-15a)
and
−20 dB
| {z }
40 kHz, air
+ −40 dB
| {z }
window
' −60 dB
| {z }
40 kHz, total
. (B-15b)
as also shown graphically in Fig. B.2.
−10 dB @ 20 kHz
−20 dB @ 40 kHz
< −40 dB
< −50 dB @ 20 kHz
< −60 dB @ 40 kHz
𝑥 = 30 m
Air, 37% RH
Figure B.2: Notional sketch for the attenuation of a 20 kHz and a 40 kHz focused acoustic
wave, emanating from a source at a distance x = 30 m, propagating through air with a
37% RH, transmitted through a glass window pane into an adjacent interior of a building.
Acoustic energy flux spreading as a spherical wave across x = 30 m, from a localized
source that can be treated as a point source at the distance, with I
0
measured, say, at
x
src
= 1 m from the emitter, would have an amplitude that would spread as 1/r and
undergo an additional geometric intensity attenuation,
x
src
x
2
=
1
30
2
= −29.5 dB , (B-15c)
in an open space, i.e., excluding multi-path effects. This yields a total attenuation at least
as high as
−39.5 dB
| {z }
20 kHz, air
+ −40 dB
| {z }
window
' −79.5 dB
| {z }
20 kHz, total
, (B-15d)
and
−49.5 dB
| {z }
40 kHz, air
+ −40 dB
| {z }
window
' −89.5 dB
| {z }
40 kHz, total
. (B-15e)
See Fig. B.3.
114
JSR-21-01 DoS Embassy Incidents February 11, 2022
−39.5 dB @ 20 kHz
−49.5 dB @ 40 kHz
< −79.5 dB @ 20 kHz
< −89.5 dB @ 40 kHz
𝑥 = 30 m
< −40 dB
Air, 37% RH
Figure B.3: Notional sketch for the attenuation of a 20 kHz and a 40 kHz spherical acoustic
wave, emanating from a source at a distance x = 30 m, propagating through air with a
37% RH, transmitted through a glass window pane into an adjacent interior of a building.
In a realistic situation, the total attenuation would be bracketed by these bounding
values imposing, in either case, substantial requirements to the projection of significant
acoustic energy/power into the interior of a building or a vehicle, through closed, glass-pane
windows, on the transmitter power and size, if emission is intended to be covert/undetected.
Infrasound has a very low attenuation propagating through even moist air, with an
intensity that primarily decreases as a result of geometric dispersion, i.e., as a spherical
wave, for example, or refraction/ducting characteristics, diffuse scattering/reflection from
rough/irregular surfaces, etc. As a consequence and because of its low attenuation, in-
frasound can propagate to great distances and is useful in monitoring distant natural or
man-made explosions in the atmosphere, excitation by seismic waves, volcanic eruptions,
exploiting long-distance wave-guide ducting by the layered terrestrial atmosphere.
Generally, as a wave, sound can be focused, for example, to spatial extents scaled by
its wavelength. As a consequence, infrasound on human spatial scales cannot be localized to
meter-/sub-meter-scaled regions. In contrast, ultrasound can form beams and be collimated
(e.g., Fig. B.2), or even focused, by large-aperture reflectors or transmitters if the diameter
of their aperture, D, scaled by the sound wavelength is large, i.e., if D/λ 1, akin to
optics.
With short attendant wavelengths, ultrasound enables high-resolution medical imag-
ing. Such applications have helped establish safe operating limits and intensities, extending
115
JSR-21-01 DoS Embassy Incidents February 11, 2022
to fetuses, albeit employing direct, well-coupled means to human tissue.
B.4 Sound Transmission through a Slab
Transmission of sound at normal incidence through a slab of thickness d, e.g., a glass pane
or a wall, can be estimated in an ideal case where the transverse extent of the slab can be
taken as infinite and we have a right-going incident wave with a left-going reflected harmonic
wave (pressure) amplitude given by at x = 0,
p
0
i
and p
0
r
, (B-16a)
respectively, a right-going and a left-going wave through the slab material at x, given by,
p
0
+
e
ikx
and p
0
−
e
−ikx
, (B-16b)
with kd = 2πd/λ, and an outgoing, right-propagating transmitted wave in air, given by, as
it emerges at x = d,
p
0
t
e
ikd
(B-16c)
as depicted in Fig. B.4.
Given the incident-wave amplitude, p
0
i
, we can solve for the other four amplitudes by
matching the pressure and velocity on the left and right slab faces, with u
0
= ±p
0
/Z
ac
for
right- and left-going waves, respectively, i.e.,
p
0
i
+ p
0
r
= p
0
+
+ p
0
−
p
0
i
− p
0
r
Z
ac,air
=
p
0
+
− p
0
−
Z
ac,slab
(B-17a)
at x = 0, and
p
0
+
e
ikd
+ p
0
−
e
−ikd
= p
0
t
e
ikd
p
0
+
e
ikd
− p
0
−
e
−ikd
Z
ac,slab
=
p
0
t
e
ikd
Z
ac,air
,
(B-17b)
at x = d, or, if
ζ =
Z
ac,slab
Z
ac,air
, (B-17c)
116
JSR-21-01 DoS Embassy Incidents February 11, 2022
then
p
0
i
+ p
0
r
= p
0
+
+ p
0
−
p
0
i
− p
0
r
=
1
ζ
p
0
+
− p
0
−
(B-17d)
at x = 0, and, at x = d with ϕ = e
ikd
,
p
0
+
ϕ + p
0
−
ϕ
−1
= ϕ p
0
t
p
0
+
ϕ − p
0
−
ϕ
−1
= ζ ϕ p
0
t
.
(B-17e)
𝑝
𝑝
𝑝
e
𝑝
e
𝑝
e
𝑍
,
𝑍
,
𝑍
,
0 𝑑
𝑥
Figure B.4: Transmission of a sound wave at normal incidence on a slab, at a frequency ν,
with a wavenumber k
air
= 2π/λ
air
= 2πν/a
air
in air, through a slab of thickness d, with a
wavenumber k = k
slab
= 2π/λ
slab
= 2πν/a
slab
in the slab.
Solving (B-17e) for p
0
±
in terms of p
0
t
, we have
p
0
+
=
1
2
(1 + ζ) p
0
t
p
0
−
=
1
2
(1 − ζ) p
0
t
ϕ
2
.
(B-18a)
Substituting in (B-17d), we then have in terms of the transmitted wave
p
0
i
+ p
0
r
=
1
2
(1 + ζ) p
0
t
+
1
2
(1 − ζ) p
0
t
ϕ
2
p
0
i
− p
0
r
=
1
ζ
ï
1
2
(1 + ζ) p
0
t
−
1
2
(1 − ζ) p
0
t
ϕ
2
ò
(B-18b)
117
JSR-21-01 DoS Embassy Incidents February 11, 2022
or,
p
0
i
+ p
0
r
=
1
2
(1 + ζ) + (1 − ζ) ϕ
2
p
0
t
p
0
i
− p
0
r
=
1
2ζ
(1 + ζ) − (1 − ζ) ϕ
2
p
0
t
.
(B-18c)
Adding the two, we solve for the incident amplitude, p
0
i
required to yield a transmitted
amplitude p
0
t
, i.e.,
p
0
i
=
1
4
Å
1 +
1
ζ
ã
(1 + ζ) + (1 − ζ) ϕ
2
p
0
t
=
(ζ + 1)
2
−
ζ
2
− 1
ϕ
2
4ζ
p
0
t
,
(B-18d)
noting that p
0
t
/p
0
i
→ 1 as d → 0 and ϕ → 1, as it must. The corresponding intensities are
then given by,
p
0
i
2
=
î
(ζ + 1)
2
−
ζ
2
− 1
ϕ
2
óî
(ζ + 1)
2
−
ζ
2
− 1
ϕ
−2
ó
16ζ
2
p
0
t
2
=
(ζ + 1)
4
− (ζ + 1)
2
ζ
2
− 1
ϕ
2
+ ϕ
−2
+
ζ
2
− 1
2
16ζ
2
p
0
t
2
=
(ζ + 1)
4
− 2 (ζ + 1)
2
ζ
2
− 1
cos(2kd) +
ζ
2
− 1
2
16ζ
2
p
0
t
2
,
(B-18e)
for lossless interface transmission and reflection, and propagation through the slab medium.
As a consequence, even at ultrasonic frequencies, a typical window-pane thickness
(d ≈ 3/32
00
' 0.24 cm) is much smaller than the corresponding wavelengths in glass and the
argument in the cosine in (B-18e) will be small, i.e.,
cos(2kd) ≈ 1 − 2 (kd)
2
+ . . . . (B-19a)
In that limit, we’ll then have,
p
0
i
2
'
(ζ + 1)
4
− 2 (ζ + 1)
2
ζ
2
− 1
î
1 − 2 (kd)
2
ó
+
ζ
2
− 1
2
16ζ
2
p
0
t
2
,
=
ñ
1 +
(ζ + 1)
2
ζ
2
− 1
4ζ
2
(kd)
2
ô
p
0
t
2
,
'
ï
1 +
1
4
(ζkd)
2
ò
p
0
t
2
,
(B-19b)
118
JSR-21-01 DoS Embassy Incidents February 11, 2022
the latter relation for ζ 1.
Substituting for ζ from (B-17c), we then have,
p
0
i
2
'
ñ
1 +
1
4
Å
Z
ac,glass
kd
Z
ac,air
ã
2
ô
p
0
t
2
=
ñ
1 +
1
4
Å
ρ
glass
2πνd
ρ
air
a
air
ã
2
ô
p
0
t
2
(B-19c)
or,
p
0
i
2
' π
2
Å
ρ
glass
d
ρ
air
λ
air
ã
2
p
0
t
2
, (B-19d)
noting that ρ
glass
d is the areal density [ kg/m
2
] of the glass slab.
Substituting ρ
glass
' 2.5 × 10
3
kg/m
3
, d ' 3/32
00
= 0.24 cm, ρ
air
' 1.2 kg/m
3
, and
λ
air
(20kHz) = 1.7 cm, we then have,
π
2
Å
ρ
glass
d
ρ
air
λ
air
ã
2
' 8.2 × 10
5
, (B-20)
yielding a transmitted acoustic intensity ratio for ν = 20 kHz sound of
p
0
t
p
i
2
' 1.2 × 10
−6
' − 59 dB . (B-21)
and less at higher frequencies, or for a thicker glass plane.
Transmission of acoustic energy through an actual window with a glass pane can be
somewhat higher because the glass pane also acts as a drum membrane and can transmit
by vibrating as such as well through its mounting, frame, etc.
119
JSR-21-01 DoS Embassy Incidents February 11, 2022
C APPENDIX: Mathematical Details of the Frey Effect
For simplicity, consider a plane-parallel geometry with RF pulses normally incident from
x < 0, and the tissue of interest (brain or skin) a semi-infinite medium lying at x > 0. A
very short pulse heats this tissue (very slightly) at t = 0, resulting in an initial pressure
profile
δp(x, 0) = δp
max
e
−x/h
, x > 0. (C-1)
Here the prefix “δ” serves as reminder that the pressure perturbation is much less than
the background (atmospheric) pressure, which we presume to be uniform. We neglect
nonlinearities and treat this problem to first order in δp. Insofar as the sound speed in the
tissue is constant,
1
the general solution to the one-dimensional acoustic wave equation is
δp(x, t) = g
+
(x − t) + g
−
(x + c
s
t),
in which g
±
are arbitrary functions describing right- (g
+
) and left-going (g
−
) waves. The
corresponding velocity perturbation is
δv(x, t) =
g
+
(x − c
s
t) − g
−
(x + c
s
t)
ρc
s
,
in which ρ ≈ 1 g cm
−3
is the density of the tissue. By hypothesis, δv(x, 0) = 0 for all x ≥ 0,
and it therefore follows that g
+
and g
−
are the same function, i.e. g
+
(ξ) = g
−
(ξ) ≡ g(ξ).
A rigid boundary: In this case δv(0, t) = 0 for all t ≥ 0. It then follows that g(−c
s
t) =
g(+c
s
t), i.e. g(ξ) is symmetric about ξ = 0. Matching this to the initial condition (C-1),
one sees that for x h, the acoustic pulse propagating into the tissue has the pressure
profile
δp(x, t) =
δp
max
2
exp(|x − c
s
t|/h) . (C-2a)
The factor 1/2 comes from the fact that g(x ± c
s
t) contribute equally to δp at t = 0, but
g(x + c
s
t) becomes exponentially small at x ≈ c
s
t at t & h/c
s
.
1
We ignore the slight change in sound speed c
s
produced by the heating itself.
120
JSR-21-01 DoS Embassy Incidents February 11, 2022
A free boundary: In this case δp(0, t) ≈ 0 for t > 0, so that
2
−g(−ξ) = g(ξ), i.e. g(ξ)
is antisymmetric around ξ = 0. Instead of eq. (C-2a), the pulse propagating into the tissue
becomes
δp(x, t) =
δp
max
2
exp(|x − c
s
t|/h) × sign(x − c
s
t) . (C-2b)
To determine how much of this is audible, consider first the temporal Fourier trans-
forms of these pulses. At some fixed station x h, let τ = t − x/c
s
be a shifted time
variable, and δp(τ) the pressure profile measured at this station. The Fourier transform of
this is
δ˜p(f) ≡
∞
Z
−∞
δp(τ)e
2πifτ
=
f
s
f
2
+ f
2
s
δp
max
2π
for a rigid boundary; (C-3a)
=
−if
f
2
+ f
2
s
δp
max
2π
for a free boundary. (C-3b)
Here, again, f
s
= c
s
/2πh. Notice that in the case of a free boundary, the amplitude near
DC (i.e. f = 0) is small, because δp(τ ) integrates (almost) to zero.
The frequency response of the human ear is complicated and dependent on age, but
to demonstrate the scalings mentioned in the caveats above, a simple and mathematically
convenient low-pass filter will do, such as
H(f) =
f
2
0
f
2
+ f
2
0
. (C-4)
Here f
0
is a cutoff frequency, which might be 20 kHz for a healthy young person with clean
habits, and rather less for the author of this note. The normalization has been set so that
H(0) = 1.
2
More precisely, −g(−ξ) = g(ξ) × (1 − r)/(1 + r), with r = (ρc
s
)
air
/(ρc
s
)
tissue
1.
121
JSR-21-01 DoS Embassy Incidents February 11, 2022
The audible (low-pass-filtered) pressure profile is then
δˆp(τ) =
∞
Z
−∞
δ˜p(f)H(f)e
−2πifτ
df
=
δp
max
2
×
f
0
f
2
s
− f
2
0
Ä
f
s
e
−2πf
0
|τ|
− f
0
e
−2πf
s
|τ|
ä
if rigid; (C-5a)
=
δp
max
2
×
f
2
0
f
2
s
− f
2
0
Ä
e
−2πf
0
|τ|
− e
−2πf
s
|τ|
ä
sign(τ) if free. (C-5b)
For f
s
f
0
, the peak pressure occurs when f
−1
s
τ f
−1
0
and is reduced from its unfil-
tered value δp
max
/2 by factors ≈ f
0
/f
s
and the square of this for rigid and free boundaries,
respectively.
122
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
123
JSR-21-01 DoS Embassy Incidents February 11, 2022
D APPENDIX: BlueJay 2nd-gen RF Detector
This Appendix is designated as CUI
124
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
125
JSR-21-01 DoS Embassy Incidents February 11, 2022
E APPENDIX: Microwave Pulser
Given that there appears to have been no upsets of electronic devices (cellphones, televisions,
computer and audio equipment) at the time of the reported events, one is led to ask whether
the source could have been ultra-short pulsed RF, say with pulse widths in the nanosecond
or sub-nanosecond range. Such RF spreads its energy over a broad frequency range (∼1 GHz
or more), consequently with greatly reduced spectral energy density: so the resonances of
wiring within nearby electronic devices would be less strongly driven. Another way to
think about it is that the distributed series inductances and shunt capacitances
1
of the
victim device more easily attenuate the ultra-short pulses.
So it may be useful to conduct some animal model tests with a source of pulsed
microwaves, for example ∼ 10 GHz nanosecond-scale pulses at the kilowatt power level.
From the briefings we received, it appears that such test sources are not available, or at
least they have not been used to explore physiological effects in vivo.
We have identified commercially available technology that can satisfy these specifica-
tions; we detail these below.
E.1 Commercially Available Components
What’s needed is a source of low-level ns-pulsed microwave RF, followed by a microwave
power amplifier of adequate bandwidth to preserve the pulse.
E.1.1 The low-level pulsed RF
A first idea for an inexpensive pulsed source might be a CW oscillator, gated by an RF
switch. The oscillator is no problem, but the switch is: PIN diode switches are slow, and
even the fastest GaAs switches have rise and fall times in the tens of nanoseconds. Another
way to switch RF is to use a mixer, driving the (dc-coupled) IF port with the gating pulse:
the oscillator’s RF drives the RF port, and the gated burst emerges from the LO port.
1
Contemporary electronic devices make extensive use of ground planes in their multilayer circuit boards,
to ensure good signal integrity and to suppress coupling and radiated RF.
126
JSR-21-01 DoS Embassy Incidents February 11, 2022
This does work, but typically the RF–LO isolation in the off state is unimpressive, with
typical values of 40–50 dB. We’ve used this technique in applications where some off-state
feedthrough is acceptable. For our application here we could use a Marki Microwave MM1-
0212HSM (or -LSM) GaAs mixer (datasheet provided separately), with typical off-state
isolation of 57 dB over 2–12 GHz, driven by a low-level X-band signal generator.
A better source, if cost is not constrained, is an arbitrary function generator (“arb”);
for this task something like the Tektronix AWG70001B (datasheet provided separately),
which clocks its DAC at 50 Gsps and can generate waveforms to 20 GHz. This is high-end
instrumentation (as the saying goes, “if you have to ask the price, you can’t afford it”), but
a well equipped test lab might have one, and in any case one can always rent a unit for a
set of limited tests.
E.1.2 The power amplifier
Microwave power amplifiers tend to be narrowband, with tuned elements or cavities. But
there are available some kilowatt-scale amplifiers with 10% to 20% bandwidths.
An example of the latter is the L5850 helix traveling wave tube (TWT) from L3-
Harris: its specified band is 8.5–10.5 GHz, with 60 dB RF gain and 2 kW output power; it
is specified for “CW/Pulsed” operation, so it is running cool when driven with nanosecond
pulses at, say, audio rates (which would generate an audible Frey effect). So it takes just
2 mW drive (+3 dBm) for full 2 kW output. TWTs require high voltages (here −11 kV and
+7.7 kV), but they do the job.
2
For higher frequencies a gyrotron (a Russian invention)
could be used. For example, the VGB-8193 amplifier from CPI (Communications & Power
Industries) has 1.6 GHz bandwidth at 95 GHz, with 55 kW (peak) output power and 33dB
gain.
A solid-state amplifier uses more friendly operating voltages; a suitable unit is the
VSX3696 from CPI. It has a 10% bandwidth (9–10 GHz), 58 dB gain, and a saturated peak
RF output of 1.8 kW; it runs on a single 50 V supply (datasheet provided separately).
2
We requested a full datasheet, but have not received a response.
127
JSR-21-01 DoS Embassy Incidents February 11, 2022
These examples are based on limited searching on the Internet. Greater power levels
and shorter pulse widths can likely be obtained with commercially available components.
E.1.3 Getting to Megawatts
There is a substantial Russian literature on techniques for generating ns-scale repetitively
pulsed microwaves at megawatt peak RF levels. For example, an article by Efremov et
al. (“Generation and Radiation of High-Power Ultrawideband Nanosecond Pulses”) reports
E
p
R products of 750 kV in nanosecond L-band pulsed radiation, and antenna far-field pat-
terns of some 25
◦
FWHM (with impressively low sidelobes). This and similar publications
use nonlinear techniques of pulse shortening – spark-gap switches (singly or in sequence), or
ferrite-loaded nonlinear transmission lines, or avalanche breakdown of semiconductor diodes.
Some of these publications feature compact sources (“Small-Sized Nanosecond Source of
Powerful Wide-Band VUV–UV Radiation”)
Figures E.1–E.3 show representative data from another Efremov et al. publication
(“High-Power Sources of Ultra-wideband Radiation with Subnanosecond Pulse Lengths”).
Notable are the high radiated power levels (Fig. E.1), the sub-nanosecond RF pulse width
(Fig. E.2), and the good directivity of the antenna array (Fig. E.3).
Figure E.1: Radiated potential (E
R
· R) and drive pulse voltage of Efremov et al. (2010).
128
JSR-21-01 DoS Embassy Incidents February 11, 2022
Figure E.2: RF pulse waveforms of Efremov et al. (2010).
Figure E.3: RF radiation pattern of the array of Efremov et al. (2010).
129
JSR-21-01 DoS Embassy Incidents February 11, 2022
Additionally, some published articles from the same organization (Institute of High-
Current Electronics, Tomsk, Russia) refer to observations of “different biological objects”
to pulsed RF (e.g., Romanchenko et al., “Gyromagnetic RF source for interdisciplinary
research”)
E.2 Portable Power Sources
Could a carryable battery provide the dc power needed for a kilowatt-to-megawatt ns-pulsed
mm-wave illuminator? The average energy requirements are surprisingly modest, and in
fact there are abundant battery packs, intended for video camcorders, lighting equipment,
and the like that would do the job in a compact human-carried package.
To evaluate this, condider a baseline scenario of a P
pk
=10 kW (peak power) mm-wave
(say 95 GHz) source producing a train of τ =10 ns RF pulses at a f
rep
= 1 kHz rep rate, with
a conservative η =10% dc-to-RF efficiency. The RF frequency and pulse width are sized to
avoid detection by currently fielded surveillance devices and to allow narrow beamforming
in a compact radiator, and the rep rate is chosen in the range of perceived audio symptoms.
The average power required is P
avg
=P
pk
τf
rep
/η, or 1 watt. Scaling up to 1 megawatt
peak power raises the average dc power requirement to 100 W. For suitable battery packs
one might go to the canonical website that caters to professional photographers (B&H
Photo and Electronics Corporation: bhphotovideo.com). There are listed many choices,
for example the Bescor PRB-9NC battery belt (12V, 9A-hr), the Cool-Lux 12V 200 W-hr
LCE battery belt, or the larger (shoulder-slung) Bescor 12V 20A-hr battery pack for studio
lighting. These currently retail for $77, $330, and $215 respectively, and would power the
megawatt configuration for about an hour of continuous operation.
One could extend this theme to a mobile (van-carried) illuminator, where some two
orders of magnitude more battery energy would be feasible by simply configuring series-
parallel arrangements of the exemplar battery packs above. However, there are battery
packs intended for substantially larger power and energy requirements. For example, lead-
acid batteries intended for marine applications can be purchased (e.g., from Amazon) with
130
JSR-21-01 DoS Embassy Incidents February 11, 2022
ratings of 24V/77A-hr (thus 1.8kW-hr), or 12V/250A-hr (thus 3kW-hr), providing an order
of magnitude or more of energy storage compared with the belt-packs above. Another
example is the Tesla “powerwall,” with 13.5kW-hr capacity; it includes charging and dc-to-
ac inverters, that is, 120V/240V 60 Hz ac for charging and during discharge. Although one
can’t buy these things singly, they currently cost about $10k, and they are large and heavy
(4.5 cu-ft, 250 lb), but they establish a convenient benchmark for considering what could
be run from batteries in a vehicle.
Of course, one could as well operate a generator (from the vehicle’s engine, or an
auxiliary power generator), producing a few kilowatts of continuous power. However, given
that a human-portable battery pack would suffice, in terms of power requirements even for
a pulsed megawatt illuminator, the only essential role of a van would be to house the RF
hardware itself, which is likely to be bulkier and heavier than the power source; moreover,
one would want to conceal from sight the antenna and RF hardware.
131
JSR-21-01 DoS Embassy Incidents February 11, 2022
F APPENDIX: Classified (S)
This Appendix is classified at the SECRET Level
132
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
133
JSR-21-01 DoS Embassy Incidents February 11, 2022
G APPENDIX: Classified (TS//SCI)
This Appendix is classified at TOP SECRET//SCI Level
134
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
135
JSR-21-01 DoS Embassy Incidents February 11, 2022
H APPENDIX: Briefings
This Appendix, listing the summer 2021 briefings to JASON, is delivered as a separate
document.
136
JSR-21-01 DoS Embassy Incidents February 11, 2022
This Page Intentionally Left Blank
137
JSR-21-01 DoS Embassy Incidents February 11, 2022
References
Adair, E. R. & Black, D. R. (2003). Thermoregulatory responses to rf energy absorption.
Bioelectromagnetics, 24(S6), S17–S38.
Adair, E. R., Blick, D. W., Allen, S. J., Mylacraine, K. S., Ziriax, J. M., & Scholl, D. M.
(2005). Thermophysiological responses of human volunteers to whole body rf exposure
at 220 mhz. Bioelectromagnetics, 26(6), 448–461.
Allingham, J. R. (1964). A descriptive study of communication networks and rumor diffusion
aboard a Navy ship. PhD thesis, Citeseer.
Ascone, L., Kling, C., & Wieczorek, J. e. (2021). A longitudinal, randomized experimental
pilot study to investigate the effects of airborne infrasound on human mental health,
cognition, and brain structure. Nature Sci Rep, 11, 3190. https://doi.org/10.1038/s41598-
021-82203-6.
Backstrom, M. & Lovstrand, K. (2004). Susceptibility of electronic systems to high-power
microwaves: summary of test experience. IEEE Transactions on Electromagnetic Com-
patibility, 46(3), 396–403.
Baliatsas, C., van Kamp, I., van Poll, R., & Yzermans, J. (2016). Health effects from
low-frequency noise and infrasound in the general population: Is it time to listen?
a systematic review of observational studies. Science Total Env., 557-558, 163–169.
http://dx.doi.org/10.1016/j.scitotenv.2016.03.065.
Baloh, R. W. (1997). Neurotology of migraine. Headache: The Journal of Head and Face
Pain, 37(10), 615–621.
Baloh, R. W. & Bartholomew, R. E. (2020). Havana Syndrome: Mass Psychogenic Illness
and the Real Story Behind the Embassy Mystery and Hysteria. Springer Nature.
Barnes, F. S. & Greenebaum, B. (2007). Biological and medical aspects of electromagnetic
fields. CRC press, third edition.
Bartholomew, R. E. (2000). Re:“epidemic hysteria: A review of the published literature”.
American journal of epidemiology, 151(2), 206–207.
138
JSR-21-01 DoS Embassy Incidents February 11, 2022
Bartholomew, R. E., Wessely, S., & Rubin, G. J. (2012). Mass psychogenic illness and the
social network: is it changing the pattern of outbreaks? Journal of the Royal Society of
Medicine, 105(12), 509–512.
Bas, C. U., Wang, R., Sangodoyin, S., Choi, T., Hur, S., Whang, K., Park, J., Zhang, C. J.,
& Molisch, A. F. (2019). Outdoor to indoor propagation channel measurements at 28
ghz. IEEE Transactions on Wireless Communications, 18(3), 1477–1489.
Bass, H., Sutherland, L., Zuckerwar, A., Blackstock, D., & Hester, D. (1995). Atmospheric
absorption of sound: Further developments. J. Acoust. Soc. Amer., 97, 680–683. doi:
10.1121/1.412989.
Beh, S. C. & Friedman, D. I. (2019). Acute vestibular migraine treatment with noninvasive
vagus nerve stimulation. Neurology, 93(18), e1715–e1719.
Benbadis, S. R. & Hauser, W. A. (2000). An estimate of the prevalence of psychogenic
non-epileptic seizures. Seizure, 9(4), 280–281.
Berger, R., Ashtiani, P., Ollson, C., Aslund, M., McCallum, L., Leventhall, G., & Knop-
per, L. (2015). Health-based audible noise guidelines account for infrasound and low-
frequency noise produced by wind turbines. Frontiers in Public Health, 3(31), 1–14. doi:
10.3389/fpubh.2015.00031.
Blick, D. W., Adair, E. R., Hurt, W. D., Sherry, C. J., Walters, T. J., & Merritt, J. H. (1997).
Thresholds of microwave-evoked warmth sensations in human skin. Bioelectromagnetics:
Journal of the Bioelectromagnetics Society, The Society for Physical Regulation in Biology
and Medicine, The European Bioelectromagnetics Association, 18(6), 403–409.
Bolin, K., Bluhm, G., Eriksson, G., & Nilsson, M. (2011). Infrasound and low frequency
noise from wind turbines: exposure and health effects. Environ. Res. Lett., 6, 035103
(6pp). doi:10.1088/1748-9326/6/3/035103.
Broderick, J. E., Kaplan-Liss, E., & Bass, E. (2011). Experimental induction of psychogenic
illness in the context of a medical event and media exposure. American journal of disaster
medicine, 6(3), 163.
139
JSR-21-01 DoS Embassy Incidents February 11, 2022
Brodeur, P. (1977). zapping of America. Norton.
Cain, C. A. & Rissmann, W. J. (1978). Mammalian auditory responses to 3.0 ghz microwave
pulses. IEEE Transactions on Biomedical Engineering, (3), 288–293.
Camp, M., Garbe, H., & Nitsch, D. (2001). Uwb and emp susceptibility of modern elec-
tronics. In 2001 IEEE EMC International Symposium. Symposium Record. International
Symposium on Electromagnetic Compatibility (Cat. No.01CH37161), volume 2 (pp. 1015–
1020 vol.2).
Ceylan, E., Ozpolat, C., Onur, O., Akoglu, H., & Denizbasi, A. (2019). Initial and sustained
response effects of 3 vagal maneuvers in supraventricular tachycardia: a randomized,
clinical trial. The Journal of emergency medicine, 57(3), 299–305.
Della Sala, S. & Cubelli, R. (2018). Alleged “sonic attack” supported by poor neuropsy-
chology. Cortex, 103, 387–388.
Dieterich, M., Obermann, M., & Celebisoy, N. (2016). Vestibular migraine: the most
frequent entity of episodic vertigo. Journal of neurology, 263(1), 82–89.
Elder, J. & Chou, C. (2003). Auditory response to pulsed radiofrequency energy. Bioelec-
tromagnetics, 24(S6), S162–S173.
Elwood, J. M. (2012). Microwaves in the cold war: the moscow embassy study and its
interpretation. review of a retrospective cohort study. Environmental Health, 11(1), 1–10.
Eren, O. E., Filippopulos, F., S¨onmez, K., M¨ohwald, K., Straube, A., & Sch¨oberl, F. (2018).
Non-invasive vagus nerve stimulation significantly improves quality of life in patients with
persistent postural-perceptual dizziness. Journal of neurology, 265(1), 63–69.
Fletcher, M., Jones, S., White, P., Dolder, C., Leighton, T., & Lineton, B. (2018). Effects
of very high-frequency sound and ultrasound on humans. Part I: Adverse symptoms after
exposure to audible very-high frequency sound. J. Acoust. Soc. Amer., 144, 2511–2520.
doi: 10.1121/1.5063819.
140
JSR-21-01 DoS Embassy Incidents February 11, 2022
Foster, K., Kritikos, H., & Schwan, H. (1978). Effect of surface cooling and blood flow
on the microwave heating of tissue. IEEE Transactions on Biomedical Engineering, (3),
313–316.
Foster, K. R. (2000). Thermal and nonthermal mechanisms of interaction of radio-frequency
energy with biological systems. IEEE Transactions on Plasma Science, 28(1), 15–23.
Foster, K. R. & Finch, E. D. (1974). Microwave hearing: evidence for thermoacoustic
auditory stimulation by pulsed microwaves. Science, 185(4147), 256–258.
Foster, K. R., Lozano-Nieto, A., Riu, P. J., & Ely, T. S. (1998). Heating of tissues by
microwaves: A model analysis. Bioelectromagnetics: Journal of the Bioelectromagnetics
Society, The Society for Physical Regulation in Biology and Medicine, The European
Bioelectromagnetics Association, 19(7), 420–428.
Foster, K. R., Zhang, H., & Osepchuk, J. M. (2010). Thermal response of tissues to millime-
ter waves: implications for setting exposure guidelines. Health physics, 99(6), 806–810.
Foster, K. R., Ziskin, M. C., & Balzano, Q. (2016). Thermal response of human skin to
microwave energy: a critical review. Health physics, 111(6), 528–541.
Foster, K. R., Ziskin, M. C., Balzano, Q., & Hirata, A. (2018). Thermal analysis of averaging
times in radio-frequency exposure limits above 1 ghz. IEEE Access, 6, 74536–74546.
Frey, A. H. (1962). Human auditory system response to modulated electromagnetic energy.
Journal of Applied Physiology, 17(4), 689–692.
Frey, A. H. (1963). Some effects on human subjects of ultra-high-frequency radiation. Am
J Med Electron, 2, 28–31.
Frey, A. H. & Etchert, E. S. (1985). Psychophysical analysis of microwave sound perception.
Journal of Bioelectricity, 4(1), 1–14.
Frey, A. H. & Messenger, R. (1973). Human perception of illumination with pulsed
ultrahigh-frequency electromagnetic energy. Science, 181(4097), 356–358.
Fry, W. (1952). Mechanism of acoustic absorption in tissue. J. Acoust. Soc. Amer., 24(4),
412–415.
141
JSR-21-01 DoS Embassy Incidents February 11, 2022
Gabriel, S., Lau, R., & Gabriel, C. (1996). The dielectric properties of biological tissues: Iii.
parametric models for the dielectric spectrum of tissues. Physics in medicine & biology,
41(11), 2271.
Gournay, L. S. (1966). Human perception of illumination with pulsed ultrahigh-frequency
electromagnetic energy. The Journal of the Acoustical Society of America, 40(6), 1322–
1330.
Gowrishankar, T., Stewart, D. A., Martin, G. T., & Weaver, J. C. (2004). Transport lattice
models of heat transport in skin with spatially heterogeneous, temperature-dependent
perfusion. BioMedical Engineering OnLine, 3(1), 1–17.
Guy, A. W., Chou, C., Lin, J. C., & Christensen, D. (1975). Microwave-induced acoustic
effects in mammalian auditory systems and physical materials. Annals of the New York
Academy of Sciences, 247(1), 194–218.
Harris, C. (1966). Absorption of sound in air versus humidity and temperature. J. Acoust.
Soc. Amer., 40(1), 148–159.
Hilton, D. & Shermetaro, C. (2021). Migraine-Associated Vertigo. StatPearls Publishing.
Hirata, A., Morita, M., & Shiozawa, T. (2003). Temperature increase in the human head
due to a dipole antenna at microwave frequencies. IEEE Transactions on Electromagnetic
Compatibility, 45(1), 109–116.
Hurt, W. D. (1996). Radio Frequency Radiation Dosimetry Workshop: Present Status and
Recommendations for Future Research. Technical report, ARMSTRONG LAB BROOKS
AFB TX OCCUPATIONAL AND ENVIRONMENTAL HEALTH DIRECTORATE.
IEEE Standards, S. (2019). Ieee standard for safety levels with respect to human exposure
to electric, magnetic, and electromagnetic fields, 0 hz to 300 ghz. IEEE Std C95.1-2019
(Revision of IEEE Std C95.1-2005/ Incorporates IEEE Std C95.1-2019/Cor 1-2019), (pp.
1–312).
Jauchem, J. & Cook, M. (2007). High-intensity acoustics for military nonlethal
applications: A lack of useful systems. Military Medicine, 172(2), 182–189.
https://doi.org/10.7205/MILMED.172.2.182.
142
JSR-21-01 DoS Embassy Incidents February 11, 2022
Jiang, H. (2020). Thermoacoustic Tomography. IOP Publishing.
Jones, T. F. (2000). Mass psychogenic illness: role of the individual physician. American
family physician, 62(12), 2649–2653.
Kasprzak, C. (2013). The influence of infrasounds on humans. 20
th
International Congress
on Sound and Vibration. ICSV20, Bankok, Thailand (7-11 July 2013).
Leighton, T. (2016). Are some people suffering as a result of increasing mass ex-
posure of the public to ultrasound in air? Proc. Roy. Soc. A, 472, 20150624.
http://dx.doi.org/10.1098/rspa.2015.0624.
Leighton, T., Lineton, B., Dolder, C., & Fletcher, M. (2020). Public exposure to
airborne ultrasound and very high frequency sound. Acoustics Today, 16, 17–26.
https://doi.org/10.1121/AT.2020.16.3.17.
Lempert, T., Olesen, J., Furman, J., Waterston, J., Seemungal, B., Carey, J., Bisdorff, A.,
Versino, M., Evers, S., & Newman-Toker, D. (2012). Vestibular migraine: diagnostic
criteria. Journal of Vestibular Research, 22(4), 167–172.
Lilienfeld, A. M., Tonascia, J., Tonascia, S., Libauer, C., Cauthen, G., Markowitz, J., &
Weida, S. (1978). Foreign service health status study-evaluation of health status of foreign
service and other employees from selected eastern european posts. Final Report (Contract
number 6025-619073) to the US Dept of State.
Lin, J. C. & Michaelson, S. M. (1987). Biological effects and health implications of radiofre-
quency radiation. Springer Science & Business Media.
Lotfi, H., Faiz, B., & Moudden, A. (2013). Characterization of the acoustic impedance
of mortar using ultrasonic technique. J. Civil Eng. Res., 3(1), 46–51. doi:
10.5923/j.jce.20130301.05.
Lyons, L. (2008). Open statistical issues in particle physics. The Annals of Applied Statistics,
2(3), 887–915.
Macc`a, I., Scapellato, M., Carrieri, M., Maso, S., Trevisan, A., & Bartolucci, G. (2015).
High-frequency hearing thresholds: effects of age, occupational ultrasound and noise
143
JSR-21-01 DoS Embassy Incidents February 11, 2022
exposure. Int. Arch. Occup. Environ. Health, 88, 197–211. doi: 10.1007/s00420-014-
0951-8.
Marano, E., Marcelli, V., Stasio, E. D., Bonuso, S., Vacca, G., Manganelli, F., Marciano,
E., & Perretti, A. (2005). Trigeminal stimulation elicits a peripheral vestibular imbalance
in migraine patients. Headache: the Journal of Head and Face Pain, 45(4), 325–331.
Mariotto, S., Farinazzo, A., Magliozzi, R., Alberti, D., Monaco, S., & Ferrari, S. (2018).
Serum and cerebrospinal neurofilament light chain levels in patients with acquired pe-
ripheral neuropathies. Journal of the Peripheral Nervous System, 23(3), 174–177.
McAfee, R. D. (1962). Physiological effects of thermode and microwave stimulation of
peripheral nerves. American Journal of Physiology-Legacy Content, 203(2), 374–378.
McAffee, R. (1971). Analeptic effect of microwave irradiation on experimental animals
(correspondence). IEEE Transactions on Microwave Theory and Techniques, 19(2), 251–
253.
McGlynn, D. J. (2005). Facing the leadership challenges on decommissioning United States
Navy ships. University of San Diego.
Meregalli, C., Fumagalli, G., Alberti, P., Canta, A., Chiorazzi, A., Monza, L., Pozzi, E.,
Carozzi, V. A., Blennow, K., Zetterberg, H., et al. (2020). Neurofilament light chain:
A specific serum biomarker of axonal damage severity in rat models of chemotherapy-
induced peripheral neurotoxicity. Archives of toxicology, 94(7), 2517–2522.
Moen, B. E., Møllerløkken, O. J., Bull, N., Oftedal, G., & Mild, K. H. (2013). Accidental
exposure to electromagnetic fields from the radar of a naval ship: a descriptive study.
International maritime health, 64(4), 177–182.
Muth, C. C. & Lewis, S. L. (2018). Neurological Symptoms Among US Diplomats in Cuba.
JAMA, 319(11), 1098–1100.
National Academies of Sciences, Engineering, and Medicine (2020). An Assessment of
Illness in U.S. Government Employees and Their Families at Overseas Embassies. The
National Academies Press.
144
JSR-21-01 DoS Embassy Incidents February 11, 2022
Neff, B. A., Staab, J. P., Eggers, S. D., Carlson, M. L., Schmitt, W. R., Van Abel, K. M.,
Worthington, D. K., Beatty, C. W., Driscoll, C. L., & Shepard, N. T. (2012). Auditory and
vestibular symptoms and chronic subjective dizziness in patients with meniere’s disease,
vestibular migraine, and meniere’s disease with concomitant vestibular migraine. Otology
& Neurotology, 33(7), 1235–1244.
Neuhauser, H. (2016). The epidemiology of dizziness and vertigo. Handbook of clinical
neurology, 137, 67–82.
Nitsch, D., Camp, M., Sabath, F., ter Haseborg, J., & Garbe, H. (2004). Susceptibility
of some electronic equipment to HPEM threats. IEEE Transactions on Electromagnetic
Compatibility, 46(3), 380–389.
Page, L. A., Keshishian, C., Leonardi, G., Murray, V., Rubin, G. J., & Wessely, S. (2010).
Frequency and predictors of mass psychogenic illness. Epidemiology, 21(5), 744–747.
Persinger, M. (2014). Infrasound, human health, and adaptation: an integrating overview
of recondite hazards in a complex environment. Nat. Hazards, 70, 501–525. doi:
10.1007/s11069-013-0827-3.
Prier, J. (2017). Commanding the trend: Social media as information warfare. Strategic
Studies Quarterly, 11(4), 50–85.
Riu, P. J., Foster, K. R., Blick, D. W., & Adair, E. R. (1997). A thermal model for
human thresholds of microwave-evoked warmth sensations. Bioelectromagnetics: Journal
of the Bioelectromagnetics Society, The Society for Physical Regulation in Biology and
Medicine, The European Bioelectromagnetics Association, 18(8), 578–583.
Rizk, H. G., Sharon, J. D., Lee, J. A., Thomas, C., Nguyen, S. A., & Meyer, T. A. (2020).
Cross-sectional analysis of cognitive dysfunction in patients with vestibular disorders. Ear
and hearing, 41(4), 1020–1027.
Roos, A. J. (1943). The scuttle butt afloat: a study in group psychology. Archives of
Neurology & Psychiatry, 50(4), 472–474.
145
JSR-21-01 DoS Embassy Incidents February 11, 2022
R¨oschmann, P. (1991). Human auditory system response to pulsed radiofrequency energy
in rf coils for magnetic resonance at 2.4 to 170 mhz. Magnetic resonance in medicine,
21(2), 197–215.
Sapkota, R. P., Brunet, A., & Kirmayer, L. J. (2020). Characteristics of adolescents affected
by mass psychogenic illness outbreaks in schools in nepal: A case-control study. Frontiers
in Psychiatry, 11, 1271.
Schwan, H. P. (1971). Interaction of microwave and radio frequency radiation with biological
systems. IEEE Transactions on microwave theory and techniques, 19(2), 146–152.
Shahim, P., Hallett, M., & Chan, L. (2021). NIH study on traumatic brain injury. (unpub-
lished).
Sheth, N., Dell’Osso, L., Leigh, R., Van Doren, C., & Peckman, H. (1995). The effects of
afferent stimulation on congenital nystagmus foveation periods. Vision research, 35(16),
2371–2382.
Singer, P. W. & Brooking, E. T. (2018). LikeWar: The weaponization of social media.
Eamon Dolan Books.
Sivian, L. (1947). High frequency absorption in air and other gases. J. Acoust. Soc. Amer.,
19(5), 914–916. https://doi.org/10.1121/1.1916640.
Staab, J. P., Eckhardt-Henn, A., Horii, A., Jacob, R., Strupp, M., Brandt, T., & Bron-
stein, A. (2017). Diagnostic criteria for persistent postural-perceptual dizziness (pppd):
Consensus document of the committee for the classification of vestibular disorders of the
b´ar´any society. Journal of Vestibular Research, 27(4), 191–208.
Staab, J. P. & Ruckenstein, M. J. (2007). Expanding the differential diagnosis of chronic
dizziness. Archives of Otolaryngology–Head & Neck Surgery, 133(2), 170–176.
Stephens, R. & Bate, A. (1966). Acoustics and Vibrational Physics. Edward Arnold Pub-
lishers, London, United Kingdom.
Swanson, Randel L., I., Hampton, S., Green-McKenzie, J., Diaz-Arrastia, R., Grady, M. S.,
Verma, R., Biester, R., Duda, D., Wolf, R. L., & Smith, D. H. (2018). Neurological
146
JSR-21-01 DoS Embassy Incidents February 11, 2022
Manifestations Among US Government Personnel Reporting Directional Audible and
Sensory Phenomena in Havana, Cuba. JAMA, 319(11), 1125–1133.
Thiel, M. & Sarabandi, K. (2009). 3d-wave propagation analysis of indoor wireless channels
utilizing hybrid methods. IEEE Transactions on Antennas and Propagation, 57(5), 1539–
1546.
Thompson, P. (1972). Compressible-Fluid Dynamics. McGraw-Hill Inc., New York, NY.
Tonin, R., Brett, J., & Colagiuri, B. (2016). The effect of infrasound and negative expec-
tations to adverse pathological symptoms from wind farms. J. Low Frequency Noise,
Vibration and Active Control, 35(1), 77–90. doi: 10.1177/0263092316628257.
Ueda, M., Ota, A., & Takahashi, H. (2014). Investigation on high frequency noise in public
space — we tried noise abatement measures for displeasure people. In Proceedings, 7
th
Forum Acusticum (Krakow, Poland, 7–12 September 2014).
Vanni, S., Pecci, R., Edlow, J. A., Nazerian, P., Santimone, R., Pepe, G., Moretti, M.,
Pavellini, A., Caviglioli, C., Casula, C., et al. (2017). Differential diagnosis of vertigo
in the emergency department: a prospective validation study of the standing algorithm.
Frontiers in neurology, 8, 590.
Vass, Z., Steyger, P., Hordichok, A., Trune, D., Jancso, G., & Nuttall, A. (2001). Capsaicin
stimulation of the cochlea and electric stimulation of the trigeminal ganglion mediate
vascular permeability in cochlear and vertebro-basilar arteries: a potential cause of inner
ear dysfunction in headache. Neuroscience, 103(1), 189–201.
Vilagosh, Z., Lajevardipour, A., & Wood, A. (2018). Simulations of the penetration of 60–
300 ghz radiation into the human ear. In 2018 43rd International Conference on Infrared,
Millimeter, and Terahertz Waves (IRMMW-THz) (pp. 1–2).: IEEE.
Vilagosh, Z., Lajevardipour, A., & Wood, A. (2020). Computer simulation study of the
penetration of pulsed 30, 60 and 90 ghz radiation into the human ear. Scientific reports,
10(1), 1–10.
Weir, E. (2005). Mass sociogenic illness. Cmaj, 172(1), 36–36.
147
JSR-21-01 DoS Embassy Incidents February 11, 2022
Wells, J., Kao, C., Konrad, P., Milner, T., Kim, J., Mahadevan-Jansen, A., & Jansen,
E. D. (2007). Biophysical mechanisms of transient optical stimulation of peripheral nerve.
Biophysical journal, 93(7), 2567–2580.
West, L. J. (2016). # jihad: Understanding social media as a weapon. Security Challenges,
12(2), 9–26.
Wright, R. M. & Ramesh, K. (2012). An axonal strain injury criterion for traumatic brain
injury. Biomechanics and modeling in mechanobiology, 11(1), 245–260.
Yeh, W. Z., Blizzard, L., & Taylor, B. V. (2018). What is the actual prevalence of migraine?
Brain and behavior, 8(6), e00950.
Zasowski, T., Meyer, G., Althaus, F., & Wittneben, A. (2006). Uwb signal propagation
at the human head. IEEE Transactions on Microwave Theory and Techniques, 54(4),
1836–1845.
Zhang, X., Yarema, K., & Xu, A. (2017). Biological effects of static magnetic fields. Springer.
148
