Homeopathic pathogenetic trials
produce more specific than
non-specific symptoms: results
from two double-blind placebo
H Walach School of Social Sciences and European Office of the Samueli Institute for Information Biology, University of Northampton, Northampton,
UK; Academic Section on the Evaluation of Complementary Medicine, Institute for Environmental Medicine and Hospital Epidemiology, University Hospital
Freiburg, Freiburg, Germany.
H Möllinger Health Center Socrates, Güttingen, Switzerland.
J Sherr Malvern, Worcester, UK.
R Schneider Department of Human Sciences, University Osnabrück, Osnabrück, Germany.
We conducted two parallel, blinded homeopathic pathogenetic trials
conducted at two different sites to determine whether symptoms reported
by healthy volunteers were significantly different for homeopathic
remedies than for placebos. Study 1 used a two-armed design, testing
ozone against placebo. Study 2 used a three-armed design, testing ozone
and iridium against placebo. We found significantly more remedy-specific
symptoms in provers taking ozone or iridium than in provers taking
placebo in the three-armed trial and in both trials pooled for ozone and
placebo. We, therefore, conclude that homeopathic remedies produce more
symptoms typical for a remedy than non-typical symptoms. The results
furthermore suggest a somewhat non-classical pattern because symptoms
of one remedy appear to be mimicked in the other trial arm. This might be
indicative of entanglement in homeopathic systems.
non-specific symptoms; pathogenetic trial; placebo
The recent debate around homeopathy focuses on the question
whether homeopathy is a placebo or not (Fisher, et al., 2005;
Shang, et al., 2005; Walach, et al., 2005a; Walach, et al.,
2005b). Although homeopaths hold that their materia medica,
comprising more than 3000 remedies, is a collection of specific
remedies that have to be applied according to the rule of
matching individual symptoms of patients to symptoms proper
to the remedy, critics contend that the whole system of home-
opathy is nothing but a placebo. It may be possibly clinically
effective, but the effects are due to the activation of non-
specific effects through enhancement of hope, reduction of anx-
iety, good patient–doctor relationship, but not due to the phar-
macological properties of the remedies themselves. The bone of
contention is homeopathy’s theoretical implausibility in the
face of modern day physical and chemical knowledge because
homoeopathic remedies are frequently applied in dilutions well
beyond Avogadro’s number. This a priori implausibility makes
it difficult to just stick to empirical results and postpone theo-
retical debate, as most scientists operate according to an
implicit Bayesian scheme of reasoning. In such a scheme, the
odds are stacked high against homeopathy right from the start
so that no matter how convincing empirical results may be,
critics are unwilling to believe them, as long as a viable theory
explaining them is missing (Vandenbroucke and de Craen,
2002). However, before one sets out to construct a theory, it
is important to carefully scrutinize the empirical findings. To
the long-term observer, the homeopathic database has two
clearly distinct and contradictory features. On the one hand,
Journal of Psychopharmacology
22(5) (2008) 543–552
©2008 British Association
SAGE Publications Ltd,
Los Angeles, London,
New Delhi and Singapore
Corresponding author: Harald Walach, PhD, School of Social Sciences and Samueli Institute for Information Biology, University of Northampton, Boughton Green Rd,
Northampton NN2 7AL, UK. Email: firstname.lastname@example.org
homeopathy seems to be helpful in clinical practice. This is
documented by large observational studies and a host of case
studies compiled in homeopathic journals over the two centu-
ries of homeopathy’s existence (Becker-Witt, et al., 2004; Güth-
lin, et al., 2004; Muscari-Tomaioli, et al., 2001; Schlappack,
2004; Steinsbekk and Lüdtke, 2005; Thompson, et al., 2004;
van Wassenhoven and Ives, 2004; Witt, et al., 2005a; Witt,
et al., 2005b). On the other hand, the experimental database
of homeopathy is less convincing than practical clinical results
would have us expect. Clinical trials have difficulties, overall,
to really make a convincing point for the specificity of homeo-
pathic remedies over and against placebos. Experimental mod-
els are still in their infancy, with a large body of data lacking
stability and reproducibility in the hands of independent
researchers (Vickers, 1999). This situation has led some theor-
ists to postulate that homeopathic effects may be due to some
non-local mechanism, which is as yet badly understood (Mil-
grom, 2002, 2005; Walach, 2003; Weingärtner, 2002).
We have been using the paradigm of homeopathic pathoge-
netic trials (HPT) to investigate the issue of specificity of
homeopathic remedies, whereas at the same time being mindful
of potential effects of non-locality. The latter is a non-trivial
challenge for if homeopathic effects are due to some non-local
process, standard ways of experimental testing are doomed to
failure in the long run, as they presuppose a causal mechanism,
which might not be operative in the first place (Lucadou, et al.,
2007). To circumvent this problem, we seek ways to combine
the framework of experimental testing with the as yet little
understood boundary conditions of such non-local processes
(von Stillfried and Walach, 2006). We report here on the first
experiment coming out of a series of pilot attempts.
The experimental model used is a homeopathic pathoge-
netic trial. Such a trial follows the original scheme of homeo-
pathic remedy testing introduced by Hahnemann, the founder
of homeopathy, for researching the symptoms homeopathic
substances produce (Hahnemann, 1811, 1982). In such a trial
healthy volunteers take homeopathically prepared substances.
The volunteers note the symptoms they experience during the
trial, and the symptoms deemed specific are entered into the
homeopathic materia medica and used for prescription in
cases of illness when a patient presents with similar symptoms.
Although in the initial phases of homeopathy the substances
used in such trials were often crude and toxic, homeopaths
have later on often used substances diluted beyond Avogadro’s
number, such as C30. This is a homeopathic potency, which
has been diluted and succussed, that is, vigorously shaken, 30
times at a ratio 1:100, and hence contains 10−60mol of the
original substance, that is, is extremely unlikely to contain
any material trace of the substance at all.
Although first attempts to translate such a homeopathic
pathogenetic trial into an experimental setting were inconclu-
sive (Walach, 1993; Walach, et al., 2001), a re-analysis of the
data using grade of membership analysis showed a pattern of
data suggestive of specific effects cancelling each other out in
the cross-over design used (Walach and Kohls, 2005). We,
therefore, embarked on a new set of pilot studies, which form
the basis for the present report (Möllinger, et al., 2004; Walach,
et al., 2004).
Our studies start from a double vantage point. In one hand,
the criticism levelled against Hahnemann’s original method
likely to inflate symptoms because of desirability effects (Dan-
tas, 1996), as well as against extant pathogenetic studies,
regarding their methodology and validity (Dantas and Fisher,
1998; Dantas, et al., 2007) was taken into account. We used
rigorous blinding and randomization procedures. On the
other hand, we wanted to avoid the restrictions seen in many
modern HPTs (Fisher and Dantas, 2001; Goodyear, et al.,
1998; Walach, 1993) that used a predefined list of symptoms
and hence reduced variability, which goes against homeop-
athy’s unrestricted qualitative data capture method.
We wanted to imitate the original Hahnemannian method
as closely as possible, while using more rigor in terms of con-
trol and blinding, yet keeping the diligent, qualitative, inquisi-
tive approach that allows for a wide range of symptoms to be
perceived. This follows modern recommendations (Riley,
1994a). Such a revival of Hahnemannian method has resulted
in many new pathogenetic trials over the last two decades, pro-
ducing new remedies that are in use among modern homeo-
paths (Riley, 1994b, 1995b; Schadde, 1995; Schuster, 1995;
Sherr, 1998, 2002) (see http://www.dynamis.edu/eng/search.
htm for a database of new HPTs, together with some indicators
of methodology, availability, and source documents where
applicable). When judging the usefulness of this information
(Mortelmans, 1997), it is important to note the following:
homeopathic methodology and epistemology is circular.
Unusual and prominent symptoms produced by healthy volun-
teers are being used for prescribing (Riley, 1994a; Wehmeyer,
et al., 1996). If the prescription heals the patient, the symptom,
and by proxy the HPT, is verified (Walach and Schüppel,
1997). The proof of the symptom is not in the methodology
of the HPT as such, but in the pragmatic verification in clinical
application. Hence, criticisms of that methodology and doubts
about the results of such HPTs, although certainly warranted
in general, fails to really strike, as it neglects this circularity
between HPTs and clinical application that is the epistemolog-
ical basis of homeopathy. To highlight this point, we decided to
conduct an HPT with modern remedies that have been studied
only recently according to the standards of modern homeo-
pathic methodology. We reasoned that the database for these
remedies is new and collected to a comparatively similar stan-
dard (Riley, 1994a, 1995a; Sherr, 1994). This should allow for
a test not only of HPTs in general but also of homeopathy in
In these studies, we combined the standard method of qual-
itative data collection typical for homeopathic pathogenetic
trials with rigorous experimental controls and quantitative
analysis. Data of our pilot studies suggested both specific and
non-specific effects. We also found significant differences for
specific, remedy typical symptoms in one study. We attempted
to replicate these findings in a new set of studies. Our hypothe-
sis was that, while overall the number of symptoms should not
be different between placebo and homeopathy groups, symp-
544 Homeopathic pathogenetic trials
toms specific for the remedies tested should be different either
in single studies or in the combined data set.
We followed the methods described in detail in previous pub-
lications (Möllinger, et al., 2004; Walach, et al., 2004) and laid
out in a protocol before commencement of the study. The
study protocol was submitted to the Ethics Committee of Frei-
burg University Hospital, ethical advice was sought and
We decided beforehand on a list of 20 homeopathic reme-
dies, which were rather newly established, but considered well
studied by the homeopathic practitioners (JS and HM) and
incorporated in the Synthesis repertory (Schroyens, 1993), the
largest homeopathic database widely used by classical homeo-
paths and practitioners (cf. Table 1). We conducted two studies
run in parallel at two different sites (see Figure 1). Study 1
(director JS) used a two-armed design, testing one remedy
against placebo. Study 2 (director HM) used a three-armed
design, testing two homeopathic remedies against placebo. A
homeopathic pharmacy received a randomization code, which
determined the remedies to be studied. The remedies were cho-
sen from this list and all researchers, study directors, supervi-
sors and volunteers were unaware of the particular remedy
used and to the allocation of volunteers to groups. Although
one remedy was common to both studies, the three-armed
study used an additional remedy in the third arm, thus allowing
for pooling of both studies. The study flow and some design
characteristics are depicted in Figure 1.
Remedies were freshly prepared by Helios Pharmacy as C30
potency. For each study, a separate randomization code of ran-
dom numbers was dispatched to the pharmacy. Remedy con-
tainers and placebo containers, which were indistinguishable
containing either medicated or unmedicated globules, were
numbered and dispatched to proving directors. Placebo pills
contained one drop of 90% unsuccussed alcohol. Thus, placebo
and homeopathic medications were undistinguishable by taste
or by any conventional means of detection. The code was kept
at the pharmacy and accessible only to the director and one
assistant. There was no communication with the pharmacy
during the trial, and all stages of information release from the
pharmacy to the researchers were logged.
Iridium) were randomly selected
List of the homeopathic remedies from which two (Ozone,
Cygnus cygnus (swan)
Participant flow and essential design features.
Homeopathic pathogenetic trials545
Participants gave informed consent. They were healthy
volunteers. This was defined through a self-declaration of
being healthy, not taking any conventional, recreational, or
homeopathic drugs on a regular basis for any kind of medical
condition, except contraception, and not suffering from any
acute or chronic medical condition. There was also a quaran-
tine on homeopathic medications of 4-12 weeks previously
depending on the potency used.
Volunteers in study 1 were students of a homeopathic class
learning to become homeopathic practitioners. They were
familiar with the homoeopathic process of studying remedies
in a pathogenetic trial, but had never participated in such a
study. Volunteers in study 2 were a group of experienced
homeopathic medical doctors, who had volunteered for patho-
genetic trials on a regular basis before, and who knew them-
selves and their normal reactions very well. The volunteers
did neither know the list and type of remedies nor whether
they would receive placebo or a homeopathic remedy. Volun-
teers were instructed to report any changes in well being and
any symptoms to the directors of the study. The definition of a
symptom adopted was ‘any unusual change from normal reac-
As a memory aid volunteers kept an unstructured study
diary in which they were to note down any change they experi-
enced immediately, and were regularly instructed and con-
tacted by supervisors. The supervisors contacted the volunteers
on a daily basis to find out which, if any, symptoms were expe-
rienced. In a personal conversation, they helped to establish
whether the reported changes were potentially worth noting
or chance fluctuations of well being. This procedure followed
standard criteria for HPTs, which have also been used to estab-
lish the remedy pictures of the potential proving substances
(Riley, 1994a; Sherr, 1994). For instance, for each change expe-
rienced the supervisor would ask whether this change was
known to the volunteer or not, if known, under which circum-
stances, whether the volunteer considered this unusual or not.
Symptoms never experienced before were especially noted and
are classically considered symptoms because of the remedy.
Directors, supervisors and volunteers were blind as to the rem-
edy chosen and group allocation at all times, until data evalua-
tion was finished.
Volunteers were instructed to observe themselves for a base-
line period of 1 week. Then they started ingesting the sub-
stances by taking five globules and dissolving them slowly in
their mouth. They repeated this process at their discretion sev-
eral times during the first 3 days until they experienced symp-
toms or changes and then stopped intake. This is necessary
from the homeopathic point of view to prevent either strong
symptom aggravations or else an antidoting of symptoms.
The volunteers then kept closely observing themselves for
another 2 weeks, reporting to the director of the respective
study on a daily basis. Thus, a database of symptoms was
established that was deemed likely to be due to the ingestion
of the substance and different from known fluctuations in
well being in a particular volunteer. The database was then
reordered from head to foot according to homeopathic reper-
tory rubrics, thus eradicating temporal and individual informa-
tion. Once the database was confirmed by the study director, it
was closed and sent to the study centre, which checked for
plausibility and integrity.
After the database had been closed and confirmed, a third
person, an expert in homeopathic materia medica and an expe-
rienced practitioner, not otherwise involved in the study was
handed over the database. The study centre arranged for the
name of the remedy used to be released by the pharmacy to
this person only, without, however, releasing the group alloca-
tion information. Nobody else was privy to this information at
The materia medica experts, a different one for each study,
scrutinized every symptom of the database and determined
whether it was a symptom known to be typical for the remedy
tested. The computerized Synthesis repertory on Radar, Ver-
sion 9.9 (Archibel, BE-Namur), was used. The result of this
data evaluation procedure was the number of symptoms
deemed typical for this particular remedy and symptoms
untypical or non-specific symptoms experienced by each volun-
teer during the baseline and trial period. After this database
had been established and this fact confirmed in writing to the
pharmacy, the study centre received the code for the group
allocation and conducted the analysis.
The question to be answered was: is the number of symp-
toms typical for the remedy (remedies) tested different between
placebo and homeopathy groups? All statistical analyses were
done using non-parametric Kruskall–Wallis tests to determine
group differences, and Mann–Whitney tests to determine post-
hoc differences. The main outcome was the number of symp-
toms typical for the remedy tested in the experimental groups
during the proving phase (i.e., when the remedy was taken).
Seventeen women participated in study 1. Mean age was
28.4 years (Standard deviation (SD) 8.5; range 21-58). Ten
were randomly allocated to remedy, seven to placebo. Thirty-
six volunteers participated in study 2. Twenty participants were
women and 16 were men. Participants were on average
43.9 years old (SD 6.2; range 34-56). Eleven were randomly
allocated to remedy 1, 12 to remedy 2 and 13 to placebo.
The remedies randomly chosen from the list were ozone
(administered in both studies) and iridium as the second rem-
edy given in study 2 only. The number of the symptoms typical
for each remedy in all groups is shown in Table 2; examples of
symptoms for each substance are given in Table 3.
The statistical analysis showed no significant differences
between the numbers of typical symptoms in study 1. For
study 2, the Kruskall–Wallis analysis showed significant differ-
ences for symptoms typical for iridium during baseline
(P = 0.037), the difference being between the iridium and the
546 Homeopathic pathogenetic trials
placebo group (P = 0.012; Mann–Whitney). Also, there was a
highly significant difference between the groups in symptoms
typical for ozone during the experimental phase (P = 0.0015;
Kruskall–Wallis). The difference was strongest between the
group that received ozone and the placebo group (P = 0.0008;
Mann–Whitney), but a lesser significant difference was also
found between the iridium group and the placebo group
(P = 0.02; Mann–Whitney).
When the data from both studies were pooled for the groups
receiving ozone and placebo, clear significant differences
emerged. These are shown in Figure 2. Mann–Whitney tests
showed a significant difference between the groups for symp-
toms typical for ozone during the proving phase (P = 0.011).
Other parameters, which were used for an exploratory analysis
also yielded significant differences, namely the total number of
symptoms between groups (P = 0.045) and the total number of
ozone symptoms during proving (P = 0.026).
We adopted a strict experimental protocol to study the effects
of two newly established homeopathic remedies in 30C on
healthy volunteers in two studies similar in methodology, but
different in location and with different volunteers and study
directors. We found clear and significant differences in symp-
toms typical for the remedy tested in the second study and in
the combined data set. It is worthwhile noting that we com-
bined the careful qualitative data analysis typical for homeo-
pathic pathogenetic trials with quantitative evaluation. This
methodology presupposes that the remedy in question is rea-
sonably well known. Although we used newly established rem-
edies, such as ozone and iridium, they are considered reason-
ably well known meanwhile. Qualitative analysis tries to
document as many symptoms purportedly triggered by the
ingestion of the remedy as possible. This yields a wide array
of idiosyncratic symptoms and reactions. By matching these
symptoms with the known remedy picture of that remedy
from the materia medica, the main outcome variable of this
study is derived: the number of symptoms typical for the rem-
edy tested. Note that this is a kind of meta-variable, which
would allow for combining different experiments conducted
with different remedies.
The results we observed are very unlikely to be due to
experimental errors or biases. The whole study was carried
out strictly blinded for all parties involved. Neither study direc-
tors, nor volunteers, nor the personnel dealing with data han-
dling had any knowledge of the substances tested or of the
Number of symptoms typical and untypical for each remedy in all groups during baseline and trial period (mean, standard deviation [SD], and
Study 1Study 2
(n = 10)
(n = 7)
(n = 17)
(n = 11)
(n = 12)
(n = 13)
(n = 36)
All symptoms baseline37.2
All symptoms trial
Ozone symptoms baseline
Iridium symptoms baseline
Ozone symptoms trial
Iridium symptoms trial
Homeopathic pathogenetic trials 547
group allocation code. Randomization was completely con-
cealed. Only the pharmacy knew about the remedies tested
and the allocation code. The volunteers had no access to the
code, and the directors of the study knew only the possible
range of 20 different remedies from which the ones tested
were chosen at random. Hence unblinding is not a reasonable
explanation for our results. Even if the directors of the trial had
guessed the medication, they would have had no way of biasing
Examples of symptoms(*) reported by volunteers after ingestion of substances, before unblinding and without the knowledge of substances
Some symptoms typical for ozone
Immediately after ingestion, desire to breath deeply. Emotionally happy, feeling of more energy (NS 1)
Immediately after ingestion bitter taste, with a transient feeling of aversion against the globules, afterwards very tired (NS 1)
During the day more physical unrest than usual, had to get up more often, could not stay seated (NS 2)
Despite a lot of work during the day very fit and balanced (OS)
Feeling of lightness, like “flying”, a bit dizzy (NS)
Feeling strongly sedated, as if I had taken a neuroleptic medication (NS 3)
Ingestion: dry gum, deep inhalation (NS 3)
Bitter taste in mouth (IS)
Hard stool, violent (OS 3)
Hard stool, with blood
Diarrhoea, sudden and violent (NS 4)
Diarrhoea, abating, but flatulence follows (NS 3)
Again diarrhoea, once (NS 2)
Some symptoms typical for iridium
Excitement because of several phone call, overlaying all other problems (OS)
Mood not as continuously depressed as usual after depressing news (IS 2)
Feeling balanced (NS)
Mood good and balanced all day (OS)
Mood all day excellent and very good weather (bike tour with picknick at lake) (OS)
I say “no” to a good friend of mine’s wish. But I am ruminating about this for hours (OS 3)
Mood rather depressed, concentration not so good, dissatisfied with my work output. Trigger: problems of people close to me that
occupied me a lot as well (OS 2)
Headache behind forehead, eyes and below frontal part of head, dull, worse stooping (OS)
On awakening, dull headache (NNH) (OS)
Feeling like pressure in ears, low humming/droning in ears, subjectively hearing became worse in the morning, better in the
afternoon, worse again after 10 p.m. (NS 3)
Pressure in ears, worse right, than left; extending to mastoid (NS 2)
Light stabbing pain in larynx, left more than right, not changed by swallowing (NS 1)
Some symptoms experienced under placebo
Morning, 1 h after ingestion irritated; irritation grows until afternoon (NS 2)
Only now I realise that I have forgotten taking the globules; feel a bit stressed and forgetful; involuntarily, I stopped taking the
globules; this is what happens I guess. Unusually chaotic. Feel chased around (NS4)
In the evening sensitive against noise, irritated and angry at diner until 8 p.m., alright after things have calmed down (OS)
Slight vertigo on getting up and especially when showering (NS)
Slight migraine, left, after quarrel with husband (OS 2)
Slight pressure in head morning, after getting up (NS 1)
Burning pain in larynx after waking up, better after swallowing (OS 2)
Enlarged lymph node left, at inception of m. sternocleidomastoideus (OS 3)
Nausea after diner (OS 2)
Flatulence in abdomen after wholemeal bread for breakfast (OS 3)
NS, new symptom; OS, old symptom that is known; IS, improved symptom; indicates a healing reaction; numbers after abbreviations indicate the num-
ber of persons experiencing the symptom.
*The symptoms have been determined by individual examinations and discussions with a supervisor and by comparison with normal known status of
volunteers, after being recognized as noteworthy changes from normal or expected status. By definition, all symptoms, that is, placebo symptoms as
well, are unusual for the person in question. This material has been used by the materia medica expert to compare against the known symptoms of the
two remedies tested and a determination was made whether the symptom is either typical for ozone or iridium or for none.
548 Homeopathic pathogenetic trials
the data, as the allocation code was also concealed. When the
name of the remedy was released to the materia medica expert,
the final database of symptoms had already been deposited at
the study centre; hence, no post-hoc tampering with data was
possible. After data evaluation by the expert, there were no
data processing steps necessary that might have unconsciously
or overtly biased the results.
The fact that we have conducted two studies, slightly modi-
fied in design but using the same methodology and the same
core medications with two different study directors, different
volunteers, settings and locations, adds some robustness and
credibility to our data and reduces the likelihood of chance
findings. A clear drawback is the fact that the first study does
not reach significance independently. However, it should be
noted that the first study was smaller and hence had less
power. More specifically, the effect between the groups for
symptoms typical for ozone during proving was d = 0.60. This
effect would have required a sample size producing twice as
many symptoms to have a power of 1−β = 0.80 (α = 0.05).
Nevertheless, the effects point in the same direction and,
when pooled, yield a clearly significant result. Non-specific
symptoms are also experienced more frequently during the
proving period by the groups receiving homeopathic remedies,
not only by those receiving placebo. This might be a reflex of
lacking knowledge on part of the materia medica expert, who
in fact might have made mistakes, or it might have been due to
the fact that in those new remedies new trials still bring out new
and unknown symptoms, which then are considered non-
specific because they are not yet known.
The clearest result was produced by the study that used
experienced volunteers who had actually conducted many
such studies before, although they produced much less symp-
toms overall (study 2). This underscores the fact that distin-
guishing symptoms from normal background noise is not a
trivial task and presupposes good knowledge of one’s own
organism and potential reactions to situations potentially dis-
turbing to the system. Only then is it possible to distinguish
potentially new symptoms from rather trivial or known
outcome parameter: symptoms typical for ozone during proving; all other data are shown for reasons of transparency and tests are to be considered
Combined data set: Mean number of symptoms per group and confidence intervals; significance given according to Mann-Whitney tests. Main
Homeopathic pathogenetic trials549
individual reactions. Moreover, homeopathic pathogenetic
symptoms are sometimes minute, hardly noticeable and often
transient. This is why our methodology involved daily contacts
between provers and supervisors to discuss the experiences and
to decide which experiences are unusual and thus qualify as
symptoms. The supervisor assumes the role of a partner listen-
ing carefully and posing questions to help decide on the experi-
ences. The decisions are not always clear-cut as the changes
observed are often very small, such as a very slight sensation
of pain, or a peculiar dream, etc. (see Table 3 for examples). A
volunteer has to compare the experienced symptoms with the
normal state and take into consideration other events in order
to determine whether a change is unusual. Also, this procedure
may produce extra noise in susceptible individuals, reducing
the power of an HPT to distinguish between symptoms, and
one might consider the purely female sample of study 1 espe-
cially susceptible. This is visible in the comparatively high
number of untypical symptoms under ozone in study 1.
Our volunteers who are from an experienced group (study
2) were clearly more successful in determining what counted as
remedy symptoms. They reported far less symptoms than the
novice volunteers for whom this was the first experience (study
1). The result is a clearer and significant distinction between
symptoms typical for the remedy and non-specific symptoms.
This difference in mean number of symptoms observed
between study 1 and 2 might also be partly an artefact of
recording. Symptoms can be broken down into smaller units.
In a different vein, volunteers in study 2 may just not have
experienced more symptoms because they were of a less volatile
temperament on the study, whereas the novice volunteers were
keen on experiencing a homeopathic pathogenetic trial for the
first time. This line of reasoning is corroborated by the number
of symptoms produced in the study with novice provers. They
(P < 0.001) and during proving (P < 0.001) than their counter-
parts in study 2.
We would like to point out a noteworthy characteristic of
our data: we see significantly more symptoms typical for irid-
ium during baseline in the group later receiving iridium. This is
a completely counterintuitive finding. One might be tempted to
attribute it to chance, as the significance level is P = 0.037 and
hence not convincing enough considering the amount of tests
done. Nevertheless, it points to some element in our data,
which one could interpret as a signature of non-locality
(Walach, 2005). A similar result has been observed in another
pathogenetic trial (Lewith, et al., 2005). Also note that during
the experimental phase, although not significant, more symp-
toms typical for iridium were found in the group actually tak-
ing ozone than that actually taking iridium. Although with a
two-armed trial such a finding can be attributed to the setting
of the study and the particular style of a materia medica expert,
in a three-armed trial this is not a viable explanation. For
instance, a materia medica expert might be generally more
prone to label a symptom as ‘typical for the remedy’ in the
latter part of a study, thus increasing the count of those symp-
toms in both the control and experimental group and reducing
the separability. However, it must be remembered that only a
veridical decision to the best of the knowledge of the materia
medica expert will be an optimal strategy under conditions of
blinding and hence only an unconscious tendency is a reason-
able option. Such a scoring habit can account for the result of
study 1 only, but it cannot account for the results of study 2,
where significantly more symptoms typical for ozone compared
with placebo was found in the iridium group, but not more
iridium symptoms in the ozone group compared with placebo.
Such a pattern of distribution is, again, a signature for a non-
local effect, we suggest. Such effects are predicted by a general-
ized formalism of quantum mechanics, which predicts entan-
glement in a generalized form under certain conditions
(Atmanspacher, et al., 2002). It has been argued that homeop-
athy might be a case in question where such processes are at
work (Milgrom, 2002; Walach, 2003). Our data give at least a
hint that this idea is not completely off the mark.
By combining the data from two different trials where only
one (blinded) remedy was common, we managed to distil at
least one fact from the data: that significantly more symptoms
typical for a remedy were experienced by those taking the rem-
edy than non-specific symptoms. These typical symptoms were
most prominent in the group actually taking that remedy.
One might argue that we did not verify whether our blind-
ing was successful. This is true, but circular. Blinding is nor-
mally assessed by asking participants about assumed group
assignment. Especially when side effects are prominent, these
side effects are known to unblind patients in clinical trials. An
HPT is designed to elicit what in a clinical context would be
deemed a side effect. Hence, by the same token, it does not
make sense to ask participants of an HPT about their assump-
tion, as these assumptions would only reflect their hypotheses
about assignment as inferred from the symptoms they experi-
ence, but would not allow any true conclusion as to the real
success of blinding. As homeopathic remedies and homeo-
pathic placebos are probably the only pharmacological agents
that are really undistinguishable, in terms of appearance and
taste, as well as in terms of any known chemical and physical
properties that are measurable, any procedure to measure
unblinding would be rather non-sensical. The best guarantee
for a tight blinding in our study was a multi-level procedure
of control, documentation, and locking up of codes that was
well documented and, we believe, completely intact until the
very end of the study.
We suggest that others take up our findings and improve on
our method or verify our findings by their research. A few
words of caution are in place. If generalized entanglement
plays a role in homeopathy, then trying to pinpoint the effect
as a causal one will lead to its vanishing for very clear theoret-
ical and formal reasons laid out elsewhere (Lucadou, et al.,
2007). Hence, detection should pursue indirect paths. Several
ways have been pointed out (Walach, 2003, 2005). One obvious
method was described in this article. It entails using multi-
armed trials and combining data collected separately. How-
ever, as we still do not know enough about such effects, we
550 Homeopathic pathogenetic trials
need further information and some more exploratory work
before final conclusions can be drawn.
Some preliminary conclusions, however, can already be
drawn from this study:
1) In healthy volunteers, homeopathic remedies produce more
symptoms typical for a remedy than non-typical symptoms,
at least sometimes.
2) Volunteers taking homeopathic remedies are more likely to
experience symptoms typical for that remedy than volun-
teers taking placebo, at least sometimes.
The study was supported by a grant from the Samueli Institute, Alex-
andria, Virginia, USA. RS and HW were funded by the Samueli Insti-
tute at the time the study was conducted. The Swiss-German arm of
the study (HM) was supported by the Health Center Socrates and the
Socrates Foundation, Switzerland. We are grateful to all volunteers for
their participation, to Nadia Bakir and Misada Vince who conducted
the proving of study 1, and to Helios Pharmacy for operating the ran-
domization and concealment procedures and for donating the reme-
dies. We are grateful to Andrea Weninger who helped with the evalua-
tion of symptoms.
Atmanspacher, H, Römer, H, Walach, H (2002) Weak quantum the-
ory: complementarity and entanglement in physics and beyond.
Found Phys 32: 379–406.
Becker-Witt, C, Lüdtke, R, Weisshuhn, TER, Willich, SN (2004)
Diagnoses and treatment in homeopathic medical practice. Forsch
Komplementarmed Klass Naturheilkd 11: 98–103.
Dantas, F (1996) How can we get more reliable information from
homoepathic pathogenetic trials? A critique of provings. Br
Homeopath J 85: 230–236.
Dantas, F, Fisher, P (1998) A systematic review of homoeopathic
pathogenetic trials (‘provings’) published in the United Kingdom
from 1945 to 1995. In: Ernst, E, Hahn, EG (eds), Homoeopathy:
A Critical Appraisal. London: Butterworth-Heineman, pp. 69–97.
Dantas, F, Fisher, P, Walach, H, Wieland, F, Rastogi, DP, Texeira, H
et al (2007) A systematic review of homeopathic pathogenetic trials
published from 1945 to 1995. Homeopathy 96: 4–16.
Fisher, P, Bell, IR, Belon, P, Bolognani, F, Brands, M, Connolly, T
(2005) Letter to the editor: are the clinical effects of homoeopathy
placebo effects? Lancet 366: 2082.
Fisher, P, Dantas, F (2001) Homeopathic pathogenetic trials of
Acidum malicum and Acidum ascorbicum. Br Homeopath J 90:
Goodyear, K, Lewith, G, Low, JL (1998) Randomised double-blind
placebo controlled trial of homoeopathic proving for Belladonna
C30. J R Soc Med 19: 579–582.
Güthlin, C, Lange, O, Walach, H (2004) Measuring the effects of acu-
puncture and homoeopathy in general practice: an uncontrolled
prospective documentation approach. BMC Public Health, 6.
Hahnemann, S (1811) Reine Arzneimittellehre: Erster Theil. Dresden:
Hahnemann, S (1982) Organon of Medicine. Los Angeles: JP Tarcher.
Lewith, GT, Brien, S, Hyland, ME (2005) Presentiment or entangle-
ment? An alternative explanation for apparent entanglement in
provings. Homeopathy 94: 92–95.
Lucadou, Wv Römer, H, Walach, H (2007) Synchronistic phenomena
as entanglement correlations in generalized quantum theory. J
Conscious Stud 14: 50–74.
Milgrom, LR (2002) Patient-practitioner-remedy (PPR) entanglement:
a qualitative, non-local metaphor for homeopathy based on quan-
tum theory. Homeopathy 91: 239–248.
Milgrom, LR (2005) The sound of two hands clapping: could homeop-
athy work locally and non-locally? Homeopathy 95: 100–104.
Möllinger, H, Schneider, R, Löffel, M, Walach, H (2004) A double-
blind, randomized, homeopathic pathogenetic trial with healthy
persons: Comparing two high potencies. Forsch Komplemen-
tarmed Klass Naturheilkd 11: 274–280.
Mortelmans, G (1997) What do provings prove? A critical appraisal.
Homoeopath Links 10: 201–204.
Muscari-Tomaioli, G, Allegri, F, Miali, E, Pomposelli, R, Tubia, P,
Targhetta, A (2001) Observational study of quality of life in
patients with headache, receiving homeopathic treatment. Br
Homeopath J 90: 189–197.
Riley, DS (1994a) Contemporary drug provings. J Am Inst Homeo-
path 87: 161–165.
Riley, DS (1994b) Nicotinamide adenine dinocleotide (NAD): a prov-
ing. J Am Inst Homeopath 87: 74–78.
Riley, DS (1995a) History of Homoeopathic Drug Provings. Philadel-
phia: Homoeopathic Pharmacopea Convention of the United
Riley, DS (1995b) Proving report – Veronica officinalis. Br Homeo-
path J 84: 144–148.
Schadde, A (1995) Ozon. Eine homöopathische Studie. Munchen:
Müller & Steinicke.
Schlappack, O (2004) Homeopathic treatment of radiation-induced
itching in breast cancer patients. A prospective observational
study. Homeopathy 93: 210–215.
Schroyens, F (1993) Synthesis: Repertorium Homeopathicum Synthe-
ticum. London: Homoeopathic Book Publishers.
Schuster, B (1995) Bambus. Homöopathische Prüfung und Verifizier-
ung. Mit Kasuistiksammlung. Kronberg: Kent-Gesellschaft.
Shang, A, Huwiler-Münteler, K, Nartey, L, Jüni, P, Dörig, S, Sterne,
JAC (2005) Are the clinical effects of homeopathy placebo effects?
Comparative study of placebo-controlled trials of homoeopathy
and allopathy. Lancet 366: 726–732.
Sherr, J (1994) The Dynamics and Methodology of Homoeopathic
Provings. West Malvern: Dynamis Books.
Sherr, J (1998) Dynamic Provings, vol. 1. West Malvern: Dynamis
Sherr, J (2002) Dynamic Provings, vol. 2. Malvern: Dynamis Books.
Steinsbekk, A, Lüdtke, R (2005) Patients’ assessments of the effective-
ness of homeopathic care in Norway: a prospective observational
multicentre outcome study. Homeopathy 94: 10–16.
Thompson, E, Barron, S, Spence, D (2004) A preliminary audit inves-
tigating remedy reactions including adverse events in routine
homeopathic practice. Homeopathy 93: 203–209.
Vandenbroucke, JP, de Craen, AJM (2002) Alternative medicine: a
‘mirror image’ for scientific reasoning in conventional medicine.
Ann Intern Med 135: 507–513.
Van Wassenhoven, M, Ives, G (2004) An observational study of
patients receiving homeopathic treatment. Homeopathy 93: 3–11.
Homeopathic pathogenetic trials 551
Vickers, A (1999) Independent replication of pre-clinical research in Download full-text
homoeopathy: a systematic review. Forsch Komplementarmed 6:
Von Stillfried, N, Walach, H (2006) The whole and its parts: are com-
plementarity and non-locality intrinsic to closed systems? Interna-
tional Journal of Computing Anticipatory Systems 17: 137–146.
Walach, H (1993) Does a highly diluted homoeopathic drug act as a
placebo in healthy volunteers? Experimental study of Belladonna
C30. J Psychosom Res 37: 851–860.
Walach, H (2003) Entanglement model of homeopathy as an example
of generalizsed entanglement predicted by Weak Quantum Theory.
Forsch Komplementarmed Klass Naturheilkd 10: 192–200.
Walach, H (2005) Entangled – and tied in knots! Practical conse-
quences of an entanglement model for homeopathic reserach and
practice. Homeopathy 94: 96–99.
Walach, H, Jonas, W, Lewith, G (2005a) Letter to the editor: are the
clinical effects of homoeopathy placebo effects? Comparative study
of placebo-controlled trials of homoeopathy and allopathy. Lancet
Walach, H, Jonas, WB, Ives, J, van Wijk, R, Weingärtner, O (2005b)
Research on homeopathy: state of the art. J Altern Complement
Med 11: 813–829.
Walach, H, Kohls, N (2005) Grade-of-Membership (GoM) – analysis
as a sensitive method for evaluating categorical data – introduction
and some examples. In: Beauducel, A, Biehl, B, Bosniak, M,
Conrad, W, Schönberger, G, Wagener, D (eds), Multivariate
Research Strategies – Festschrift for Werner W. Wittmann.
Aachen: Shaker, pp. 151–172.
Walach, H, Köster, H, Hennig, T, Haag, G (2001) The effects of
homeopathic belladonna 30CH in healthy volunteers – a random-
ized, double-blind experiment. J Psychosom Res 50: 155–160.
Walach, H, Schüppel, R (1997) Homöopathieforschung – Eine Taxo-
nomie. Forsch Komplementarmed 4: 344–347.
Walach, H, Sherr, J, Schneider, R, Shabi, R, Bond, A, Rieberer, G
(2004) Homeopathic proving symptoms: result of a local, non-
local, or placebo process? A blinded, placebo-controlled pilot
study. Homeopathy 93: 179–185.
Wehmeyer, A, Heger, M, Riley, DS (1996) Homöoapthische Arznei-
mittelprüfungen – Grundlagen und Praxis. In: Hornung, J (ed),
Forschungsmethoden in der Komplementärmedizin. Über die Not-
wendigkeit einer methodologischen Erneuerung. Stuttgart: Schat-
tauer, pp. 32–43.
Weingärtner, O (2002) Über die wissenschaftliche Barbeitbarkeit der
Identifikation eines “arzneilichen Gehalts” von Hochpotenzen.
Forsch Komplementarmed Klass Naturheilkd 9: 229–233.
Witt, C, Keil, T, Selim, D, Roll, S, Vance, W, Wegscheider, K,
Willich, SN (2005a) Outcome and costs of homoeopathic and con-
ventional treatment strategies: a comparative cohort study in
patients with chronic disorders. Complement Ther Med 13: 79–86.
Witt, CM, Lüdtke, R, Baur, R, Willich, SN (2005b) Homeopathic
medical practice: long term results of a cohort study with 3981
patients. BMC Public Health 5: 115.
552 Homeopathic pathogenetic trials