ChapterPDF Available

Kuhn on Discovery and the Case of Penicillin



The chapter gives an exposition of Kuhn's theory of discovery and discusses how it applies to the discovery of penicillin. It is suggested that the penicillin case requires some modification of Kuhn's approach to discovery.
Kuhn on Discovery and the Case of Penicillin
By Donald Gillies, University College London
This is the final author version of a paper which was published in Wenceslao J. Gonzalez
and Jesus Alcolea (eds.) Contemporary Perspectives in Philosophy and Methodology of
Science, netbiblo, 2006, pp. 47-63.
1. Kuhn’s Theory of Discovery in Science
2. The Discovery of Penicillin Phase 1: Fleming’s Work
3. Why Fleming abandoned his hope that Penicillin would be a ‘perfect antiseptic’
4. The Discovery of Penicillin Phase 2: The Work of Florey and his Oxford team
5. Suggested Modifications of Kuhn’s Theory in the light of the Penicillin example
1. Kuhn’s Theory of Discovery in Science
Kuhn is of course most famous for his theory of scientific revolutions. However
in this paper I want to consider his theory of discovery in science. This theory is
connected to his theory of scientific revolutions, but is nonetheless somewhat separate
and very interesting in its own right. Kuhn first published his theory in a paper entitled:
‘The Historical Structure of Scientific Discovery’. This appeared shortly before The
Structure of Scientific Revolutions in the same year (1962). It has been reprinted in the
collection The Essential Tension from where I will take my quotations. Much of the
material in the paper was used in The Structure of Scientific Revolutions. It reappears
largely in Chapter VI and Chapter X, p. 114 of that work. I will base my account of
Kuhn’s theory both on his paper (1962a) and his book (1962b).
The plan of my own paper is as follows. I will begin in this section by
expounding Kuhn’s theory of discovery in science. I will then go on to describe a
famous discovery which is not considered by Kuhn – namely the discovery of penicillin.
I will give a historical account of this discovery in sections 2-4. Finally in section 5 I will
consider how well Kuhn’s theory fits this example. In some respects the fit is very good,
and the example of the discovery of penicillin may be said to support some of Kuhn’s
general ideas on the subject of discovery very well. On the other hand the fit is not
perfect and some modification of Kuhn’s theory is needed to take account of the case of
Kuhn is concerned to criticize the view of scientific discovery as (1962a, p. 165):
‘a unitary event, one which, like seeing something, happens to an individual at a
specifiable time and place.’ Such an account he thinks applies at best to a relatively
unproblematic kind of scientific discovery in which theory predicts a new sort of entity
such as radio waves, and this entity is subsequently detected experimentally. There is,
however, a different kind of scientific discovery which Kuhn refers to as ‘troublesome’,
and to which the account definitely does not apply. In such cases an entity is discovered
which was not predicted by theory, and whose existence takes scientists by surprise.
Often there seems to be an accidental element in such discoveries. This is how Kuhn
himself makes the distinction (1962a, pp. 166-7):
‘The troublesome class consists of those discoveries – including oxygen, the
electric current, X rays, and the electron – which could not be predicted from accepted
theory in advance and which therefore caught the assembled profession by surprise. …
there is another sort and one which presents very few of the same problems. Into this
second class of discoveries fall the neutrino, radio waves, and the elements which filled
empty [47] (Numbers in square brackets are the page numbers of the published version.)
places in the periodic table. The existence of all these objects had been predicted from
theory before they were discovered, and the men who made the discoveries therefore
knew from the start what to look for.’
Kuhn’s theory is concerned mainly with discoveries of the ‘troublesome’ class.
His main point is that such discoveries involve at least two steps, namely recognizing that
something is, and recognizing what it is; or, to put the matter another way, observing
something novel, and providing a theoretical explanation of that novelty. Because such
discoveries are a complex process, they do not take place at an instant, and often more
than one person is involved. As Kuhn says (1962a, p. 171):
‘ … discovering a new sort of phenomenon is necessarily a complex process which
involves recognizing both that something is and what it is. Observation and
conceptualization, fact and the assimilation of fact to theory, are inseparably linked in the
discovery of scientific novelty. Inevitably, that process extends over time and may often
involve a number of people. Only for discoveries in my second category – those whose
nature is known in advance – can discovering that and discovering what occur together
and in an instant.’
Kuhn illustrates his theory by the examples of the discovery of oxygen, of the
planet Uranus, and of X rays. For this brief account of his views, I will confine myself to
the example of the discovery of Uranus. Kuhn writes (1962a, p. 171):
‘On the night of 13 March 1781, the astronomer William Herschel made the following
entry in his journal: “In the quartile near Zeta Tauri … is a curious either nebulous star
or perhaps a comet.” That entry is generally said to record the discovery of the planet
Uranus, but it cannot quite have done that.’
Indeed it cannot, because the discovery of Uranus was the discovery of a new planet
unknown to previous astronomers. However, Herschel does not mention a planet, but
speaks only of a ‘nebulous star or perhaps a comet’. He does, however, observe that the
object is ‘curious’. Kuhn says (1962b, p. 114):
‘On at least seventeen different occasions between 1690 and 1781, a number of
astronomers, including several of Europe’s most eminent observers, had seen a star in
positions that we now suppose must have been occupied at the time by Uranus. One of
the best observers in this group had actually seen the star on four successive nights in
1769 without noting the motion that could have suggested another identification.
Herschel, when he first observed the same object twelve years later, did so with a much
improved telescope of his own manufacture. As a result, he was able to notice an
apparent disk-size that was at least unusual for stars.’
The key point here is that ordinary stars are not magnified in size by a telescope however
powerful. They are so far away that they become effectively points of light rather than
disks. The effect of a more powerful telescope is to make stars brighter and hence more
visible rather than bigger. If a celestial object is increased in size by a telescope it must
be something in the solar system such as a planet or comet, or a nebula. A nebula
consists of [48] a large collection of stars and so magnification in the sense of an
increased separation of these stars becomes possible. Herschel would have been very
familiar with this, and, as he observed a magnification of the star, he at once recognised
this as ‘curious’. He could have concluded that the object was a planet, a comet, or a
nebula. In fact he rejected the correct one of these three possibilities and concluded that
what he had seen was either a comet or a nebula. It was now easy to distinguish between
these two possibilities. A comet would move against the background of the stars, while a
nebula would remain fixed. Herschel observed his curious object on two further
occasions, namely 17 and 19 March. The object moved, and he therefore concluded that
it must be a comet. He therefore announced to the scientific community that he had
discovered a new comet. Kuhn now continues the story as follows (1962a, p. 172):
‘ … astronomers throughout Europe were informed of the discovery, and the
mathematicians among them began to compute the new comet’s orbit. Only several
months later, after all those attempts had repeatedly failed to square with observation, did
the astronomer Lexell suggest that the object observed by Herschel might be a planet.
And only when additional computations, using a planet’s rather than a comet’s orbit,
proved reconcilable with observation was that suggestion generally accepted. At what
point during 1781 do we want to say that the planet Uranus was discovered? And are we
entirely and unequivocally clear that it was Herschel rather than Lexell who discovered
The example of the discovery of Uranus illustrates perfectly the two main claims
which Kuhn makes about discoveries of his ‘troublesome’ class. These are that (1) such
a discovery is ‘necessarily a complex process which involves recognizing both that
something is and what it is’, and (2) that ‘inevitably, that process extends over time and
may often involve a number of people’. The main features of the story seem to be the
following. First of all Uranus was seen by astronomers no less than seventeen times
before Herschel’s crucial observation, but none of these astronomers realised that there
was anything unusual about what they had seen. Herschel, in contrast to these
predecessors, did realise that he had seen something which in his own words was curious.
This is an example of what Kuhn refers to as (1962a, p. 173): ‘the individual skill, wit, or
genius to recognize that something has gone wrong in ways that may prove
consequential.’ But Kuhn also points out that Herschel’s success was the result not just
of greater ‘wit or genius’, but depended on his having a better telescope. We can add that
background knowledge also played a crucial part here. Herschel’s better telescope
enabled the magnification in size of Uranus to become more obvious, but it was his
background knowledge which enabled him to realise that this was significant. An
amateur without this background knowledge would not have made the discovery. But
although Herschel realised that there was something unusual, he misinterpreted what he
had seen as a comet. Thus the discovery was only complete when Lexell, as a result of
elaborate theoretical calculations, did finally identify Uranus as a planet. So the
discovery of Uranus did definitely extend over time, and did involve more than one
person. We could perhaps say that Herschel discovered that there was something new
and of interest, while Lexell discovered what it was. The example is [49] also in
agreement with the ideas of Fleck who was a major influence on Kuhn. Fleck in fact
writes (1935, p. 76):
‘If any discovery is to be made accessible to investigation, the social point of view must
be adopted; that is, the discovery must be regarded as a social event.’
This completes my account of Kuhn’s theory of discoveries of the ‘troublesome’
class. I will now begin my examination of whether the discovery of penicillin is in
accordance with Kuhn’s model. The first point to notice is that the discovery of
penicillin agrees with Kuhn in that it took place over a quite long period of time, and
involved several people. We may in fact distinguish two phases in the discovery. The
first phase was the one which involved Alexander Fleming. Fleming, as the result of a
chance observation of a contaminated Petri dish, discovered the existence of a new
substance which powerfully inhibited a variety of pathogenic bacteria. Indeed some
might regard this as constituting the discovery of penicillin. However, to me it does not
seem that this in itself was all that there was to the discovery, because Fleming did not
discover that penicillin could be used as what we would now call an antibiotic. In fact he
had reasons for supposing that penicillin would not work as an antibiotic, and he himself
used penicillin for another purpose. This brings me to the second phase in the discovery
of penicillin which involved Howard Florey and his team at Oxford. They were the ones
who showed that the substance which Fleming had discovered could be used successfully
as an antibiotic. It will be seen that there is an analogy here to the contributions which
Herschel and Lexell made to the discovery of Uranus. However, I will return to
philosophical analysis in section 5 of the paper. In the next 3 sections (2, 3, and 4), I will
give a brief historical account of how penicillin was discovered.1
2. The Discovery of Penicillin Phase 1: Fleming’s Work
It was early in September 1928 that Fleming noticed an experimental plate in his
laboratory which had been contaminated with a penicillium mould. If, however, we are
to understand his reaction to this fateful event, we must first examine some of the
research which Fleming had carried out previously. There were in fact two episodes
which had influenced Fleming in a crucial fashion. The first of these was Fleming’s
experiences in the First World War, and this will be described in section 2.1. The second
was Fleming’s discovery in 1921 of an important biochemical substance which was
named lysozyme. This will be dealt with in section 2.2. An understanding of how these
two episodes had prepared Fleming’s mind will enable us to understand why Fleming
acted as he did when he stumbled on penicillin itself, and this will be described in section
2.3. [50]
2.1 Fleming’s experiences in the First World War Fleming spent most of his career
carrying out research in bacteriology in the inoculation department of St Mary’s Hospital,
London. This department was headed until his retirement in 1946 by Sir Almroth
Wright. When the First World War broke out in 1914, Wright, Fleming and the rest of
the department were sent to Boulogne to deal with the war wounded, and, in particular, to
try to discover the best way of treating infected wounds. At that time wounds were
routinely filled with powerful antiseptics which were known to kill bacteria outside the
body. Fleming, however, soon made the remarkable discovery that bacteria seemed to
flourish in wounds treated with antiseptics even more than they did in untreated wounds.
The explanation of this apparent paradox was quite simple. In an untreated wound the
bacteria causing the infection were attacked by the body’s natural defences, the white
cells, or phagocytes, which ingested the invading bacteria. If the wound was treated with
an antiseptic, some bacteria were indeed killed, but the protective phagocytes were also
killed, so that the net effect was to make the situation worse than before. Wright and his
group therefore maintained (quite correctly) that wounds should not be treated with
antiseptics. They advocated the earliest possible surgical removal of all dead tissue, dirt,
foreign bodies, and so forth, and the irrigating the wound with strong sterile salt solution.
The medical establishment of the day rejected this recommendation, and so the superior
treatment was accorded only to those directly in the care of Wright and his team.
2.2 The discovery of lysozyme After the war, Fleming returned to the inoculation
department in London, and here in 1921 he discovered an interesting substance which
was given the name lysozyme. Lysozyme was capable of destroying a considerable range
of bacteria, and was found to occur in a variety of tissues and natural secretions. Fleming
first came across lysozyme while studying a plate-culture of some mucus which he took
from his nose when he had a cold. He later discovered that lysozyme is to be found in
tears, saliva, and sputum, as well as in mucus secretions. He extended his search quite
widely in the animal and vegetable kingdoms, and found lysozyme in fish eggs, birds’
eggs, flowers, plants, vegetables, and the tears of more than fifty species of animals.
Lysozyme destroyed about 75% of the 104 strains of airborne bacteria and some other
bacteria as well. Moreover, Fleming was able to show that, unlike chemical antiseptics,
even the strongest preparations of lysozyme had no adverse effects on living phagocytes,
which continued their work of ingesting bacteria just as before. From all this, it seemed
that lysozyme was part of many organisms’ natural defence mechanisms against bacterial
infection. Lysozyme had only one drawback. It did not destroy any of the bacteria
responsible for the most serious infections and diseases. The hypothesis naturally
suggested itself that the pathogenic bacteria were pathogenic partly because of their
resistance to lysozyme.
If we put together Fleming’s research on war wounds and his research on
lysozyme, a problem situation emerges which I will call the ‘antiseptic problem
situation’. On the one hand, the chemical antiseptics killed pathogenic bacteria outside
the body, but were less effective for infected wounds, partly because they destroyed the
phagocytes as well. On the other hand, the naturally occurring antiseptic lysozyme did
not kill the phagocytes, [51] but also failed to destroy the most important pathogenic
bacteria. The problem then was to discover a ‘perfect antiseptic’ which would kill some
important pathogenic bacteria without affecting the phagocytes. The work on lysozyme
suggested that such antiseptics might be produced by naturally occurring organisms.
2.3 Fleming stumbles on penicillin This then is the background to Fleming’s work on
penicillin. Fleming actually stumbled on penicillin while he was carrying out a fairly
routine investigation. He had been invited to contribute a section on the staphylococcus
group of bacteria for the nine-volume A System of Bacteriology which was being
produced by the Medical Research Council. Fleming’s contribution did indeed appear in
the second volume in 1929. Staphylococci are spherical bacteria which are responsible
for a variety of infections. For example, the golden-coloured Staphylococcus aureus is
responsible for skin infections such as boils and carbuncles, as well as for a variety of
other diseases. While reading the literature on staphylococci, Fleming came across an
article by Bigger, Boland, and O’Meara of Trinity College, Dublin, in which it was
suggested that colour changes took place if cultures of staphylococci were kept at room
temperature for several days. This interested Fleming, because the colour of a
staphylococcus can be an indicator of its virulence in causing disease. He therefore
decided to carry out an experimental investigation of the matter with the help of
D.M.Pryce, a research scholar.
The staphylococci were cultured in glass dishes, usually 85 mm in diameter,
known as Petri dishes. These dishes were filled with a thin layer of gelatinous substance
called agar to which enough nutrients could be added to allow the microbes to multiply.
Using a platinum wire, some staphylococci were spread across the surface of the agar,
and the plate was then incubated at a suitable temperature (usually 37oC), to allow the
microbes to multiply. After this period of incubation, the dish was set aside on the bench,
and was examined every few days to see if changes in the colour of some of the
staphylococci could be observed.
While this investigation was continuing, Pryce left the laboratory in February
1928 to start another job, but Fleming continued the work on his own throughout the
summer. At the end of July Fleming went off for his usual summer holiday, leaving a
number of culture-plates piled at the end of the bench where they would be out of the
sunlight. Early in September (probably on 3 September) when Fleming had returned
from his holiday, Pryce dropped in to see him. Pryce found Fleming sorting out the pile
of plates on his bench. Discarded plates were put in a shallow tray containing the
antiseptic Lysol. This would kill the bacteria, and make the Petri dishes safe for the
technicians to wash and prepare for use again. Fleming’s tray was piled so high with
dishes that some of them were protruding above the level of the Lysol. Fleming started
complaining about the amount of work he had had to do since Pryce had left him. He
then selected a few of the dishes to show to Pryce. More or less by chance he picked up
one in the tray of discards but above the level of the Lysol. According to Pryce’s later
recollection, Fleming looked at the plate for a while, and then said: ‘That’s funny.’ The
plate was in fact the famous penicillin plate.
This is how Fleming himself described what happened in the paper which he
published in June 1929 (p. 226):
‘While working with staphylococcus variants a number of culture-plates were set aside
on the laboratory bench and examined from time to time. In the examinations [52] these
plates were necessarily exposed to the air and they became contaminated with various
micro-organisms. It was noticed that around a large colony of a contaminating mould the
staphylococcus colonies became transparent and were obviously undergoing lysis.’
Fleming’s photograph of the original penicillin plate is reproduced in Plate 1, and
it is easy to follow his description when examining the photograph.
The colonies of staphylococci are the small circular blobs, and the contaminating mould
is very obvious. Near the mould the staphylococci become transparent or disappear
altogether. They are obviously, as Fleming says, undergoing lysis, which means the
dissolution of cells or bacteria. From his observation of the plate, Fleming inferred that
the mould was producing a substance capable of dissolving bacteria. The mould was
identified as being a Penicillium. At first it was incorrectly classified as Penicillium
rubrum, but later it was found to be the much rarer species Penicillium notatum. Fleming
accordingly gave the name penicillin to the bacteriolytic substance which he thought was
being produced by the mould.
The events described so far may make it look as if Fleming’s discovery was
simply a matter of luck. Indeed, there is no doubt that a lot of luck was involved. Hare
subsequently tried to reproduce a plate similar to Fleming’s original one, and found to his
surprise that it was quite difficult (see Hare, 1970, pp. 66-80). The general effect of
Fleming’s plate could be produced only if the mould and the staphylococci were allowed
to develop at rather a low temperature. Even room temperature in the summer would
usually be too high, but here the vagaries of the English weather played their part. By
examining the weather records at Kew, Hare discovered that for nine days after 28 July
1928 (just when Fleming had gone on holiday!), there was a spell of exceptionally cold
weather. In addition to this, Hare concluded that [53] (1970, p. 79): ‘ … the plate cannot
have been incubated in the usual way.’ A final point is that the strain of penicillium
which contaminated Fleming’s plate is a very rare variety, and most penicillia do not
produce penicillin in sufficient quantity to give rise to the effect which Fleming observed.
How did such a rare mould find its way into Fleming’s laboratory? The most likely
explanation is a curious one. There was at that time a theory that asthma was caused by
moulds growing in the basements of the houses in which the asthmatics lived. This
theory was being investigated by the scientist C. J. La Touche in the laboratory
immediately below Fleming’s, and La Touche had as a result a large collection of moulds
taken from the houses of asthma sufferers. It seems probable that penicillium notatum
was one of these moulds.
There is no doubt then that a great deal of luck was involved in the discovery of
penicillin. Yet it still needed creativity and insight on Fleming’s part to seize the
opportunity which chance had presented to him. Nothing shows this more clearly than a
comparison of Fleming’s reaction to the contaminated plate with that of his colleagues in
the laboratory (including the head of the laboratory, Sir Almroth Wright) when he
showed it to them. With characteristic candour, Hare describes the complete lack of
interest shown by himself and the others (1970, p. 55):
‘The rest of us, being engaged in researches that seemed far more important than a
contaminated culture plate, merely glanced at it, thought that it was no more than another
wonder of nature that Fleming seemed to be forever unearthing, and promptly forgot all
about it.
The plate was also shown to Wright when he arrived in the afternoon. What he
said, I do not recollect, but … one can assume that he was no more enthusiastic – he
could not have been less – than the rest of us had been that morning.’
Fleming was by no means discouraged by his colleagues’ cool reaction. He took
a minute sample of the contaminating mould, and started cultivating it in a tube of liquid
medium. At some later stage he photographed the plate, and made it permanent by
exposing it to formalin vapour, which killed and fixed both the bacteria and the mould.
Fleming kept the plate carefully, and it is now preserved in the British Museum. The
whole episode then is a perfect instance of the famous claim made by Pasteur in his
inaugural lecture as professor at Lille in 1854 when he said that: ‘In the field of
observation fortune favours only the prepared mind.’2 Let us now examine how
Fleming’s mind had been prepared to appreciate the significance of his contaminated
culture plate.
It is interesting in this context to compare Fleming with Herschel. Herschel
needed a prepared mind to realise that a celestial body which was magnified by his
telescope was something unusual. Specifically he needed the background knowledge that
ordinary stars were not magnified by a telescope but only made brighter, and that, if
something was magnified by a telescope, it had to be either an object within the solar
system or a nebula. However, these bits of knowledge would have been part of the
background of [54] any competent astronomer. The background knowledge which made
Fleming appreciate the significance of the penicillin plate was, by contrast, rather unusual
since it was not possessed by his very competent colleagues. It is not difficult, however,
to see how this background knowledge which was specific to Fleming arose from his
earlier researches.
Fleming, during his researches on lysozyme, had over and over again observed a
substance produced from some naturally occurring organism destroying bacteria. This
was a phenomenon with which he was very familiar. However, it would have struck him
immediately that there was something new and curious about the penicillin case because
the bacteria being inhibited were pathogenic staphylococci, whereas lysozyme only
destroyed non-pathogenic bacteria. From the time of his work on the healing of wounds
in the first world war, Fleming had been aware of the problem of finding a ‘perfect
antiseptic’ – that is an antiseptic which would kill pathogenic bacteria without destroying
the phagocytes. In the light of his knowledge of this problem, it is reasonable to suppose
that, when he saw the penicillin plate, he conjectured that the mould might be producing
a ‘perfect antiseptic’.
The assumption that Fleming made such a conjecture is borne out by his
subsequent actions. Fleming grew the mould on the surface of a meat broth, and then
filtered off the mould to produce what he called ‘mould juice’. He then tested the effect
of this mould juice on a number of pathogenic bacteria. The results were encouraging.
The virulent streptococcus, staphylococcus, pneumococcus, gonococcus, meningococcus,
and diphtheria bacillus were all powerfully inhibited. In fact, mould juice was a more
powerful germicide than carbolic acid. At the same time, mould juice had no ill effects
on phagocytes. Here at last seemed to be a ‘perfect antiseptic’. Indeed in his 1929 paper,
Fleming wrote (p. 236): ‘It is suggested that it [penicillin] may be an efficient antiseptic
for application to, or injection into, areas infected with penicillin-sensitive microbes.’
However, at this point a series of difficulties began to emerge for Fleming and his
colleagues who were working on penicillin. These difficulties led Fleming to the
conclusion that penicillin would not after all be the kind of ‘perfect antiseptic’ which he
had been hoping to find. He did, however, find another important use for penicillin.
There is a certain analogy here to Herschel who identified his curious heavenly body as a
comet rather than a planet. This is why Herschel’s work was only the first phase in the
discovery of Uranus, and a second phase carried out by Lexell was needed to establish
the existence of a hitherto unknown planet. In the same way Fleming’s work was only
the first phase in the discovery of penicillin, and a second phase carried out by Florey and
his Oxford team was needed to establish that penicillin was after all a ‘perfect antiseptic’
– what we would now call a ‘powerful antibiotic’. In the next section I will discuss the
reasons which led Fleming and his team to give up the conjecture that penicillin would be
a perfect antiseptic.
3. Why Fleming abandoned his hope that Penicillin would be a ‘perfect antiseptic’
Fleming did not leave behind a diary or detailed notebook setting out the reasons
behind his changes in research strategy, so that these reasons have to be inferred from
what he did, and naturally this can lead to differences in opinion. There are, however,
three factors [55] which most historians would agree might have influenced Fleming in
abandoning his early attempts to demonstrate that penicillin was a powerful antibiotic.
The first of these was the fact that some of the results of tests carried out by Fleming and
his collaborators seemed to indicate that penicillin would not work against bacteria when
injected into the body. In section 3.1, I will discuss these ‘counter indications’. A second
factor was that there were considerable difficulties in isolating penicillin from mould
juice. These problems will be considered in section 3.2. The third factor was that neither
Fleming nor any of his colleagues carried out what is known as an animal protection
tests. By contrast, tests of this kind were done by Florey and his Oxford team. This issue
will be discussed in section 3.3. We now come to another surprising twist in the story of
the development of penicillin. Although Fleming seems to have given up his initial hope
that penicillin might be a perfect antiseptic, he did not abandon penicillin altogether
because he found another use for it. In section 3.4 I will explain what this use was, and
why it had a very positive influence on the further development of research into
3.1 The Counter-Indications Although the results of Fleming’s first tests were
encouraging, further experiments gave reasons to doubt whether penicillin would be an
effective systemic antibacterial agent. First of all Fleming discovered that while
chemical antiseptics killed microbes in a few minutes, his mould juice took several hours
to do so. Then on 22 March 1929, Fleming’s collaborator Craddock injected 20cc of
penicillin into the ear vein of a rabbit. 30 minutes later a blood sample showed that
almost all penicillin activity had disappeared. So if penicillin required about 4 hours to
kill bacteria, but had disappeared 30 minutes after injection, it looked as if it could not
work. Another finding of Craddock made the situation look even worse. Craddock
discovered that penicillin lost about 75% of its activity in mixtures containing blood
serum. Now as there is a great deal of serum in infected areas, this again strongly
suggested that penicillin would not work as a ‘perfect antiseptic’ if injected into the body.
So we see that Fleming had good reasons for giving up his ‘perfect antiseptic’
hypothesis, but was he too Popperian in doing so? This example perhaps shows that
Popper is too fiercely insistent on the need for scientists to give up hypotheses which
appear to have been refuted.
3.2 Difficulties of Isolation Another problem facing Fleming was that of isolating the
active ingredient (penicillin) from mould juice. Fleming was a bacteriologist not a
chemist, and it could be argued that the chemical problems of isolating and storing
penicillin were what caused him to abandon his research on it. This theory seems to me
false, however, because 3 skilful chemists worked in Fleming’s laboratory around this
time, namely Craddock, Ridley and Holt. As we shall see, between them they took most
of the key steps for the extraction of penicillin which were later carried out by the Oxford
team. This leads me to think that, if Fleming had retained his belief that penicillin might
be a ‘perfect antiseptic’, the chemical difficulties of extraction could have been
overcome. However, he is unlikely to have thought it would be worth taking a lot of time
and trouble to extract something which would not work. In other words, the counter-
indications probably influenced Fleming more than the chemical difficulties of extracting
penicillin from mould juice. [56]
3.3 Absence of an Animal Protection Test There is another important factor connected
with Fleming’s early work on penicillin. Neither he nor his collaborators ever performed
an animal protection test. This is a test in which an animal, e.g. a mouse, is infected, and
then injected with the drug being investigated to see if it cures the animal. Craddock, as
we have seen, carried out an experiment on a rabbit, but this was not an animal protection
test in the sense just defined.
It is in this connection that the discovery of the sulphonamide drugs in 1935 was
very important for the further development of penicillin. These drugs were discovered in
Germany as a by-product of the activities of the giant chemical company I.G.Farben.
The discovery was made by a team headed by Gerhard Domagk, who was born in 1895
and appointed at the early age of thirty-two as director of research in experimental
pathology and bacteriology in the institute attached to the I.G.Farben works at Elberfeld.
Domagk and his team had huge laboratories in which they routinely tested compounds
produced by the firm’s industrial chemists on thousands of infected animals to see if the
compounds had any therapeutic value.
The I.G.Farben chemists Hoerlin, Dressel, and Kothe produced a rich red dye
which was very effective with protein materials such as wool and silk. This was known
as prontosil rubrum. Domagk and his team then discovered that this same compound
possessed the definite ability to cure mice infected with haemolytic streptococci.
Domagk published this finding in 1935, but referred back to experiments carried out in
Now the interesting thing about this case is that the pharmaceutical value of
prontosil rubrum could not have been discovered without the use of animal protection
tests for the simple reason that prontosil rubrum does not inhibit bacteria in Petri dishes
(in vitro). It is only when prontosil rubrum is injected into living creatures (used in vivo)
that it acts as an effective agent against bacteria.3 This suggested that penicillin too,
despite the discouraging in vitro results, might work in vivo.
A personal difference between Fleming and Florey may also have been important
here. Fleming was the deputy of Almroth Wright who cast scorn on random experiments
of the I.G.Farben type and argued that a good scientist should proceed by making
deductions from a few carefully chosen tests. There is much to be said for Almroth
Wright’s approach, but, in this instance, it led to the wrong conclusion. Moreover,
Almroth Wright and Fleming almost never conducted animal experiments. They worked
in vitro, but not in vivo. Florey, on the other hand, had been trained at working on
physiology through animal experiments, and was very skilled at experimental animal
surgery. For him, animal experiments were a matter of routine.
3.4 Why Fleming Nonetheless Preserved the Penicillin Mould Despite abandoning his
hopes that penicillin might be a ‘perfect antiseptic’, Fleming nonetheless continued the
cultivation of the penicillin mould and the production of mould juice. This was
extremely fortunate since the mould (penicillium notatum) was a very rare [57] type of
penicillium, and most penicillia do not produce penicillin in the same quantity, if at all.
If Fleming had ceased cultivating the mould, it would have been difficult to restart doing
so. He continued the cultivation because he had found another use for mould juice.
The main source of income of the inoculation department where Fleming worked
was the production and sale of vaccines. There was indeed an efficient unit for producing
vaccine (a vaccine laboratory, as it was then called) within the walls of the department,
and Fleming had been in charge of the production of vaccines since 1920. In particular, a
vaccine was made against Pfeiffer’s bacillus (bacillus influenzae) which was believed to
cause influenza and other respiratory infections. It was difficult to isolate this bacillus
because cultures were apt to be swamped by other micro-organisms. Fleming, however,
had discovered that penicillin, despite its effect on so many virulent bacteria, left
Pfeiffer’s bacillus unaffected. By incorporating penicillin into the medium on which he
was growing Pfeiffer’s bacillus, he could eliminate the other germs, and produce good
samples of the bacillus itself. Fleming in fact used this method for preparing the
influenza vaccine in his vaccine laboratory for this purpose every week after its
discovery. Significantly, the title of Fleming’s 1929 paper on penicillin was: ‘On the
antibacterial action of cultures of a penicillium with special reference to their use in the
isolation of B. influenzae’. Because of this application of penicillin, cultures of the mould
were established at the Lister Institute, Sheffield University Medical School, and at
George Dreyer’s School of Pathology at Oxford. Thus, when Florey and his team
decided to take up again the question of whether penicillin might be a ‘perfect antiseptic’,
they were able to find samples of Fleming’s strain of penicillium notatum just down the
corridor in the Dreyer School of Pathology where they were working. This then is an
appropriate moment to turn from Fleming to Florey and the Oxford team.
4. The Discovery of Penicillin Phase 2: The Work of Florey and his Oxford team
As we have seen, Fleming worked for a while on lysozyme, and this led on to his
later work on penicillin. Curiously enough the head of the Oxford team, the Australian
Howard Florey, followed the same route. In section 4.1 it will be explained why Florey
got interested in lysozyme, what research he and his team carried out on it, and why this
research suggested that it might be useful to move on to investigate penicillin. Then in
section 4.2 I will describe the Oxford team’s work on penicillin between 6 Sept. 1939 and
27 March 1943. It was this work which established that penicillin was, after all, a very
effective antibiotic.
4.1 The Oxford Team also started with lysozyme Curiously enough the Oxford team
also started by working on lysozyme, and moved on from there to penicillin. Howard
Florey first got interested in lysozyme from his studies on mucus and its function.
Lysozyme is found mainly in mucus-containing body fluids. At any rate on 17 January
1929 he sent various organs and tissues from experimental rats which had been killed to
Fleming, presumably for help in assaying the quantity of lysozyme they contained.
In the late 1930s, Florey was joined in Oxford by Ernst Chain, a Jewish refugee
from Nazi Germany, and an expert biochemist. During the academic year 1938-9, Chain
worked on lysozyme with Epstein, an American D.Phil student and Rhodes scholar.
They [58] confirmed that lysozyme is an enzyme – a polysaccharidase, and then looked
for the substrate in the bacterial cell wall which it broke down. This turned out to be N-
acetyl glucosamine. This result was published in 1940.
While working on lysozyme, Chain surveyed the literature on other natural
antibacterial substances which might be worth investigating. This is how he came across
Fleming’s 1929 paper on penicillin. This was in Vol. 10 of the British Journal of
Experimental Pathology, while Fleming’s papers on lysozyme were in Vols 3 & 8, and
Florey’s in Vol. 11. Chain at first thought that penicillin was a kind of mould lysozyme,
a bacteriolytic enzyme on which he could repeat his investigation of lysozyme.
Interestingly he did not realize at the time that Fleming was still alive.
After discussions with Florey, the two of them decided to investigate penicillin,
and Florey prepared a grant application to support Chain.
4.2 The Oxford Team’s work on penicillin (6 Sept. 1939 – 27 March 1943) Britain
declared war on Germany on 3 Sept. 1939, and on 6 Sept. 1939 Florey sent in the grant
application. Here is an extract (taken from Macfarlane, 1979, p. 299):
‘Filtrates of certain strains of penicillium contain a bactericidal substance, called
penicillin by its discoverer Fleming, which is especially effective against staphylococci,
and acts also on pneumococci and streptococci. There exists no really effective substance
acting against staphylococci in vivo, and the properties of penicillin which are similar to
those of lysozyme hold out promise of its finding a practical application in the treatment
of staphylococcal infections. Penicillin can easily be prepared in large amounts and is
non-toxic to animals, even in large doses. Hitherto the work on penicillin has been
carried out with very crude preparations and no attempt has been made to purify it. In
our opinion the purification of penicillin can be carried out easily and rapidly.
In view of the possible great practical importance of the above mentioned
bactericidal agents it is proposed to prepare these substances in a purified form suitable
for intravenous injections and to study their antiseptic action in vivo.’
In October the Medical Research Council approved Chain’s grant for £300 per
annum, with £100 expenses for three years. This was not enough but Florey got an
additional grant from the Rockefeller Foundation of £1000 for the initial cost of
equipment and £1670 per annum for 5 years. With this money, the work was able to go
The first step was to purify penicillin which Florey had rather optimistically said
in his research proposal ‘can be carried out easily and rapidly’. In fact it was a difficult
task. The first result was that penicillin is soluble in alcohol. This had been shown
earlier by Craddock and Ridley – Fleming’s collaborators. This was interesting in that is
showed that penicillin was not a protein. Chain confirmed this by demonstrating that
penicillin would pass through micropore filters that retained proteins. So penicillin was
not an enzyme and had a relatively small molecule. This must have disappointed Chain,
but it did show that penicillin might be injectable without producing the allergic reactions
due to foreign proteins.
The next result was that penicillin could be extracted by ether if the mixture was
made acidic. This too had been discovered by a Professor of Biochemistry at the London
School [59] of Hygiene and Tropical medicine (Harold Raistrick) in 1930, but Raistrick
had been unable to get the penicillin back from the ether. This problem was solved by
another of Florey’s team at Oxford: Norman Heatley.
In March 1940 Heatley discovered back-extraction. The idea was simple. If an
acidified solution was needed to make penicillin dissolve in ether, perhaps an alkaline
solution would cause it to come out of the ether. Heatley shook ether, containing
penicillin, with alkaline water, and, sure enough, the penicillin passed back into the
water. Curiously enough this had already been discovered by Holt working in Fleming’s
laboratory in 1934, but had never been published.
Partially purified penicillin could now be prepared, and Florey conducted a
systematic series of administration and toxicity tests in April & May of 1940. Penicillin
was injected into rats, mice, rabbits, and cats with no ill-effects, though it was rapidly
excreted and had largely disappeared from the bloodstream in 1 or 2 hours. (Penicillin is
in fact toxic to guinea pigs. So it is fortunate that these were not used.) Penicillin was
also shown to be harmless to the delicate white cells known as leucocytes. Fleming’s
findings regarding the inhibitory effect of penicillin on a wide range of pathogenic
bacteria were confirmed, and it was realised that penicillin worked by blocking the
normal bacterial process of cell-division.
So far the Oxford team had largely repeated Fleming’s work, albeit more
systematically and on a larger scale. The next step was their crucial innovation. On
Saturday 25 May 1940, they carried out the first mouse protection test. 8 mice were
injected by Florey at 11am intraperitoneally with 100 million streptococci. 4 (the
controls) received no further treatment. Of the remaining 4, 2 (Group A) received an
injection of 10 mg of penicillin subcutaneously at 12 pm and no further treatment, while
the remaining 2 (Group B) received 5 injections of 5 mg at 2 hour intervals starting at 12
pm. The results are shown in figure 1.4
The 4 control mice were all dead within 16 hours. The 2 group A mice survived 4 and 6
days. One group B mouse survived 13 days and the other 6 weeks +.
The result of this test was so striking that it could leave little doubt that penicillin
was a highly effective antibiotic. However it was only the first of a series of animal trials
carried out by the systematic Florey. These prepared the ground for the first clinical trial,
but since a man is 3,000 times the weight of a mouse, the production of penicillin had to
be greatly increased. Florey tried to interest several pharmaceutical firms but with little
success. They were all occupied with seemingly more urgent war work. Undaunted,
Florey decide to turn his laboratory into a factory, and was greatly helped in this by the
innovative and resourceful Heatley. In 1941, the extraction of penicillin was improved
by the use of column chromatography, which yielded penicillin ten times purer. This was
necessary for the clinical trials, which were carried out by Fletcher, a doctor who worked
with Florey.
The first clinical trial took place on 12 February 1941. This case illustrates the
ravages which bacterial infections could cause in the pre-penicillin era. The patient was a
43 year old policeman, Albert Alexander, who had become infected after pricking
himself on a rose bush. The infection had spread over his face, and his left eye had to be
removed on 3 February. However, the infection spread further into his right shoulder and
lungs. On 12 February, Fletcher gave him an intravenous injection of 200 mg of
penicillin, followed by 100 mg at 3 hour intervals. The patient improved immediately.
Within 4 days his temperature dropped to normal and his appetite returned.
Unfortunately though, the penicillin was running out even though it was being extracted
from his urine and re-injected. Without further penicillin, his improved health continued
for a while, but then he relapsed and died on 15 March 1941.
The third patient, Percy Hawkins, a 42 year old labourer, was a more unequivocal
success. He had a 4 inch carbuncle on his back. Beginning on 3 May 1941, he was given
200 mg of penicillin every hour for five hours, then 100 mg hourly. After 7 May this
dose was reduced to 50 mg. By 10 May the carbuncle had almost completely
disappeared. This was a striking case of a successful treatment of a localised
staphylococcal infection.
Of course these clinical trials were only the start. Howard Florey continued the
work helped by is wife Ethel who was a doctor. On 27 March 1943, they published a
paper in The Lancet describing 187 cases of sepsis treated with penicillin. This
established beyond doubt the efficacy of penicillin as an antibiotic.
5. Suggested Modifications of Kuhn’s Theory in the light of the Penicillin example
Let us now return to Kuhn and see how well the work of Fleming and the Oxford
team fits his theory of discovery in science. It is certainly a striking confirmation of
Kuhn’s claim (1962a, p. 171):
‘ … the discovery of scientific novelty … extends over time and may often involve a
number of people.’
And of Fleck’s claim (1935, p. 76):
‘… discovery must be regarded as a social event.’ [61]
This feature of scientific discovery is still largely unrecognised by the general public.
Alexander Fleming is famous everywhere as the discoverer of penicillin, while the name
of Howard Florey and the work of the Oxford team remain largely unknown.
This attribution of credit in the popular imagination goes back to the time (August
1942) when news first leaked out of the new ‘miracle cure’ and the press got interested.
A key factor was Sir Almroth Wright’s letter to The Times of 31 August 1942 which ran
(Macfarlane, 1979, p. 349):
In the leading article on penicillin in your issue yesterday you refrained from
putting the laurel wreath for this discovery round anyone’s brow. I would, with your
permission, supplement your article by pointing out that, on the principle palmam qui
meruit ferat it should be decreed to Professor Alexander Fleming of this laboratory. For
he is the discoverer of penicillin and was the author of the original suggestion that this
substance might prove to have important applications in medicine.’
In the outburst of press interest which followed, Fleming and Florey handled
things very differently. Fleming saw reporters and gave interviews. Moreover this dour
and laconic Scotsman proved to be a great favourite with the general public. Florey on
the other hand refused to see any reporters and sent them away. It was natural then that
the press should concentrate on Fleming whose name was the only one which became
known to the public. The scientific community was more judicious. When a Nobel prize
for the discovery of penicillin was awarded in 1945, it was divided between Fleming,
Chain, and Florey. This was certainly quite reasonable, but it might well be argued that
further members of the Oxford team, such as Norman Heatley, could have been included.
While the discovery of penicillin confirms Kuhn’s views on the social nature of
scientific discovery, it diverges from other aspects of his account. Kuhn essentially
considers two types of discovery, viz. (i) the unproblematic class, in which an object is
predicted and later discovered (e.g. radio waves), and (ii) the troublesome class, in which
something strange is observed and its true nature only later elucidated. The discovery of
penicillin does not fit exactly into either class. Penicillin was certainly not predicted, but,
before its discovery, the concept of a perfect antiseptic had been formulated, and Fleming
was on the look out for such an antiseptic without, however, any definite conviction that
he would find one. It could be said that penicillin, or something similar, was hoped for
rather than predicted.
In Kuhn’s ‘troublesome class’ of discoveries, more than one scientist may be
needed because the process requires (i) the initial observation of something unusual (e.g.
Herschel’s observation of a curious celestial object), and (ii) the elucidation of the nature
of what was observed (e.g. Lexell’s claim that Herschel’s celestial object was a planet).
In the penicillin case, however, more than one scientist was necessary because Fleming’s
initial hypothesis regarding penicillin, viz. that it was a perfect antiseptic appeared to be
refuted by experiments, so that further work was needed to show that the initial
hypothesis was correct after all. Thus, in addition, to Kuhn’s two classes, we might
introduce a third class of discovery in which researchers are looking for something with a
particular set of properties. The discovery then consists of two stages: (i) becoming
aware of something which might have the required set of properties, and (ii) the
demonstration that it really does have these properties. [62]
1. My account of the discovery of penicillin is largely based on Hare (1970) and Macfarlane (1979 and
1984). Hare’s book is partly an eye witness account since he was working in the same laboratory as
Fleming when Fleming made his discovery. Macfarlane’s two books are excellent historical works which
are informed by a deep scientific knowledge of the area.
2. In Pasteur’s original French, the quotation runs: ‘Dans les champs de l’observation le hasard ne favorise
que les esprits préparés’. This is slightly ambiguous since ‘le hasard’ in French can mean ‘luck or fortune’
as it is translated here, or ‘chance’ in the statistical sense.
3. Further details about the discovery of prontosil rubrum, and the explanation of why it works only in vivo
and not in vitro, are to be found in Gillies, 1993, pp. 48-53.
4. This figure is taken from Macfarlane, 1979, p. 314.
FLECK, LUDWIK, Genesis and Development of a Scientific Fact, 1935. English
Translation by Fred Bradley and Thaddeus J. Trenn, Chicago University Press, Chicago
and London, 1979.
FLEMING, ALEXANDER, “On the Antibacterial Action of Cultures of a Penicillium,
with Special Reference to their Use in the Isolation of B. Influenzae,” British Journal of
Experimental Pathology, v. 10, 1929, pp. 226-236.
GILLIES, DONALD, Philosophy of Science in the Twentieth Century. Four Central
Themes, Blackwell, Oxford UK & Cambridge USA, 1993.
HARE, RONALD, The Birth of Penicillin and the Disarming of Microbes, George Allen
and Unwin, London, 1970.
KUHN, THOMAS, The Historical Structure of Scientific Discovery, Science, v. 136,
(1962a), pp. 760-764. Reprinted in THOMAS S. KUHN, TheEssential Tension. Selected
Studies in Scientific Tradition and Change, The University of Chicago Press, Chicago
and London, 1977, Chapter 7, pp. 165-177.
KUHN, THOMAS, The Structure of Scientific Revolutions, The University of Chicago
Press, Chicago and London, 1962b.
MACFARLANE, GWYN Howard Florey. The Making of a Great Scientist, Oxford
University Press, Oxford, 1979.
MACFARLANE, GWYN Alexander Fleming. The Man and the Myth, Chatto &
Windus, The Hogarth Press, 1984.
ResearchGate has not been able to resolve any citations for this publication.
Nature is the international weekly journal of science: a magazine style journal that publishes full-length research papers in all disciplines of science, as well as News and Views, reviews, news, features, commentaries, web focuses and more, covering all branches of science and how science impacts upon all aspects of society and life.