In September 1928 Alexander Fleming returned from holiday and began to sort through the mess in his laboratory. At ﬁrst he discarded a culture plate that had been lying in the open air for some weeks. On it a blob of mould had interfered with the development of the staphylococci that had been sown on the jellied broth. Glancing at it again, Fleming rescued the plate from the bath of disinfectant in which it was about to be immersed. Six years earlier he had discovered a substance called lysozyme, a substance found in tears, saliva, and mucus, which had shown a similar capacity to kill oﬀ bacteria. Lysozyme, it had turned out, had little eﬀect on those bac- teria that cause dangerous diseases, but Fleming’s experience with it meant he only needed a glance at his contaminated plate to recognize that something important might be happening, for on this plate the unknown mould was killing an organism which was a common source of dangerous infections, a staphylococcus. It was straightforward to establish that the mould was a member of the Penicillium family, and that it was active against numerous danger- ous bacteria. Fleming could easily show that it did no harm to white blood cells: this was important because the laboratory he worked in, headed by Almroth Wright, had long been committed to the idea that the key to eﬀective treatment was to mobilize the body’s own capacity for defence. Fleming himself, during the First World War, had studied infections in soldiers’ wounds and had argued that con- ventional antiseptics both killed oﬀ white blood cells faster than they killed bacteria, and failed to penetrate into the jagged interstices of gunshot wounds: they were, he thought, positively fostering infection. He could also straightforwardly show, by injecting the broth derived from his mould into a very small number of mice and rabbits, that it was not toxic. And he could also show that it quickly lost its anti- bacterial eﬀect when mixed with digestive juices: there would be no point in taking it as a pill. Fleming was surely moving towards injecting penicillin (as he was soon to call his ‘mould broth ﬁltrate’) into infected animals to see if it would cure them. He had long worked with salvarsan, which was the ﬁrst drug eﬀective against syphilis, a disease that Fleming had exten- sive experience of treating in private practice. But by April 1929 he seems to have lost all interest in injecting penicillin into the blood- stream. Penicillin took around four hours to kill bacteria; but tests showed that both in animals and in the test tube it ceased to be active in blood after two hours. This seems to have persuaded him that it would be pointless introducing penicillin into a diseased body. This left the alternative of applying penicillin topically to local infections. He used it successfully on a case of conjunctivitis, and applied it to carbuncles with mixed results. An attempt to treat a case of septicaemia was a failure. The possibility that penicillin might have a future use as an antiseptic was mentioned in Fleming’s ﬁrst and only major publication on his new discovery, which appeared in 1929. He wrote: ‘It is suggested that it may be an eﬃcient antiseptic for applica- tion to, or injection into, areas infected with penicillin-sensitive microbes.’ But between 1930 and 1940 Fleming made no eﬀort to develop a clinical use for penicillin. Throughout this period, however, he employed it regularly for the one use that was outlined in his key publication. While penicillin killed many bacteria, it did not kill a bacterium called Pfeiﬀer’s bacterium, which some (including Fleming) thought might be the main cause of inﬂuenza. Pfeiﬀer’s bacterium was very diﬃcult to cultivate because it was usually con- taminated by other bacteria that grew more vigorously and over- whelmed it. Fleming found that if he took a sample of mucus and spread it over a petri dish treated with penicillin, then he could grow a pure sample of Pfeiﬀer’s bacterium, because it was immune to penicil- lin, while the bacteria that normally overwhelmed it were sensitive to it. Fleming was happy with this discovery because the laboratory in which he worked was funded by the production of vaccines. Almroth Wright in eﬀect ran a private company within St Mary’s Hospital in London; from its income he ran a couple of wards and paid the salaries of a small staﬀ. The enterprise seems also to have been person- ally proﬁtable for Wright and his close associates, including Fleming. Wright had discovered the vaccine against typhoid, a conventional prophylactic vaccine, but from his starting work at St Mary’s in 1902 until his retirement in 1946 his main preoccupation was producing vaccines that would be given to people after they had become infected and would stimulate the body’s defences –– the model was Pasteur’s vaccine against rabies, which was injected after the victim had been bitten by a rabid dog. Wright and Fleming (Fleming had charge of the production of vaccines from 1920 on) produced vaccines against acne, boils, inﬂuenza, gonorrhoea, tuberculosis, and cancer. The modern view would be that all these vaccines were totally ineﬀectual: the whole proﬁtable business was founded on a failure to carry out adequate controlled trials to see if people receiv- ing Wright’s vaccines did better than those not receiving them. Wright had been involved in a controversy over the statistics he had used to show that his typhoid vaccine (which really was eﬀective) worked, and he carefully avoided subjecting his new vaccines to proper tests. Instead he invented a spurious measure, the opsonin index, of the body’s resistance to infection, and claimed this measured the improvement resulting from his vaccinations. Fleming thus made a very good living out of selling what were, in eﬀect, sophisticated quack remedies. Penicillin, he hoped, would enable him to produce an ‘improved’ inﬂuenza vaccine. Although Fleming recognized that penicillin might possibly have a therapeutic use, he was far too interested in the production of vaccines to waste much time exploring the possibility. A few dis- couraging ﬁndings, and he dropped all work on it. He was also quite uninterested in the problem of how to produce purer, stronger sam- ples of his new drug. Two students of his, Ridley and Craddock, did astonishingly able work, under horribly primitive conditions (they worked on tables in a corridor, and had to go to the next ﬂoor to ﬁnd running water), to produce a purer drug. They evaporated a broth made from the penicillin under a vacuum, and dissolved the penicillin in alcohol, in the process purifying it further. Where their ﬁrst prepar- ations were highly unstable, they discovered that they could make the penicillin stable by adding acid. Fleming seems to have had virtually no interest in their work. He misreported some of their ﬁndings in his ﬁrst publication, and later claimed that the problem of producing stable penicillin had proved insoluble. When others set about pro- ducing penicillin in a purer and more stable form, they had to rediscover everything that Ridley and Craddock had discovered because Fleming never mentioned their work to later investigators. Fleming himself was quite happy using the impure penicillin broth, which was perfectly adequate for the production of uncontaminated samples of Pfeiﬀer’s bacillus. In September 1939 (to be exact on 6 September, three days after the declaration of war), eleven years after Fleming’s discovery, Howard Florey, in Oxford, began to seek funding for penicillin research –– for a year or so his colleague Ernst Chain had been cultivat- ing penicillin derived from Fleming’s original strain. Florey and Chain were engaged in a systematic search to ﬁnd biological agents (rather than the chemical agents already developed into salvarsan and prontosil) which would be capable of killing the bacteria that caused fatal infections, and penicillin was only one agent on their shortlist of promising substances. In May 1940 they had enough puriﬁed penicil- lin to carry out a straightforward experiment: injecting penicillin into four mice that had been infected with streptococci –– four others were infected but not given penicillin. The results were dramatic, for the mice treated with penicillin survived in good health, and those not treated died. They were conﬁrmed when the experiment was repeated the next day. From the ﬁrst, the Oxford team (two professors and seven researchers) were convinced they had a discovery of the foremost importance –– they smeared the spores of Fleming’s strain of penicillin into the linings of their coats, so that if the Germans invaded they could preserve their raw material. In August 1940 they published the results of their animal experi- ments; a year later they published the results of the ﬁrst trials on humans, although the quantities of penicillin available meant that it had been possible to treat only ﬁve patients. All had shown astonish- ing improvements, although two had died. The most striking case, perhaps, was the cure of a boy of 14 with staphylococcal septicaemia from an osteomyelitis of the left femur, a condition previously almost invariably fatal. In August 1941 the results of these trials were pub- lished, and the race was on to produce penicillin in commercial quan- tities. By June 1943, penicillin production in the US was enough to treat 170 cases a month; a year later it was enough to treat 40,000 cases a month, or all the battleﬁeld casualties of the Allied invasion of Europe; and within another year it was enough to treat a quarter of a million patients a month. That year, 1945, Fleming, Florey and Chain shared the Nobel prize for medicine. A medical revolution had taken place within the space of four years. From the beginning, Fleming and his associates sought to claim for him the credit for the therapeutic use of penicillin. In September 1940, after the publication of Florey and Chain’s animal experiments, Fleming himself wrote to the British Medical Journal pointing out that he had foreseen a therapeutic use in his 1929 paper. In September 1941, after the publication of the ﬁrst clinical trials, Almroth Wright wrote to The Times (which had published a leading article on penicil- lin) to claim the credit for the discovery for Fleming. Fleming, who had abandoned work on the clinical use of penicillin within months of his ﬁrst discovery, happily answered calls from the press. He rapidly became famous throughout the world as the discoverer of penicillin, while Florey and Chain were left in obscurity. Fleming’s contribu- tion only began to be placed in proper perspective with the publication of Ronald Hare’s The Birth of Penicillin in 1970. The story I have just told is now a familiar one. It was Florey and Chain, not Fleming, who demonstrated the clinical value of penicil- lin, and they and their associates who began to solve the problems of producing penicillin on an industrial scale. But their key experiment of May 1940 could have been carried out by Fleming, who certainly had, particularly as a result of the unappreciated work of Ridley and Craddock, an adequate supply of penicillin to inject into mice. Had he done this experiment in 1929 literally millions of lives could have been saved, lives that were lost without an adequate broad-spectrum antibiotic. (Some writers have claimed that the technology of freeze- drying was essential for the Oxford work, and was not available in 1929; but Ridley and Craddock’s work shows that Fleming could have managed without it.) If Fleming deserves the credit for recog- nizing the action of penicillin on his contaminated dish, he also carries the responsibility for this delay. The situation would thus appear straightforward: Fleming discovered penicillin; Florey and Chain ﬁrst put it to eﬀective use. The question of the relative contribution of Fleming on the one hand, and Florey and Chain on the other to the revolution repre- sented by modern drug therapy has however distracted attention from an even more puzzling and diﬃcult question. In what sense can Fleming be said to have discovered penicillin? Contamination of bacterial cultures by moulds takes place all the time. In 1871 Sir John Burdon Sanderson reported that moulds of the Penicillium group would prevent the development of bacteria in a broth exposed to the air. In 1872 Joseph Lister established that the growth of Penicillium glaucum would kill oﬀ bacteria in a liquid culture. He at once saw the possible clinical application of the phe- nomenon. He wrote to his brother saying ‘Should a suitable case present, I shall endeavour to employ Penicillium glaucum and observe if the growth of the organisms be inhibited in the human tissues.’ He never published his results, so we do not know how far and how long he pursued the question, but we do know that in 1884 a patient of Lister’s, a young nurse, was suﬀering from an infected wound. Various chemical antiseptics were tried without success, and then a new sub- stance was used. She was so astonished and so grateful at her seem- ingly miraculous cure that she asked Lister’s registrar to write the name of this substance in her scrap-book. It was penicillium. Why did Lister keep this success to himself? There is, I think, only one possible explanation. Throughout the 1870s and 1880s he was struggling to win acceptance for the principle of antiseptic surgery. He lacked the energy or the resources to embark on a new campaign while the germ theory itself remained so widely contested. In 1895 Vincenzo Tiberio in Naples injected extracts of penicillium moulds into infected animals, the experiment ‘ﬁrst’ performed by Florey and Chain in 1940, though his results were nothing like as striking as theirs. In 1897, a young French army doctor called Duch- esne described similar experiments in a thesis. His preliminary results were certainly striking; unfortunately he died of tuberculosis before he could carry out further trials. Fleming was blissfully ignorant of all this previous work. Had he known of it he might have been less quick to claim the credit for the discovery of a new substance. Gwyn MacFarlane, whose ﬁne book on Fleming is my source for this information, tries to play down its signiﬁcance. He argues that the mould Fleming discovered in 1929 was a rare strain of Penicillium notatum. Where many strains of penicillium are completely inactive, including most strains of Penicillium notatum, Fleming’s strain was peculiarly powerful. Exhaustive studies in the early 1940s were to ﬁnd only two more powerful strains among hundreds tested. Fleming’s rare strain of P. notatum was far more active than any used by Burdon-Sanderson, Lister, Tiberio, Duchesne, and many others from 1870 onwards. If any one of these had been lucky enough to have been visited by the mould that alighted on Fleming’s plate in 1928, they too would have discovered penicillin and might possibly have taken it further than he did . . . Here Macfarlane states as a fact what is at best a statement of probability. We do not know exactly what strains Burdon Sanderson, Lister, Tiberio, Duchesne and the others worked with. We do know however that none of them had diﬃculty ﬁnding active strains of penicillium, and that in every case the activity of the penicillium was suﬃciently marked to suggest that it had clinical potential. Others too watched penicillium killing oﬀ bacteria. Tyndall, for example, had noted in test tube after test tube ‘the struggle for existence between the Bacteria and the Penicillium’, although he had not grasped the potential signiﬁcance of what his eyes had seen. Thus it would seem fair to say that Lister and Duchesne had both independently dis- covered penicillin, and had taken it somewhat further than Fleming did, and that there was nothing remarkable in Fleming’s initial identiﬁcation of penicillium as an antibiotic. Moreover the idea behind the research project of Florey and Chain, the idea that one could ﬁnd biological agents capable of killing oﬀ infectious diseases (what we now call antibiotics), was not a new one. Lister had immediately recognized the potential of penicillium in 1872. Arnaldo Cantani had used bacteria painted on the throat of a sick child to reduce her fever in 1885, and had stated the principles involved. So the larger project envisaged by Florey and Chain, that of research on what was initially called bacterial antagonism, was not original –– one of the major purposes of a book published by George Papacostas and Jean Gaté in 1928, Les Associations microbiennes, was to collect together in one place the information on this. Here again we encounter the same phenomenon that we have encountered so often before. The true puzzle about penicillin is why it was not brought into medical use ﬁfty years earlier. Florey and Chain discovered an eﬀective antibiotic within months of starting looking for one; there is no reason to think that a similar achievement was beyond a Pasteur or a Lister.