At 10:20 on the morning of Tuesday 12 April 1955, in the Rackham Building of the University of Michigan at Ann Arbor, Thomas Francis stood up before some four hundred scientists and physicians to read the result of the largest medical experiment that had ever been run. In the press room along the corridor more than two hundred reporters and cameramen had been waiting since early morning. Men with guards came to the door carrying the report in corrugated boxes, and never got through it; the crowd pulled the copies out of their hands, and the men began throwing them into the air. The NBC television network had already broken the embargo, at 9:22, from the campus.

The sentence everyone was waiting for was in the summary’s conclusions. “On this basis it may be suggested that vaccination was 80 to 90 per cent effective against paralytic poliomyelitis, that it was 60 to 70 per cent effective against disease caused by Type I virus and 90 per cent or more effective against disease from Type II and Type III virus.” The basis was 200,745 children who had been given three injections of the vaccine the summer before, and 201,229 who had been given three injections of salt water, and nobody who had examined any of them had known which was which.

That afternoon, in Washington, the Secretary of Health, Education, and Welfare signed licences for six companies to make the vaccine and sell it. By the following day the government’s Laboratory of Biologics Control had released thirteen lots for use: two from Eli Lilly, three from Parke, Davis, two from Pitman-Moore, and six from the Cutter Laboratories of Berkeley, California.

Who ran it

Three organisations ran the experiment, and a fourth made the thing that was being tested.

The first was the National Foundation for Infantile Paralysis, the charity that Basil O’Connor, Franklin Roosevelt’s former law partner, had built out of the March of Dimes. It paid for the vaccine’s development, bought the vaccine for the trial, ran the trial through its county chapters and the state health departments, and in August 1954, before anyone knew whether the vaccine worked, ordered up to 27 million cubic centimetres of it from the manufacturers so that they would stay in production. Its Vaccine Advisory Committee set the standards the vaccine was made to.

The second was Jonas Salk’s Virus Research Laboratory at the University of Pittsburgh, where the vaccine had been devised. Salk was forty in the spring of 1955. His vaccine was poliomyelitis virus of all three types, grown in minced monkey kidney, killed with formaldehyde, and injected. Between May 1953 and March 1954 he had given his own laboratory’s vaccine to 5,320 volunteers, and in the spring of 1954 he gave the manufacturers’ vaccine to 7,507 children around Pittsburgh as the last check before the trial began.

The third was the Poliomyelitis Vaccine Evaluation Center at the University of Michigan, under Francis, who had been Salk’s teacher. His deputy was Robert Korns of the New York State Department of Health. The Center borrowed epidemiologists from the Public Health Service’s Communicable Disease Center in Atlanta, statisticians from the Bureau of the Census, and the university’s tabulating machines, and twenty-seven laboratories did its blood tests.

The vaccine itself was made by industry. In February 1954 the Foundation sent its provisional standards to ten pharmaceutical firms and invited them to make vaccine for the trial. Five accepted: Parke, Davis of Detroit, Eli Lilly of Indianapolis, Wyeth of Marietta, Pennsylvania, Pitman-Moore of Zionsville, Indiana, and the Cutter Laboratories of Berkeley, a family firm whose president was Robert K. Cutter. Sharp and Dohme of Philadelphia joined later. Over all of them stood the Laboratory of Biologics Control, a laboratory of the National Institutes of Health under William Workman, which licensed biological products under an Act of 1902 and had never before licensed anything like this.

The Public Health Service came into the story properly only in its last act. Its Surgeon General was Leonard Scheele; the Secretary over him was Oveta Culp Hobby, who had commanded the Women’s Army Corps in the war; and in Atlanta the chief of the Communicable Disease Center’s Epidemiology Branch was Alexander Langmuir, who on 28 April 1955 was given a surveillance unit and put a young Epidemic Intelligence Service officer named Neal Nathanson in charge of it.

What they were trying to find out

The hypothesis had two parts, and the whole of this entry is about the seam between them.

The first part belonged to Francis, and it was a question about an effect: did three injections of formaldehyde-killed virus prevent paralytic poliomyelitis in children? It could only be answered by giving the vaccine to a great many children in a polio season and comparing them with a great many who had not had it, and the comparison was only worth making if the two groups were alike in everything but the vaccine and nobody who diagnosed a case knew which group the child was in. The health departments of several states insisted on that, and in the end about half the trial was run their way.

The second part belonged to the manufacturers and the Laboratory of Biologics Control, and it was a claim about a process: that the vaccine could be made safe by the method of making it. The Public Health Service’s technical report of June 1955 put the method in one sentence: “Dr. Salk has stated that the process of inactivation follows the course of a first order reaction yielding a straight line when the log of the viral concentration is plotted against time.” Measure the fall of live virus for the first two or three days, draw the line, and extend it. “The original concept of vaccine preparation was that the process itself assured a wide margin of safety. Safety tests for the vaccine were conceived to detect both mass contamination resulting from accidents in manufacture, and residual live virus which the process was intended to eliminate.”

The two parts had been joined in writing before the trial began. On 4 February 1954 the chairman of the Foundation’s Vaccine Advisory Committee wrote to the Journal of the American Medical Association that every reasonable safeguard had been built into the standards, and that “since three separate laboratories will be carrying out the safety tests, chances of error will be remote indeed.” In April 1954, with the Public Health Service’s concurrence, the committee’s statement authorising the trial said that “the possibility of infectious activity remaining in any vaccine meeting the specifications and Minimal Requirements has been reduced to a point below which it cannot be measured by practicable laboratory procedures.”

That last sentence is the hinge. The trial was designed to measure the vaccine’s effect to within a few percent. The vaccine’s safety was defined as something that could not be measured, and the test for it was designed accordingly.

What they built

The trial first. The Foundation’s original plan was what the statistician Paul Meier later called the observed control approach: offer the vaccine to every second-grader in a participating school, and compare their polio with that of the first- and third-graders around them. It was cheap and simple, and the health departments of several states saw its two flaws at once: a doctor diagnosing a feverish seven-year-old would know whether the child had been vaccinated, and the families who volunteered would differ from the families who did not. “Certain health departments took a firm stand,” Meier wrote: “they would participate in the trial only if it were such a well-designed experiment.” So in roughly half the areas every volunteer in the first three grades was given three injections from a vial that carried only a code number, half the vials holding vaccine and half salt water, assigned by chance, and no one in the clinic, the family or the laboratory knew which until every case had been classified. The companion piece “Salt water” tells how the design was argued and what it found.

The study population, in 44 states, came to 1,829,916 children. In the placebo areas there were 749,236 of them: 200,745 vaccinated, 201,229 given placebo, 338,778 whose parents had not volunteered. In the observed areas there were 1,080,680: 221,998 second-graders vaccinated and 725,173 first- and third-graders watched. Every reported case went to a physical therapist for a muscle examination at the time and again at two months, and to one of the laboratories for virus and antibodies, and was classed as not polio, doubtful, non-paralytic, paralytic or fatal before the code was broken. The injections were given between April and June 1954, and the polio season that followed was a worse one than usual in the trial areas.

Now the vaccine. The provisional standards had been drafted between December 1953 and February 1954 by Workman and by Joseph Smadel of the Army Medical Service Graduate School, from Salk’s account of what he did in his own laboratory. Each type of virus was grown separately in monkey kidney tissue, to a titre of a million to ten million infective units per millilitre, filtered, and inactivated in formaldehyde at one part in four thousand, at 37 degrees. The three single-strain pools, each tested for live virus in tissue culture on a sample equal to one-thousandth of its volume, were mixed into a trivalent pool, which was tested again, on another thousandth, and in eighteen monkeys.

The straight line governed the timing. Starting from a million infective units per millilitre, the measured fall reached one unit per millilitre somewhere between 48 and 72 hours, and the line was drawn on through where nothing could be measured: one unit per million millilitres at six days, one unit per million million at nine. “This concentration is of such a low order that its significance can be grasped only when the numbers are reduced to common terms,” the technical report wrote. “The prediction means that if the reaction proceeds as indicated in Figure 1, there will be a concentration of living virus in the order of one tissue culture infective unit in each million tons of inactivated viral suspension at the end of nine days.” The recommended treatment was three times the period to the intercept, which came to nine days. The tissue-culture test was expected to turn negative somewhere in the second half of that time, and it was permitted, if it did not, to go on heating the pool until it did.

The first commercial lots did not follow the line. In March 1954, within weeks of the start of production for the trial, live virus was found in four of the first six lots that reached the Laboratory of Biologics Control. The Foundation’s committee put the trial back four weeks, tightened the processing, and “established the requirement that a series of consecutive lots with negative tests must be produced by a manufacturer to demonstrate consistent performance before acceptance of his product.” The triplicate testing, by the maker, by Salk and by the government laboratory, found live virus in eleven lots from three firms; only two of the eleven were caught by all three laboratories, and eight by one alone. By 25 April 1954 the runs had turned consistently negative, and that, the report said a year later, “was accepted at the time as indicating that the early difficulties in inactivation had been resolved with experience.” The trial went ahead in May with vaccine from Parke, Davis and Lilly only. Cutter, Wyeth and Pitman-Moore had made lots that passed, and they were not used.

Between the trial and the licence came a year in which the makers geared up for a market. In May 1954 the Laboratory of Biologics Control issued prospective “Minimum Requirements” for anyone who wished to make and store vaccine against the day of licensing. In the autumn the lots returned from the trial were found to have lost potency, the preservative merthiolate was blamed, and several million doses already made were thrown away. From January 1955 the makers sent the government laboratory the detailed records of each production lot, the protocols, with samples. By 12 April forty protocols had come in. The laboratory had fully tested six of the lots, in monkeys and in tissue culture, had run tissue-culture tests on about half the rest, and had examined the others on paper.

On the morning of the Francis report, a group of consultants met Workman in Ann Arbor to advise on licensing. They recommended it unanimously; Workman accepted, Scheele approved, and Hobby signed. The technical report gives the reasons: established firms, each holding from 44 to 83 product licences already, each plant inspected in the winter, a year’s experience, and the field trial. “Although the record of satisfactory inactivation of the virus had at first been poor for the three initial producers, it appeared that improved techniques in processing had by this time corrected the difficulty.” And: “The field trial evidence itself was interpreted to indicate that the commercial product was safe.” By 27 April forty lots and ten and a half million cubic centimetres had been released, and nearly five million had gone into arms, about ninety percent of them the arms of first- and second-graders in school clinics run with vaccine the Foundation had bought. Cutter’s went to clinics in Arizona, California, Idaho, Nevada, New Mexico and Hawaii, and through drugstores to private doctors in at least twenty-five states.

Six cases

On 25 April 1955 an infant with paralysis of both legs was admitted to the Michael Reese Hospital in Chicago. The child had been given Cutter vaccine in the buttock on 16 April, and the paralysis had begun on the 24th. The Chicago Board of Health reported it to the Public Health Service that day. At noon on 26 April a case came in from the Napa Valley in California, a child injected about a week before; that evening four more came in from widely separated parts of the state, and a sixth Californian case during the night. All had received Cutter vaccine ten or eleven days earlier. In each the paralysis had begun in the limb that had been injected. “This was the beginning of the Cutter incident,” Nathanson and Langmuir wrote eight years later, when they had counted all of it.

That evening a telephone conference was held between the staff of the National Institutes of Health and Francis, Smadel, William Hammon of Pittsburgh and Howard Shaughnessy of the Illinois state laboratories. The expected number of cases among unvaccinated children was worked out; it was smaller than six. “The advisors agreed that action should be taken, but that it should be limited to Cutter vaccine. They did not believe there was justification for suspending all poliomyelitis vaccination.” On the morning of 27 April Scheele asked the Cutter Laboratories to recall all their vaccine, and the firm did, notifying every distributor within the hour. The same day the National Institutes of Health sent two scientists to Berkeley to go through the plant’s processing and testing, and a sanitary engineer after them to look for any accident that might have put live virus into a finished lot. On 28 April Scheele ordered the surveillance unit set up in Atlanta, and every state was asked to name a reporting officer and report every case of poliomyelitis in a vaccinated person.

Two questions, the report said, “then urgently required answers”: what had caused the association, and whether the same thing could happen with the other five firms’ vaccine. On 29 and 30 April eleven virologists and immunologists met at Bethesda; by then there were seventeen cases, all but one Cutter’s. The group found the recall justified, found the continuation of vaccination with the other firms’ products warranted, and recommended a review of the Minimum Requirements. On 30 April Workman told the manufacturers he could not approve any more lots until the review was done.

What the review found was not a Cutter secret. A six-member subcommittee met the makers’ technical men on 5 and 6 May; there were now 38 Cutter cases and six among children given Lilly’s vaccine, though only the Cutter cases showed the paralysis in the injected limb. “The group learned from the manufacturers that inactivation sometimes failed to eliminate all tissue culture or monkey infectivity for no immediately apparent reason. There was evidence also that safety tests were less sensitive and consistent than was desired; on some occasions pools of virus, each containing a single type and each negative after inactivation, gave positive tests when combined into polyvalent pools.” The protocols the government had been reading since January described only the lots the makers had chosen to submit; the records of the lots that had failed “were available, up to that time, only in the producers’ establishments.” Some of the subcommittee wanted the national programme postponed indefinitely; others wanted to go on. At four o’clock on the morning of Saturday 7 May Scheele announced that he was recommending that all vaccination stop until each manufacturer’s records had been gone through, plant by plant. Salk, arriving at Bethesda that afternoon for the meeting that would write the new rules, told the reporters it was “a lovely day.”

What it looked like from inside

Inside the plants, what the inspectors found is set out in the technical report’s tables, with the five firms reduced to the letters A to E. Between them they had put 809 single-strain pools through the first tissue-culture test, and 11 percent had been positive: 21 percent at Laboratory A, 19 at B, 2 at E. Of 127 trivalent pools, each made from three pools already passed as free of live virus, 13 percent were positive on their first test, and at Laboratory A a third. “The process of inactivation did not always follow the predicted course,” the report concluded; in practice the fall of live virus was “often characterized by a curvilinear rather than a straight line relationship.”

Then it did the arithmetic the requirements had never done. A sample of one-tenth of one percent of the pool meant that the chance of catching a given concentration of virus depended on whose pool it was. If the limit were five infective units per litre, Laboratory A, testing 60 millilitres of a single-strain pool, had a 70 percent chance of catching a contaminated one; Laboratories C and F, testing up to 1,600, better than 99.9. The companion piece “Laboratory A” sets out the tables.

Inside the surveillance unit it looked like a map filling in. By the end of May there were 113 cases of poliomyelitis among five and a half million vaccinated people, 69 of them, 59 paralytic, among the 409,000 who had received Cutter vaccine, which was an attack rate of 17 per 100,000 in a month against less than one for the other four firms. Among the Cutter cases the interval from injection to paralysis ran from five to twenty days with its median on the ninth, which is the distribution of a common-source outbreak and almost exactly the distribution that David Bodian at Johns Hopkins had recorded in 32 monkeys injected in the calf with the Mahoney strain. In 43 of 55 cases with complete records the paralysis had begun in the injected limb; among the ten Lilly cases, in one. Forty-seven household contacts of vaccinated children had come down with polio, 35 of them contacts of Cutter children, at intervals of about two incubation periods. And in Idaho, where 32,000 children, nearly eighty percent of the first two grades, had been given Cutter vaccine and nothing else, there was polio in April and May in a state that normally had none until July. In the ten weeks from 17 April to 25 June the state had never reported more than eleven cases in any year from 1950 to 1954. In 1955 it reported 88, and 84 of them were among vaccinated children, their families and their playmates. Two of the first eight children reported, by 28 April, were from Pocatello, Idaho; one of the eight was dead.

Inside a family it looked like this. On 16 April, in a suburb of Baltimore, a baby girl was given one injection from one of the Cutter lots. She never became ill; type 1 virus was found in her stool. Twenty-three days later her mother came down with poliomyelitis. The family had seen two other families nearby, each with small children; in one of those the father, and in the other the mother and a child, fell ill 34 to 49 days after the baby’s injection, and virus was recovered from six of the eight people in the two houses. Their children played with the children of two more families, in which the last two cases came, 37 and 40 days after the injection. Six cases in one block in three weeks, five of them paralytic, from one shot that had made nobody sick.

Inside the Public Health Service it looked like a programme coming apart in public. Parke, Davis’s plant was inspected on 11 May and its lots recleared on the 13th; Scheele, called back before a House committee that morning, told it “that the deliberations which led to the recommendation to postpone temporarily the vaccination program in no way concerned the efficacy of the Salk vaccine itself,” and that “our attention is directed only at the problems of manufacture which have arisen in the transition from a pilot plant basis of operation to mass production.” Lilly’s lots followed on the 15th. Wyeth’s were held, because five children given one Wyeth lot had fallen ill, three of them in the injected limb; the firm withdrew the rest of the lot. On 18 May President Eisenhower was asked about Senator Morse’s charge that Hobby was guilty of gross incompetency; he said she had been “highly efficient.” A reporter noted that she had told the Senate Labor Committee that no one could have foreseen the public demand for the vaccine, and asked what had made it hard to foresee. “I don’t know to what she is referring,” the President said. On 23 May a monkey in Louis Gebhardt’s laboratory at the University of Utah, injected with a suspect Cutter lot from the Idaho health department, was paralysed, and type 1 poliovirus was recovered from its spinal cord. On 10 June the technical report went to Hobby. On 13 July she resigned, for what she called personal reasons of a high order; her husband was ill in Houston.

And inside the room at Ann Arbor on 12 April, by one account given under oath two and a half years later, it had looked like a clock. Shaughnessy, one of the consultants who advised Workman that morning, testified for Cutter’s defence in Oakland in December 1957 that the group had been given seventy-five minutes to reach its decision and had asked for at least another hour. “Mrs. Hobby was waiting on the other end of the line and wanted an answer by 4 o’clock,” he said. The committee had learned just before it met that the Foundation had tested the vaccine and found it satisfactory. “We had to decide largely on the basis of what had been given to us orally.”

What happened

The suspension lasted from 8 May to 13 May. Small amounts of vaccine were released through the summer under the amended requirements, but most of the school programmes that had begun in April were put off until the autumn, because health officers were no longer sure of the vaccine and were not willing to inject children as the polio season rose. Hammon, alone among the consultants, put on record that the new rules did not give an adequate margin of safety and that no more vaccine should be released until the production methods and the epidemiology had been studied further.

The new rules abandoned the line. A single-strain pool now had to give two consecutive negative tissue-culture tests, three days apart, on samples of at least 500 cubic centimetres each, whatever the size of the pool; one reinactivation was permitted after a positive test, and only one; the trivalent pool had to pass on 1,500 cubic centimetres; and the maker had to test the final bottles of every lot. The objective was stated as a number for the first time: no more than five live units per litre, which a lot passing all three tests would exceed “less than once in 100,000 times.” The protocols were to include every lot started. At the end of May a standing Technical Committee on Poliomyelitis Vaccine was created under Shannon, with Bodian, Francis, Salk, Richard Shope and Smadel; on 9 June the Laboratory of Biologics Control became a Division of Biologics Standards, with inspectors in the plants while vaccine was being made. Scheele’s letter sending the report to Hobby said what the Service thought had gone wrong: “Events which in the traditional course of scientific development would have covered years were telescoped into months and, as a result, both successes and failures have been magnified.”

What had gone wrong at Cutter in particular the report could not say. The on-site inspection had ended on 12 May; the laboratory tests on the Cutter lots were not finished; a separate report was promised and delivered on 25 August, and the New York Times printed its substance the next morning. The incident, the June report said, “may have been due to vaccine which, by deficiency in the tests themselves or by error in the performance of the tests, passed the established safety procedures and yet contained sufficiently large amounts of live virus to infect a small number of recipients. It may also have been due to some break in manufacturing which led to contamination after testing had been completed.”

The full count took eight years. Nathanson and Langmuir published it in three papers in the American Journal of Hygiene in 1963. Between 17 April and 30 June 1955, 260 cases of poliomyelitis had been reported in association with the roughly 400,000 inoculations of Cutter vaccine given between 13 and 27 April: 94 in the vaccinated, 126 in their families, 40 among other contacts; 192 of the 260 were paralytic. Sorted by the eight production pools the vaccine had come from, two of the pools, numbered 19468 and 19764 and used for about 120,000 inoculations, carried attack rates of 112 and 123 cases per 100,000 among vaccinees and their families; the other six ran from nine to 32, which was about what coincidence would have produced. Type 1 virus was eventually recovered from seven of the eight filling lots drawn from the two bad pools. In the routine safety tests run by Cutter and by the government before release, 74 monkeys had been injected with vaccine from those lots and none had been infected; it took 461 monkeys in eighteen laboratories, injected by every route and pretreated with cortisone and X-rays, to find virus in 46.

The cases that could be laid to the two pools came to 60 among the vaccinated, 51 of them paralysed, and 89 in their families, 74 paralysed. Five of the vaccinated children died and five of the family contacts. Among the children, 41 of the 50 injected limbs had been the first to go. Of the 70,000 schoolchildren who had received the two pools, roughly half had been susceptible to type 1 virus, and the small surveys that were done suggested that 10 to 25 percent of the vaccinated had become infected and were passing virus in their stools. The spread into the community, the authors judged, had been limited, “resulting in not more than 100 to 200 additional cases of poliomyelitis.” Between 1955 and 1961, under the amended requirements, more than 400 million doses of vaccine were given in the United States, and the surveillance that had been built in the last week of April 1955 found not one case of inoculation poliomyelitis among them.

After

Anne Elizabeth Gottsdanker was five years old in April 1955 and lived in Santa Barbara, where her father taught psychology. James Randall Phipps was a small boy in Monrovia. Each was given a Cutter injection from a sealed bottle a doctor had bought at a pharmacy, and each came down with poliomyelitis. Anne lost the use of one leg and part of the other and her spine curved. Their parents sued the Cutter Laboratories in the Superior Court of Alameda County, with Melvin Belli as their lawyer, for $385,000, and the case came to trial in Oakland late in November 1957 before Judge Thomas Ledwich. It lasted 28 days. Cutter’s defence was that its vaccine had passed every test the government required, and that if there had been live virus in it the fault lay with the testing procedures all five firms had followed; Shaughnessy’s seventy-five minutes were given in evidence for the defence.

The jury of eight women and four men deliberated for fifteen hours and came back on 17 January 1958, eleven to one, with a verdict for the children: $131,500 for Anne and $15,800 for James, and $8,300 to the parents in special damages. With it they handed up a statement they had written themselves. They had considered negligence first and had “from a preponderance of the evidence concluded that the defendant, Cutter Laboratories, was not negligent either directly or by inference.” But: “With regard to the law of warranty, however, we feel that we have no alternative but to conclude that Cutter Laboratories came to market … vaccine which when given to plaintiffs caused them to come down with poliomyelitis, thus resulting in a breach of warranty. For this cause alone we find in favor of plaintiffs.” One juror, Mrs Anita Steiner, told the reporters: “We felt obligated to follow the law, whether we agreed with it or not.” The judge said: “You can call it a directed verdict.” Belli called it the first verdict “which makes a manufacturer stand behind its product in the good old American manner.” Robert Cutter said the firm would appeal, and that it was “most appreciative that the jury took such an unusual action as to express their firm belief that there was no carelessness or incompetence on the part of Cutter Laboratories.”

The District Court of Appeal affirmed on 12 July 1960. California had held since 1939 that the maker of a food was liable to whoever ate it, without any contract between them, if the food was not wholesome. “The vaccine is intended for human consumption quite as much as is food,” Justice Draper wrote. Cutter had argued that its printed directions, which said the vaccine was “prepared in accordance with the requirements of the National Institutes of Health of the U. S. Public Health Service,” were an express warranty that displaced any implied one. The court noticed that the directions had left out one word: “The word ‘minimum’ which is used by the Public Health Service to modify the word ‘requirements’ in its own statement is significantly omitted.” To the argument that liability without fault would retard the development of medicines it answered that the warranty here was only “an assurance that the product will not actively cause the very disease it was designed to prevent,” and that if public policy wanted vaccines exempted, that was for the Legislature. By April 1962, by the firm’s own circular to physicians, 54 of 60 suits had been settled for more than three million dollars, a million more than its insurance. Bayer bought the company in 1974.

Hobby left office on 1 August 1955. Francis’s final report came out in 1957 and confirmed what he had said at Ann Arbor. Salk’s vaccine was given to tens of millions of children over the next six years and was then displaced by live vaccines taken by mouth; Meier, looking back in 1972, thought that “the momentum of acceptance had been broken” in May 1955. The salt-water children of 1954, the Foundation said that week, would be among the first to be offered the real thing.

What it taught

The experiment is remembered as a triumph with an accident attached. It is more useful read as two experiments that were run on different standards and joined at a signature.

The trial tested a product; the licence assumed a process. Francis’s placebo arm answered, to within a few percent and beyond reasonable argument, whether three injections of a particular vaccine, made by two firms under triple testing in the spring of 1954, prevented paralysis. The licence of 12 April 1955 extended that answer to six firms, to lots made a year later under a different preservative, and to a manufacturing process whose consistency nobody had measured, because “the field trial evidence itself was interpreted to indicate that the commercial product was safe.” The trial could say nothing about that; its controls had been injected with salt water, not with Cutter’s April lots. The number that proved the vaccine was read as if it also proved the factory, and the proof was so good that it stopped the question. The reindeer count in “Sixteen Reindeer” did the same thing in Alaska, and the income experiments in “Eighteen Hundred Boxes” were read for a marriage effect they had not been built to measure, and the reason is the same each time: a clear answer to the question you can measure is the most persuasive argument there is for not measuring the one you cannot.

A safety margin that lives in a theory is not in the vaccine. The whole safety case was a straight line extended into a region where nothing could be seen, to one live particle in a million tons. The tests were built on that premise: they were to catch gross accidents, because the process was taking care of the rest. When the line bent, and the makers’ own records showed it bending in a fifth of the pools at two firms, the test was all that stood between the line and the child, and it had been designed as a check on something else. The arithmetic the technical report did in May 1955 could have been done in 1953; it needed only the Poisson distribution and an honest statement of what concentration the test was meant to exclude. Nobody wrote that number down until after the children were paralysed. A test that has never been asked what it would miss is a ritual.

The regulator sees the lots it is shown. The protocols that the Laboratory of Biologics Control read through the winter described the lots the makers meant to sell. The lots that had failed, been reheated, passed on a second try and gone into trivalent pools, or been thrown away, were in the makers’ own files, and the requirements did not ask for them. Every firm had such lots. The government licensed consistency it had no way of seeing, on the strength of inspections of buildings and a record of 44 to 83 other products. The Minimum Requirements were, as the court later noticed, minimum, and the word was the first thing the directions on the box left out.

The instrument that worked was the oldest one. Nothing in the plants caught the two bad pools; the routine monkeys were clean and tissue culture found a single particle in a litre. What caught them was a hospital in Chicago, a health department in California, the interval from injection to paralysis, and which arm. Within two days of the first report the Public Health Service had an expectation to compare six cases against, and within ten it had a surveillance unit that could tell one maker’s lots from another’s by their attack rates. The five hallmarks Nathanson and Langmuir wrote down, an unseasonal outbreak, cases clustered in vaccinees and their households, rates that differ by lot, onsets four to fourteen days after injection, paralysis in the injected limb, are the instrument on which every killed-virus vaccine since has been watched. The 1976 programme described in “One in a Hundred Thousand” was stopped by the direct descendant of that unit, reading the same kind of count. The complication is the one that entry is honest about: the unit could count what the vaccine did only because the vaccine had been given to 400,000 children first.

Blame was found nowhere and liability everywhere. The jury found that Cutter had not been negligent, and the record supports them: its released lots passed every test on the books, and the books were the same for everyone. The technical report found faults common to the industry and could not say what was unique to Cutter. The Public Health Service took “part of the blame itself,” as the Washington Evening Star put it the day the report came out, and wrote new rules. The courts then held the company to a warranty it had never written, on the ground that the vaccine had caused the disease it was sold to prevent, and said that if the country wanted it otherwise the Legislature could say so. That is a coherent answer to the question of who pays. It is no answer at all to the question of how a lot that passes its tests can paralyse a child, which the new requirements answered instead, with a number: five particles per litre, missed less than once in a hundred thousand times.

Check the story before you learn from it. The vaccine was 80 to 90 percent effective at Ann Arbor, 60 to 90 in the Public Health Service’s chronology, and 60 to 80 in the observed areas. The trial had 1,829,916 children or “well over a million,” and “more than 450,000” or 422,743 of them vaccinated, depending on which grades and which areas are counted. Cutter’s vaccine went into 400,000 or 409,000 people. The cases among them number 60, 69 or 94, and the paralysed 51, 56 or 59, depending on whether one counts the two pools, the preliminary reports, or the whole incident; the dead are ten in the final count and five in the first; the ill with fever and headache are a third of one survey or forty thousand in later books. The jury gave $147,300, or $139,000 plus $8,300.

What the record supports without argument is the sequence. A trial designed so that its answer could be trusted gave its answer on a Tuesday morning; a process whose safety rested on a line nobody could see past was licensed that afternoon; thirteen days later the line turned out to have bent in two pools out of eight; and the thing that found it was not any test on the vaccine but the vaccine’s effect on the children who got it, counted by people who had built, in the same week, the instrument to count them with.

Sources

• Thomas Francis, Jr., Robert F. Korns, Robert B. Voight, Morton Boisen, Fay M. Hemphill, John A. Napier and Eva Tolchinsky, An Evaluation of the 1954 Poliomyelitis Vaccine Trials: Summary Report, Poliomyelitis Vaccine Evaluation Center, University of Michigan, 12 April 1955; printed as Part 2 of the American Journal of Public Health, vol. 45, no. 5, May 1955. The study population and its subdivisions, the case counts, the plan of study, the effectiveness estimates and the sentence “On this basis it may be suggested…” are quoted here from the summary as the Associated Press printed it verbatim on 12 April 1955 (see the Evening Star below) and as Meier tabulates it; the front matter, participant lists and appendix tables were read in the Global Health Chronicles copy. Global Health Chronicles, PMC: I. Plan of Study, PMC: II. Results, PMC: III. Summary of Estimates
• United States Public Health Service, Technical Report on Salk Poliomyelitis Vaccine, June 1955, with the Surgeon General’s letter of transmittal to the Secretary of 10 June 1955. Chapters 1 to 5 (the field-trial production problems, the licensing, release by protocol, the withdrawal of Cutter vaccine, the suspension and the plant inspections), chapters 6 to 10 (the process, Salk’s straight line and the “million tons”, the analysis of the safety tests, the manufacturers’ records as Laboratories A to E, the amended requirements), chapter 11 (the epidemiology to 27 May), Appendix A (the chronology from July 1953 to 27 May 1955) and Appendix D (the advisers). Read in the National Archives copy at Atlanta as scanned by the Global Health Chronicles; the figure is its Figure 1. Global Health Chronicles
• “Technical Report on Poliomyelitis Vaccine,” Public Health Reports 70, no. 8, August 1955, pp. 738 to 751: the summary, the epidemiological chapter and the calendar of the technical report as the Public Health Service reprinted them. CDC Stacks. stacks.cdc.gov
• Neal Nathanson and Alexander D. Langmuir, “The Cutter Incident: Poliomyelitis Following Formaldehyde-Inactivated Poliovirus Vaccination in the United States during the Spring of 1955,” I. Background, II. Relationship of Poliomyelitis to Cutter Vaccine, III. Comparison of the Clinical Character of Vaccinated and Contact Cases, American Journal of Hygiene 78, no. 1, July 1963, pp. 16 to 81. The first cases, the chronology, the 260 cases, the eight production pools and the two high-rate pools, the virus isolations, the attack rates in households, the Baltimore block, Idaho, the incubation periods, the injected limb, the ten deaths, the 400 million doses, and the settlements of 1962. Read in the David J. Sencer CDC Museum’s copy. Global Health Chronicles
• Alexander D. Langmuir, Neal Nathanson and William Jackson Hall, “The Surveillance of Poliomyelitis in the United States in 1955,” paper for the American Public Health Association, Kansas City, 15 November 1955 (published in the American Journal of Public Health 46, 1956, pp. 75 to 88), and Poliomyelitis Surveillance Reports nos. 30 and 31 of 17 and 21 June 1955. The surveillance unit’s organisation, the Idaho and Arizona reports and Gebhardt’s isolation. CDC Stacks. paper, report 30, report 31
• Gottsdanker v. Cutter Laboratories, 182 Cal.App.2d 602, 6 Cal.Rptr. 320, District Court of Appeal of California, First District, Division Two, 12 July 1960, Draper, J. The verdicts, the jury’s statement, the food analogy, the omitted word “minimum” and the public-policy argument. biotech.law.lsu.edu
• Evening Star (Washington), 12 April 1955, pp. A-1 to A-4 (Alton Blakeslee’s Associated Press report from Ann Arbor, the text of the summary report, the press-room scene, the NBC embargo, Salk by telephone, Dr Harold Kennedy of Fairfax); 13 April 1955, pp. A-1 and A-6 (the licensing and the first shipments); 28 April, p. A-1 (the withdrawal, the eight children); 30 April, p. A-1 (the conference of experts, 300,000 Cutter vaccinations); 7 and 8 May, p. A-1 (Scheele’s announcement at 4 a.m., “a lovely day”); 13 May, p. A-1 (Scheele before the House committee); 16 May, pp. A-1 and A-6 (Hobby’s report to the President); 10 June, p. A-1 (the technical report released); 13 July, p. A-1 (Hobby’s resignation); 12 December 1957, p. A-10 (Shaughnessy’s testimony); 18 January 1958, p. A-2, and 19 January 1958, p. A-3 (the verdict, the jurors, the judge, Belli and Cutter). Library of Congress, Chronicling America, read through loc.gov’s full-text service. 12 April p. 1, 12 April p. 4, 28 April, 7 May, 13 May, 10 June, 12 December 1957, 18 January 1958
• Dwight D. Eisenhower, The President’s News Conference, 18 May 1955, 10:32 a.m., Executive Office Building. Arrowsmith’s question on Senator Morse, Donovan’s on Hobby leaving, and Emory’s on “no one could have foreseen the public demand.” The American Presidency Project. presidency.ucsb.edu
• Paul Meier, “The Biggest Public Health Experiment Ever: The 1954 Field Trial of the Salk Poliomyelitis Vaccine,” in Judith M. Tanur and others, eds., Statistics: A Guide to the Unknown, 1972. The observed-control and placebo-control plans, the health departments’ stand, the volunteer effect, Table 1 adapted from the summary report, and the epilogue. webpages.math.luc.edu
• Paul Meier, “Safety Testing of Poliomyelitis Vaccine,” Science 125, 31 May 1957, pp. 1067 to 1071, and Jerome Cornfield, Max Halperin and Felix Moore, “Some Statistical Aspects of Safety Testing the Salk Poliomyelitis Vaccine,” Public Health Reports 71, 1956, pp. 1045 to 1056: the later statistical treatments of the sampling problem, cited by Nathanson and Langmuir; not read.
• Report on Poliomyelitis Vaccine Produced by the Cutter Laboratories, Public Health Service, 25 August 1955; not reached (the University of Michigan’s copy is behind a verification page). Its findings are known here as the technical report promised them and as Nathanson and Langmuir summarise them.
• Bayer AG, company history: the purchase of Cutter Laboratories in 1974. bayer.com