The Boston experiments had one obvious weakness, and everyone involved could see it from the start: the volunteers had all been living inside the epidemic. The second series, three thousand miles away, was supposed to be the answer to that. It was run at the same time, on men who had been locked away from the disease.

Who ran it

Surgeon George W. McCoy of the Public Health Service was the director of the Hygienic Laboratory in Washington, the federal research laboratory that would later become the National Institutes of Health. He had held the post since 1915. Rosenau, running the Boston half, had held it from 1899 to 1909. The laboratory’s past director and its present one were working the two ends of one question in the same weeks.

With him was Lieutenant De Wayne Richey of the Navy Medical Corps. They worked at the United States Quarantine Station on Angel Island in San Francisco Bay, with laboratory help from Karl F. Meyer of the Hooper Foundation at the University of California, and took their donors from the San Francisco Hospital through Dr R. G. Broderick.

Rosenau, describing their work from a podium in 1919, called the place Goat Island. Goat Island is the old name for Yerba Buena, where the volunteers came from; the station was on Angel Island. It is a small slip and it has been repeated ever since.

The men

Fifty volunteers came from the enlisted personnel of the United States Naval Training Station at Yerba Buena. The report sets out what made them different from the Boston men in four sentences:

They had been under quarantine for a month. At no time had influenza occurred in the station. With 4 exceptions, all of the 50 men had experienced one or more of the exanthemata during childhood. Only 5 of them … gave a history of a possible influenzal attack before 1918.

One man, number 41, said he had been struck down in October 1918 by a severe head cold from which he recovered completely; close questioning “failed to reveal probability of influenza.” They were aged 18 to 23, in good physical condition, with white cell counts between 6,800 and 13,400. Forty-seven of the fifty gave a negative Wassermann reaction and the other three were doubtful.

So here, at last, was a group of susceptible men: young, healthy, with no history of the disease, kept behind a quarantine line for a month while the pandemic went past outside. That was the point of them.

And then there is the paragraph that follows, which the report prints without any apparent sense that it is dismantling the experiment it introduces:

During the second week in October, 1918, the entire personnel of the station, including the volunteers received a vaccine subcutaneously of which 1 c. c. contained — B. influenzae, 5,000,000,000; Pneumococcus Type I, 3,000,000,000; Pneumococcus Type II, 3,000,000,000; Pneumococcus Type III, 1,000,000,000; Streptococcus haemolyticus, 100,000,000.

Three doses, forty-eight hours apart. Every volunteer in the San Francisco series had, a month before the experiment, been injected with five billion Pfeiffer’s bacilli per dose, three times over — the very organism the experiment was about to test as the cause of influenza. The reactions, the report says, “were very slight or were absent.”

The experimenters also surveyed what was already living in the volunteers’ noses and throats before they began: Pfeiffer’s bacillus in about a quarter of them, pneumococci in 78 per cent, haemolytic streptococci in 70 per cent, a gram-negative diplococcus in all but two. Whatever else these men were, they were not sterile.

The design

The volunteers were sent across to Angel Island ten at a time and split into groups of five in separate quarters. Within each group of five, four men were inoculated and the fifth was left alone, living with them, as a contact control. That is a feature the Boston series never had: in every San Francisco experiment there is a man in the room who got nothing, to show whether whatever happened to the others could spread.

The material was collected in the city and carried across the bay, kept warm in transit. The interval between taking it from a patient and putting it into a volunteer ran from two to six hours, and the report says plainly what it thought of that: “Care was taken to control every step and it is to be regretted that the time interval between donors and volunteers, which varied from two to six hours, could not, under the circumstances be shortened.”

Each dose was divided: half the material left raw, half drawn through a Berkefeld filter, so that the two could be given to different men. Cultures were taken of both. The filtrates, in every experiment, grew nothing at all; the raw material grew Pfeiffer’s bacillus, haemolytic streptococci, pneumococci and a gram-negative diplococcus. That pairing — a sterile filtrate and a crowded raw suspension — is how the question of whether the agent passed a filter was supposed to be settled.

The donors

The Boston donors were sailors. The San Francisco donors were whoever the city hospital had.

Donor I was a girl of nine, ill for twenty-six hours, admitted after contact with influenza patients at home, with a temperature between 101 and 103.2 degrees Fahrenheit and a pulse that ran to 140.

Donor II was an infant of one year, swabbed twenty-four hours after the onset of his illness. His mother and sister were in the hospital with influenza when he was admitted. The report describes his illness as an uncomplicated attack.

Donor III was a hospital apprentice of twenty-two who had been given the same vaccine as the volunteers in October, and who was taken suddenly ill “after shaving several influenzal patients a day or so previously,” with headache, backache, chills, nosebleed and a horror of light.

Donor IV was a nurse of twenty-one, ill for twelve hours when her secretions were taken — the freshest material in either series that autumn.

Donor V was a nurse of twenty-seven, ill for twenty-four hours when her blood was drawn. Three days later she developed bronchopneumonia, and recovered.

What happened

Eight experiments were done between 11 November and 3 December 1918.

Thirteen men had filtered nasopharyngeal secretions sprayed or dropped into the nose. Thirteen had the unfiltered material. Ten men lived alongside them as contact controls. Two had filtrate put into the conjunctival sac of each eye with a dropper and sprayed in with an atomiser. One had two cubic centimetres of filtrate injected under the skin of his left arm. One had five cubic centimetres of fresh citrated blood from an influenza patient injected under the skin. Four were given a pooled suspension of eight living strains of Pfeiffer’s bacillus into the nose and throat, and four more the filtrate of the same suspension.

The summary is one sentence long: “In no instance was a clinical case of influenza produced.”

What did happen was tonsillitis. Three of the men who received the unfiltered secretions came down with acute lacunar tonsillitis — the crypts of the tonsils filling with a creamy exudate, temperatures to 103 degrees, prostration slight, recovery in four days. In Experiment IV, numbers 31 and 32 both developed it within thirty-six hours, and the organism cultured from their tonsils was “an apparently pure culture of the same type of haemolytic streptococcus as was encountered in the control cultures of the donor’s unfiltered secretions.” The report’s own verdict: “There was no reason for believing that these were attacks of influenza.”

In Experiment II, number 11 developed it fifty hours after his inoculation, and the same streptococcus came out of his tonsils as had gone into them. He had arrived at the station with markedly enlarged tonsils and a history of similar attacks.

And in every case the contact controls — the fifth man in each group of five, living and eating alongside — stayed well. The report notes it after Experiment IV in a sentence that is quietly devastating about what the experiment was able to demonstrate: “The other members of the group failed to contract the disease, despite the fact that they were in constant association with the affected volunteers.”

What it was worth

The San Francisco series mattered to the Boston series immediately, and Rosenau’s team used it in their discussion the moment they had to explain themselves:

It is possible that all our volunteers resisted infection because of a natural or an acquired immunity. If this be true, then we have an indication of a much higher degree of immunity to this disease than is generally assumed. The fact that our colleagues in the San Francisco studies … failed to reproduce the disease in volunteers who had not been exposed in the recent pandemic suggests that the immunity of our volunteers was at least not the sole controlling factor.

That is a fair argument, and it was the best one available in 1919. It is also weaker than it looks. The San Francisco men were susceptible by history, but the number of them actually inoculated with patients’ material was twenty-six, not fifty; their material sat for two to six hours in transit, which is longer than most of the Boston material; their donors were a child, an infant, two nurses and an apprentice rather than a ward full of sailors; and all fifty had been given three doses of a vaccine containing five billion Pfeiffer’s bacilli, which means that if the bacillus had been the cause, the one group assembled to test it had been vaccinated against it a month before.

Rosenau’s own summary of the pairing, to the American Medical Association in June 1919, was shorter and more honest than any of this: “Perhaps there are factors, or a factor, in the transmission of influenza that we do not know.”