The Beech-Nut Packing Company sold bacon, and Americans in the 1920s mostly ate a small breakfast — coffee, juice, a roll.
Bernays did not commission a bacon advertisement. He asked his agency's physician a question: is a heavy breakfast better for health than a light one?
The doctor thought it probably was, on the reasoning that the body loses energy overnight. Bernays then asked him to write to a large number of physicians — around five thousand — putting the question, and to report the results.
Somewhere over four thousand agreed that a hearty breakfast was preferable. The finding was released to newspapers as a health story: four thousand five hundred physicians urge a bigger breakfast, with — in the coverage Bernays arranged — bacon and eggs mentioned as an example.
Bacon sales rose. The American breakfast has looked like that ever since.
Three mechanisms are in play and they are stacked deliberately.
Authority laundering. Bernays never made a health claim. He manufactured a third party — the medical profession — whose credibility carried it, which is chapter 7's mechanism executed at scale and with no falsehood anywhere in the chain. The doctors were real, the question was really asked, and the answers were really given.
Evasion of reactance. Brehm's theory holds that a perceived attempt to restrict or direct behavior produces a motivational push in the opposite direction, and advertising is recognized as such an attempt and discounted accordingly. A news story about medical opinion is not, so it is processed by the peripheral route without the resistance an advertisement triggers.
And repetition into consensus. A single survey reported in hundreds of newspapers becomes, for a reader who encounters it in three of them, a settled fact — chapter 25's illusory truth effect, produced by syndication rather than by argument.
The survey itself would not survive a moment's methodological scrutiny. A question asked of doctors, with no controls, no definitions and no data, is an opinion poll of a profession, and physicians in 1925 had no evidence about breakfast composition. The conclusion was not established; it was elicited.
What makes the case worth studying is that it is entirely honest at every individual step and entirely misleading in aggregate. Nobody lied. A norm was installed in a population by arranging the sequence in which true things were said.
Cross-ref: Attack — Speak With Borrowed Authority