"...deduced, correctly, that the Brazilians - perhaps via their Germanophile Foreign Minister Lauro Muller - had managed to find an agent at the Signaldienst telegraph stations in either Aruba, Costa Rica, or both. This was a coup, to be sure, but not quite as dangerous as it could have been. For one, by the last months of 1915, there was not all too much that the United States was transmitting that could have affected Brazil, what with the Caribbean effectively an American lake and only a few scattered companies of trainers and volunteers on the ground in Argentina. Indeed, the American position in the Caribbean was strong enough that a debate had begun within the Navy of how to restore cables to Haiti and Nicaragua safely and effectively in order to stop relying on the Germans, who Yardley had always known as an open secret were "reading the mail."
Before that, though, Yardley's small tests of Brazilian ships in the water, with the coordination with the Naval Intelligence Office, did prove that somebody in either Aruba or Costa Rica (it was hard to determine which, exactly, because it was still hard to send messages to one without the either) was passing information on American ship movements off the coast of Guyana to the Brazilians to help their patrols near the mouth of the Amazon avoid the Americans. Despite the fact that Stimson strongly disliked Yardley - the cryptologist's practice was, to an old-fashioned brahmin like the Secretary of War, ungentlemanly and "Y" was never the easiest personality to manage as a superior anyways - he set his dislike for the man aside when presented with a novel idea that, perhaps, helped speed up the end of the war in South America and cemented a place for the Cypher Bureau, and the Office of Strategic Services thereafter, for good. [1]
What Yardley proposed was classic ruse de guerre. The Navy and Army would, using the Aruba transfer station, begin sending signals to Nicaragua to prepare for a major "Mesopotamian Expeditionary Force" to be supplied and staffed through the Canal, in order to place over a hundred thousand American soldiers at Mar del Plata and Buenos Aires in tandem with an expansion of the Naval squadron off the Argentinean coast to begin a general offensive to liberate not only the Mesopotamian Argentina but push into Brazil and perhaps seize Uruguay as well. The messages were coded, but weakly, making them just lucrative enough to steal and pass along to Brazil but also easy enough for Brazilians to decipher.
They were also absolute fiction. Even as Mexico called for a ceasefire and prepared to exit the war, the United States was exhausted after a long, brutal year of fighting and had as its strategic objectives by year end little more than securing a foothold north of the Rappahannock, taking Knoxville and Huntsville and consolidating its position in West Texas. There were plans to put men in Argentina sometime in 1916, but those plans kept getting pushed further and further back, and War Department estimates suggested that the earliest a field force of sufficient size, strength and morale could be sent to the River Plate in the kinds of numbers needed to actually be useful was probably May or June. Brazil, however, did not know this, but they did know that with the Canal clear for use and both Chile and Mexico exiting the war, the linkages of the Bloc Sud were now very weak, and the idea that eight or more divisions of American men might be arriving as soon as late December or early January did what it was intended to - it spooked military leadership, especially once rumors of an imminent direct American intervention began trickling down to junior officers and enlisted men.
These false deployment plans are a large part of what eventually, in early 1916, persuaded Brazil to seek peace with Argentina and the United States, and likely thus saved tens of thousands of lives on all sides from the meatgrinder on the Parana River. Yardley's work was validated, even begrudgingly by many, and the Cipher Bureau was, surprisingly to him, kept open even after the war as signals intelligence revealed yet again its immense value on the modern battlefield..."
- Shadow Wars: A History of Espionage and Counterintelligence
[1] This is largely true - Stimson was ambivalent of the somewhat-still unproven signals intelligence bureaus generally, but especially not in peacetime - the quote of "Gentlemen do not read each others' mail!" is attributed to him. This wasn't necessarily just a Stimson thing, either - the British rolled up MI-8 in the late 1920s, as well.