The first purpose-built aircraft carrier—HMS Hermes, commissioned in 1924—was a modest vessel of 10,850 tons carrying approximately 20 aircraft. The same year, Washington Naval Treaty tonnage limitations forced the United States and Japan to convert two battlecruiser hulls into carriers (USS Lexington and Saratoga, IJN Akagi and Kaga). These converted ships—33,000–36,000 tons, capable of operating 90+ aircraft—demonstrated for the first time that carrier air power might have strategic reach. The battleship admirals dismissed the possibility.
The Doctrinal Development
The Imperial Japanese Navy and the United States Navy developed carrier doctrine independently during the 1930s and arrived at similar conclusions: carrier aircraft could strike an enemy fleet before the enemy's guns could engage—making standoff range the carrier's decisive tactical advantage over the gun-armed battleship. The IJN's simulated carrier strike against Pearl Harbor in 1941 war games had concluded it was feasible; the actual attack revealed that risk estimates had been too conservative.
The Pacific War Confirmation
Midway proved the thesis: four Japanese fleet carriers were destroyed by carrier-based dive bombers before a single surface ship came within gun range of another. Pearl Harbor was carried out by carrier aircraft without the battle fleet ever leaving Japanese waters. The battleship's role contracted to shore bombardment; the carrier defined naval power for the rest of the century. American shipyards ultimately built 24 Essex-class fleet carriers during the war.
The Contemporary Carrier
The Nimitz-class carriers (1975–2009) and Gerald R. Ford class (2017–present) represent the apex of carrier development: 100,000 tons, 75–90 aircraft, nuclear propulsion, with carrier strike groups that constitute self-contained offensive military power. Their vulnerability to anti-ship ballistic missiles—developed specifically to hold carriers at risk—is the modern carrier debate. No carrier has been seriously threatened in combat since World War II, leaving the vulnerability question empirically unresolved.