"Many thanks to old Cadmus, who made us his debtors,
by inventing one day the capital letters."
—from an old rhyme, according to Oscar Ogg
When I decided to write about acronyms, it occurred to me to wonder where capital letters came from—or more precisely, where and when the idea was formed of having two sets of letters, one set for ordinary words and the other set for special situations like the starts of sentences. The search led me through several fascinating volumes on the history of the alphabet and of writing. Conclusive statements on the subject were hard to come by, but here's how I understand it:
The uppercase letters we use in English today (the Roman alphabet) were inherited from (surprise!) the Romans, though we've added a few to the original set. The Romans derived their letters from the Greek alphabet (possibly via the Etruscans); the Greeks got it from the Phoenicians. Both the Greeks and the Romans apparently had a set of less formal letters used for inscriptions on wax or papyrus, supplementing the letters used in stone carvings; however, no sources I've found indicate that the two sets of letters were mixed the way that we mix cases today.
Hand-written (not carved) Roman letters later developed a more rounded form known as "uncials" (which word derives either from uncialibus litteris, inch-high letters, or from uncus, crooked). As they spread across Europe, uncials became smaller and rounder, a form known as half-uncials or semiuncials. Irish monks wrote in a particularly nice semiuncial script, which a scholar from York named Alcuin learned a derivation of. Around 780, Alcuin found himself under Charlemagne's patronage, where he proceeded to create a script now known as Caroline (or Carolingian) which mixed majuscules and minuscules in much the way modern writing does, using majuscules at the beginnings of sentences; hence the term capital, from Latin caput or capitis, "head." The Carolingian minuscules looked nice and were designed for clarity; everyone else wanted in on the action, so the lowercase letters spread.
(Side note on terminology: "majuscule" and "minuscule" (note the correct spelling) refer to the size of the letters (big and small); this terminology can cause confusion when the same letter comes in varying sizes. I attempted to read the Dr. Seuss ABC book to a young friend some months back; he correctly identified a "big G," but when I asked him to point to a "little G," he pointed to a smaller capital G... "Capital" implies headings (and has no good opposite), and "uppercase" and "lowercase" are former obsolecisms, referring as they once did to the cases of type used by printers in the days before computer typesetting. Faced with this lack of terms which accurately describe the differences between the two alphabets, I can only shrug and use all of the terms interchangeably.)
Besides headings and the heads of sentences, capital letters are also used in acronyms—makes sense, what with acronyms being formed from the heads of words. Technically speaking an acronym is not just an abbreviation formed from initial letters, but a word formed from initial letters, pronounced as if it were a real word; AIDS is an acronym, but IBM (because it's pronounced /aI bi Em/ rather than /'Ib @m/) is a mere abbreviation, sometimes known as an initialism.
The distinction between acronyms and abbreviations, though, is not as clear-cut as one might wish. Many abbreviations can either be pronounced as words or spelled out: AWOL can be spelled out or pronounced /'eI wAl/; VRML can be spelled out or pronounced /'vR mL/. I tend to ignore the distinction, and just use the term "acronym" loosely to refer to any term spelled in capital letters taken from the beginnings of words.
In the world of computers, capital letters have really come into their own, making the acronym distinction even fuzzier. From the early days of ENIAC ("Electrical Numerical Integrator And Calculator"), we progressed to UNIVAC ("UNIVersal Automatic Computer," more concerned with an interesting-sounding name than with taking only the first letter of each word), and subsequently into dozens of capitalized or partly capitalized words with little regard to whether they're made up of bits of other words or not. In the realm of computer languages, for instance, COBOL is for COmmon Business-Oriented Language; FORTRAN for FORmula TRANslation; and PASCAL shouldn't be capitalized at all (and, to be fair, often isn't), seeing as how it stands for nothing but the name of French mathematician Blaise Pascal. Then again, COBOL is sometimes written Cobol, as though it weren't an acronym. More recently, companies (especially computer companies) have taken to running words and bits of words together but only capitalizing the first letter in each piece. You can make up examples by stringing together techie-sounding syllables: CompuTech, InterObSoft, TeleSysComCorp. The technical term for intercalary capitals is BiCapitalization, but I prefer Mykle Hansen's term StudlyCaps.
By the way, any time you want to find expansions of acronyms and abbreviations on the Web, you might try the acronym and abbreviation server, Babel (computer acronyms), or the Jargon file (aka The New Hacker's Dictionary).
David Diringer, The Alphabet, vol. 1, Funk & Wagnalls, NY, 1968.
Albertine Gaur, A History of Writing, Charles Scribner's Sons, NY, 1984.
Alexander & Nicholas Humez, ABC Et Cetera, David R. Godine, Publisher, Boston, 1985.
Georges Jean (tr. Jenny Oates), Writing: the Story of Alphabets and Scripts, Thames and Hudson Ltd, London, 1994.
Oscar Ogg, The 26 Letters, The Thomas Y. Cromwell Company, 1961.
For further reading, check the 411 section in any Dewey-Decimal-organized library.