There is nothing inherently wrong with beginning a sentence with any conjunction, including but. Some stylists have deprecated it, and some English teachers still enforce it; but it has never been anything more than a house rule. It has been continually ‘violated’ by writers—including many of the Best Writers—in every register and every age of English literature.
I’ve never looked to see where and when this “rule” arose, but the history of similar ‘superstitions’ (the characterization is Fowler’s) suggest that it surfaced in the 18th century as a perfectly reasonable recommendation to use the device sparingly, and hardened into a doctrine among 19th century schoolmasters and schoolmarms.(see below)
It appears that educators use this “rule” to stamp out two common tendencies among young writers: 1) sustaining the flow of narrative and exposition by beginning virtually every sentence with and or but or so, and 2) making this a license to isolate sentence fragments as full sentences.
It’s basically what I call a ‘baby rule’, forbidding a given practice until the practitioner is experienced enough to know when it’s safe to use it. I don’t think there’s any need for it around here; even our rank beginners are grown-ups, too sophisticated in their own languages to need that sort of restriction.
(But if you want something real to grouse about, I suggest you take aim at the absurd practice, now epidemic in business writing, of following conjunctions with commas.)
Catija asks for examples from the Best Writers. Here are three, from the past three centuries, by authors who cannot conceivably be accused of innovative barbarism or vulgar colloquiality:
… The western isle might be improved into a valuable possession, and the Britons would wear their chains with the less reluctance, if the prospect and example of freedom were on every side removed from before their eyes.
But the superior merit of Agricola soon occasioned his removal from the government of Britain … —Edward Gibbon, The Decline and Fall of the Roman Empire
Everybody, too, would be willing to admit, as a general proposition, that the critical faculty is lower than the inventive. But is it true that criticism is really, in itself, a baneful and injurious employment ; is it true that all time given to writing critiques on the works of others would be much better employed if it were given to original composition, of whatever kind this may be?
—Matthew Arnold, The Function of Criticism at the Present Time
Significant though that breakthrough was, however, and though he passed another civil rights bill in 1960, liberal antagonism toward him had softened scarcely at all since the bills were weak, only meagre advances toward social justice, and because his championing of them was regarded by most liberals as mere political opportunism: an attempt to lessen northern opposition to his presidential candidacy.
But although the cliché says that power always corrupts, what is seldom said, but what is equally true, is that power always reveals.
—Robert Caro, The Passage of Power
My conjecture that the prohibition of sentence-initial coordinating conjunctions would be found to have originated among 18th-century normative grammarians appears to be false. Goold Brown (The Grammar of English Grammars, 1851), who takes great pleasure in discovering and exploring “disputed points”, does not mention this matter, but states flatly that “The period is often employed between two sentences which have a general connexion, expressed by a personal pronoun, a conjunction, or a conjunctive adverb [emphasis mine].”
The “rule” does arise in educational circles. George Payn Quackenbos, in Advanced Course of Composition and Rhetoric, 1857, 88, is the first author I find taking issue with the practice:
[I]t is not proper to place a period immediately before a conjunction which closely connects what follows with what precedes. This is frequently done in the translation of the Scriptures, where we have verse after verse commencing with and; but it is not authorized by good modern usage. In such cases, either the passage so introduced ought to form part of the preceding sentence, and be separated from it only by a colon or semicolon; or else, if this is impracticable on account of the great length or intricacy it would involve, the following sentence should be remodelled in such a way as to commence with some other word. These remarks apply to all conjunctions that form a decided connection between the parts; such as merely signify to continue the narrative, and imply no connection with what precedes, may without impropriety introduce a new sentence.
As the substance of the preceding paragraph, we may lay down the following general rule, remembering that there are occasional exceptions:—A sentence should not commence with the conjunctions and, for, or however; but may do so with but, now, and moreover.
M. Barrett’s remarks on Misuse of the Word “And” in The American Educational Monthly for April, 1870, 159-60, elevates Quackenbos’ measured deprecation of sentence-initial and to a frank prohibition [the spelling is his]:
”And” is a conjunction, whose office is explained in the etymology of the word “conjunction”, i.e. to join together. Says Webster, “it signifies that a word or part of a sentence is to be added to what preceeds.”
The period indicates a completion, and is used to separate that which preceeds from what follows. It is properly called a full stop.
Then, it is evidently improper, inconsistent, and contradictory to commence a sentence after a period with the conjunction and. The one tells us to divide, the other tells us to unite, and both at the same time.
This is answered in the May issue, 204-5 by ‘S.W.W.’, who points out that the argument ‘proves too much’, is ‘sophistical’, and is ‘opposed to the practice of scholarly writers’ such Steele, Addison, Junius, and Macauley, and concludes
That the conjunction and is often improperly used, not only at the commencement of sentences but elsewhere, we admit. But the idea that a sentence should never begin with it is absurd. It would be quite as sensible and worthy of consideration to insist that a sentence should never begin with but or nor.
(You can read Mr. Barrett’s heated rejoinder in the July issue, 289-93.)
By 1885 both the “rule” (at least with respect to and) and the resistance seem to have reached Scotland, where a schoolgirl in Louisa M. Gray’s Mine Own People) complains
”My essay was best again,” said Gretchen. “Well, you and I, Anna, who know what a poor piece of composition it was, may guess what the others were like! But would you believe it?—Mr. Anderson objected to one of my sentences, because it began with ‘And!’ Now, wasn’t it ridiculous? Anna, isn’t it perfectly allowable to begin a sentence with ‘And ?’ Don’t the very best authors do it?”
I’m on Gretchen’s side.