[begin page 665]

When Samuel Clemens turned forty at the end of 1875, he found himself a long way from the little river town in Missouri where he had grown up in genteel poverty. He was now a prosperous citizen of Hartford, Connecticut, and a literary celebrity known to the world as “Mark Twain,” the author of The Innocents Abroad and Roughing It. He had an open invitation to write for the Atlantic Monthly, edited by his friend William Dean Howells,1 and he was well on his way to having “the largest audience of any English writer above ground,” as Richard Watson Gilder put it nine years later.2

Clemens and his wife, Olivia (who celebrated her thirtieth birthday three days before he celebrated his fortieth), had recently built an elaborate brick mansion in Hartford,3 where they lived in modern splendor with their two young daughters, Susie and Clara, and a bevy of household servants.4

[begin page 666] Despite this profound transformation of his personal circumstances, Clemens still had vivid memories of, and even intermittent correspondence with, the people in Hannibal, where he had lived from age four to age seventeen.5 He also had vivid memories of the Mississippi itself, and of the river-men in whose company he had spent more than four years (1857–61) becoming “a good average St. Louis and New Orleans pilot.”6

As recently as May 1875 Clemens had finished a rambling series of articles for the Atlantic called “Old Times on the Mississippi,” which drew extensively on his memories of those piloting days. And in July of that year he finished his first draft of The Adventures of Tom Sawyer, begun sometime in 1872, which drew even more extensively on memories of his Hannibal childhood.7 On 5 July he mentioned completing this draft in a letter to Howells:

In July 1876, exactly one year after he thought of telling such a tale “in the first person,” and even as he was reading proof for the first chapters of Tom Sawyer, Clemens began work on the sequel. As predicted, he decided to tell it in the first person—and not in Tom’s voice but in the voice of Tom’s friend, Huckleberry Finn, described in chapter 6 of Tom Sawyer as the “idle, and lawless, and vulgar,” young outcast who horrified all the respectable mothers of St. Petersburg.9 More than seven years would pass [begin page 667] before he submitted the finished copy of this novel to his publisher. During those years he twice stopped work on it, for periods of four and three years, respectively; he completed and published three other full-length books;10 he welcomed a third child, Jean, into the family; and he twice changed publishers, in the end creating his own firm, Charles L. Webster and Company.

The typewritten copy of his book, which Clemens thoroughly revised before submitting it to that publisher, is lost. But the complete holograph manuscript from which the typescript was made is available for study—its first half, long assumed lost, having recently been reunited with the second half. Study of this manuscript, upon which the present edition is based, reveals how the text was altered before publication, by the author himself—often in previously unsuspected ways—and by his typists, typesetters, proofreaders, and publisher, sometimes with (but more often without) his active agreement. It also reveals, in greater detail than ever before, exactly when, and by what stages, Clemens composed and then meticulously revised his masterpiece.

From the moment it was published, Huckleberry Finn prompted questions about how and when Clemens wrote it. Brander Matthews and Jeannette Gilder were only the most prominent of the author’s contemporaries to ask him about these matters directly. Interviews with Clemens, and [begin page 668] statements by him in letters and other documents, are important evidence for determining the how and when. But it has been the gradual emergence and critical interpretation of physical evidence—notes, proofs, and manuscripts—which have proved decisive in the effort to understand the timing and the stages of composition for Huckleberry Finn.

Clemens’s biographer, Albert Bigelow Paine, who had access both to the author himself and to his private papers, published the first, brief chronological account of the book’s composition in 1912. Between 1912 and 1938, only a handful of critics wrote anything about the making of Huckleberry Finn, and of these, only one made use of the known manuscript, which was then still missing its first half.11 When DeVoto succeeded Paine as editor of the Mark Twain Papers in 1938, he welcomed the opportunity to examine, for the first time, the documents he hoped would permit him to speak with real authority about how Huckleberry Finn was written. In 1942 he published the result of his investigation, first as an introduction to a text of the novel, then in Mark Twain at Work. DeVoto’s remarkably detailed effort to describe and date the stages of composition was revolutionary, and was welcomed as “indispensable to scholars.”12 But over the years it was also gradually amended by those who reinterpreted the documents he had, and took advantage of evidence he did not.

Clemens was not especially forthcoming in his letters or notebooks [begin page 669] about the course of composition, and much of what he did say is ambiguous. Little additional evidence of this kind has been found since Paine wrote that Clemens composed Huckleberry Finn in three stages or stints: (a) “about half” in the summer of 1876; (b) an unspecified amount in the summer of 1880 (“writing alternately on The Prince and the Pauper and on the story about Huck Finn”); and (c) all the rest in 1883.13 Paine cited only two letters as evidence to support his dating of the composition. His manifest confidence in his report suggests that he had another primary source: Clemens himself. In any case, his very general description would ultimately be proved right, but only when independent documentary evidence was found to confirm it.

DeVoto, who added greatly to our detailed picture of how the text evolved (he was the first to appreciate the “working notes,” which he found in the Mark Twain Papers), rejected Paine’s middle stage because he said it lacked persuasive documentation.14 He opted instead for only two stages: (a) chapters 1 through 16 (except for chapters 12–14) up to “where Huck’s raft is rammed by a steamboat,” written in July and early August of 1876; and (b) all the rest of the book—chapters 12–14 and all of chapters 17–end—written in the summer of 1883. DeVoto was also the first to realize that a typescript (which he never saw, but which he understood was intended for the typesetter) must have been made of the first part of the book.15

Sixteen years later, Walter Blair (working, like DeVoto, from only the 1883 half of the manuscript) reanalyzed the physical evidence provided by the working notes, comparing their ink and paper to those which Clemens used in letters and other manuscripts found in widely scattered archives. He suggested that Huckleberry Finn had been written not in two stints, but most likely in four: (a) the first sixteen chapters (absent 12–14) in the summer of 1876, as DeVoto had first said; (b) the last three paragraphs of chapter 16 plus chapters 17 and 18 in 1879–80; (c) chapters 19 through 21 sometime between 1880 and 1883; and finally (d) chapters 12–14 and 22–end during the summer of 1883. Blair further argued that there had been two partial typescripts (both lost) made for the printer from the holograph manuscript. The first “consisted of that part of the book written before” the summer of 1883, the holograph for which he thought had been [begin page 670] “destroyed.” The second was of that part written during the summer of 1883, and “the autograph version of this was preserved . . . in the Buffalo Public Library.” Blair added that his own redating and refinement of the course of composition made it clear to him that both Paine and DeVoto, along “with most other students of Twain, overestimate the influence of Clemens’s Mississippi River trip of 1882 on the novel.”16

Thirty years later, Blair and Fischer, for the Iowa-California scholarly edition of the novel, had access to more thorough collations of the manuscript and first edition than had ever existed before, and they also had access to documentary materials donated in 1977 to the Vassar College Library.17 They used both to refine the dating of the four stints originally posited by Blair. They also proved that there must have been three partial typescripts, two of which were revised by Clemens. They further suggested that two complete versions of the finally revised text—a copy for the illustrator and a copy for the printer—were assembled from these several typescripts, and that the holograph manuscript of the first half of the book had never been sent to the printer. They could not, however, advance any theory to account for the disappearance of that first half.18

Despite these gradual improvements and refinements in the dating of composition, important questions remained, such as exactly which parts of the text were influenced by Clemens’s 1882 river trip. Everyone agreed that only the missing first half of the manuscript was likely to resolve such questions, and its loss, however mysterious, was widely assumed to be permanent.

Then in the first days of 1991 a truly remarkable rumor swept through the rare book and manuscript world, eventually reaching the Mark Twain Papers in Berkeley. It was said that the first half of the Huckleberry Finn manuscript, which had not been seen for more than one hundred years, had in fact been found in a Hollywood attic, and that it was being evaluated for auction by Sotheby’s in New York. Another rumor had it that a number of manuscript dealers were pooling their funds to come up with the winning bid (estimated to be as high as $1.5 million) while planning to recover their investment by reselling the manuscript, page by page. In January 1991 Sotheby’s telephoned the editors of the Mark Twain Project and confirmed that it had recently authenticated a manuscript of some six hundred pages which was indisputably the missing first half of Huckleberry [begin page 671] Finn. Sotheby’s wanted to know if there was any evidence in the files of the Mark Twain Papers which bore on Clemens’s disposition of this first half, and which might therefore establish who could properly claim to own it. There was such evidence, but it was understood only after these six hundred plus pages had been found, and partly because of where they had been found.

In November 1885, nine months after he published the first American edition of Huckleberry Finn, Clemens received a letter at his home in Hartford from James Fraser Gluck (1852–97), a lawyer and a curator for the library of the Young Men’s Association in Buffalo. Clemens had never met Gluck, but he had been a member of that association when he lived in Buffalo in 1869–71, as had his friend and partner on the Buffalo Express, Josephus N. Larned (1836–1913), who was now superintendent of the library. Gluck wrote to Clemens, in part:

Clemens answered this letter on the day he received it, 11 November, saying in part: “I will comply, as far as I can, with the greatest pleasure—that is, to the extent of 50 per cent, of a MS. book (‘The Adventures of Huckleberry Finn.’)” He was referring to the second half of Huckleberry Finn—the half written last, in 1883. “I have hunted the house over, & that is all I can find,” he explained, adding that the first half of the manuscript must have been sent to “the printers, who never returned it.”20 Gluck acknowledged this letter on 12 November, offering to show Clemens “how pleasant an abiding place” in the library had been set aside for “the much-abused Huckleberry Finn. When Boston & Concord desert him then the home of the Presidents shall take him up.”21 This half of the manuscript was duly shipped and its receipt acknowledged on 14 November 1885: “The manuscript of Huckleberry Finn arrived this afternoon and was at [begin page 672] once placed in the Library vault.” Clemens annotated the envelope simply, “MS for Buffalo.”22 On the envelope of Gluck’s first letter he wrote, “Sent what was left of Huck Finn.”23

There the matter stood, and might have remained, except that some nineteen months later, probably soon after 24 June 1887, when the Clemenses arrived in Elmira for their usual summer stay, Clemens came across some more of his old manuscript. He promptly sent it to his business agent in Hartford, Franklin G. Whitmore (1846–1926), and instructed him to have it sent to Buffalo. Larned, superintendent of what was by then called the Buffalo Library, acknowledged its receipt on 5 July:

Just a few days later, Gluck also sent an acknowledgment: “Please accept my thanks for your kindness in forwarding the first part of Mss. of ‘Finn’ which Mr Larned has just rec’d. The whole can now be bound and placed on exhibition.”25

Despite these letters of acknowledgment from library officials, the first half of the manuscript was neither deposited nor put on display there. As far back as its records go, the library has no indication that it owned anything except the second half, the one sent in 1885. That explains why the 1887 letters were not correctly understood until 1991. The pages at Buffalo actually consisted of chapters 22 through “Chapter the last” (chapter 43), plus sixty pages from chapters 12–14. The assumption had always been, therefore, that Clemens sent chapters 22 through 43 in 1885, then found [begin page 673] and sent a mere sixty pages from “the first part of Mss.” in July 1887. If he had found and sent the whole “first half of the Ms. of ‘Huckleberry Finn’ ” (as Larned’s letter certainly seemed to say), surely that large stack of pages would still be in the library.

This view of what must have occurred was understandable, but mistaken. In October 1990 one of James Fraser Gluck’s granddaughters, Barbara Gluck Testa, opened a trunk in her attic and found a stack of 665 manuscript pages in Clemens’s handwriting, together with a blue envelope labeled “MSS. of Huck Finn by Mark Twain.”26 She had found the great bulk of the first half, the part sent by Clemens and acknowledged by Larned and Gluck in July 1887. Clemens himself had been wrong about the printers’ having failed to return it (none of the holograph manuscript had ever gone to the printer). Having unsuccessfully searched the Hartford house in November 1885, Clemens must have found the first half belatedly among his papers in Elmira, in late June 1887. And when, ten years later, James Gluck died suddenly at the age of forty-five, he evidently had the pages of the first half still in his possession. They were presumably gathered up and stored along with his other papers, and these eventually passed to his daughter and then, in 1961, to his granddaughters. Exactly why he still had them instead of the library may never be known. What is clear is that the pages stayed within the Gluck family, tucked away in that trunk, until rediscovered by chance in the fall of 1990.27 The complete manuscript of Huckleberry Finn is now housed in the Mark Twain Room of the Buffalo and Erie County Public Library in Buffalo, New York (NBuBE).

The survival of the first half of the manuscript (here abbreviated MS1 for the whole, but MS1a and MS1b for its two constituent parts) confirms that Clemens kept it in his own possession, and that he must therefore have sent typewritten copy to his publisher instead of the holograph. It also shows that Paine’s 1912 account was essentially accurate, even though seemingly based on little documentation: there were just three (rather than two, or four) major stints or stages in the composition (1876, 1880, and 1883). Moreover it demonstrates for the first time exactly what parts of the text comprised each of these stints, since each can be identified from physical evidence: different writing materials used at different times. MS1 [begin page 674] consists of two physically distinct parts that can be confidently identified as the product of the first two periods of composition. MS1a, drafted during the summer of 1876, extends from the beginning of the story up through the first part of chapter 18 (through line 11, page 146 of this edition), lacking only the second half of chapter 12 through chapter 14. MS1b, drafted between mid-March and mid-June 1880, extends from where MS1a ends through chapter 21 (page 188 of this edition), where the mob is preparing to lynch Colonel Sherburn for shooting Boggs. Both MS1a and MS1b were written (although they were not finally revised) before Clemens’s 1882 river trip, and could not have been directly influenced by it. The holograph housed in the Buffalo library since 1885, called MS2, is also physically distinct, and was written during the summer of 1883.

Beginning in 1874 and continuing as an almost unbroken habit for fifteen years, Clemens and his family spent at least part of the summer just outside Elmira, New York, at Quarry Farm, the home of Olivia Clemens’s sister and brother-in-law, Susan and Theodore Crane. There, isolated on a hilltop, in a small gazebolike octagonal study built especially for him by Sue Crane,28 Clemens wrote daily on whatever project interested him at the time. In 1874 he worked on the first half of Tom Sawyer, and in 1876 he read the proofs of that book and began a new project—Huckleberry Finn.

His method of working on books was well established by 1876. As he explained to Jeannette Gilder just two years after the publication of Huckleberry Finn, in a letter so frank he decided not to send it, it was his “habit to keep four or five books in process of erection all the time, & every summer add a few courses of bricks to two or three of them; but I cannot forecast which of the two or three it is going to be. It takes seven years to complete [begin page 675] a book by this method.”29 In 1890 he described this process in even more detail to his friend, the writer and critic Brander Matthews, during a summer stay at Onteora, New York. Matthews reported that

Clemens had earlier described how he hit a snag during the composition of Tom Sawyer. Having written a chapter that was “a failure, in conception, moral, truth to nature & execution,” he used for the first time what became a standard metaphor of his composition process: “I had worked myself [begin page 676] out, pumped myself dry.”31 And in 1906, with a lifetime of writing and self-observation behind him, he again recalled the circumstance and described his method:

Huckleberry Finn was written over seven years in three widely separated stints. There is ample textual evidence of adjustments made to early portions of the book, aimed at providing the “necessary preparation” for later episodes. From the beginning, Clemens had the idea of writing this story of a boy’s adventures in the first person—“autobiographically—like Gil Blas”—as he told Howells in 1875. Brander Matthews recognized a two-fold debt, in method and in form, to Le Sage’s Gil Blas (originally published in 1715–35) and mentioned it to Clemens:

Clemens was probably being disingenuous, if not simply forgetful. As his 5 July 1875 letter to Howells reveals, he had Gil Blas specifically in mind as a model for his sequel to Tom Sawyer. In fact, he had read Gil Blas in 1869 and commented on it in a letter to his then-fiancée, Olivia L. Langdon:

[begin page 677]

The other book that he “could not admire” was one that, like Gil Blas, undoubtedly influenced the creation of Tom Sawyer and Huckleberry Finn: Thomas Bailey Aldrich’s semiautobiographical Story of a Bad Boy, published in 1869. Aldrich described his Tom Bailey as “a real human boy, such as you may meet anywhere in New England, and no more like the impossible boy in a story-book than a sound orange is like one that has been sucked dry,” and involved him in various mischievous escapades.35 Writing to an unidentified correspondent in 1891, Clemens asserted that he had been drawn to “the boy-life out on the Mississippi” as a subject for literature by the “peculiar charm” it held for him, as well as for a second, more pragmatic, reason—the competition was less intense in the “boyhood field” of literature.36

Ultimately, of course, Adventures of Huckleberry Finn was not driven by literary models, nor by simple nostalgia. In creating the book, Clemens drew on his memories, his reading, his humor, his wide sympathies, his anger at cruelty and injustice, and his interest in discovering how to recreate the voices of the Mississippi Valley and how to reimagine the place and time so that it seemed profoundly real to his readers.

Settled in at Quarry Farm by the middle of June 1876, Clemens first tried to work on what he called his “pet book”—perhaps the same work referred to as the “double-barreled novel” in a letter to Howells.37 But his [begin page 678] work on this manuscript stalled in early July. By early August he was writing enthusiastically to his friends about a new project. “We are up here at the farm for the summer,” he wrote to his friend and English agent, Moncure D. Conway (1832–1907), on 1 August:

A few days later he wrote to Mary Mason Fairbanks: “I am tearing along on a new book & can’t interlard a vacation, being warned against it by the fate of my pet book, which lies at home one-third done & never more to be touched, I judge. Destroyed by a vacation. The mill got cold & could not be warmed up any more.”39 Finally in a letter of 9 August to Howells, he explicitly identified the new project:

Writing his new book as a first-person narrative—“Huck Finn’s Autobiography”—meant that Clemens had to create a transition from Tom Sawyer because the new book was both a sequel and a departure. In the very [begin page 679] first lines of the manuscript he distinguished Huck’s voice from the narrator of Tom Sawyer: “You will not know about me, without you have read a book by the name of ‘The Adventures of Tom Sawyer.’ That book was made by Mr. Mark Twain, & he told the truth, mainly.”42 In the second paragraph, Huck summarizes the conclusion of Tom Sawyer, making clear to the reader that the new book picks up just where Tom Sawyer left off, as Tom outlines his plan to form a gang of robbers and convinces Huck that he must return to the Widow Douglas’s care if he wants to be “respectable” enough to join the gang. Although it is evident that the recapitulation of events in the first chapter of Huckleberry Finn comes directly from the text of Tom Sawyer as Clemens published it, some of the detail about Huck’s life at the widow’s may derive from material he deleted while revising the present chapter 35 of Tom Sawyer or from a (lost) final chapter he wrote in 1875, which traced the subsequent history of his characters.43

The gang’s adventures are the subject of chapter 3 of Huckleberry Finn. In chapter 4, however, Clemens hints at an altogether darker narrative when Huck’s violent and drunken father, Pap Finn, reappears in his life (as Huck himself predicted he would in chapter 25 of Tom Sawyer).44 Saved from the widow’s “sivilizing” but trapped by his father’s violence and greed, Huck decides to escape. In chapter 7, he stages his own death and flees down the river in a salvaged canoe, deciding to stop at Jackson’s Island. He soon encounters a friend who is now a runaway slave, “Miss Watson’s Jim” (chapter 8). Jim’s plight as a fugitive slave introduces a subversive new element to the plot—slavery, a subject hardly touched upon in Tom Sawyer. In the next chapters, Jim tells a long “ghost” story about his experience with a cadaver in a medical college;45 the island floods; Huck [begin page 680] and Jim salvage “a little section of a lumber raft” (chapter 9); and Huck goes ashore disguised as a girl to “find out what was going on” (chapter 10). When he discovers that some townsmen are intending to search Jackson’s Island, he returns with the urgent news that “They’re after us!” and together he and Jim flee downstream on the raft (chapter 11). After they drift downstream for several days, Huck reports in chapter 15 (which in this draft immediately followed the first part of chapter 12), that they will raft down the river to Cairo, “at the bottom of Illinois, where the Ohio river comes in,” and “sell the raft and get on a steamboat and go away up the Ohio amongst the free States, and then be out of trouble.” With this decision, Clemens brought together two narrative strands: the river journey would provide the opportunity for a series of adventures both on the river itself and on shore, while the relationship of the outcast boy and the runaway slave would provide the moral dilemma of the protagonist. The plot would turn on whether Huck would help Jim to freedom, or turn him over to the authorities as he was trained to do. Although Clemens had Huck and Jim decide on a course that would take them north up the Ohio River, it seems unlikely that he ever seriously considered that line of development. His working notes show no evidence of plans to send the two fugitives up the Ohio, whereas both the manuscript and the working notes show his concern with finding plausible reasons to keep them drifting south on the Mississippi.46 In what became chapter 16, Huck and Jim—not yet aware they have already passed Cairo in the fog—begin to wonder how they will recognize it, and Huck secretly boards a passing lumber raft, hoping for information. The “raft episode” runs for more than fifty manuscript pages, with its own cast of characters, and with singing, dancing, fighting, and an embedded ghost story. Huck returns to Jim on their own raft, and Huck devises a lie to save Jim from capture in an encounter with a “skiff” of slave-hunters. Soon after, their journey is suddenly disrupted: the raft is smashed by a steamboat, and Huck and Jim are separated in the river. Jim’s plight is unknown, and the story thereafter follows Huck’s adventures on land as he joins the Grangerfords, one of two feuding families. Huck first encounters the Grangerfords on page 397 of MS1a. In what became chapter [begin page 681] 17 and the first part of chapter 18, the family and their house are described in detail and the stage is set for the feud.

When, on 9 August, Clemens reported to Howells that he had “written 400 pages” of the new book,47 he still had about a month of his working summer season left. Even though he made no other reference to Huckleberry Finn in his letters of late August and early September, he seems to have continued to work on the book, albeit not exclusively: on 8 August he began compiling his “Record of the Small Foolishnesses,” unconscious witticisms of his two young daughters, and by 23 August he had written a sketch, “The Canvasser’s Tale,” which he sent to Howells for publication in the Atlantic.48 Despite these distractions, he probably finished the entire 446 pages of MS1a before he left Elmira on September 10 or 11.49

MS1a consists of manuscript pages 1 through 446 (1.1–80.29 and 99.1–146.11 in this edition) and comprises chapter 1 through the middle of chapter 12, and chapter 15 through the middle of chapter 18.50 The pages are all written in black ink on Crystal-Lake Mills stationery, with numerous revisions in pencil. There are some minor variations in handwriting and density of ink that might signal small breaks in composition, but the only evidence suggestive of a longer pause occurs on page 362, where his marginal notes refer to an as yet unidentified literary work.51 MS1a comprises just over a third of the full manuscript, which totals almost fourteen hundred pages. Clemens provided chapter breaks throughout, although only the first nine chapters were numbered. (Ultimately, he would reconfigure most of the chapter breaks before publication. The chapter numbers mentioned here and throughout the introduction are those of the first edition.)

Without access to the first half of the manuscript, and with the evidence of Clemens’s own reference to having completed 400 pages by 9 August, DeVoto and Blair speculated that he broke off his first writing stint where the steamboat collides with the raft in chapter 16. They assumed the collision [begin page 682] occurred about page 400. (We now know it occurs on page 394. On manuscript page 400, Huck, claiming to be “George Jackson,” is admitted to the Grangerford house.) They reasoned that Clemens was stymied by technical problems: how was he to convincingly explain Huck and Jim’s continued southward drift on the river, and what form should subsequent adventures take? Clemens’s impatience with these problems “blew the book from the writing table to a pigeonhole.”52 The evidence of the manuscript now makes clear that his work on the book was not stalled by these problems in 1876: he continued his story by shifting the action to the shore, and he was well into the Grangerford episode (chapter 18) before he stopped work.53 It would be 1880 before he returned to the manuscript.54

When Clemens put aside his manuscript in August or September 1876, he stopped rather abruptly at the top of page 446 in the middle of a conversation at the Grangerford house, where Huck asks Buck Grangerford to explain what a feud is. The six lines of inscription at the top of that page were the last of the manuscript written in black ink.55 Huck’s question apparently remained unanswered for more than three years, as Clemens worked [begin page 683] on other literary projects. Chief among them were The Prince and the Pauper, begun in earnest in the fall of 1877, and A Tramp Abroad, the record of his time in Europe from April 1878 through August 1879. A Tramp Abroad—and a myriad of other concerns—occupied him for a number of months after his return from Europe. His letters of late fall and winter 1879–80 are filled with the aggravation of finishing that book; details of its production and its early reception; reports of his progress on the manuscript of Prince; comments on his brother Orion’s literary work; references to his short scatological work, 1601; concerns about his investments in Kaolatype; and concerns for Olivia’s deteriorating health during her pregnancy with Jean.56 Early in 1880, Clemens finished his work on A Tramp Abroad and then turned his attention back to The Prince and the Pauper for several weeks, probably putting that manuscript aside around mid-March 1880. It was only then that he had an opportunity to return to Huckleberry Finn, and the physical evidence of the manuscript suggests that he did so.

MS1b is the shortest of the three physically distinct sections of manuscript: it comprises pages 447 through 663, which became the remainder of chapter 18 through the end of chapter 21 (“ ‘Well . . . with.”, 146.12–188.16 in this edition). It is written in the same purple ink and on the same white wove paper as pages 367 through 414 of The Prince and the Pauper manuscript,57 and matches ten of the eleven pages of Group 2 of the working notes for Huckleberry Finn.58 Blair believed that Clemens might have written chapters 19 through 21 at some point between mid-June 1880 and mid-June 1883, but that time span can now be narrowed considerably: the newly discovered manuscript shows that those chapters were written, like the second half of chapter 18, in the purple ink that Clemens used only until mid-June 1880 and then not again for several years. Although it is possible that MS1b pages were written anytime between [begin page 684] November 1879 and the middle of June 1880, when Clemens was known to be using the combination of purple ink and wove paper, the most likely time is when he took a break from The Prince and the Pauper in mid-March 1880. Clemens’s own 1883 reference to the book as “a big one that I half-finished two or three years ago,” confirms that he recalled a writing stint in 1880 or 1881.59 On the basis of this varied evidence, we assign the composition of the MS1b pages to the period from March to June 1880.

On 6 May 1880 Clemens wrote to Howells that he had “knocked off” writing and did not “intend to go to work again till we go away for the summer, 5 or 6 weeks hence.”60 But at Quarry Farm that summer his time was for the most part devoted to his work on The Prince and the Pauper. Except for the draft of the “Notice” page, which he added to the beginning of the Huckleberry Finn manuscript toward the end of June (or possibly later), the manuscript was put aside for a second stretch of years.61 Despite this break in composition, he clearly had both books in mind when he wrote a letter to his sister dated by Paine “near the end of the year” 1880:

The “verbal agreement” with publisher James R. Osgood was overruled: The Prince and the Pauper was eventually published separately, and Clemens apparently did no further work on Huckleberry Finn in 1880.63

In the 217 pages Clemens completed in the spring of 1880 he brought the Grangerford-Shepherdson feud to its bloody conclusion and reunited [begin page 685] Huck and Jim, sending them on down the river where they encounter two fugitive confidence men, the king and the duke (chapters 18 and 19). The introduction of these characters provided the motivation for Huck and Jim’s continued travel south on the Mississippi: the king and the duke commandeer the raft by implicitly threatening to turn Jim in. The king makes a profitable appearance at a camp meeting, the duke plans a Shakespearean performance, and they stop at a “little one-horse town in a big bend,” Bricksville, Arkansas, and witness the deadly confrontation between the drunken Boggs and Colonel Sherburn (chapters 20 and 21). At that point, the last page of MS1b (663)—just about the mid-point of the book—the purple-ink inscription ends.64 Over the next three years, Clemens may have tinkered with his manuscript: there are numerous pencil revisions throughout MS1a and MS1b that cannot be assigned any clear date, many of them seemingly part of a concentrated and thorough revamping of dialect usages.65 It seems likely, however, that he did not go past page 663 of the Huckleberry Finn manuscript until 1883.

Between the fall of 1880 and the spring and summer of 1883, various events and projects affected the completion of Huckleberry Finn. The first was the death on 28 September 1880 of Clemens’s publisher, Elisha Bliss, Jr., of the American Publishing Company. Long dissatisfied with Bliss’s management of his books, Clemens signed a contract with James R. Osgood, a sociable Boston publisher with an impressive list of English and New England authors and a wide circle of literary friends. Osgood, however, proved to know little about subscription publishing. Over the next few years, his friendship with Osgood strained by disappointment over sales of his books, especially Life on the Mississippi, Clemens began to contemplate becoming his own subscription publisher.

Another project, a long-deferred trip to the Mississippi River region,66 finally became reality in April 1882. Clemens, accompanied by Osgood and a hired stenographer, Roswell H. Phelps, traveled first to St. Louis, and [begin page 686] from there, over three weeks, down the Mississippi by steamboat to New Orleans. There he spent time with George Washington Cable (1844–1925), whom he had first met in Hartford in June 1881, and with Joel Chandler Harris (1848–1908), whose dialect stories he had first encountered in 1880.67 The travelers then returned to St. Louis. Clemens continued upriver to Hannibal, where he visited with old friends, and as far north as St. Paul, Minnesota.68

Clemens’s observations and impressions, coupled with his extensive reading for the trip, formed the basis for forty-six of the sixty chapters of Life on the Mississippi (chapters 4 through 17 reprinted his 1875 Atlantic Monthly series, “Old Times on the Mississippi”). The influence of the trip on the second half of Huckleberry Finn is undeniable, but its relationship to the first half was not well understood until the discovery of MS1. Because Life on the Mississippi was published two years before Huckleberry Finn, and because so little was known about the composition of chapters 17 through 21 of the latter book, it was generally accepted that passages in those chapters—the Grangerford-Shepherdson feud and the description of the Grangerford parlor, in particular—were reworkings of incidents and descriptions already composed for Life on the Mississippi.69 In fact, the order of composition was the reverse: the passages in Huckleberry Finn came first and were then reworked for Life on the Mississippi.

The return to the river certainly revived Clemens’s youthful memories, but it also stoked his anger. The advances in the social and legal status of blacks achieved during Reconstruction, while under siege from the beginning, had been further deteriorating since 1876, and racial violence against the former slave population was increasing. With unhindered mob rule, intimidation, arson, and lynchings, the justice system in the South seemed to have entirely broken down.70 The adult Clemens, a “northernized ex-Missourian” as Walter Blair styled him,71 had in “Old Times on the Mississippi” and in The Adventures of Tom Sawyer nostalgically recreated the times, places, and voices he knew in his youth. In 1882 he was confronted with the reality of riverside scenes and towns, with the speech [begin page 687] and manners of the postbellum population, and with the failure of Reconstruction and reform. Walter Blair discussed the discernible impact of these “contradictory perceptions” on Huckleberry Finn: “Contrasting the past with the present, as he was bound to do, the author found the past both more glamorous and more sordid.” He was struck by the “unlovely aspects of Southern life,” and generalized those perceptions into a view of human nature as a whole. “The parts of the novel written in 1883 show the author well on his way to accepting the view of mankind that would lead him in his last years to speak of ‘the damned human race.’ ”72

Once back in Hartford in late May 1882, Clemens immediately began work on Life on the Mississippi. In mid-July he and his family relocated to Quarry Farm. Despite intermittent health problems that interfered with his work all summer,73 he made some progress. Between 31 July and 15 September he started two Elmira typists, Harry M. Clarke and Jakob B. Coykendall, on the job of copying the holograph manuscript of Life on the Mississippi. They typed at least to the middle of chapter 44.74 This would be the first of his books for which he would submit typewritten printer’s copy to the publisher. Clemens’s decision to start using typewritten copy must have been, to some extent, the result of his own recent successful attempts to use the machine. He had acquired a Remington all-capitals typewriter early in 1882 (similar to, but not identical with, the machine that Clarke and Coykendall used in Elmira), and he began typing some of his own letters and also dictating some to a secretary who then typed them.75 (His earlier experiences, with an all-capitals machine purchased [begin page 688] in November 1874, had been distinctly unsuccessful: he quickly became disillusioned with the balky machine and gave it away.)76

On 15 September, from Elmira, Clemens wrote Osgood about Life on the Mississippi: “Am mailing you another chapter or two. . . . Book nearly done, now. Is mainly in the hands of the copyist. Will send you the seven (reprint) chapters, revised and corrected presently—the ones first illustrated by the artist. Also, title-page, etc. (‘Life on the Mississippi’), so that you can hurry up your canvassing specimen.” Osgood, himself just back from a summer in Europe, wrote to Clemens on 22 September pressing him for matter for the salesmen’s prospectus or canvassing book.77 Clemens probably also had the “raft episode” from his Huckleberry Finn manuscript typed at this time: he had decided to include it in chapter 3 of Life on the Mississippi, and it would appear in the prospectus. With all the pressure to complete Life on the Mississippi, however, it seems unlikely that he had the rest of MS1 of Huckleberry Finn typed at this time.

Back in Hartford at the end of September, Clemens continued to struggle with Life on the Mississippi. While attempting to complete the book by writing new chapters, he found himself distracted by also having to respond to the publisher’s editing and deletions in the portion already completed.78 To add to Clemens’s difficulties, the woman he employed in Hartford to type the remainder of his manuscript succumbed to scarlet fever in December, further delaying the book.79 On 15 January 1883, however, [begin page 689] Clemens wrote George Washington Cable, “I have just finished my book at last.”80

Despite Clemens’s frustrating experience with Life on the Mississippi, the new technology of the typewriter transformed his way of working. It allowed him to make a clean copy of his manuscript, greatly facilitating his continued revision of the text. He described the process in a letter of recommendation for Harry M. Clarke:

Although revision and proofreading on Life on the Mississippi continued briefly in late January and early February 1883,82 Clemens was able to report to Howells on 1 March:

In all probability, Clemens’s “labor” included a return to Huckleberry Finn. He may have made new revisions to his manuscript (MS1a, 1–446, and MS1b, 447–663) at this time, mostly in pencil,84 but most of his revisions to this first part of the text would be made after he had it typed.

[begin page 696]

Clemens referred to TS1 in a marginal note in MS1a and in his 1883 working notes (Group 3, notes 3–1, 3–2, and 3-4, in Mark Twain’s Working Notes, pp. 503–5) when he was reviewing what he had written so far. In addition, he referred to TS1, to the raftsmen’s passage from LoM, and to TS2 in the notebook he kept in late 1884 when he was choosing readings for his upcoming lecture tour. Between April and July 1884, Kemble annotated his illustration boards with references to TS1, TS2, and TS3. All the known references are listed below: in the first column, the typescript page, followed in parentheses by the page and line in the present edition; in the second column, the words mentioned by Clemens or Kemble; and in the third, the source of the information, whether from Kemble’s notations or from Clemens’s (SLC) marginal notes, working notes, or notebook (as published in N&J3).

TS1

13 (14.12–28)	“old Finn supposed to”	SLC, 3-1
19 (19.27–20.7)	“GaveSold $6000 to Judge”	SLC, 3-1
36 (39.16–17)	“Log raft. . . Plank raft 12 × 16.”	SLC, 3-1
43 (47.17)	“ ‘Bessie’ or Becky?”	SLC, 3-1, 3-2
62 (61.3–14)	“Huck’s father in floating house”	SLC, 3-1
64 (63.1–16)	″	SLC, 3-1
68 (67.8–12)	“I started . . . quarters”	MS1a, 235
70 (69.33–70.1)	another reference to “Huck’s . . . house”	SLC, 3-1

TS2 interpolation at TS1 page 81

81–10 (88.23)	“ ‘Hello whats up’ ”	Kemble
81–15 (92.7)	“ ‘We turned in & slept like dead people’ ”	Kemble
81–19 (96.2–13)	“ ‘The story of Solomon’ ”	Kemble

TS1

83 (100.24)	“ ‘Amongst a lot of snags’ ”	Kemble
84 (100.29–103.3)	“waking Jim”	SLC, N&J3, 60
86 (102.18)	“Asleep with one arm across the oar”	Kemble

LoM tear sheets interpolated at TS1 page 89

89½ (107.1)

“Raftsmen fight”

SLC, N&J3 , 60

TS1

90 (123.30–124.3)	“Troubled conscience & smallpox”	SLC, N&J3, 60
90 (124.20–22)	“Jim has wife & 2 children.”	SLC, 3-1
95 (127.10–15)	“$40 from men”	SLC, 3-1
100 (132.29)	“Betsy”	SLC, 3-1
101 (133.1)	“Shepherdson”	SLC, 3-1
105 (136.25–137.26)	“Art & Bible”	SLC, N&J3, 60
106 (137.12)	″	SLC, N&J3, 60
109 (140.1)	“Grangerford”	SLC, 3-1
136 (163.7–165.3)	“Duke & K”	SLC, 3-1
143 (170.21–172.8?)	“let ’em tell these adventures”	SLC, 3-4
147 (174.6–30?)	“Another ref” follows “Duke & K”	SLC, 3-1

TS2 begins at 160–1

160–64 (241.20)	“ ‘There! Royal Nonesuch, Brickville’ ”	Kemble
Ch. 33 (271.1 Ch. 31)	“ ‘All right, I’ll go to hell.’ ”	SLC, N&J3, 61
120 (285.27–30)	“Then he makes a graceful bow.”	Kemble
191 (341.2–3)	“I doan go from heah widout a doctor”	Kemble
194 (345.2–3)	“I nearly ran into him”	Kemble

TS3 references to chapters 18–20 of TS3, the retyped version of TS1

123 (144.15)	“ ‘Pretty soon . . . galloping down the road’ ”	Kemble
134 (152.9)	“ ‘A couple of young chaps behind a wood pile’ ”	Kemble
149 (163.22)	“ ‘I am the late Dauphin’ ”	Kemble
156 (169.23–170.2)	“ ‘The King as Juliet’ ”	Kemble

In mid-June 1883 the Clemens family once again traveled to Elmira for the summer, and by the end of the month they were at Quarry Farm, with Clemens enthusiastically at work. “The three summer months being my chief working time,” he wrote to Charles A. Dana on 19 July, “I slave it without losing a day while we are here. I have written one small book, & am far along in a bigger one—& shall finish it if I don’t run around any.”92 (The “small book” was his long-winded burlesque, “1,002. An Oriental Tale.” The “bigger one” was Huckleberry Finn.) The next day, he shared his good spirits in a letter to Howells:

Clearly, the “tank” had refilled.94 With his invention and enthusiasm in full swing, Clemens produced 695 new manuscript pages before summer’s end. This final half of the manuscript (MS2) was written on Old Berkshire Mills stationery in blue ink. He also accumulated fourteen pages of working notes for this section of the manuscript.95 His summer’s work included sixty pages that he wrote, had typed, and eventually inserted into TS1 at page 81, comprising the adventure aboard the sinking Walter Scott and Huck’s conversations with Jim about the ways of royalty, Frenchmen, and cats (the second half of chapter 12 and all of chapters 13 and 14).

Aside from that long insertion, he picked up the main thread of his narrative [begin page 693] with page 160 (MS2), which he presumably numbered to follow the last page of TS1 (159), not the last page of MS1b (663).96 He had left off three years earlier with a lynch mob coming to get Colonel Sherburn. The chapter originally concluded that “they was too late. Sherburn’s friends had got him away, long ago,” and he afterwards added a direction to himself, “No, let them lynch him” at the bottom of his MS page.97 Instead, when he began the 1883 manuscript with Colonel Sherburn confronting the Bricksville mob, Clemens poured some of his own disgust at lynchings and mob rule into Sherburn’s speech. Then Huck attends a circus, and the king and the duke successfully perform the “Burning Shame” (which Clemens revised before publication to “Royal Nonesuch”).98 Two long and detailed episodes complete the book: the account of the king and the duke’s attempt to swindle the Wilks heirs, and the “evasion” chapters, in which Tom Sawyer and Huck rescue Jim from the Phelps farm. Clemens was able to write to Howells on 22 August, throwing in a joking reference to Olivia as “the boss”:

A little more than a week later, on 1 September 1883, Clemens reported to his publishers the result of his summer’s efforts. To James R. Osgood, he wrote: “I’ve finished ‘1,002’ (Arabian Nights Tale,) & likewise ‘The Adventures of Huck Finn’; had written 50,000 words on it before; & this summer it took 70,000 to complete it.” (MS1 and MS2 are actually closer in length: about 55,000 words and 58,000 words.) To his English publisher, Andrew [begin page 694] Chatto, whose firm had published all of his major books since The Adventures of Tom Sawyer, he added his assessment of the novel: “I’ve just finished writing a book; & modesty compels me to say it’s a rattling good one, too—‘Adventures of Huckleberry Finn.’ (Tom Sawyer’s comrade.)”100

[begin page 713]

---

This chart shows in stages the documents Mark Twain wrote, had copied by typewriter, and revised as he prepared Adventures of Huckleberry Finn for publication. When he submitted the book to his publisher in April 1884, there was only one complete copy, here designated “Mark Twain’s Copy,” essentially a composite of two typescripts (TS1 + TS2), which had been made from the two halves of the manuscript at different times and by different typists. By the end of May 1884, the first typescript (TS1) had been retyped, incorporating late revisions by both Mark Twain and William Dean Howells (TS3). A new, completely revised copy of the text was now assembled—TS3 + the duplicate of TS2—to serve as the “Printer’s Copy” for the first American edition.

All page references to the manuscript (MS1a, MS1b, and MS2) correspond to the actual MS in the Buffalo and Erie County Public Library (NBuBE). All references to chapters are to those in the published book. All information about the typescripts (TS1, TS2, and TS3), none of which is known to survive, is inferred from references to them by Mark Twain or the illustrator of the book, Edward Windsor Kemble, and from reconstructed models of them at the Mark Twain Project based on those references and other documentary evidence. In these models, TS1 is 159 pages long; TS2 consists of 216 pages (numbered 160-1 through 216) and includes an additional sequence of pages (numbered 81-1 through 81-22) intended to be inserted into TS1 at page 81; and TS3 is 175 pages long.

---

[begin page 714]

1876:

Mark Twain begins Huckleberry Finn at Quarry Farm in July. By September his working copy consists of:

MS1a	pages 1–446 became chapters 1–12½, 15–18½

1880:

Mark Twain brings the MS to the halfway point, most likely in Hartford from March through mid-June 1880, by which time his working copy consists of:

MS1b	“Notice” added in late June or after
MS1a	pages 1–446
MS1b	pages 447–663 became chapters 18½–21

1882–83:

The raft episode (MS1a pages 309–62) is typed for inclusion in the printer’s copy of Life on the Mississippi (LoM), probably by Harry Clarke in Elmira in September 1882. By December 1882 the raft episode is typeset and printed tear sheets are available. MS1a and MS1b are typed (TS1), probably in Hartford between October 1882 and early May 1883. At that time, Mark Twain’s working copy consists of:

TS1	“Notice”
TS1	pages 1–159 with raft episode tear sheets at TS1 page 89

1883:

Mark Twain completes the rest of the book (MS2) at Quarry Farm between mid-June and 1 September, then has MS2 typed (TS2) with a carbon copy.

MS2	title page
MS2	pages 81-A-1 through 81-60 became chapters 12½–14
MS2	pages 160–787 became chapters 22–43
	↓
	TS2	title page
	TS2	pages 81-1 through 81-22
	TS2	pages 160-1 through 216

By late September 1883 the holograph manuscript has been entirely superseded by Mark Twain’s new working copy: TS1 + TS2.

[begin page 715]

1883–84:
Mark Twain’s Copy

Mark Twain’s complete typescript (TS1 + TS2) has 45 chapters (2 more than the first edition because the raft episode is still in place). He adds a “dedication” (which he later replaces with the “Explanatory”). In April 1884 he sends the complete typescript to Webster and Company so that the artist can begin work. This copy is returned to Mark Twain by late September 1884.

TS2	title page
TS1	“Notice”
MS	“dedication” possibly added at this stage
TS1	pages 1–81
TS2	pages 81–1 through 81–22
TS1	pages 82–89
LoM	raft episode tear sheets
TS1	pages 90–159
TS2	pages 160–1 through 216

April–July 1884:
The Artist’s Copy

The illustrator, E. W. Kemble, initially uses Mark Twain’s copy to plan drawings for the first 12 chapters. In late May, after TS1 is retyped to create TS3, the raft episode is dropped; Kemble makes no drawings for it. In June and July he consults first TS1, then TS3, and then one of the copies of TS2. He finishes all illustrations by 12 July and returns all book copy to Webster and Company. His drawings are processed into electroplates between 29 May and 29 July.

May–August 1884:
The Printer’s Copy

The “ghost” story is dropped from chapter 9 before TS1 is retyped to create TS3; the raft episode is dropped in late May. The printer’s copy for the text—TS3 + TS2—is assembled, so that typesetting can begin in early June. The “Explanatory” is probably added about this time. The picture captions, running heads, and tables of contents and illustrations are ready by early August.

TS3	title page + “Notice” + “Explanatory”
TS3	pages 1–89 chapters 1–12½
TS2	pages 81–1 through 81–22 chapters 12½-14
TS3	pages 90–175 chapters 15–21
TS2	pages 160–1 through 216 chapters 22–43

The printing of the first American edition begins by late August 1884.

In mid-June, before the illustrations for chapters 13–20 were ready for the printers, galley proofs began to arrive at Webster’s office. Each galley held [begin page 722] text divided into three successive pages stacked vertically, with the electrotype plates for the illustrations (or at least blocks of the correct size) in place. Because of the intermingling of pictures and text, type could not be efficiently set until the illustrations had been marked for reduction to size and either electroplated (with the electroplates mounted type-high for insertion in the type) or dummied in the form of correctly sized blocks (pending final production of the actual electrotypes).175 When the typesetting of Huckleberry Finn began, Kemble had finished the drawings for the first twelve chapters (each of which he drew double size, to be reduced photographically), and the printers had most or all of the corresponding electroplates in hand. These early proofs—variously called “galleys,” “slips,” or “revises”—could have included no more than those twelve chapters, and probably included less.

[begin page 723]

The three pages in each galley proof showed pictures and text only. They lacked running heads, “page titles,” picture captions, and final page numbers. Running heads, “the fixed or general title of the volume,” appeared on left-hand pages and were distinct from “page titles,” which appeared on right-hand pages and were based on the specific contents of that page, and thus could not be composed until the pages were “final.”176 At this early stage the captions for the illustrations had also not been fully prepared, and the number of pages of “front matter” which would precede the text was not known, so page numbers remained indeterminate. (The front matter, paged in sequence with the text, would include a table of contents which repeated the page titles, and a list of illustrations which repeated the captions. Both would of course require the actual page numbers as well, but could be set up pending those final numbers, and the numbers [begin page 724] added at the last moment.) The proofs sent to Webster were called “revises” (second galleys) because the first galley proofs had already been read by the printer’s proofreader, very probably with his copyholder reading aloud from the printer’s copy, and then the type was corrected—at least that was the standard practice.177 Once the “revises” or “author’s” galley proofs arrived at the Webster company, they were checked again before they were sent to Clemens, as Webster’s letters make clear.

Webster evidently ordered two sets of author’s galleys for these early chapters, one for Clemens and the other for Howells. (Howells’s clarification of what he had meant in offering to read “proof” had evidently not been passed along to Webster.) Howells may have felt some irritation when he received these proofs for the early chapters. He wrote to Webster on 16 June: “You need not send me proof of Huck Finn. I read the copy so carefully that a good proof-reader’s revision is all that is now necessary.”178 In the midst of final preparations for the annual move from Hartford to Elmira, Clemens also wrote Webster, chastising him for sending the proofs by letter rate rather than manuscript rate.179

Webster sent “more proof” on 25 June.180 By then the typesetters had certainly reached chapter 12, and possibly chapter 13—that is, twenty-five to twenty-eight galleys in all. Clemens complained to Howells on 28 June: “My days are given up to cursings—both loud & deep—for I am reading the H. Finn proofs. They don’t make a very great many mistakes; but those [begin page 725] that do occur are of a nature to make a man curse his teeth loose.”181 Somewhat bewildered, Howells replied on 2 July: “Why need you read the Huck Finn proofs? I went over the printed copy i.e., the typescript so carefully that a good office reading was all that was necessary. If I’d supposed they were going to send them to you I would have read them again myself.”182

For the time being, the question of who was to read the proofs was left in abeyance. Other procedures, however, for completing the illustrations and making electrotypes of them, and also for setting the type, were going smoothly. Kemble had to stay well ahead of the printers, continuing to work in sequence, and he did so. On 25 June he wrote Webster: “I shall be down Saturday noon i.e., 28 June with thirty or forty pictures. I would like to draw $200.00 or more. Will you be in. I would like to talk with you concerning the rest of the pictures.”183 By the time Kemble arrived with thirty-five pictures, Webster had new instructions from Clemens about an illustration he had already approved (and that had probably been processed by the electrotypers on 19 or 21 June):

The word “texas” was removed, probably by the engravers, from the electrotype of this illustration, but it was not replaced with “Walter Scott,” no doubt because that would have been more complicated or more expensive or both (see the illustration on page 91). Belated corrections were rare, however, since Clemens’s suggestions had been transmitted through Webster early enough to be incorporated before engraving. Of the 28 June batch of thirty-five drawings, sent to Clemens immediately after Webster received them, Clemens said on 1 July, “Kendall’s Kemble’s pictures are mighty good, now.”185 At this point, Kemble had completed at least one hundred and sixteen illustrations, or those for two-thirds of the book (up to about page 241, “How to Find Them”). He was well ahead of the typesetters: plates for his most recent drawings would not be ready until 14 July.186

[begin page 726] Toward the end of June, Webster began to consider what to do about captions for the pictures. Although Clemens had criticized drawings that did not please him, so far as the evidence shows, Kemble alone chose their subjects, quoting a relevant portion of the text below his drawings to suggest captions. Sometime before 5 July, Webster, evidently working from Kemble’s tentative captions and preparing copy for the compositors, must have asked Clemens whether captions taken from the text needed quotation marks (referred to as “quotations”). That he did not send Clemens copy at this time is indicated by the reply on 5 July: “As a general thing, no quotations are needed. In the instance mentioned they were not necessary.”187 Webster wrote again on 9 July with his plan for captions (what he called “titles”): “In regard to quotations: I will put the titles in, plain, then when proof goes to you, if you think they are needed you can put them in. . . . I send you more proof today.”188

The captions that ultimately appeared in the book often differ significantly from Kemble’s working captions, but only some of these differences are traceable to Clemens. He had of course seen Kemble’s working captions when he approved the original drawings, and he could have asked to see the printer’s copy for captions before it went to the printer, but he probably waited to review them in proof, as Webster suggested. He had followed just this procedure during production of The Prince and the Pauper.189 When he did see proof, however, he must have supplied some captions that depart from the text in ways characteristic of him—for instance, “Solid Comfort” (page 30), which he liked so much that he had it repeated in chapter 26 of A Connecticut Yankee.190 Nonetheless, the captions show the work of Kemble and Webster as well as Clemens. Kemble, as a staff cartoonist on the New York Graphic and Life magazine, had experience in writing captions and was capable of doing so for Huckleberry Finn, though in his working captions he had been careful to quote the text. [begin page 727] Webster’s letters to Clemens show that he wanted detailed guidance from the author, which makes it seem unlikely that he would volunteer completely rewritten captions on his own. Yet Webster clearly did either modify many of the working captions or supply new captions (and later, page titles) for Clemens to edit in proof. Lacking both printer’s copy and edited proof for these captions, one can only speculate about the extent of the various principals’ contributions.191

Webster’s letter of 9 July said that he was sending “more proof,” but he did not actually send it until the following day, when he again wrote:

[begin page 728] If by 25 June the compositors had set twenty-eight galleys, reaching the end of chapter 12, this additional batch of eleven galleys would have brought the total to thirty-nine, or two pages into chapter 17 (that is, past the point where the raft episode was omitted from chapter 16).

Having received on 10 July a promise from Kemble to bring the “remainder of the sketches” by the end of the week, Webster reported: “I will send you the last batch of pictures I hope by Saturday.”193 Kemble delivered his final batch of drawings on Saturday, 12 July, and was paid $585, the balance of his $1,000 fee. Webster sent them to Clemens the same day, calling them “another batch of good pictures.” On 15 July Webster picked up “1002 Sheep backs & 1002 ½ mor. backs of Huck”—the binding samples that were to be pasted on the inside covers of the salesmen’s prospectuses—and, presumably having received the last batch of illustrations back from Clemens, promised to have “everything, cuts and all, in the printers hands early next week.”194 He was too optimistic by a matter of two weeks: the final fifty-seven illustrations were not electroplated until 29 July, and Webster did not finish preparing the front matter until the first week in August.195

In mid-July Webster’s time was taken up with matters other than production of Huckleberry Finn, despite his continuing task of transmitting pictures and proofs between author, printer, and electrotyper. In addition to Webster’s increased responsibilities at home because of the birth of his son, Samuel Charles Webster, he had been asked by Clemens to: (1) intercept a statue by Karl Gerhardt that was being mistakenly shipped to Hartford; (2) search for reference books on the West (Clemens was then writing “Huck Finn and Tom Sawyer among the Indians”); (3) look into some financial matters; (4) “heave” his “surplus energies” into Clemens’s “to-be-patented portable calendar”; (5) hire someone to fix the furnace in the Hartford house; (6) negotiate a contract with John T. Raymond for production rights to “Colonel Sellers As a Scientist”; (7) work on the manufacture of the history game Clemens had invented the previous summer; and (8) prepare a detailed contract with James B. Pond for the reading tour Clemens was about to undertake with George Washington Cable, in part to publicize the new book.196 Webster was also making arrangements to hire general agents to sell Huckleberry Finn.

[begin page 729] But production was still proceeding smoothly. In mid-July the printer began to produce “foundry proofs” for the first eleven or twelve chapters—that is, he transferred the corrected standing type from the vertical three-page galleys to horizontal two-page “foundry chases” in preparation for casting the printing plates.197 The foundry proofs from these chases were also called “page proofs,” or simply “pages.” They were much closer to the final book pages than the rudimentary pages in the three-page galleys because they now contained the running heads and page titles, as well as picture captions and probably even page numbers.198 Webster, hoping to avoid an explosion, sent these first foundry proofs to his uncle on 22 July, and carefully explained: “I send you by mail today a batch of paged proof that you have already corrected. I send this for the reason that I have inserted the titles of the pagesi.e., the running heads and the page titles and picturesi.e., the captions, and if you wish to make any changes it must be done before the pages are electrotyped.”199 The proofs may have included a title page, which Clemens would have seen when he opened the package, but he probably did not immediately examine the rest. For almost three weeks, he had been spending “every other day” in the dentist’s chair having “one or two teeth gouged out & stuffed . . . a couple of hours a day,” and he still dreaded another few days of treatment. On his “off days,” however, he was at work on “Huck Finn and Tom Sawyer among the Indians,”200 and his plan for it caused him to ask for a change in the subtitle of Huckleberry Finn. On 24 July he wrote:

The printing had not begun. When the galleys sent to Clemens on or about 16 June and those sent on 25 June had been returned, the typesetters could have corrected the type and paged only as far as the end of chapter 12, and they would not electroplate the pages until they had been approved in [begin page 730] the form of foundry proofs. Even if Clemens had read and returned the “11 galleys of proof” sent him on 10 July (chapter 13 through the beginning of chapter 17), foundry proofs for the pages could not have reached him in the package Webster sent on 22 July. There would hardly have been enough time for him to return galleys to Webster, for Webster then to have added the picture captions, running heads, and page titles, and for the typesetters in turn to make the corrections and set the new matter. By 22 July some portion of the text following chapter 17 probably had been set in galleys: electrotypes of the illustrations were ready up through the first four pages of chapter 28 by 14 July, and the compositors had reached chapter 26 well before 7 August. But Clemens had not read these galley proofs. Moreover, the final portion of the book (at least chapters 26 through “Chapter the last”) was not yet in type. So on 26 July Webster had no difficulty agreeing to “alter the title-page.”202 At the same time he asked Clemens, “Please send back those paged proofs that I sent you as there are some corrections on them for the printer that we must have.”203 The reference to “corrections” on these foundry proofs shows that Webster’s firm had instituted its own proofreading of the pages before they were sent to Clemens. The foundry proofs clearly gave the author an opportunity to read and alter the captions and running heads, whether or not he did so. Clemens, who may already have had some of the galleys between chapters 17 and 26 on hand, reminded himself on the envelope of Webster’s 26 July letter, “Return the pages.”

By 6 August Clemens had certainly received galleys up to chapter 26, for on that day Webster transmitted a new batch containing most of chapters 26 through 29. Smarting from an impatient complaint about the history game, Webster wrote:

Because Webster had now given his attention to the captions, and the printers had all of the electrotyped illustrations for the book, it was possible for the compositors to include captions as well as pictures in the galleys—clearly impossible for the first chapters—and they evidently did [begin page 731] just that, continuing, however, to set the type in galleys (“slips”), each with three pages stacked vertically. Although Clemens had now seen the picture captions for the first twelve chapters in foundry proof, this would have been his first opportunity to edit the captions for these new chapters. He did not do so, however, for he got only as far as opening the package before losing his temper. Remembering Howells’s offer on 2 July to read the proofs, he repacked them and sent them to Howells on 7 August:

On the same day, Clemens explained to Webster what he had done:

Although he planned to be traveling the rest of the month, Howells gracefully agreed to take on the job, and he wrote Clemens on 10 August: “If I had written half as good a book as Huck Finn, I shouldn’t ask anything better than to read the proofs; even as it is I don’t. So send them on; they will always find me somewhere.”207 Two installments found him the next day in Boston, and he returned them to Webster two days later.208

Clemens, meanwhile, having skipped the galleys he packed off to Howells on 7 August, regained his composure and began to read the succeeding batch, which arrived soon after. First, though, he finished reading the galleys he had up to the beginning of chapter 26. He then read one later, out-of-sequence galley and the remainder of this newest batch, which began with chapter 30, after which he conscientiously asked Webster for the galleys he had passed on to Howells, which fell between the beginning of chapter 26 and chapter 30:

Before reading this letter, Webster received more galleys from the printers, who wrote: “Will you please return us some of the galleys you have as we are compelled to come to a stand still for want of type. With these two galleys we send you, you now have about 90 pages.”210 With so much type standing, they had exhausted the font and could not continue to set type until some of the pages had been plated, thus releasing the type. Webster sent this letter to Clemens with the note “Please send if possible,” and added:

Webster must have received the corrected galleys from Clemens on 12 or 13 August, and on 14 August he reported the arrival of the package from Howells: “I have received from Mr Howells those galleys that you wanted returned, properly corrected, so that I need not send them to you again.”212 Clemens came to the same conclusion about the galleys he had missed reading, but again complained about the printer’s proofreading:

If we calculate from the first page of the text proper in the first edition (page numbered 17), and count three pages for each galley, the “missing galleys” that Clemens never proofread included first edition pages 221–35 (“slips” or galleys 69–73), and pages 239–59 (“slips” or galleys 75–81).214 Although he probably later saw foundry proofs for these pages, they would afford less latitude for revision. Any major revision in galley proof could [begin page 733] only have been introduced on the galleys by Howells, but such independent revision is very unlikely, despite his “carte blanche.”215 Besides, Clemens’s lack of concern about the galleys indicates that he was not eager to make substantive revisions of his own, but only to insure the accuracy of transmission from printer’s copy to type. His complaint that the proofs other than the “first 2 or 3 chapters” were “never read by copy at all” shows that he expected the printer’s proofreader to make this comparison. It also suggests that he was in a position to check because the printer returned copy with the galleys, and it further suggests that he was confident Howells would read against copy. Therefore, most revision between the manuscript and the first American edition must have been made on the typescript. Moreover, except for his concern with accuracy, the new matter in the proofs (captions and running heads) did not greatly concern him.

By mid-August the typesetting had begun again. At least the first twelve chapters had been approved in foundry proof and presumably plated. With Howells’s and Clemens’s recent submissions, galleys running through chapter 30 had been returned to Webster. By now the printers had not only all the captions and running heads, but the table of contents and list of illustrations, which Webster had finished preparing the first week of August. Now able to paginate the front matter, the printers began including page numbers along with pictures, captions, running heads, and page titles in all subsequent galleys. By mid-August, too, seeing that Clemens was resigned to reading proof himself instead of depending on Howells, Webster sent him galleys for the rest of the book, and quite possibly some additional foundry proofs as well. On 23 August Webster wrote, doubtless with some relief, “I have sent you the last batch of galley proofs.”216

The fifty pages of foundry proofs which survive show very little revision.217 But the author did make a few alterations in response to queries [begin page 734] from a Webster in-house editor or proofreader who was evidently reading for sense as well as accuracy. On page 160 of these proofs, this reader circled the word “canoe” in the phrase “took the canoe” and wrote in the margin “Qy see p 129
canoe lost”—referring to the point in chapter 16 where Huck and Jim discover that “the canoe was gone.” Clemens crossed out the query and altered Huck’s “took the” to “found a,” a change that was made in the type before plating (see the illustration on page 427). On page 164, the editor circled the identification of a speaker as “Baldy” (the king) and asked whether he should in fact be identified as the duke. Clemens so altered it, and the change was also incorporated in the type before plating. On page 188, the editor questioned whether the crowd following the “long lanky man” ought to be “stooping . . . to watch him mark the places on the ground” rather than “stopping,” as it was then in type. Clemens, probably remembering that he had written “stooping” in his manuscript, made the correction (see the illustrations of the proofs below). All three corrections appeared in the first American edition, corrected in the type before plates were cast.

With the return of these finally corrected pages and also the newly typeset final portion of the book then still in galleys, which Clemens “cursed his way through” by the end of August, Webster was finally able to authorize J. J. Little to begin printing and manufacturing the prospectus. He wrote Clemens on 30 August: “The prospectuses will be ready in a week or so, now. The book is being beautifully printed and will please you.”218

As it turned out, Webster was overly optimistic: the prospectus would not be ready for another month. In the meantime, he turned to other matters. In his letter of 23 August, he had enclosed an advertisement placed by the Frank Coker News Company of Alabama which offered unauthorized copies of seven of Mark Twain’s books in paper covers for fifty cents each, noting, “Something must be done about it soon.” He had also enclosed one of the Webster company’s “new circulars to agents, with prizes for selling books on the new plan.” Evidently having second thoughts about the “new plan,” Webster reverted to the matter in his letter of 30 August, hoping finally [begin page 735] to settle the terms for canvassers and gain Clemens’s approval of his list of incentives. Describing it as “quite an important matter,” Webster enclosed for reference a portion of the circular, headed “MAGNIFICENT AND UNPARALLELED OFFER TO CANVASSERS. A CHANCE TO MAKE MONEY FOR ALL” (see Publisher’s Advertisements, 1884–1891 for a facsimile of the full circular, issued by one of Webster’s general agents). In the same letter, he explained that the American Publishing Company was refusing to cooperate in the plan to sell Tom Sawyer along with Huckleberry Finn at a reduced price.219

Clemens, who had been in the dental chair or ill a good deal of the summer, and had just “cursed” his way through the last proofs, had had enough. The season in Elmira was nearly over and he now needed to get ready for his return to Hartford, then for the speaking tour he was to begin in November. Webster’s letter provided an excuse for a salutary explosion, which is worth quoting in full:

Clearly Clemens was in no mood to participate in business decisions, still less to ponder the consequences of abandoning the plan to market Tom Sawyer and Huckleberry Finn together. Although the original motive for omitting the raft episode from Huckleberry Finn was now moot, to restore the episode at this time would require having new pictures drawn and plated; new type set and proofread; new captions, page titles, and contents written and typeset; the illustrated chapter headings for each succeeding chapter altered to reflect new chapter numbers; and the plates for the remainder of the book newly made or altered to conform with the new chapter headings and new foliation. In short, to restore it would halt production, reimmerse everyone in book-making details, delay publication for some weeks, and cost a good deal of money. Even if restoring the episode crossed Clemens’s mind, he must have dismissed it as unfeasible. Webster was eager to get the prospectuses in hand and the printing and binding of the book underway so that he could leave on his cross-country trip to contract with general agents. If he remembered the raft episode, he restrained any impulse to mention it (“you want to think twice before you send me irritating news”).221

[begin page 737]

In fact, before Webster left on his trip he hoped to have, in addition [begin page 738] to the prospectuses, a complete set of printed sheets for the book. His accounts for August and September are filled with amounts paid for items necessary for printing, production, and sales.222 Hoping to settle other outstanding matters, he wrote Clemens on 1 September: “Isn’t it about time we began to think about copyrighting the new book? Shall I attend to it or are you going to?” Assuring Clemens again that the book was being “beautifully printed,” Webster promised to send him “a set of sheets” as soon as possible.223 By 7 August Webster already had a “dummy” (that is, blank sheets equal in number to the leaves of the book) bound, which Clemens had seen, but the promise of true sheets was still not fulfilled by 20 September, when Clemens wrote requesting “an unbound copy of Huck Finn.” The copyright application was not submitted until 3 December.224

Although he had not suggested pulling apart the book to restore the raft episode, Clemens did have a new idea that might cause a delay—a second frontispiece. A young sculptor, Karl Gerhardt (1853–1940), recently arrived back in the United States after three years of art studies in Europe supported by the Clemenses, had made a clay bust of him in Elmira and cast it in plaster, and the author proposed to include a picture of it in the book. He wrote to Webster on 8 September, enclosing a photograph taken by Elisha M. Van Aken of Elmira:

[begin page 739] Although Webster could hardly have welcomed a delay, he had no choice but to investigate, and his response, written on 13 September, was all Clemens could have hoped for:

Clemens replied on 15 September, reminding Webster, “Be sure & attach the words ‘From the bust by Karl Gerhardt.’ ” By 17 September Webster had “made a contract with the Heliotype people,” and two days later he sent “by express a bottle of Heliotype ink” to Elmira for Mark Twain to use in inscribing his signature, a facsimile of which was to be used as the caption to the photograph.227 The ink was delayed but evidently caught up with Clemens in Elmira before 23 September, or soon thereafter in New York or Hartford.228 Webster paid J. J. Little and Company sixty-nine dollars for “ptg. and electrotyping pages for 1000 prospectuses” on 23 September, but the new frontispiece, which had to be tipped into each prospectus or book, could not have been ready for several days.229 The author’s heliotyped signature appeared on the frontispiece along with the picture of the bust and his proposed caption identifying the sculptor.230

[begin page 740] Meanwhile, on 19 September, about three weeks after printing began, Webster reported that he had sent “advance sheets” to Chatto and Windus, Clemens’s English publisher, and that he was “getting up a set of plates the same as ours” for Dawson Brothers, Clemens’s Canadian publisher.231 This report may mean that the printers had already completed a set of electrotype plates, and even perhaps that imposed pages (or folded and gathered sheets of each signature) for the entire book were ready. That seems to be what Clemens himself assumed when he replied on 20 September that he wanted “an unbound copy of Huck Finn” sent. The surviving Webster company accounts show no printer’s bill and therefore do not specify a completion date, or spell out charges for plating and printing. But at least one complete set of printing plates must have been included in the printer’s contract, and that set had doubtless been manufactured in sequential batches since the middle of August, when revised foundry proofs were returned. Collation shows that the sample pages in even the earliest prospectuses must have been printed from plates rather than from standing type. (The prospectuses also contain material, such as advertising, not included in the book.) These prospectus plates, which were typically duplicated from plates already made for the book, allowed simultaneous printing and ensured that the plates of pages selected for the prospectus would not undergo unequal wear. It follows that the book must have been plated by 23 September, the date of the bill for “ptg. and electrotyping” of the prospectus.

The corrections that Clemens marked on the foundry proofs were made in the standing type before plates were made: both the prospectus and book have the correct reading on page 164 (“the duke” instead of “Baldy”; 163.23 in this edition). One other correction, which had been overlooked in proof, was made after the first printing of the prospectus: the incorrect reading “base” was altered to “race” on page 174 (173.4 in this edition). The corrected reading appeared in even the earliest impressions of the book and in the late impressions of the prospectus. The implication is that [begin page 741] the error was first noted in the prospectus and immediately brought to the attention of the printers (in late September or early October), who were able to correct and re-plate the book page. Three other errors were not corrected until at least November, after the first impression of the book was completed (and after the duplicate plates for the Canadian issue were made): “Decided” was changed to “Decides” in the table of contents (xxxv.13 in this edition); the incorrect page “88” was changed to “87” in the list of illustrations; and “was” was changed to “saw” in the text on page 57 (41.13 in this edition). In addition, sometime after the prospectus and a portion of the first impression were printed and bound, the date on the copyright page was changed from 1885 to 1884 to reflect the earlier submission of the book for copyright. Between the initial printing and 1891, all other changes in the plates were to correct type batter or plate wear, with one exception. A discovery in November 1884 that one of the illustrations had been tampered with to make it obscene caused a halt in production and a flurry of corrective activity.

At the time the offending illustration was first noticed, Webster was away in the West, contracting with book agents. When he left in late September (he had already made brief trips to see some general agents in Washington, D.C., Baltimore, and Philadelphia earlier in the month), he was confident that he had indeed “engineered the book through all its critical periods, and everything is in working order.”232 The prospectus would soon be in print with the new frontispiece tipped in, ready for delivery to the general agents; the advertising campaign was begun; the New York canvassers were ready to begin their work; and the first impression of the book may have been completed—some copies would be bound before the middle of November. Perhaps as early as the first week of November, when he was in San Francisco, Webster was informed of the discovery of the obscene illustration in the prospectus. The picture of Uncle Silas on first-edition page 283 (page 281 in this edition) had been altered so that it appeared his genitals were exposed.233 Webster later said that no more than [begin page 742] 250 copies had “left the office . . . before the mistake was discovered.” According to the New York Tribune, he “telegraphed at once to this city and had the publication stopped.” The World reported that “the prospectuses were called in. The page containing the cut was torn from the book, a new and perfect illustration being substituted.”234 According to Merle Johnson, who interviewed the printer, J. J. Little, only pages torn from the prospectuses were called in (and salesmen faced dismissal unless they complied). Johnson’s version was most likely correct—return of the entire prospectus would have stopped the canvass, whereas removing the page allowed it to continue.235 Furthermore, from his talk with Little, Johnson deduced that “thousands of copies of the book were in the plant at the time the marred plate was discovered . . .; that a new sheet was run off with a reëngraved plate to eliminate the damage, and that these sheets i.e., pages from the sheets were tipped in, using the stub of the excised page. In the unbound copies whole signatures were printed and supplied in the regular manner of binding.” The bibliographical evidence fully supports this account.236

Webster’s search for the culprit led him to suspect that the damage had occurred after the entire page (with type and illustration) was plated but before printing had begun. On 28 November he explained to a reporter from the New York Tribune why he believed the plate had been altered at that stage:

The first Canadian issue of Huckleberry Finn, printed from a duplicate set of the Webster company plates and published by Dawson Brothers on 10 December 1884, shows the illustration in its repaired state. The first English edition, published by Chatto and Windus on 10 December 1884, likewise shows the repaired illustration. Clemens made no changes to the text of either the Canadian issue or the English edition.

The first American edition’s official publication date was 18 February 1885. The first impression of 30,000 copies and most of a second impression of 10,000 copies were exhausted in less than a month. On 14 March 1885 Webster reported that he had ordered “paper for 10000 more books and shall print them right away, this will make 50,000 printed.”238 The book sold steadily thereafter; Webster continued to have the printers run off new impressions of the book as needed. Signatures from these impressions were in turn given to the binders, who produced new books essentially on demand. Because not all signatures from the earlier impressions were bound into books before the new impressions were printed, unbound [begin page 744] signatures from early impressions shared warehouse space with later ones. This procedure resulted in the binders’ assembling books with signatures from earlier and later impressions, which for over a hundred years has caused consternation to bibliographers comparing “points” in an attempt to determine the earliest copies of the first edition.239 The last known issue of the first edition was dated on the title page 1891. No evidence of any authorial change has been found after the first impression was printed.

Before the first American edition was published, three excerpts from Huckleberry Finn appeared in the Century Illustrated Monthly Magazine of December 1884, and January and February 1885. In the mid-1880s the Century enjoyed an extraordinary expansion of its popularity and influence.240 In January 1885 its circulation was estimated at 180,000 copies per month, with some issues reprinted several times to meet subsequent demand; in April 1885 the cover advertised a “first edition” of 225,000 copies.241 According to Arthur John, during the next ten years the magazine “reached a pinnacle of prestige and influence unprecedented in American magazine history.”242

The excerpts from Huckleberry Finn were edited by Richard Watson Gilder, probably with the help of one or both of his Century colleagues—Robert Underwood Johnson, first associate editor, and C. C. Buel, second assistant editor.243 In 1932 Bernard DeVoto characterized Gilder as a liberal [begin page 745] “civilizing force” whose “intelligence, courage, and integrity are beyond question,” but asserted that his editing of Huckleberry Finn for the Century, especially his suppressions of parts of the text, expressed contemporary “standards of gentility.” Critics since DeVoto have made much of Gilder’s editing, but until 1988, when the first Mark Twain Project scholarly edition of Huckleberry Finn was published, the relationship of the magazine selections to the book was not clearly understood, partly because the magazine selections were published before the book. Without ascertaining what the Century’s setting copy had been, DeVoto assumed that Mark Twain acquiesced in the editing, but subsequently “restored his own text when the book appeared.”244 In 1955 Arthur L. Scott thought that Gilder had made his selections “from Mark Twain’s manuscript,” and that the Century’s editing was of little consequence to the author, since he was simultaneously publishing his book (presumably from another “manuscript” or setting copy). In 1970 Herbert F. Smith, noting that the first English edition had been published in December 1884, wrote that “Gilder was working from the manuscript of a novel already in print! . . . It is even possible, since there is no evidence either for or against it, that Gilder worked from a galley proof of the novel. Such an arrangement served as a [begin page 746] constant reminder to the editor, if he needed one, that whatever was published in the magazine would have no influence on the text of the book.”245

In fact, the Century excerpts were not set from manuscript or from galley proof but, as Gilder’s letters show, from final book pages—most likely folded and gathered sheets (unbound signatures), marked to incorporate Gilder’s changes. Although Gilder or his editors proposed the major omissions and changes, Clemens evidently saw proof of what they proposed to publish, because the Century text exhibits some evidence of authorial change. There was no subsequent need to “restore” readings to the first American edition since they had not been affected by the Century revision: the first impression of the book was already printed (even though not formally published) before the last two episodes appeared. The author willingly acquiesced in the extraction and editing of his text for a magazine audience.

Clemens seems not to have pursued his idea of serializing Huckleberry Finn in the Century or the New York Sun after speaking with the editors, Gilder and Charles A. Dana, respectively, in September 1883. Both editors must have expressed their desire to publish something by Mark Twain, however, and in August, Clemens did give the Sun a brief sketch, but insisted it be published anonymously.246 Gilder, however, had not forgotten Huckleberry Finn. Perhaps in late September 1884 when Clemens stopped off in New York on the way to Hartford from Elmira, Gilder approached him, hoping to revive the idea of publication in the Century in advance of the book. Clemens gave permission for a short excerpt, and probably gave Gilder a set of folded and gathered sheets from which to choose it.247 Gilder had read at least through chapter 18 by 10 October, when he wrote asking for a much longer excerpt than Clemens had offered:

In his eagerness to secure a large portion (“half or three quarters”) of Huckleberry Finn for the Century, Gilder was replying to objections Clemens had already expressed, and also trying to forestall any new ones that might arise from his offer. Clemens’s “rule” must have been that he would not allow periodical publication of a significant portion of a book to precede book publication: a short extract might awaken interest, but a longer one would undercut book sales. Although Gilder began by asking him to break this “rule” by delaying book publication, four paragraphs later he offered to let publication by subscription precede or coincide with publication in the Century.249 In the remainder of the letter he set out some of the criteria that would guide him in editing the text for a Century audience:

What Gilder meant by the “extra sheets” is uncertain. If he had been given only a partial set of folded and gathered sheets, he might have meant that the remainder had arrived. More likely, however, he had been given a complete set, and had requested a second. (Since the book pages were printed on both sides, two sets would be needed to allow him to cut and paste copy for the Century typesetter.)

Clemens replied in a telegram (not known to survive) on 10 or 11 October. Apparently he was not upset by Gilder’s proposed editing of Huckleberry Finn for the Century. His major concern seems to have been financial—what effect this partial magazine publication would have on sales of the book. Despite his “rule,” he apparently agreed at least to experiment, [begin page 749] at his “former” rate of pay. He evidently did not suggest a way to edit the feud chapter as a self-contained episode.

Gilder, who had preferred the feud episode for the December issue if allowed only one excerpt, was now faced with selecting from the whole narrative. He therefore dropped the idea of using Huckleberry Finn in the December issue, and wrote Clemens on 11 October about a new plan:

Soon after Clemens sent his telegram, however, he had second thoughts. On 11 or 12 October he wrote Gilder a long letter (also lost) explaining his change of heart. Presumably worried that extended selections in more than one issue of the Century would look like full serialization and therefore undercut book sales, he withdrew his consent to the experiment. He apparently renewed his offer of a single episode, of any length, before book publication. In response, Gilder returned to his original idea of using the feud episode for the supposedly “closed up” December issue. He devised a way to edit it that resolved his earlier worry about dovetailing it in “without making it a mere extract from the book.” On 13 October he wrote Clemens:

Gilder’s solution to the problem of editing the feud chapters for the Century turned out to be a simple one after all. To set the scene he chose the opening paragraphs of chapter 19, describing Jim and Huck’s life on the river, and then planned to use the end of chapter 16, and chapters 17 and 18 in full. His letter shows that he (or his deputy on the staff) made the editorial choices assuming Clemens’s consent, but without his guidance. Gilder was evidently primarily concerned with making the episode coherent to a magazine reader who was not familiar with the earlier part of the book, and for that purpose he omitted the “paragraph about the snakeskin”: “Anybody that don’t believe, yet, that it’s foolishness to handle a snake-skin, after all that that snake-skin done for us, will believe it now, if they read on and see what more it done for us” (130.4–6). Clemens supplied a paragraph identifying Huck and Jim, and by 17 October it had been typeset, having undergone at least two minor modifications by Gilder. It appeared in the December Century as follows:

Gilder had altered the opening sentence of the paragraph and made a few more cuts in the story:

[begin page 751] Gilder’s letter evidently put an end to negotiations, at least temporarily. Neither he nor Clemens expected further episodes from the book in the Century. The December issue went to press and was later distributed with no announcement of further episodes from Huckleberry Finn for January.259

Although no further correspondence about the December episode has been found, it is likely that Clemens saw not only “the first page . . . sent to press,” but other proofs as well. It was customary for the Century to send author’s proofs,260 and only time constraints might have forced Gilder to abandon his usual practice. If Clemens did see proof for this episode, however, he made little or no detectable alteration to it. Collation of the December Century against the first edition reveals only twenty-seven small substantive changes (and changes in accidentals affecting dialect) other than those already mentioned by Gilder. Three of them, while possibly authorial, do not constitute enough of a case to establish Clemens’s intervention. The change from “hands” to “hand” at 133.33 could as easily have been made by an editor or compositor. Similarly, the change from “was” to “is” at 143.33 (in which the substitute reading is clearly inferior to the original) is most likely a compositor’s error. The change from “him” to “it” at 142.28 (making “no frivolishness” refer only to the Colonel’s mahogany cane rather than to the Colonel himself) so trivializes the meaning that it is unlikely to be authorial. The remaining changes are all attributable to editorial intervention: the process of excerpting, the imposition of house style, and the sophistication of nonstandard grammar and dialect to more standard usage.261 Consistency of dialect spelling was also imposed, not only by the Century editors but by the proofreaders and compositors in the De Vinne print shop, which produced the magazine. Theodore De Vinne’s manual asserted that dialect “must be made uniform in its spelling, even if it is irregular in copy. Different abbreviations or clippings of the same word by the same speaker or writer should not be [begin page 752] passed.”262 The Century changes in dialect reflect this instruction: whenever dialect forms were corrected, they were made to match nearby spellings of the same word (for instance, the change from “ ’m” to “ ’em” at 149.31, and five changes from “ ’um” or “ ’m” to “um” at 150.34–151.2).263

Two additional extracts from the book did subsequently appear in the Century. Just when Clemens gave his consent for them remains uncertain. Negotiations may have been reopened as early as October when Gilder sent the proofs of the December episode; an agreement had certainly been reached by 9 November, when Clemens wrote Webster from Providence, where he was performing on his “Twins of Genius” speaking tour with George Washington Cable: “Gilder of the Century said to me, ‘We are not only indebted to you for a good chapter for our next number, but are profoundly indebted to you for unearthing a gem of an artist for us.’ ” But the selection of material probably did not occur until later in the month, when Gilder came to one of Clemens’s New York readings at Chickering Hall and talked with him afterwards.264 Clemens read “King Sollermun,” from chapter 14 of Huckleberry Finn, at the evening performance on the eighteenth, and at the matinee on the nineteenth, noting on his own program that he added “de Bank” from chapter 8 as an encore both times.265 These two dialogues between Huck and Jim make up the second (January) episode in the Century, and Gilder’s hearing them read aloud may well have prompted their selection. In any case, the decision about content must have occurred well before the end of November in order for the episode to be included in the January issue. Clemens and Gilder evidently agreed on a third extract at the same time, for the February episode was announced in the January Century.266

[begin page 753]

Clemens did not furnish a prefatory note for the January episode,267 but unlike the December one, it shows clear evidence of authorial revision. Jim’s summing-up remark at 57.14–15 (“I wisht I had de money, I wouldn’ want no mo’ ” in the first American edition) is replaced by “But live stock’s [begin page 754] too resky, Huck—I wisht I had de eight hund’d dollars en somebody else had de nigger,” a change only Clemens could have made.268 Out of ten other substantive or dialect changes, only one seems likely to be Clemens’s: a revision from Jim’s “old” in the first American edition to “ole” in the Century.269 The Century editors, who had made an effort in the December episode to substitute the house spelling “gwine” for Clemens’s “gwyne,” seem to have nearly given up the effort in the January episode, substituting “gwine” for “gwyne” only two out of ten times. They did, however, pull the apostrophes from Clemens’s “ain’t” and “hain’t,” each of which occurs once. Either Clemens objected to the respelling or the effort was finally given up as unproductive, for in the next episode they changed neither “gwyne” nor the apostrophes.

The February episode, entitled “Royalty on the Mississippi: As Chronicled by Huckleberry Finn,” was by far the longest of the three extracts and contained the most substantive change. Comprising the greater part of chapters 19–28, with an additional bit from chapter 29, it began with an editorial note to place the characters: “See The Century for December and January. The negro Jim is escaping on a raft from slavery in Missouri, and Huck Finn is running away from a drunken and cruel father.—Ed.”270 In all, there were 107 substantive changes (and accidental changes affecting dialect), more than half of them simple omissions from the text. Analysis of their content—especially that of the omissions—suggests strongly that they originated with Gilder or the Century editors. In his letter of 10 October, Gilder had promised not to “mutilate” Mark Twain’s book, but proposed three categories of change: (1) alteration of material “not adapted to our audience”; (2) omissions on account of space; and (3) omissions determined by both space and suitability. Alterations in the February episode for the most part fit these categories. Those specifically aimed at the Century audience include, as DeVoto, Arthur L. Scott, and Herbert F. Smith noted, not only corrections of Huck’s grammar and softening of “coarse” expressions, but deletion of descriptive passages or incidents that might offend refined taste: Huck’s statement that he and Jim were “always naked, day and night, whenever the mosquitoes would let us” (157.37–38); the preacher’s wild address and the king kissing the girls at the camp meeting (172.5–16, 173.13–17); the “signs of a dead cat” (198.7) and other references to smells; and the matter-of-fact references to Peter [begin page 755] Wilks’s corpse and coffin (230.20–24).271 Also in this category are the Century’s omission of the “biggest line of all” on the Royal Nonesuch handbill, “LADIES AND CHILDREN NOT ADMITTED,” and the duke’s comment, “There, . . . if that line don’t fetch them, I don’t know Arkansaw!” (195.6–9).

Gilder’s concern about extracting the feud episode—“it may be somewhat awkward to select as the story runs in and out”—must have also applied to the episodic nature of the ten-chapter selection about the king and the duke. Omissions for space were of particular necessity here. Thus he dropped passages that do not advance the plot: the death of Boggs; Huck’s visit to the circus (chapters 21, 22); Jim’s story about his daughter ’Lizabeth (chapters 23); Huck and Joanna’s discussion of life in England (chapter 26); and the description of the undertaker (chapter 27). “Suitability” must also have played a part in the selection of some of these passages. Of course there were other reasons to cut, among them relevance—a passage might lose its meaning when extracted from the context of the book: thus the omission of the snakeskin reference in the December extract, and of the reference to Pap (at 165.15–17) in the February extract.272

Gilder’s responsibility for these changes is confirmed by his 1886 reply to a letter from a Century reader, a “superintendent of public schools in a distant part of the West,” who criticized the magazine for publishing its selections from Huckleberry Finn:

Clemens apparently did give his “full consent” to the editing of Huckleberry Finn for a magazine audience. Although no letter about Huckleberry Finn survives, the author’s letters to the Century editors about other works he published there show that he expected such editing for magazine publication as a matter of course, at times even asked for and welcomed it. In August 1885 he wrote one of the editors before submission of his next contribution, “The Private History of a Campaign That Failed,” commissioned by the Century:

The editing of the Century episodes of Huckleberry Finn in almost every instance originated with the editors and was a requirement of the magazine’s format and its implicit contract with its audience.

Although the raft episode as published in Life on the Mississippi in 1883 drew some scattered comment by reviewers, the extracts from Huckleberry Finn published in the Century were the first to reach a general audience and to elicit published (and unpublished) comment on the book. Clemens was on his speaking tour with George Washington Cable, reading selections from his book, when the first Century extracts appeared. Although most of the newspaper notices of the Century were bland or perfunctory, Clemens may have been slightly alarmed at the tone of others, especially since these came on top of the bad publicity he was getting for suing a Boston bookseller who had offered Huckleberry Finn for sale before the Webster company had even completed its canvass.275 On 1 February 1885 the Boston Herald, in reviewing the February Century (with the episode about the king and the duke, “Royalty on the Mississippi”), wrote that it “has a trifle of ‘too muchness of that sort of thing,’ which is the prevailing [begin page 758] characteristic of this kind of writing. It is pitched in but one key, and that is the key of a vulgar and abhorrent life.”276

After publication of The Gilded Age in 1873, Clemens had become convinced by the persistence of an unfair accusation printed in an early unfavorable review that “the latest review of a book was pretty sure to be just a reflection of the earliest review of it. That whatever the first reviewer found to praise or censure in the book would be repeated in the latest reviewer’s report, with nothing fresh added.”277 He therefore tried to stage-manage the reviews of Huckleberry Finn by making sure that the first were favorable and published in influential journals. On 23 January, more than a month after the first English edition of Huckleberry Finn was published and three weeks before the American publication, he began to issue instructions to Webster about review copies: “A day or two after the book issues, you want to send a cloth copy to the prominent journals & magazines of the country.—but none to the N.Y. Perhaps you better send to the prominent magazinesnow (with unbound copies to make extracts from.)”278 Three days later he repeated the gist of this instruction, but expressed the hope that Gilder could “review it in next Century.”279 The next day, 27 January, he refined this strategy: “The following is a positive order: Send no copy of the book to any newspaper until after the ser Century or the Atlantic shall have reviewed it. I make an exception in New York.” He added: “What we want is a favorable review, by an authority—then immediately distribute the book among the press.”280 When Webster replied that Gilder could not get a review into the March issue of the Century or even guarantee one for April, Clemens began weighing still another plan, “sending out 300 press copies early—say Feb. 23d—without waiting for the magazines—Heavens & earth! the book ought to have been reviewed in the March Century & Atlantic!—how have we been dull enough to go & overlook that? It is an irreparable blunder.”281 On 10 February, just eight days before the official publication of the first American edition, Clemens settled on what proved to be the final plan, which Webster followed. It began by singling out three powerful New York dailies and one venerable weekly magazine:

Webster responded on 14 February, reminding Clemens that “you told me in the start that press notices hurt the last book before it was out & that this year we would send none until the book was out.” He assured Clemens that he had sent off “copies (bound & unbound)” to the newspapers and journals listed in Clemens’s letter. Although it seems unlikely that Webster followed them with the “300 press copies,” most of the named publications did review the book, as did numerous others, which probably received their review copies from local agents.283 Clemens’s English publisher, Chatto and Windus, also apparently sent out review copies.284 In addition to the formal reviews in English and American journals, Clemens received a virtual flood of mail from family and friends—including many fellow writers—to whom he had sent complimentary copies of the book.

The English reviews were for the most part favorable. An unsigned review in the Athenaeum, probably by William Ernest Henley, found the book to be “Mark Twain at his best”: “For some time past Mr. Clemens has been carried away by the ambition of seriousness and fine writing. In ‘Huckleberry Finn’ he returns to his right mind, and is again the Mark Twain of old time. . . . Jim and Huckleberry are real creations.”285 The American critic Brander Matthews, writing in the London Saturday Review, noted the technical achievement of the book: “The skill with which the character of Huck Finn is maintained is marvellous. We see everything through his eyes—and they are his eyes and not a pair of Mark Twain’s spectacles. . . . one of the most artistic things in Huckleberry Finn is the sober self-restraint with which Mr. Clemens lets Huck Finn set down, without any comment at all, scenes which would have afforded the ordinary writer matter for endless moral and political and sociological [begin page 760] disquisition. . . . In Tom Sawyer we saw Huckleberry Finn from the outside; in the present volume we see him from the inside. . . . Jim is an admirably drawn character.”286

Friends and family and many early reviewers easily recognized and praised the “real” and “documentary” character of the book—that is, the real people behind the fictional characters, the real speech and manner and popular culture of Hannibal and of the Mississippi Valley generally—and they remarked on the moral of its plot, doubtless provoked by the disclaimer in Mark Twain’s “Notice.” On 17 January 1885 Clemens’s sister-in-law wrote from Keokuk, Iowa: “Sam I have just finished Huck Finn. It simply amazes me to see how you kept up the dialects and the underlying moral lesson without a particle of apparent effort. It is real, to me.”287 John Milton Hay, Lincoln’s former private secretary and a long-time friend and admirer of Clemens’s, wrote to him after publication:

Some of the earliest reviews took the same tone. The Hartford Times praised the novel for “the fidelity with which it paints the characters and the scenes” it depicts and said that it “is a good book, and it does teach a certain moral, notwithstanding the author’s disclaimer; it teaches, without seeming to do it, the virtue of honest simplicity, directness, truth.”289 The Hartford Courant review (probably written by Charles Dudley Warner, or possibly by Charles Hopkins Clark) said:

The Courant’s review was one of several to praise the psychological insight in the portrayal of Huck’s struggles with his conscience:

The New York Sun reviewer agreed. In describing the incident in chapter 16 where Huck nearly turns Jim in, he noted that “Huck’s moral nature began to experience a singular reawakening. A conscience that was sufficiently elastic on the subject of mendacity, and that never kicked when Huck stole chickens or watermelons, . . . was strongly agitated by the thought that here he was helping a slave to escape to freedom.” After quoting the episode from chapter 31, where Huck determines to turn Jim in and then finds he cannot pray because “You can’t pray a lie,” the reviewer wrote: “Although this seems like an audacious burlesque of religious sentiment, reaching quite to the limits of the permissible, the reflections attributed to Huckleberry on the enormity of his transgression are probably as true as anything else in the book to the Missouri creed of forty years ago.”291

But other critics reacted to the book with hostility and contempt. Where the positive reviews had seen “realism,” they saw “coarseness,” “bad taste,” and “grotesqueness.” Where the positive reviews had seen the conflict between conscience and training as a serious and sophisticated “study of human nature,” they saw “irreverence” and immorality, and questioned the appropriateness of the book for young readers. The New York World subtitled its review, “ ‘Humor’ of a Very Low Order—Wit and [begin page 762] Literary Ability Wasted on a Pitiable Exhibition of Irreverence and Vulgarity,” and called the book “cheap and pernicious stuff”:

The Boston Advertiser similarly charged the book with vulgarity and irreverence: “Here and there are snatches of Mark Twain’s best work, which could be read over and over again, and yet bring each time an outburst of laughter; but one cannot have the book long in his hands without being tempted to regret that the author should so often have laid himself open to the charge of coarseness and bad taste.”293

Reviews from the newspapers in San Francisco tended to echo these same concerns. The Examiner wrote:

The San Francisco Bulletin raised the question of whether the book was appropriate for children: “The author starts out by telling his juvenile readers that there are some lies in his book—that most people lie, and that it is not very bad after all. Of course the warning is timely that persons attempting to seek a moral in the story should be banished.” Although the reviewer recognized Mark Twain’s “genuine” vein of wit, he concluded that “there is very little of literary art in the story.”295 The San Francisco Chronicle objected to these criticisms:

Clemens continued to get personal letters praising his book. William Livingston Alden, a columnist and editorial writer on the New York Times, who had written on 28 February to say that he had enjoyed the Century episodes “more than I ever enjoyed any magazine articles anywhere,” wrote him again on 15 March: “I have just read Huck through in course. It is the best book ever written.”297

In the middle of March 1885, the Concord, Massachusetts, Library Committee decided not to circulate a copy of Huckleberry Finn that had been ordered for the collection, and the story of this rejection, widely published in the newspapers, set off a debate about the book, with reactions appearing in hundreds of newspapers all across the country. One of the most complete stories, quoting members of the committee, was published in the St. Louis Globe-Democrat of 17 March:

Clemens was at first unruffled by the controversy. He wrote to Webster on 18 March: “The Committee . . . have given us a rattling tip-top puff which will go into every paper in the country. They have expelled Huck [begin page 764] from their library as ‘trash & suitable only for the slums.’ That will sell 25,000 copies for us, sure.”299

Although some critics found the objections of the committee easy to mock, others did not. For them, the question then became whether the book was truly “very coarse,” of a “low grade of morality,” and “trash of the veriest sort.” The Boston papers agreed almost entirely with the committee, and the Springfield (Mass.) Republican wrote:

As Clemens followed the controversy in the newspapers over the following weeks, he lost his equanimity. He became so furious at the Springfield Republican and Boston Advertiser for what he saw as a personal attack on him as well as his book that he wrote a “Prefatory Remark” for future editions of Huckleberry Finn, and wanted to send copies of the altered book to “all the New York & Boston papers, & to a scattering few western ones.”301 It stated:

He had dropped the idea by the next day, when he wrote Webster: “Livy forbids the ‘Prefatory Remark’—therefore, put it in the fire.”303

Across the country, many defenses of the book were published. One of [begin page 765] them, a sober but insightful editorial response to the controversy, appeared in the San Francisco Chronicle (which had already reviewed the book) on 29 March:

Clemens’s sister, Pamela, then living in California, probably sent him the San Francisco Chronicle editorial and asked him about it. On 15 April he responded, again dismissing the library controversy: “The Chronicle understands the book—those idiots in Concord are not a court of last [begin page 766] resort, & I am not disturbed by their moral gymnastics. No other book of mine has sold so many copies within 2 months after issue as this one has done.”305

Joel Chandler Harris wrote privately to Clemens that Huckleberry Finn’s “value as a picture of life and as a study in philology will yet come to be recognized by those whose recognition is worth anything. It is the most original contribution that has yet been made to American literature.”306 And later Harris wrote for publication that “there is not in our fictive literature a more wholesome book than ‘Huckleberry Finn.’ It is history, it is romance, it is life. Here . . . we see people growing and living; we laugh at their humor, share their griefs; and . . . we are taught the lesson of honesty, justice and mercy.”307 Clemens thanked him “for the good word about Huck, that abused child of mine who has had so much unfair mud flung at him. Somehow I can’t help believing in him, & it’s a great refreshment to my faith to have a man back me up who has been where such boys live, & knows what he is talking about.”308 Franklin Benjamin Sanborn, one of the founders of the Concord School of Philosophy and an advocate of social reform who helped to organize the National Prison Association, published one of the most eloquent defenses of the book in the Springfield Republican:

Although Clemens did not respond publicly to the criticism of his book at the time, he was clearly disturbed by the charges of immorality that had been leveled against it. On his 1895–96 lecture tour, he did provide an indirect public response in the introduction to one of his reading selections from Huckleberry Finn, “Small-pox & a lie save Jim” (Huck’s struggle with his conscience in chapter 16). In his draft of the passage he laid out his own interpretation of the central conflict in his book, saying that “in a crucial moral emergency a sound heart is a safer guide than an ill-trained conscience”:

By 1896, when the third American edition was published, most critical opposition to the book had apparently evaporated. Critics began comparing Mark Twain to the “classic” authors, and they regularly used superlatives in descriptions of the book. Punch called it “a bit of the most genuine and incisive humour ever printed,” a “great book,” and a “Homeric book—for Homeric it is in the true sense, as no other English book is, that I know of.”311 The Critic called it a “masterpiece” which the editors (Jeannette [begin page 768] and Joseph Gilder) read again “with even more zest and appreciation than before.”312 In 1897, Brander Matthews compared Mark Twain to Cervantes “in that he makes us laugh first and think afterwards”:

In the twentieth century, the book’s reputation continued to grow among writers and critics. In 1900 William Archer wrote in the London Morning Leader, in an essay on the subject of moral parables in contemporary literature, about the passage in chapter 31 where Huck struggles with his conscience:

Clemens called Archer’s article “compact & virile,” and commented:

In 1902 the book was adapted into a play by Lee Arthur, who was commissioned by the producer Charles B. Dillingham. In early August, Clemens read the script (based on both Tom Sawyer and Huckleberry Finn, but true to neither), which prompted him to suggest to Dillingham switching the names of the boys to show something of their relationship in Tom [begin page 769] Sawyer, since he apparently found neither boy true to character: “You see, the public will expect to see these two boys as they familiarly know them: Tom Sawyer, ostentatiously smart & inventive & always boss; Huck Finn, humble, timid, ignorant, uninventive, Tom’s willing slave and enthusiastic admirer.”317 Dillingham apparently made little or no change to the play, but later invited Clemens to a rehearsal, which Clemens attended, after which he wrote performance notes for the lead actor who played Huck:318

The play sentimentalized Huck and gave him a girlfriend, Amy Lawrence. According to one of the actors, Walter C. Kelly, it also included “a chorus of forty white-satined pierrots assisting Huck Finn to sing a march song entitled ‘I want to be a drummer in the band.’ ”320 It had its premier performance on 11 November 1902 in Hartford, where it ran for five days. Although the Hartford and New York newspapers were for the most part complimentary, the New York Dramatic-Mirror called it a “dreadful fiasco.”321 According to Kelly, on opening night, “Slowly the proof of this assault on this classic dawned upon the audience and the final curtain fell to complete silence, and a polite but angry audience made their way into the night.”322 After twenty-five performances in Philadelphia, where it was badly reviewed, it closed in Baltimore.323

The play’s failure had no apparent effect on the book’s reputation. In a 1907 letter to Clemens, George Bernard Shaw recounted a visit to William Morris:

But questions continued to be raised locally about the novel: its irreverence, its appropriateness for children, and its morality, as it came under scrutiny by library boards and religious organizations. In August 1902 it was “excluded” from the “list of books for boys”325 by the Denver Public Library, the first such banning since 1885, according to Clemens. In a letter to the Denver Post he ascribed the banning in part to local supporters of General Frederick Funston, a “hero” of the Spanish American War about whom he had recently written a highly uncomplimentary article.326 After explaining that his wife was seriously ill, Clemens wrote:

[begin page 771] Although the Denver ban was lifted even before Clemens’s letter was published in the Post, within days Huckleberry Finn was apparently banned again, this time in Omaha, Nebraska.328 Six weeks later Clemens received a letter of support from Gertrude Swain, a twelve-year-old from Greeley, Nebraska, clearly written in response to newspaper reports of the ban and public condemnation of the book by a minister: “I have read Huck Finn, about fifty times. Papa calls it my Bible, I think it is the best book ever written, and I don’t think it would hurt any little boy or girl to read it. I think it would do lots of them a lot of good. I don’t think that preacher knew what he was talking about.”329 Clemens responded, “I would rather have your judgment of the moral quality of the Huck Finn book, after your fifty re-readings of it, than that of fifty clergymen after reading it once apiece. I should have confidence in your moral vision, but not so much in theirs, because it is limited in the matter of distance, & is pretty often out of focus.”330

In 1906 news of yet another “banning” circulated in the press. Although Huckleberry Finn had long been on the shelves of the adult fiction section of the Brooklyn Public Library, in November 1905 the superintendent of the Children’s Department, “a conscientious and enthusiastic young woman,” ordered it removed from the children’s shelves. A Brooklyn branch librarian, Asa Don Dickinson, “begged” his colleagues to reinstate it, but the children’s librarians refused, responding “in effect that Huck was a deceitful boy; that he not only itched but scratched; and that he said sweat when he should have said perspiration.”331 Dickinson wrote Clemens of the deliberations and asked for his help. Clemens responded privately:

Dickinson showed Clemens’s letter to the children’s librarians, but it did not convince them to change their minds—instead they decided to suppress the letter and to exclude The Adventures of Tom Sawyer as well. In March 1906 the Brooklyn Public Library issued “an order withholding Mark Twain’s ‘Huckleberry Finn’ and ‘Tom Sawyer’ from children considered by the library authorities to be under the age of discretion.”333

In the decades after Clemens’s death in 1910, critics began to argue successfully for Huckleberry Finn’s status as literature worthy of being taught and studied in the academy. The book became what Clemens would never have predicted for it—assigned reading in colleges and in secondary schools. Perhaps in part because of this new association of the book with secondary education (which was concurrent with efforts which began in the 1950s to integrate the schools nationwide), the challenges to it increased in the 1950s. In addition to the objections mounted by religious fundamentalists, who disliked the book’s evident skepticism towards religious doctrine (among other things), there were challenges on racial grounds. These were mainly based on objections to the book’s language—in particular, the pervasive use of the word “nigger”—but also on discomfort and impatience with the book’s ironic condemnation of racism, and fear that the portrayal of the racial politics of the pre-Civil War South would affect the present-day treatment of African-American children.334 These challenges continue today: Huckleberry Finn was fifth on the list of most challenged or banned American books in the 1990s.335 In defense of the book, writers and educators have pointed out that instead of encouraging [begin page 773] racial prejudice, the book shows the effects of early indoctrination—of the “ill-trained conscience,” as Mark Twain put it—and they argue that most readers cannot fail to understand its underlying message: it advocates, as its author did, racial justice.336 Its “voice” and deadpan humor and its ironic confrontation with complex issues of class and culture and race have come in some way to represent the United States, for better or worse. Writer and broadcast journalist Charles Kuralt declared on television in the early 1980s, “If I had to say as much about America as I possibly could in only two words, I would say . . . ‘Huck Finn.’ ”337

In the last half-century, scores of critical books and articles devoted to Huckleberry Finn have been produced. In the academy, the book has been variously interpreted, argued over, and analyzed. Its place in the “literary canon” has come to seem so unshakeable that one critic, Jonathan Arac, has argued that the book has been “hypercanonized,” preventing accurate assessment and interpretation of it.338

Many American writers have weighed in with an opinion about Huckleberry Finn, most of them acknowledging a debt to Mark Twain, perhaps most famously Ernest Hemingway, who wrote in Green Hills of Africa, “All modern American literature comes from one book by Mark Twain called Huckleberry Finn.”339 The way Huck tells his story—in language that resembles actual speech, without any direct comment from the author, and that incorporates both black and white vernacular idioms— [begin page 774] seems to preserve the past while providing something immediate and modern and liberating. In 1970, analyzing the influence of “the spoken idiom of Negro Americans,” Ralph Ellison wrote:

In 1996 Toni Morrison wrote:

Writers all over the world have also acknowledged a debt. One Russian writer and critic, Dmitry Urnov, still a boy during World War II, recalled in 1986: “When the enemy was advancing on Moscow, my mother and I were evacuated from the city to a safe area in the Ural Mountains. We could take only a few necessities. Among these necessities was a copy of ‘Huckleberry Finn,’ with the original Kemble illustrations. I learned it by heart.”342 Japanese author Kenzaburo Oe said he found his greatest inspiration in Huckleberry Finn and that it was the book that convinced him to become a writer.343

Sales figures for Adventures of Huckleberry Finn are difficult to pin down, but it is safe to put the total in the millions.344 Well over a hundred editions of the book are currently in print in the United States alone. It endures on compact disk, in e-book format, in more than a score of audio tapes, and in nearly a dozen adaptations for film and video.345 In the 1980s [begin page 775] it inspired a hit Broadway musical, Big River, which is still performed regularly in the United States.346 It has been translated into more than fifty-three languages, and has appeared in more than seven hundred foreign editions; it is currently in print in scores of languages worldwide.347

When Clemens submitted his text of Adventures of Huckleberry Finn to Charles L. Webster for typesetting and publication he had, during the previous fifteen years, published several books, mainly through two publishers: among them were The Innocents Abroad; Roughing It; The Gilded Age; Sketches, New & Old; The Adventures of Tom Sawyer; and A Tramp Abroad with the American Publishing Company; The Prince and the Pauper and Life on the Mississippi with James R. Osgood and Company. The publication history of these books shows that Mark Twain understood the collaborative nature of the production process and depended upon and welcomed the contributions of his illustrators and publishers. For the most part, he allowed the artists to choose what to illustrate; he allowed them and his publishers to write captions for the illustrations and supply descriptions of the contents of the book for page headings and tables of contents; and he allowed his publishers to choose the paper, typeface, format of the pages, and bindings for his books. But in all cases, in particular those involving illustrations and text, he reserved the right to approve or modify the result. He sought outside readers, especially William Dean Howells, to suggest improvements or corrections to his text, and he expected printer’s proofreaders or publisher’s readers to discover errors and suggest corrections. Adventures of Huckleberry Finn, the first volume published by the new Webster company, followed the same pattern. But Clemens had regularly been frustrated and angry with the difficulty of maintaining control of those aspects of his text which mattered greatly to him, a problem which recurred with every book and every publisher, even his own company. As the text was copied and recopied by typists and typesetters and passed through the hands of the printer’s proofreaders and others, errors were introduced, styling was imposed, and slowly but inexorably the text began to diverge from the author’s copy.

This newly edited text is an unmodernized critical edition designed to recover, as much as possible, Mark Twain’s specific intentions for the text at the time he submitted the printer’s copy to his publisher in 1884. A critical text, according to modern practice, must place before the reader not only the text itself but the evidence and reasoning used by the editor to establish [begin page 776] it. As a first step the editor designates a “copy-text,” the form of the text to be edited—usually the manuscript or first printing—which, because it is least corrupt, provides the most satisfactory basis for establishing a text free from unauthorized readings. The editor agrees to follow the copy-text in every particular except where he considers emendation justified or required. And he agrees to report and defend all such emendations, so that a reader may if he chooses reconstruct the base from which the editor has departed. Central to the theory of copy-text is W. W. Greg’s distinction between substantives and accidentals. Greg recognized that authors are less likely to revise their accidentals (such as punctuation and spelling) than the wording of their work, whereas copyists, proofreaders, and house editors are likely to alter the formal features of a work while, for the most part, respecting the substantives. This house styling is likely to prevail, since the economics of publishing put great pressure on authors to acquiesce in it. In general, therefore, the theory of copy-text proposes that an editor adopt as copy-text the text closest to the author’s hand, incorporating as the author’s revisions most of the substantive variants appearing in later authoritative texts, while, on the whole, adhering to the copy-text’s accidentals.348 Unauthorized changes made by copyists, editors, and compositors are excluded from the text, while authorized changes in the printer’s copy and proofs, along with simple corrections supplied by the editor, appear in the text as emendations and are so recorded. The result is a text that is as close as the evidence permits to the text the author wanted. Despite the general applicability of this theory, in practice each text provides a unique set of historical circumstances. For instance, in the text of Adventures of Huckleberry Finn, the author himself revised large classes of accidentals (dialect spelling, among them) that are by their nature also highly subject to mistranscription and styling.

Clemens’s complete holograph manuscript for Adventures of Huckleberry Finn survives and serves as copy-text for this edition. The first portion of the manuscript, MS1, comprises nearly half (49 percent) of the text. [begin page 777] Written in two separate stints, MS1 can best be broken down into two sequential sections, MS1a and MS1b, each with a distinct history and physical appearance. The second portion of the manuscript, MS2, comprises the remainder (51 percent) of the text. Although it is possible that Clemens had an amanuensis copy of some portion of MS1, this copy (if it existed) was replaced by a typescript and was not part of the line of transmission from manuscript to printer’s copy to published text. Clemens’s working typescript made directly from MS1 (TS1) is lost, as are the typescript copies TS3 and TS2, which were subsequently made of TS1 and MS2, respectively, and which then in combination served as printer’s copy for the first American edition (A). Although a handful of A proofs with half a dozen autograph corrections or revisions survive, most of the proofs on which Clemens had an opportunity to revise and correct his text are lost. And he did not again revise it: all editions later than A are now known to derive directly or indirectly from it, without any infusion of new authority.349

Of the major documents or portions of documents now known to have comprised the original transmission from manuscript into print, only the MS (MS1a, MS1b, and MS2) and A survive. Barring only the raft episode from MS1a, which was omitted from A but published in Life on the Mississippi (LoM), there were two major lines of descent—one from MS1 (MS1a and MS1b) and one from MS2—for discrete parts of the text. With the raft episode reinstated, as it is in this edition, there are three lines of descent, thus:

1. MS1a and MS1b (copy-text) → TS1 → TS3 → A

Clemens wrote MS1a during the summer of 1876; he wrote MS1b almost four years later during the spring of 1880. He then put aside the combined MS1a and MS1b for almost three years, returning to it only [begin page 778] for some revision—primarily of dialect—and review. Between October 1882 and early May 1883, he had his manuscript typed, almost certainly on an all-capitals typewriter. This typescript, TS1, entirely superseded MS1a and MS1b; thereafter he made his revisions on TS1, presumably deleting and interlining his changes in the usual way, but also interpolating new sections in holograph. In 1884 he had TS1, now heavily revised, retyped to provide clean copy for the typesetters. This new typescript, TS3, was most likely made (probably with a carbon copy) on an italic typewriter that included both capital and lowercase letters. It also incorporated William Dean Howells’s “corrections,” which have not been further identified, but would not have included any major revision undertaken without Clemens’s consent. TS3 ultimately served as printer’s copy for this part of the text of A. Neither TS1 nor TS3 survives, but the manuscript from which they derived, discovered in 1990, serves as copy-text for this portion of the book: xxxi title through 80.29; from 99.1 through 106.32; and from 123.21 through 188.16. For these sections of the book, the primary source of emendation is A.

2. MS1a raft episode (copy-text) → LoM TS → LoM

For essentially practical and commercial reasons, Clemens allowed his publisher to omit the raft episode from A, but because those reasons no longer obtain, and because no other authorial reason for omitting it has so far been documented, it is restored here. Around September 1882, before TS1 was created, Clemens had the raft episode separately typed so that he might include it in chapter 3 of Life on the Mississippi (LoM). The episode, along with the typescript that had been made from the holograph manuscript for LoM, was corrected and revised to some extent by Clemens (LoM TS), and that revised typescript served as printer’s copy for LoM, published in May 1883. When Clemens had Huckleberry Finn typed (TS1), sometime between October 1882 and early May 1883, he probably just inserted printed tear sheets of the raft episode from LoM into the typescript, numbering them as an interpolation and making any new revisions directly onto them. The episode was probably not retyped when TS3 was made. Because the printer’s copy for Huckleberry Finn is lost, the revisions Clemens may have made for it are likewise lost. None of these documents survives except MS1a and LoM; MS1a is copy-text from 107.1 through 123.20. For this section of the book, the primary source of emendation is LoM.

3. MS2 (copy-text) → TS2 → A

Clemens wrote MS2 and had it typed during the summer of 1883. Unlike TS1, this new typescript (TS2) almost certainly comprised both ribbon and carbon copies. It was typed by Harry Clarke, who also typed “1,002. An Oriental Tale,” on a typewriter with capital and lowercase letters. Although MS2 included a long interpolation about the Walter [begin page 779] Scott for insertion in TS1, it consisted principally of the concluding twenty-two chapters. Clemens must have extensively revised the ribbon copy and then presumably brought the carbon copy into accord with it. Ultimately TS2 also incorporated “corrections” by Howells, and one of its duplicate corrected copies served as printer’s copy for this part of A. Neither the ribbon nor the carbon copy of TS2 has survived, but MS2 has, and is therefore copy-text from 80.30 through 98.7 (the Walter Scott interpolation) and from 189.1 through the end (362.11). For this section of the book, the primary source of emendation is A.

Because Clemens for the most part supervised both the typing and the typesetting of his text, it is usually the case that changes in wording between the MS and A originated with his revisions on TS1, TS2, or on the proofs of A. Generally speaking, the larger and more substantial the change, the less likely it is that anyone but the author would have the nerve—not to mention the imaginative power—to create it. For instance, the addition of the following thirty-five words in A (chapter 23) is assigned to the author’s revision on TS2 because of length and literary quality:

Omissions of comparable size can likewise be identified as the author’s work. For instance, when A omits MS1a’s “ghost” story (see Three Passages from the Manuscript) it seems evident that nobody but Clemens would have the temerity to delete it, and when A omits MS2’s comparison of publishers with kings (“Of course most any publisher would do that, but you wouldn’t think a king would. If you didn’t know kings”), it seems obvious that Clemens recognized the implausibility of Huck’s saying such a thing and so removed it. Even simple substitutions can often be recognized as authorial, although their purpose may be more difficult to characterize: “Stuff” for “Bosh”; “town” for “village”; “nation” for “mischief”; “lick” for “big-bug” for “swell”; “wallowed” for “rolled”; “bothered” for “flusticated”; “unfavorable to” for “dead agin”; “be in a bad fix” for “lose his life”; “laughed their bones loose” for “laughed themselves hoarse.” No one but the author is likely to have made such verbal changes, many of them directed at the most appropriate word choice or idiom for the speaker, and they are accordingly adopted from A as emendations of the MS.

On the other hand, when variant substantives between the MS and A create problematic readings in A, and when there is more reason than usual to suspect transcription error by the typist or the typesetter, the A reading may be rejected in favor of the MS. Among the errors here attributed [begin page 784] to the typist are words or phrases omitted in the first edition that were in fact critical to the sense of the passage. For instance, when Huck is still at the widow’s and Pap Finn is suing to get him back (chapter 6), the first edition reads:

But collation shows that MS1a, the manuscript for this passage, reads:

No one but Clemens is likely to have changed “raw-hiding” to “cowhiding,” or to have altered Huck’s original but too literate “as if” to “like,” but the largest change here does not appear to be authorial. It was evidently made by the typist, or possibly the typesetter, who inadvertently dropped the clause that explained the occasion for Huck’s borrowing money from Judge Thatcher (“all winter, the old man would lay for me and catch me, and then”), probably because he skipped from the first “and then” to the second. The necessary phrase is restored in the present edition. Similarly, in chapter 13, where Huck rows a skiff toward shore in a desperate attempt to find help for the thieves he and Jim have just abandoned on the Walter Scott, A reads as follows:

Collation shows that between the second and third sentences of this passage, MS2 had the following:

Since the likelihood is strong that the typist inadvertently omitted one or both of these sentences (the danger of an eye skip from the “I” of “I floated” to the “I” of “I skimmed” was increased by their positions at the end and beginning of lines in the manuscript), the omission from MS2 is judged an error, not a deliberate revision, and the A reading rejected. This decision resolves the minor puzzle that had been created by the change in A. It is now clear exactly where Huck was when he “skimmed around” in search of the watchman: he was on the deck of the ferryboat, not in his skiff, as A seemed momentarily to imply.

It is useful to distinguish, in addition, two groups of variant substantives (Similar substitutes and Small words, below) which are more difficult than usual to assign confidently to the author or to his typists and typesetters.

[begin page 785] Similar substitutes. These are variants in A which are extremely close to the MS in their spelling, pronunciation, or meaning, and which may be unique or nearly so in the text as a whole: for example, “bounded” in A for “bounced” in the MS; “Rouse” for “Roust”; “Confound” for “Consound”; “skipped out” for “slipped out”; “sling” for “fling”; “pisonest” for “piousest”; “begone” for “bedone”; and “Laws alive” for “Land alive.” Authorial revisions and corrections with this degree of subtlety occur in the MS. For instance, in MS2 “clean up the things” replaces “clear up the things”; “mud-stripes” replaces “mud-strips.” But variants of this kind between the MS and A are also suspect because variation can so easily have arisen from transcription error alone. For example, the letters k and l are next to each other on a typewriter keyboard, so that the typist could have mistakenly transformed “slipped out” into “skipped out.” In preparing the 1985 and 1988 text, with the evidence of only a unique reading in MS2, the editors suspected transcription error, and trusted the copy-text reading. With the discovery of MS1, analysis of variants within it and between it and A provided more instances of the same or analogous changes, which in aggregate could have come only from the author.354 However, when the likelihood of misreading by the typist is great, the copy-text is respected. For instance, the top of Clemens’s f in MS2 “fling” is formed like his usual s, and the typist could easily have misread it as “sling” (194.4; see the textual note).

To add to the complication, the collation of manuscript and typescript for LoM and for “1,002. An Arabian Tale” shows that Clemens was most apt to overlook unauthorized variation of this kind when the new reading approximated the original, or was at least plausible. In one unpublished chapter of LoM, the typist mistranscribed “human spirit” for the manuscript’s “humane spirit”; “tables” for “table”; “whiskey” for “whisky”; “fell” for “feel”; “favorable” for “favourable”; “unpracticed” for “unpractised”; and “customs” for “custom”; none corrected by the typist (or by Clemens, although unlike the “1,002” typescript the LoM TS has no authorial marks). In the “1,002” typescript, he failed to restore his manuscript reading when the typist substituted “proposed” for “purposed” (twice); “not” for “naught”; “continued” for “continuing”; and “mine” for “my.” And even when he did correct such errors on the typescript, he corrected for sense without consulting the original manuscript: thus, when the typist replaced “exuded” with “extended,” Clemens changed the typescript to “discharged.” Such variants between the MS and A must, therefore, be assessed individually on the particular merits of the evidence, but they are naturally more suspect than substitutions like A’s “howl” for the MS’s “yelp.”

Small words. These are variants in A which add or omit small words [begin page 786] that are often elided in ordinary speech or skipped inadvertently in transcription—usually conjunctions, prepositions, articles, or pronouns. Because they might be deliberate elisions to signify a dialect, or merely inadvertent or “helpful” omissions or additions, they are typically difficult to assign. Throughout the MS the author did in fact revise his text in respect to such words, and not infrequently. For instance, he added the first “as” to Huck’s “fix a day as soon as they could”; he added “of” to “outside of the shadows”; he deleted “to” from “said it would help to cure him”; and he deleted “that” from “Jim thought that it was a good idea.” He added the second “to” in Huck’s “I can’t imitate him, and so I ain’t agoing to try to”; he added the second “he” in “it was only one dog, but he made a racket for a million; and he kept it up, right along”; he deleted “of” from “It warn’t of no consequence”; and he deleted the first “that” in “and that he’d been sleeping like that.”

Nevertheless, similar variants between the MS and A are not necessarily authorial. Clemens’s typists were prone to omitting or changing such words, as the collations of the typescripts for LoM and for “1,002” show. In the small selection of surviving LoM TS pages (which show no authorial corrections), the typist omitted “and,” “the,” “her,” and “is” (and also substituted “the” for “a,” “this” for “the”). In the “1,002” typescript, which shows both unauthorized additions and deletions by the typist, either the typist or the author corrected three-fourths of these errors, but both typist and author overlooked omitted “again,” “be,” “first,” “that,” and “the,” as well as adding “the” in “the Tabernacles” and “to” in “vouchsafed to.” Some of these typing errors probably led to authorial corrections which did not, however, restore the original reading. When Clarke omitted “his,” Clemens supplied “the”; when Clarke typed “a blaze of” for manuscript “ablaze with,” and “a sound of” for “the sound of,” Clemens corrected to “aflame with” and “a sound as of,” respectively. Without access to the Huckleberry Finn typescripts, of course, revisions prompted by typing errors cannot be readily identified, but such variants in A have been assigned to the author when one or more of the following holds true:

(1) The A variant is identical with or very like demonstrable revisions in the MS. When A adds “of” to “outside of the house” or “to” to “helped a nigger to get his freedom,” the A variants have been judged authorial because each is similar to revisions in the MS, such as the “of” added to “come off of the steamboat.”

(2) The A variant is repeated, either in the same sentence or in the text as a whole. When A twice omits “of” from the MS “banging of tin pans and blowing of horns,” the change seems more likely to be deliberate authorial revision than a double oversight by the typist, who might be expected to drop one but not two such words in an eight-word phrase. Similarly, variants in A which either add or delete “that” occur so frequently throughout the text that they are treated as a class of authorial revision and all adopted. [begin page 787] A omits the MS “that” thirty-nine times in phrases like “why don’t deacon Winn get back the money that he lost on pork,” “I see in a minute that there warn’t much chance,” “he never suspicioned that I was around,” or “I see, straight off, that he pronounced.” A also substitutes “the” for “that” ten times and adds “that” three times. Such changes are too numerous and too pervasive in their effect, as well as too character-specific, to have been created either at random or deliberately by a strong-willed typist or typesetter, and are in fact only explicable when understood as the author’s effort to make Huck (and sometimes other characters) less formal in speech.

(3) The A variant is not only repeated, but belongs to an unmistakable pattern of purposeful change within the speech of one or more characters. The direction of this change may even be diametrically opposed to the direction of change evident in the MS. For instance, in the MS Clemens often revised away from conventional usage by adding the article “a”: the first “a” in Huck’s “a half a mile wide” and “a” in Huck’s “kind of a harrow,” Tom’s “what kind of a show,” and Aunt Sally’s “not a one of us.” Variants in the portion of A based on MS1a continue this trend, although the portion based on MS1b has an equal number of variants tending toward the conventional as toward the unconventional. The latter part of A, based on MS2, more often shows the reverse trend, dropping the MS “a” from “half a second,” “kind of a cold,” and “a half an hour,” or dropping it while also adding a new element: MS2 “puzzled kind of a way” becomes “puzzled-up kind of way” in A; MS2 “kind of a general” becomes “kind of generl” in A. These last two phrases contain manifestly authorial fine-tuning in the variant “puzzled-up” for “puzzled” and the nonstandard pronunciation “generl” for “general,” and together they suggest that the coincident omission of the indefinite article is part of that revision. In spite of the great susceptibility of “a” to inadvertent omission by typist and typesetter, this pattern suggests authorial omission of “a” in all cases where the MS differs from A.

The authority of variant accidentals is discussed in the following sections: Dialect spelling, Emphasis variants, Punctuation and other accidentals, and End-line dashes. Identification of authorial corrections and revisions in spelling, emphasis, and punctuation is dependent on the recognition of purposeful change.

Dialect spelling. The dialect of each character has a number of distinctive features that are meant to be recognizably consistent, or nearly so. Change between the MS and A toward greater consistency of dialect is therefore one of the grounds for emendation, on the assumption that achieving consistency was the author’s purpose. Such change has often the effect of making the pronunciation or the eye dialect (spelling not affecting pronunciation) for any character or characters more consistent [begin page 788] with respect to some reasonably specifiable “rule.” The patterns of change which are taken to represent authorial purpose can, however, be rather complex, so that an account of the assumptions and reasoning brought to bear on some typical cases may be of use here.

We assume, along with Carkeet, the character-specific nature of the dialects, even though some dialects are spoken by two or more characters. We also assume that any “rule” must have the purpose of making the final representation in A as consistent as possible with respect to some recognizable aspect of pronunciation or spelling, even if in applying that rule the author has not perfectly achieved the intended result. This requirement does not preclude the possibility that rules may specify different pronunciations in different circumstances, or even that a character may use several different forms of a word interchangeably, apparently regardless of circumstances. Implicit rules may be quite simply deduced and successfully demonstrated, as in “the king says ‘jest’ for ‘just,’ ” but they may also be rather complicated and harder to apply consistently, as in “Jim tends to drop the final t of a contraction when the next word begins with a consonant (as in ‘ain’ dat’ or ‘ain’ no’), but to retain the t when the next word begins with a vowel or is emphatic (as in ‘ain’t any’ or ‘ain’t dat’).”355

The supposition that Clemens revised his dialect for greater consistency with respect to such rules is grounded in the changes he introduced within the MS itself. For example, the king does say “jest” for “just” in thirteen out of the fourteen times he uses the word in the MS. In ten of these cases Clemens wrote “jest”; in one he wrote “ju,” immediately correcting it to “jest”; in two he wrote “just” but later altered the manuscript to “jest.” When, therefore, the king’s unique MS form, “just,” appears in A as “jest,” it is reasonable to suppose that the author corrected the errant form on his typescript or proofs of A. It is extremely unlikely that the form was changed by the typist or typesetter to achieve uniformity in spelling, even though uniformity was usually part of the typesetter’s responsibilities.356 In these cases no ordinary rule of uniformity could possibly have been applied by the typesetter, since, for example, Aunt Sally—in contrast with the king—says “jest” in six out of eleven instances in the MS, but uniformly says “just” in A. Only the author would undertake to make different characters use alternative forms of the same word, each more consistently [begin page 789] in A than in the MS, while also making one of these characters use a different form in A than in the MS.

For the king’s speech, in fact, most variant forms in A continue a trend toward what may be described as “more nonstandard forms used more consistently,” which begins within the MS itself. A has “agin” for MS “again”; “kin” or “k’n” for “can”; “yit” for “yet.” So when the king says “git” for “get” six times in the MS (including two times corrected from “get”), but A renders five of these as “git” and one as “get,” it is more reasonable to suspect a typist’s or compositor’s error or sophistication than the author’s deliberate retreat from consistency already achieved, and the MS reading is therefore retained.

On the other hand, A variants leading to greater consistency of a standard form in the king’s speech are also adopted as emendations. The king says “if” fifteen times and “ef” four times out of the nineteen times he uses the word in the MS. In A, however, each of the four instances of “ef” is changed to “if.” These four variants in A are attributed to the author because they generate greater consistency in the king’s use of a word that has a nonstandard spelling or pronunciation in dialects of other characters. Again there is virtually no likelihood that this greater consistency was imposed by anyone but the author, especially because no one else would undertake to make the king say “if” more consistently, while, for example, simultaneously making Jim say “ef” (or “ ’f”) more consistently. The movement toward greater consistency even of a standard usage in the dialect of a given character is a reliable sign of authorial revision, despite a more sweeping general trend toward nonstandard usage.

Clemens’s revision of the accidentals can sometimes be detected even when it does not aim at making a character use the same form of a word in every instance.357 The rule governing Jim’s retention of terminal t in contractions has already been mentioned. In the MS Jim also uses three past-tense forms of the verb “to be”: ““ ’uz,” “wuz,” and “was” (and its contracted form, “ ’s”). In MS1 Jim says only “was” (ninety-nine times), but in MS2, where Clemens introduced the other two dialect forms, he carefully distinguished between eye-dialect “wuz” and the less emphatic but differently pronounced “ ’uz”; and he tended to revise so that Jim says either “ ’uz” (five times) or “wuz” (seven times) nearly as often as he says “was” (fourteen times). In MS2 he changed four instances of “was” to “wuz,” two instances of “was” to “ ’uz,” and one instance of “wuz” to “ ’uz.” When [begin page 790] other instances of “was” remaining in MS1 or MS2 appear in A as “ ’uz” (twenty times) and “wuz” (twenty-four times), these variants are recognizably authorial, even though Jim still also says “was” in A (twenty-two times).358

One rather weak but useful generalization that emerges from this and several similar cases is that when variants in A move toward nonstandard forms, they are less likely to have been caused by the typist or compositor than if they moved toward standard forms. But even where the rules, so far as we understand them, do not seem to specify which of several forms is preferred in a given situation, authority can be recognized. When he was about halfway through composition, Clemens wrote a note to himself about two dialect equivalents for “of” in Jim’s speech—“er” and “o’ ”—either as a direction to make revisions in MS1 or as a reminder of the revisions he had recently made: twenty-six changes to “er” and five to “o’,” leaving twenty “of” and seven “ ’n” (as in “out’n”).359 In MS2, in addition to “o’,” “of,” “er,” and “ ’n,” Clemens introduced another form: Jim also says “un.” Although these five different forms are sometimes associated with specific contexts, to all appearances they are used interchangeably. Jim says “outer me” and “out’n de rain”; “sight o’ trouble,” “pack er k’yards,” and “coat o’ arms”; “all of a sudden” and “half un it”; “kiner smilin’ ” and “kine er time.” The MS contains some revision of these various usages (“coat o’ arms” was originally “coat er arms”), but no conclusive pattern. Nevertheless, when Jim says “of” in the MS but “uv” in A seven times, the change to eye dialect is recognizably authorial even though this form never occurs in Jim’s speech in the MS. The A variant “uv” is recognizably authorial because (a) it moves toward nonstandard usage; (b) it is consistent with Jim’s undifferentiated use of five or six distinct forms for the same word; and (c) it is consistent with demonstrable revisions in other words of Jim’s speech toward eye dialect.

Emphasis variants. Clemens carefully marked his MS for emphasis, using single underlines for italics, double underlines for small capitals (or capitals and small capitals), and triple underlines or block capitals for full capitals: “And they call that govment!”; “You talk like an Englishman—don’t you? It’s the worst imitation I ever heard. You Peter Wilks’s brother”; “h-wack!”; “The Balcony Scene”; “you mustn’t bellow out ROMEO! that way.” Of such emphasis markings, italics were the most consistently transmitted to A.

The overall reliability of transmission of italics makes clear that (a) for the most part the typescripts must have transmitted underlines, and (b) [begin page 791] the more direct the transmission from MS to A, the more reliable it was: 63 percent of MS1a italics, transmitted through TS1 and then TS3, are preserved in A; 80 percent of MS1a italics in the raft episode, transmitted through LoM TS, are preserved in LoM; 45 percent of MS1b italics, transmitted through TS1 and then TS3, are preserved in A; and 93 percent of MS2 italics, transmitted through TS2, are preserved in A. But accuracy of transmission from MS to A is not an altogether reliable standard, for as carefully as he revised his italics in the MS, Clemens continued to revise them on the intervening typescripts. For instance, in the MS itself, he changed Huck’s “till I was afeard I had made a mistake” to “till I was afeard I had made a mistake” and Jim’s “No, sah, I doan’ want no sich” to “No, sah, I doan’ want no sich.” When in A, Jim’s “Well, I believes you, Huck. I—I run off” becomes “Well, I b’lieve you, Huck. I—I run off,” and the duke’s “Quick sales and small profits” becomes “Quick sales and small profits,” it is clearly the author’s hand at work. Even subtler changes, such as variants in words with divided emphasis, show the same care in recreating the sound of speech and are presumed to be the author’s: the omission of italics from the MS “gimme” in the Bricksville loafer’s phrase “gimme the chaw, and you take the plug” in A, or the change of emphasis from two syllables in the MS “inter” to one syllable in the king’s “the Hebrew jeesum, to plant, cover up; hence inter” in A. Although variants where italics have been eliminated altogether cannot by any satisfactory means be divided into authorized and unauthorized changes, by far the largest proportion of variants can be identified with some confidence as authorial. Therefore, italic variants from the MS to A (and LoM) are treated as a class: all A variants are adopted unless there is a particular reason to suspect mistranscription or deliberate unauthorized styling.

One exception to this rule governs contractions, where Clemens most often emphasized only the first part, as in: “that’ll,” “I’m,” and “they’d.” Although one such form from MS1a and four from MS2 appear in A unaltered, for the most part such contractions appeared in A styled to full italics: “that’ll,” “I’m,” and “they’d.” Because Clemens may have intended some dialectal effect, however subtle, and because such forms are so easily mistranscribed by the typist and typesetter or styled by the printer’s proofreader, the MS is presumed to be a better guide to his preference than A. The copy-text form has therefore been respected wherever such variants appear.

No small capitals or full capitals survived in the portion of the text transmitted from MS1 through TS1 and TS3. A combination of circumstances probably accounted for it. In all likelihood, the TS1 typist was thwarted by the limitations of the all-capitals typewriter. He or she may have adopted a system of marking similar to the very imperfect one used by the LoM typist, who marked italic words and single capital letters with an underline but had no mechanical way to show small capitals or full capitals [begin page 792] (and had to choose between showing the capital or italics if a word had both). A typical sentence from the LoM typescript demonstrates the system: “SPRINGFIELD, MASSACHUSETTS WAS A VILLAGE OF LITTLE FRAME HOUSES WITH PORTICOES OF HUGE WOODEN CORINTHIAN AND IONIC COLUMNS.”360 Either the typist or the author would have had to hand copy all additional MS emphasis markings onto the typescript. Such copying was apparently not done. When TS1 was given to William Dean Howells for retyping (TS3), the typist could copy only what was on the page. Neither Howells nor his typist had access to the manuscript to verify the emphasis markings, and Clemens did not see this section of the book again until it was in proof. In the manuscript of the duke’s first handbill (MS1b), Clemens carefully distinguished levels of emphasis by marking in the manuscript for full capitals, capitals and small capitals, and italics (see the facsimile pages in Manuscript Facsimiles, pages 572–73). None of these markings survived into the first American edition. In contrast, in the second handbill, whose text was transmitted directly from MS2 to TS2 (typed with capitals and lowercase letters), he was able to either preserve or improve upon his MS styling. For styling of small capitals and full capitals in the text transmitted from MS1 through TS1 and TS3 only, the copy-text is a better guide to Mark Twain’s preference than the first American edition, and its reading is respected.

Punctuation and other accidentals. The punctuation of the copy-text has in general been retained, except where variants in A or LoM are manifestly or probably authorial. Several categories of variants of punctuation and other accidentals in A have been identified and adopted on the assumption that they result from Clemens’s revisions on his typescript. They include variants associated with substantive reworking of sentences, a small number of variants linked to or dependent upon emphasis revisions, and changes from two sentences to one and vice versa, all analogous to demonstrably authorial revision in the manuscript. For instance, in chapter 1, where Huck is describing the difference between the Widow Douglas’s and Pap’s cooking methods, the MS reads “In a barrel of odds and ends it is different. Things get mixed up, and the juice kind of swaps around and the things go better.” In A, these two sentences are made one by the substitution of a semicolon after “different,” and lowercase “things” for “Things.” This change is typical of a class of about forty variants between the MS and A. Often such variants are linked to other, clearly authorial, substantive revision. But even simple variants such as this one are analogous to alterations Clemens made in his MS (see, for instance, the alteration at 354.4). Similarly adopted from A are the addition or deletion of line spaces (often linked to changes in chapter division) and new paragraphing. Except where a probable misreading of the manuscript or inadvertent [begin page 793] introduction of error by the typist or compositor provides the more likely explanation of the variants, such A variants are assumed to be authorial.

End-line dashes. Clemens’s manuscript contains numerous instances of short dashes following periods at the ends of lines. These dashes are a holdover from his days setting type in columns for newspapers, where a convention existed of filling a short line with a dash to avoid confusing the end of a line with the end of a paragraph. When such dashes in his manuscripts were transcribed, they appeared in the middles of lines, punctuating his text in a way he did not intend. Some survived the process of transcription, editing, and typesetting in The Innocents Abroad (1869), Roughing It (1872), The Adventures of Tom Sawyer (1876), The Prince and the Pauper (1881), and Life on the Mississippi (1883).361 In Huckleberry Finn, out of eighty-three uncanceled end-line dashes inscribed in the 1876 and 1880 manuscripts, MS1a and MS1b, two survive in A and three in LoM, but most were deleted before publication, quite possibly by Clemens on the typescript. In fact, it may have been his experience with TS1 that led him to carefully search for and cancel all ten of his end-line dashes in his 1883 manuscript, MS2, before it in turn was transcribed.362 All instances are routinely emended out in this edition.

A is the sole authority for substantives as well as accidentals in the matter added during production—the table of contents, list of illustrations, and captions, which are emended to agree with the text when quotation marks indicate an intention to quote. For the portions of text excerpted in the Century Magazine, which were set from proofs of A and specially revised by the author, the Century has collateral authority.363 While most variant substantives in the Century were imposed either by the magazine’s editors or the author specifically for the purpose of publishing extracts, a few revisions are adopted as corrections, or on the grounds that they were made by [begin page 794] the author and intended as literary improvements quite apart from the needs of the magazine printing. Since the printing of A was in progress during the time Clemens saw and revised the Century proofs, he did not have the opportunity to incorporate these late improvements in the text of his book. For the raft episode, where MS1a is copy-text, chapter 3 of the first American edition of Life on the Mississippi is the only authoritative source of emendation. It supplies the illustrations and their captions. The captions and list of illustrations are emended in this edition to agree with the text when quotation marks indicate an intention to quote.

A full description of the relevant texts, a discussion of problematic readings in the copy-text and in the edited text, a list of every departure from the copy-text in this edition and of variants in the authorized editions, a comprehensive record of the author’s manuscript revisions, and a guide to compound words hyphenated at the ends of lines in this edition can be found in the following sections of the Textual Apparatus—Description of Texts, Textual Notes, Emendations and Historical Collation, and Alterations in the Manuscript—each of which is described in brief on the page immediately following this introduction and more fully in the headnote to each section.

December 2002　　　　　　　　　　　　　　　　　　V.F. and L.S.