University of Virginia Library

Dating Through Calligraphy: The Example of "Dover Beach"
by
S. O. A. Ullmann

Dating manuscripts can be puzzling. Although it is sometimes possible to assign dates on the basis of references in the text to specific historical or biographical events, all too frequently there are no such clues or those that exist lead to dating within a period of only five, ten, or more years. Generally this is unsatisfactory, especially for an edition of letters or for dating the poems of a writer like Matthew Arnold, who composed most of his poetry within a decade and who published his three most important volumes of poetry within five years. To deal adequately with such a writer's development precise dates for manuscripts are essential.

Calligraphy has long been used to assign anonymous manuscripts to their historical period and to determine whether a manuscript was written during a writer's youth, maturity, or old age. A relatively recent development, however, is the attempt to refine this procedure so that dating within a year or even less becomes possible. Such precision can be achieved because the handwriting of most people, and especially of those creative people who regularly write by hand, is constantly changing.[1]

The use of calligraphy to date letters, when combined with references in the text, has enabled scholars to make remarkably precise attributions. For example, a letter may refer to an unspecified death in the family or to the publication of some unnamed work. If the letter can be dated within a year by means of handwriting, it is sometimes possible, on the basis of this internal evidence, to determine the exact date on which the letter was written. Madeline House and Graham Storey, editors of the monumental Pilgrim Edition of Dickens' letters, have

found that calligraphy can sometimes even enable them to correct Dickens' own misdatings, not uncommon especially at the beginning of a new year.[2]

Two independent investigations using calligraphy to date manuscripts of nineteenth-century writers reveal important similarities in method: the first, an analysis of the handwriting of Emily Dickinson from 1850-86;[3] the second, the edition of Dickens' letters already referred to, which makes great use of calligraphy, especially in the first volume where more than two-thirds of the letters were originally undated (I, xxiii). These two studies, together with the present one, use individual letters of the alphabet as the basic unit of analysis. All three studies also usually differentiate among examples of the same letter, depending upon their position at the beginning, middle, or end of a word. Although Dickens' editors found that his signature and a few common words and names were especially useful for dating, they tended to concentrate on the initial or final letters of these words and names.

A number of variables complicate the task of comparing samples of calligraphy. Manuscripts written with a pencil differ from those written with a pen. Even a change in pen or pen point affects the calligraphy. Health, mood, and fatigue likewise influence handwriting, as does the purpose for which the manuscript was written. The calligraphy of a formal letter or a fair copy of a poem will normally differ from that used in dashing off a note to a friend or writing a first draft. No wonder many believe it impossible to use handwriting as a basis for precise dating.

These variables, however, merely make the task of analyzing calligraphy difficult, not impossible. Although comparisons among homogeneous manuscripts naturally produce the most reliable results, valid comparisons are possible among heterogeneous materials. Careful examination of dated samples will reveal the nature of the difference between fair copies and rough drafts. A single letter written informally but containing a long quotation, normally written more carefully, can enable one to observe the differences. Mrs. Ward found, for example, that Emily Dickinson's calligraphy in her late work sheets is a throwback to the handwriting she used earlier in fair copies (I, l-li). For Matthew Arnold, on the other hand, the reverse is true. Fair copies are

PLATES

Text of ["Dover Beach"] Plate I

• 1 The sea is calm tonight.
• 2 The tide is full; the moon lies fair on the French coast
• 3 Upon the Straits: [the cliffs of deleted] the light
• 4 Shines & is gone: the cliffs of England
• 5 Glimmering & vast: out in the tranquil bay. sweet
• 6 Come to the window, hush'd is the night air.
• 7 Only from the long line of spray
• 8 Where the sea meets the moon blanch'd sand
• 9 Listen, you hear the grating roar,
• 10 Of pebbles which the waves suck back & fling bank'd
• 11 At their return, up the steep strand again
• 12 Cease and begin and then begin mournful
• 13 With regular cadence [s deleted?]low, and bring
• 14 The eternal note of sadness in
• 15 Sophocles long ago
• 16 Heard it on the Ægean, and it brought
• 17 In to his mind the turbid ebb and flow
• 18 Of human misery: we
• 19 Find also in the sound a thought
• 20 Hearing it by this distant Northern Sea
• 21 The Sea of Faith
• 22 Was once too at the full and round Earth's shore
• 23 Lay like the folds of a bright garment furl'd:
• 24 But now I only hear
• 25 Its melancholy long withdrawing roar
• 26 Retreating to the breath
• 27 Of the night wind down the vast edges drear
• 28 And naked shingles of the world— Ah love [l deleted] &c

regressive in their calligraphy; while his rough drafts are innovative and reveal the direction in which his handwriting is developing. Analysis of different types of dated samples will reveal what kind of allowance needs to be made when comparing heterogeneous documents. The apparent chronological difference between a fair copy and a rough draft by Arnold is about a year. Fortunately, a few tell-tale signs always reveal that his fair copies are of later date than most of their features would otherwise lead one to suppose.

Precision in dating is of course partly a function of the amount of dated material available. Without at least three thousand words a year in dated manuscripts accurate results will be difficult to obtain.

Each manuscript can best be analyzed on a separate sheet of ruled paper on which one or more lines is provided for each letter of the alphabet and for any commonly used symbols, such as the "&." Theoretically every occurrence of every letter should be traced on this sheet, using a lighted box with a glass top. Because transcribing even a single manuscript in this way would be extremely time-consuming, one makes many shortcuts in practice. To make them intelligently, prior study of the calligraphy is essential. Some letters, perhaps the "i's" and "o's," vary little from year to year. Other letters appear too infrequently to be useful, for example, capital "Q," "K," "X," "V," and "Z." Medial letters, except for distinctive combinations such as the "dg" of "Dover Beach," can normally be eliminated, thus saving considerable time. They generally show less variation from year to year than initial or final letters, and of course they are affected by both the letter that precedes and the one that follows them. An examination of medial "d," "f," "g," "h," "p" and "y" in hundreds of pages of Arnold manuscripts yielded little that was helpful for dating. After deciding which letters and symbols to analyze, one can duplicate a number of lined forms with the appropriate letters indicated in the left-hand margin. Some blank lines should be left for later additions.

Often it is not only easier but also more useful to subdivide the examples of an individual letter into categories. Once these have been identified, it is enough to trace one example of each type and then simply note how often each occurs. For a single manuscript a handful of categories for any one letter should be adequate. Finer distinctions are unnecessary if, through preliminary examination, one has chosen categories that represent the significant variables. Obviously judgment based upon experience is necessary in deciding how many different variables to use and which letters to analyze. For example, one of Arnold's letters selected for analysis in the "Dover Beach" manuscript

was his initial "n." At first only the beginning of the letter seemed important. Was the stroke vertical or horizontal, did it hook or loop, was it large or small? Later it became apparent that the angle between the first major downstroke and the next upstroke was highly significant. After years during which the upstroke largely retraced the downstroke, Arnold began to spread the letter so that it came to have two arches instead of one.

Although a letter may vary from document to document and from year to year, it sometimes varies in a random way and is therefore useless as a clue to dating and can be dropped from further analysis. Arnold's final "y's" are extremely variable, but the variations depend more upon his mood and his purpose than upon the date. In Arnold, initial letters are the most revealing, but for other writers final letters are the most significant. Capital letters are always more useful than their relative rarity would lead one to suppose.

Once a representative group of documents has been transcribed, the sheets should be studied carefully letter by letter to discover which provide clues to dating. One will then be in a better position to proceed more selectively in transcribing the remaining manuscripts.

In examining the letter "s," for example, one would like to find variant "s1" occurring in 1850, "s2" in 1851, and "s3" in 1852. Unfortunately this never occurs. What is more likely is that "s1" and "s2" will both appear in 1850, though perhaps with "s1" dominant. By 1851 the proportions may be reversed with, in addition, an occasional "s3." By 1852 "s3" may have become dominant and "s1" may have almost disappeared. Even this sequence is far too simple as a description. More than three major variants normally appear simultaneously, and changes rarely begin on New Year's Day. Within a period of months a writer may drastically alter the way he forms a letter and then may make no change in it for years. Yet although few letters retain their importance for dating throughout a writer's career, in any given year some of the letters—initial "s," capital "L," final "g," for example—will differ significantly from those of the year before. These, then, become the major indicators for determining whether a manuscript was written during that year. Thus Mrs. Ward lists for each year the distinctive features of Emily Dickinson's handwriting (I, li-lix). Although it sometimes happens that a variant will disappear only to reappear some years later, the over-all configuration of the other letters will indicate whether this is the first or second appearance of the variant.

In this kind of analysis of calligraphy similarities are much more

important than differences. Differences are the norm, similarities the exception. Therefore if every major variant of a single letter in an undated manuscript has its counterpart in the dated manuscripts of a given year, one has a clue to the dating. If all the letters of the manuscript fall into categories occurring within that year, the date of the document is established. Despite the fact that each individual letter is unique in some small way, if a manuscript is to be convincingly assigned to a particular year, the pattern of all its letters must be appropriate to the categories for that year, and unusual variants must be of a kind that are at least possible for that date. For instance, if the staff of a letter in successive years formed an angle of 30°, 60°, and 90° with the horizontal, a staff at an angle of 40° could easily be dated by its place in the sequence, even though no other examples of a staff at this angle existed in dated documents. Because patterns for all the letters, surprisingly, do fit in this way, it is usually possible to date manuscripts with a high degree of confidence.

Because my interest in calligraphy grew out of a need as editor to date the many separate sheets of poems, notes, and reflections that constitute what is usually referred to as Arnold's "Yale Manuscript," I have naturally selected an Arnold manuscript, that of "Dover Beach," to illustrate the procedure described above. "Dover Beach" was an obvious choice, not only because it is Arnold's most famous poem, but also because it is the only one of his three or four best poems that survives in manuscript, now in the Ashley Collection of the British Museum. Moreover, ever since Tinker and Lowry first described the manuscript in 1935,[4] its dating has been a subject of increasing controversy. Warren Anderson speaks for many students of Arnold's poetry when he says of the poem: "Unique in its fame, it is also uniquely difficult to date."[5] Even with the aid of handwriting, it presents more than the usual number of problems, and is probably the most difficult to date of the more than a hundred individual Arnold manuscripts that I have dated. This draft of the poem is in pencil, most of it very faint, so faint in fact that Tinker and Lowry remark: "The lines have been so rubbed as to be almost illegible; indeed, it is even possible that an effort was made to erase them after they had been copied."[6] This cautious supposition, however, is mistaken. Arnold's

usual practice, as illustrated in Plate I, was to draw lines through rough drafts of a poem, if he did not destroy them altogether, once the work had been published. Copies of published poems that survive without these disfigurements are usually fair copies (sometimes those used by the printer) that Arnold presented to friends or members of his family. The "Dover Beach" manuscript shows no erasures of the kind one would expect if Arnold had tried to obliterate the poem. Although Plate I is more legible than the original, it has been reduced by about 16% and is still difficult to read for those unfamiliar with Arnold's handwriting. A transcript has been provided, for no accurate text of the manuscript is available in print. Occasionally it is difficult to be certain of the punctuation.

The untitled manuscript contains only the first twenty-eight lines of the poem and ends, "Ah love &c" (after "love" there are very faint remains of an "l," indicating that Arnold started to write the next word, "let," but decided to stop and add "&c"). The poem is on the front (the back is blank) of the second of two separate sheets that at one time were folded vertically and horizontally (see Plate I) and inserted into Thomas J. Wise's copy of the first edition of Arnold's Empedocles on Etna, and Other Poems, now in the possession of the British Museum. At the top of the second sheet are the last seven lines of Arnold's notes in ink for "Empedocles on Etna," taken from the second volume of Simon Karsten's edition of Philosophorum Graecorum veterum praesertim qui ante Platonem floruerunt operum reliquiae (Amsterdam, 1838), pages 3-78. (All except the first eight and a half lines of Arnold's notes are transcribed by Tinker and Lowry, Commentary, pp. 289-90. There are two errors in their transcription of that part of the notes illustrated in Plate I: "Anchities" (l.2) should be "Anchitus," and "Orpheuslike" (l.6) should be two words, not one.) "Dover Beach" fills the remainder of the page in a way that clearly indicates that it was written after the notes that precede it.

In June 1849 Arnold showed "Empedocles on Etna" to J. C. Shairp, a Balliol contemporary. Shairp's comments, in a letter of June 30th to Arthur Hugh Clough, indicate that enough of the poem had been written by then for Shairp to recognize that Arnold seemed to be using Empedocles' "name and outward circumstances" simply "for the drapery of his own thoughts." Because Arnold's notes are the kind he would have made at an early stage in his work on the poem, they cannot date from later than the spring of 1849. A letter from his mother to his brother Tom indicates that early that year Arnold still intended to write his long-contemplated tragedy, Lucretius, but he

must soon thereafter have decided to incorporate some of the work for it into "Empedocles on Etna." In an outline of his poetic plans for 1849 ("Yale Manuscript," fol. 25r), probably jotted down in the fall of 1848, Arnold includes both "chew Lucretius" and, under the heading "compose," "Empedocles—refusal of limitation by the religious sentiment." It seems probable, then, that the notes on Karsten date from early in 1849. (An analysis of the handwriting of these notes, a relatively easy manuscript to date, places them between mid-1848 and mid-1849, thus supporting the usual date based upon the evidence given above.) The draft of "Dover Beach" must, therefore, have been written between early 1849, when Arnold made the notes on Karsten, and July 1867, when the poem was published in Arnold's collection of New Poems. One scholar has placed the date as early as September 1848-October 1849,[7] and another, as late almost as the day of publication,[8] though a majority consider the early fifties the most likely period of composition.[9]

Hardly anyone who has dealt at length with the poem has been able to avoid confronting the question of its date. Arguments have been based on its connection with the notes for "Empedocles on Etna" and the likely interval between the notes and the poem;[10] and on Arnold family tradition connecting it with Arnold's wife, the possibility that it refers to Arnold's visit to Dover on his honeymoon in June (and again in October) 1851, and the likely interval between the event and the poem.[11] Other critics arrive at a date based upon the work's poetic maturity or upon its similarity in theme, imagery, and outlook to one or more of Arnold's other poems, especially "Human Life" and "Stanzas from the Grande Chartreuse," though there is disagreement about whether its composition preceded or followed the latter.[12] Some critics date the poem on the basis of Arnold's

debt to the work of other poets, for example, to Browning's "Andrea del Sarto" of 1855.[13] Others find the poem indebted to Clough's The Bothie of Tober-na-vuolich,[14] while, on the contrary, still another critic sees Clough's "Say Not the Struggle Naught Availeth" (written by October 1849) as an answer to Arnold's poem (Robertson, pp. 919-26).

If the poem was, in fact, written late in 1848, Professor Kenneth Allott ought to have placed it 30th in his chronologically arranged edition of the poems. On the other hand, if it was written early in 1867, it should be numbered 121 of the 129 poems Arnold published during his lifetime. In other words, settling the date of the poem involves deciding whether it was composed at the outset of Arnold's career as a poet, near its close, or somewhere between these extremes. Arnold's calligraphy provides a reliable basis for settling the problem, or at least for limiting its scope.

Ideally this study ought to present numerous examples of Arnold's handwriting arranged chronologically along with the sheets of analysis based upon them, but this has not proven feasible. Instead, Plate II[15] contains primarily a selection from the "Dover Beach" manuscript of variants sufficiently rare that they appear within only a limited period, usually no more than a year or two. Virtually all the available letters, diaries, and poems for the years 1849-55 have been drawn upon for comparison.[16] Manuscripts for other years were eliminated because they had too little in common with the handwriting of "Dover Beach." Where the similarity between a dated variant and one in "Dover Beach" is sufficiently close, a "+" has been entered on the chart along with the month of the matching document. Personal judgment, of course, cannot be eliminated from the process of deciding whether the similarity is "sufficiently close." Matching requires enough familiarity with Arnold's handwriting for one to recognize unusual formations. Sometimes the special features that account for the rarity of an individual

letter are small (as in the "L" illustrated in Plate II), while at other times they are large and obvious (as in S1).

In addition to relatively rare forms, Plate II includes examples of initial "s" and "w" labelled "norm." These represent categories that include more than half of each of these initial letters in the "Dover Beach" manuscript. These categories are in the majority in at least one document for each of the years marked with a "+." Arnold used these categories throughout the 1849-55 period, but only in the years indicated were they ever the norm, as they are in "Dover Beach." Because these two forms involve categories, precise matching is not required.

Something needs to be said as well about the great majority of letters not illustrated in Plate II. Analysis of the "Dover Beach" manuscript began with an attempt to find matches for all of the capital letters and for about twelve lower-case letters—in all, almost sixty categories. Elimination of both the unique and the commonplace led to the selection reproduced in Plate II. Most of the letters in the manuscript proved to be of almost no significance for dating because they can be readily matched in documents throughout the 1849-55 period. A few letters are genuinely unique, unmatched in any other document, for example the "S" in line 21, as well as the "Th" formation beginning the same line (perhaps the result of Arnold's changing his mind as he was writing the word). There are at least eight unique variants in these twenty-eight lines, an unusually large number, appropriate for so original a poem. In style, content, and handwriting, its originality consists of small but significant variations on familiar patterns.

The following brief comments point out the most important features of the individual letters and combinations illustrated in Plate II. (All lower-case examples are of initial letters, except for the "dg" combination, which occupies a medial position, and the "ds" combination, a terminal position.)

• 1. d (l.27): the major downstroke is at least slightly convex, and there is no loop at the top of the letter.
• 2. dg (l.27): the vertical of the "d" descends in a curve to form the vertical of the "g." There are small loops at both top and bottom of the combination.
• 3. ds (l.23): the "s" is rounded and extends slightly below the base of the "d."
• 4. G (l.5): the top of the letter is comparatively small, and the first part of the stroke (the upper lip) is on a line with the major downstroke.

  ---

  32

  
  
  
  The main axis of the letter slants forward less than 30° from the vertical.
• 5. L (l.9): the top has a rounded hook to the left and shows no tendency to loop. At the base of the letter the top of the loop is to the left of the vertical and has a major vertical axis, with the downstroke reaching its low point after crossing the major downstroke.
• 6. n (l.6): the initial stroke forms a rounded hook (the apparent full loop is the result of a smudge). The major upstroke diverges at an angle of approximately 45° from the base of the initial downstroke and reaches a pointed peak.
• 7a. S1a (l.3): loops at top and bottom with a forward slant.
• 7b. S1b (l.20): loop at bottom only with a forward slant. (In 1851 the initial upstroke is too short.)
• 8. S2 (l.4): hook with rounded bottom. (In 1853 the hook is too narrow and in 1855 too broad.)
• 9. s1 (l.1): the lower-case version of S1a.
• 10. s2 (l.14): a very slight overlap or loop at the top with a rounded hook at the bottom that just begins to circle back on itself.
• 11. s3 (ll.7, 8 twice, 10, 11 twice, 22, 28): the lower-case version of S2.
• 12. Th (l.14): a wide hook with a slight tendency to loop at the start. The "T" is linked to the "h" rather high up.
• 13. W (l.22): the initial downstroke slants less than 45°. The final upstroke makes an angle of about 60° with the vertical. (The "line" about mid-way down the initial stroke is a blemish in the paper.)
• 14. w (ll. 10 twice, 18, 25, 28): both halves of the letter are at least slightly rounded. The left side of the letter is at least two-thirds the width of the right side.

The period from January 1851 through April 1852 includes examples of all the formations illustrated in Plate II. Because the letters not illustrated are commonplace throughout the period and because the eight unique exceptions mentioned above, though not found in any other manuscript, are variants of a kind that one would expect to find at the time (see p. 27), it seems reasonable to suppose that the manuscript was written during this fifteen-month period. This conclusion is of course based upon the total configuration of the calligraphy and not merely upon the letters illustrated in Plate II. Although dating depends upon finding similarities, a date cannot be established unless it is possible to account for all the forms in a manuscript. It is not enough that some forms can be matched in contemporary manuscripts. One must be able to demonstrate the possibility of all of the forms occurring within the assigned period.

One very useful clue to dating Arnold's handwriting not illustrated in Plate II is the lower-case "th" combination, in whatever position

it occurs. During the late forties Arnold often formed this combination without lifting his pen or pencil to make the crossbar of the "t." On average during the late forties every fourth "th" is formed in this way. In 1850 there are some manuscripts without this formation. From 1851 on, it disappears almost completely. In the last lines of Arnold's notes on Karsten, which precede "Dover Beach" (see Plate I), there are three examples of the uncrossed "th," appropriate for the period 1848-49. The formation does not occur at all in "Dover Beach," lending support to a date after 1849 and probably after 1850. There are combinations and variants that do not appear and disappear quite so dramatically. The proportions within the major categories of initial "f," for example, reinforce the conclusions reached on the basis of the examples illustrated in Plate II.

As additional dated manuscripts become available, it should be possible to date "Dover Beach" with greater precision, though it is highly unlikely that the manuscript will be assigned to a date outside the 1851-52 limits. If a computer were to be used, refinements would be possible, even though the results would still depend upon the judgment of a human analyst.

An additional piece of evidence, based in part upon handwriting, reinforces the 1851-52 date. Arnold pencilled the following sentence at the top of the recto of the first leaf containing the notes based on Karsten: "Man has an impulse for happiness: he sees something of it, hears traditions of much of it: he thinks therefore he ought to have it: what is true is, he may have it if he can." From their location, squeezed into the top margin, it is obvious that these words also were written after the notes for "Empedocles." As Tinker and Lowry point out, the sentence is, "plainly, related" to one of the cancelled stanzas from "Empedocles" in the "Yale Manuscript" (Commentary, p. 290n.), and obviously must have preceded the composition of the stanza. The cancelled stanzas can be dated, on the basis of calligraphy, between February 1851 and February 1852. Thus it is probable that the sentence was written in 1851. Because both the sentence and "Dover Beach" are in pencil, they are relatively easy to compare. Although the sample is too brief to be dated definitely, the handwriting shows many marked resemblances to that of "Dover Beach." If one accepts 1851 as the date for the sentence, '"Dover Beach" must fall within the 1850-52 limits, since the resemblance is too marked for them to have been written more than a year apart.

If instead of focusing on the calligraphy one looks at the manuscript as a whole, still more clues to its dating emerge. Almost everyone

who has commented on it has pointed out that the final words, "Ah love &c," seem to suggest that the last nine lines of the published poem were composed before the first twenty-eight. The implication is that Arnold wrote the first twenty-eight to lead up to a conclusion that he had already arrived at. Although Tinker and Lowry point to the absence of any reference to the sea or tides in the conclusion (Commentary, p. 175), few readers consider the poem flawed structurally. In fact, Professor Dwight Culler, far from believing that the poem strains to link two diverse sections, considers the absence of the sea in the last lines "the very point of the poem."[17]

After my work on the calligraphy of the manuscript had been completed, Professor L. A. Beaurline suggested that the manuscript is very likely an "intermediate draft" of the poem and that there must have been an earlier version of the entire poem, not simply of the last nine lines. He based his hypothesis upon two observations: the revision of line 3, which suggests an eye-skip to line 4, and the absence of revisions in lines 15-28, which contain only one verbal variant from the final text.

Careful examination of the manuscript supports this hypothesis. When Arnold was copying the second half of line 3, his eye may have strayed to the second half of line 4, as Professor Beaurline suggests; or he may have been relying upon his memory and have momentarily forgotten the sequence of clauses. It is also possible that he may have decided to insert a clause that did not appear in the earlier draft.

In line 4 the manuscript omits the verb "stand," which appears in the published versions. Though "England" may seem to satisfy the need for a word to rhyme with "sand" and "strand," it does so by producing a feminine ending, the only one in the poem. Arnold again almost certainly copied his first draft hurriedly and inaccurately, or relied upon a faulty memory of his earlier draft.

The manuscript also seems to suggest that Arnold revised line 12 by adding the word "again," but the meaning of the sentence would have been incomplete without it. Arnold probably began to put the word at the end of the line and then recalled that in his earlier draft he had inverted the normal word order for the sake of the rhyme, if for no other reason.

We, of course, cannot determine in how many ways this version improved upon the previous one. Despite the absence of internal revisions in twenty-three of the twenty-eight lines, the pencilled draft

is obviously not a fair copy. Though the poem seems to be squeezed on the page and Arnold has left no room for the final nine lines, he uses the space well. Except for the first few lines, the first two sections of the poem are so evenly spaced that Arnold must have known how many lines he had to fit on the page. Although the third stanza is written across the side of the page and may seem to be an afterthought, it gives the clearest proof that Arnold knew before he began this section how many lines he needed to make room for. If the third section had been only six lines long, like the second section, he could have begun it further to the right and then the text would have been interrupted only by line 10. But he began exactly where he had to in order to fit in eight evenly-spaced lines.

Still, one wonders why Arnold didn't write the final nine lines on the back of the sheet, since it is blank. Perhaps he was satisfied with the last section and intended to use a fresh sheet for a fair copy in ink. Between completing the British Museum draft and preparing a fair copy for the printer some time before 1867, six verbal changes and numerous alterations in punctuation and capitalization were yet to be made.

If the manuscript is an intermediate draft of the poem, perhaps produced by copying a previous draft or, more likely, by revising it at least in part from memory, it is unlikely that much time elapsed between the two versions. Arnold could hardly have relied upon his memory after a lapse of more than a month or two, and a few days or less seems a more likely interval. If the earlier draft were no longer fresh in his mind, he would probably have copied it more carefully than he did and would not have relied upon his memory.

The dating by calligraphy suggests strongly that, to the extent that the poem reflects events in Arnold's life, the woman in the poem must be his wife. Since the first draft apparently included not only "the darkling plain" but also the references to the moonlit scene at Dover, it is unlikely that the poem was conceived before late June 1851, when according to Arnold's unpublished diary he and his bride first visited Dover together (Allott, Poems, p. 240). Arnold of course need not have written the first draft at Dover. The scene may well have been recollected in tranquility, perhaps late that summer or after a second brief stopover at Dover, October 8, 1851, reviving earlier memories, or during the Christmas season, with the first few discouraging months of school inspecting (October-December) behind him.

If, then, one combines biographical evidence with clues furnished

by the manuscript in general and calligraphy in particular, a reasonable date for the British Museum draft of the poem is between late June 1851 and April 1852.[18]

Notes