The New Edinburgh Edition of the Collected Works of Robert Louis Stevenson

News from the volume editors: Prince Otto

with 10 comments

by Robert Irvine

Editing a Stevenson novel can involve some very small matters as well as some big questions. Robert Irvine describes how one of the smallest points of all—the hyphen—raises questions about historical usage.

First choices
I have been working recently to establish a ‘copy text’ of Stevenson’s 1885 novel Prince Otto. A ‘copy text’ is a particular instance of a text which is taken as a base-line by the editor, against which variations in other versions of the text can be listed, and variations from which in the final published version must be justified. We have chosen the first book edition to perform this function for the New Edinburgh Edition. So my first task as editor is to ensure that the electronic copy text on my screen conforms in all aspects to the text published by Chatto and Windus in 1885. In principle, no editorial decisions are to be made at this stage: where there are mistakes, even an obvious printing error like the omission of a quotation mark, those remain in the copy text, to be corrected when the text is edited and the correction noted.

That pesky hyphen
No editorial decisions to be made in principle at this stage: but one set of editorial decisions is, in fact, unavoidable. In transcribing prose, we pay no attention to line-endings in the text from which we are transcribing, line-endings in prose being dictated by space available on the page, and nothing more. To preserve the line-endings in the transcription of a prose text irrespective of the size of your new page would be to turn it from prose into verse. But to make the most efficient use of the length of line available to him the type-setter of the printed text will sometimes split words at the end of a line with a hyphen. Usually the transcriber of the copy text can ignore these hyphens and restore the complete word. The problem comes when the word that has been split across two lines might have been hyphenated to start with. Deciding whether or not to preserve the hyphen in the copy text in such cases is no longer a case of simply preserving what is on the page in front of you, but requires reference to other sources of information: requires, that is, an editorial decision.

For example, towards the end of book II chapter vii, Princess Seraphina complains of Otto’s unusual interference in his government, ‘To-/ morrow, you will be once more about your/ pleasures’ (p. 144; line endings marked by obliques). So that first word should be transcribed as ‘Tomorrow’, yes? But no: because by the time the reader has reached p. 144 they will know that Stevenson, or perhaps Chatto’s typesetter, always use ‘to-morrow’, ‘to-day’ and ‘to-night’, so this hyphen needs to be preserved in the copy text. Reference to the Oxford English Dictionary (OED) reveals why this is the case: this is the normal nineteenth-century way of writing these words.

Historical hyphens
In many cases, the decision is made easy in this way by the presence elsewhere in the text of the same word, not split across a line-ending, which we can take as our model. Stevenson, at the end of Book I, describes Otto’s palace: ‘The modern pillared front, the ball-/ room, the great library’ and so on (p.57). Is a room for dancing a ‘ball-room’ or a ‘ballroom’? The latter form is usual for us. But on the following page, Otto hears music ‘from the ball-room’, so this is clearly the intended form. Once again, OED confirms that this was the usual form in the nineteenth century; as with ‘to-morrow’, the hyphen was dropped in the twentieth.

In the two examples I have given so far, I have called on the OED as a reassuring presence, explaining Stevenson’s (or Chatto’s) forms with reference to nineteenth-century norms. But we might be unwise to rely on OED to legitimate editorial decisions. Near the start of the novel we are told that across Otto’s dominions ‘the imperial high-/road ran straight sunward, an artery of travel’ (p.6). All examples in OED give either ‘high-road’ or two separate words, ‘high road’, suggesting the hyphen should stay. But early in the next chapter we have ‘highroad’ (p.10), and again a few pages later, and consistently throughout the rest of the novel. So in our copy text the hyphen goes from the instance on p.6.

Casting the net wider
The real problem arises when there is no other use of the word in Prince Otto for us to take as our guide. In this case the editor has no choice but to look at other versions of the text, anticipating what ought to be the next stage of the editing process, while taking scrupulous note that s/he is doing so. Earlier in 1885, Prince Otto was published in seven episodes in Longman’s Magazine; the different format means that line-breaks occur in different places, allowing the same instance of a particular word to appear undisturbed. Using Longman’s we can decide on ‘milkpail’ (Chapter xii, p.202) and ‘lamplight’ (Book II, Chapter ii, p.271), although all of OED’s examples are either hyphenated or two separate words; on ‘farmhouse’, although the hyphenated from seems the normal nineteenth-century one in OED; and ‘mile-stone’ (Book III, Chapter i, p.255) although OED’s last hyphenated example is from 1746. Near the start of the second chapter of the book, that same highroad ‘lay down-/ hill’ from Otto. This is odd, because elsewhere in the novel we only have ‘down hill’ as two separate words, the only model which we are prevented from following by that hyphen. But we do have ‘uphill’ (Book III, Chapter i, p.254), with no hyphen; and Longman’s confirms a decision to take this as a precedent by giving us the second chapter ‘downhill’ as one word.

A final case: near the end of the second chapter, Otto is visibly shaken by what he has heard about himself from an old peasant, ‘re-/ waking’ his host to hospitable thoughts (p.23). ‘Rewake’ seems an odd word to twenty-first century ears, and while OED offers it without a hyphen, it also notes that it is ‘chiefly poet. in modern use.’ The prefix ‘re’ is followed by a hyphen throughout the text where the verb to which it is attached begins with a vowel, even where OED suggests it is not necessary, as in ‘re-awoke’ (p.160); but also in ‘re-enter’, ‘re-echoed’, ‘re-enacted’, and ‘re-arranged’. We are also given ‘re-read’ (p.155). But perhaps that is a special case, as, at the end of the first chapter, Kuno and his companion ‘rejoin their comrades’, and Otto does so at the beginning of Chapter x; in Chapter vi Otto offers to ‘reconduct’ Seraphina back to the court, and asks Gondremark to do so on the next page; Seraphina’s courage ‘rekindled’ in Book III, Chapter I; and so on. These examples set a general pattern where a hyphen is not used where the following verb begins with a consonant, and we only need to refer to Longman’s for a final reassurance that ‘rewaking’ is the form required in the second chapter.


Written by pennyfielding

13/07/2011 at 9:32 pm

10 Responses

Subscribe to comments with RSS.

  1. This is a fine discussion of the tricky subject. Thanks for this – and wish us luck with the same issues in the Essays, though we also have to contend with versions of texts stretching over many years, such as “An Old Scotch Gardener”, first published in 1871 but then reprinted in 1887.

    R L Abrahamson

    13/07/2011 at 10:18 pm

  2. Very interesting contribution, especially since we’ve been working on the first set of copy-texts for a small number of essays. ‘to-/morrow’ came up for us too and we too solved it by reference to ‘to-day’ and ‘to-night’ in the same text.

    There are a couple of other classes of eolh (end-of-line hyphen) problems that we have come across with the essays:

    1. undecidability because of free variation, e.g. ‘beech-/woods’ (‘An Autumn Effect’) where later in the same text we have both ‘beech-woods’ and ‘beechwoods’. In this case we kept the two alternative forms later in the text, and for earlier the example with eolh we used a hyphen (though we could just as justifiably have chosen no hyphen). However, on the principle that no decision on the CT should go unrecorded, we have placed among the metadata at the head of the text, two sections:

    {end-of-line hyphenation changes made while proofing: }
    {end-of-line hyphenation, dubious cases noted while proofing: }

    and noted the ‘beech-/woods’ case in the second of these.

    2. undecidability because of lack of evidence: in ‘Jules Verne’s Stories’ we have ‘story-/teller’. The hyphen was first kept because of ‘lack of evidence’ in the same text and noted as a ‘dubious case’; then I used the Google N-Gram (ask me if you want to know about this useful tool) and saw that for British English texts both 1800-2000 and 1870-1879 (the decade of the text) ‘storyteller’ was the clear favourite, so I changed the text accordingly and put a note in ‘hyphenation changes made while proofing’.

    THEN I looked at the context, the very last words of the text’:

    ‘…what a deal of new matter will be at the disposal of the dormitory story-teller’

    and started to think again… Maybe ‘storyteller’ would be someone who composes stories and someone who reads them out aloud (as in this case) just _had to be_ a ‘story-teller’. I looked at the OED and found lots of examples (in fact a majority there have a hyphen, even though Google N-Grams must give a better sampling), so for semantic-syntactic reasons (‘story-teller’ meaning ‘someone who reads stories out loud’ seems a nonce-use (it’s not listed in the OED) and therefore more suited to the hyphenated form), comforted by the examples in the OED, I put the hyphen back in again and inserted a comment again in the ‘dubious cases’ metadata section.


    14/07/2011 at 3:08 pm

  3. Hi Bob, You write ‘my first task as editor is to ensure that the electronic copy text on my screen conforms in all aspects to the text published by Chatto and Windus in 1885. In principle, no editorial decisions are to be made at this stage: where there are mistakes, even an obvious printing error like the omission of a quotation mark, those remain in the copy text, to be corrected when the text is edited and the correction noted.’

    We are of course talking of letters and punctuation, though the transcription is probably already going to be lacking a lot of formatting: caps/small caps, size of font, spacing before tall punctuation. I have been noting these in the metadata with ref to tagging of special paragraphs title, off-set quotation, epigraph, signature at end of essay.

    Your rule about transcribing the CT first before making emendations is useful, however, as among the metadata at the top of the transcription doc, I’ve put (perhaps mistakenly) the following (apart from the e.o.l.h. points I mentioned before):

    {variant spellings noted in the text while proofing: ***}
    {emendations made while proofing: ***}
    {possible emendations noted while proofing: ***}

    It’s that second one which breaks the principle you give above: first create an accurate transcription of the CT, then think about emendations.

    But let’s look at all the examples I’ve put in this section so far:

    1. after ‘ploughlands’ (p.75) there is a comma that is below the line but clearly a comma in form so has been inserted here [that seems innocuous enough]

    2. italicized punctuation after italic title names (also italicized parentheses around such names) changed to vertical (note that some may remain); [this is formatting – and we’re accepting that we lose some of this anyway – but it’s noted so can be reconstructed if necessary]

    3. title in text “The Two Travellers” changed to The Two Travellers [in italics] to maintain consistency with the rest of the document; AM not sure about this – deserves discussion. [This is again formatting, but is not moving in the direction of modern norms that will probably be edition norms, but consistency within the text]

    And actually that is all I have placed in this section of metadata so far–so perhaps I haven’t broken the rule after all. In the section ‘possible emendations noted’ I’ve placed and started a discussion about ‘develope’, ‘‘for awhile’ etc. – possible cases, including those I’m convinced should be changed (not the case of the two cited, which I take as acceptable variant forms). Any comments anyone on any of the above examples?


    15/07/2011 at 8:58 am

  4. The approach that Robert Irvine describes in his discussion of Prince Otto – tidying up the copy-text before collation – seems to me badly misguided. And I’m not sure that Richard Dury’s workaround of recording some or all such choices in ‘metadata’ is any better.

    Both approaches seem to me to remove from the textual record (and therefore from the reader’s ability to reconstruct) details that reflect the process of the physical translation of Stevenson’s manuscripts to print.

    It would seem to me that preserving this record is of the utmost importance in a serious scholarly edition – if not in details of house style such as the italicization of book titles, certainly in details that affect the literary result, such as the spelling and compounding of words.

    When the copy text is tidied up in the manner described by Irvine, no one (including other editors of other works by Stevenson) will ever be able to know, from the edition that results, whether a given usage – for example, to-morrow, broken by a hyphen – never occurs in the text or does occur, although only as a break at the end of one or more printed lines.

    All record of usages other than the chosen one is simply swept away at the outset, in the name of tidying up the copy text so that it looks good on the editor’s screen. Or, in Dury’s method, details of these decisions are put into some netherworld of metadata that may or may not be readily accessible to readers and that is treated differently from other editorial emendations.

    In both approaches, editorial decisions – to follow or not to follow the majority of usages in the current document, for example, or to follow the OED or an earlier periodical version – are being made and not publicly recorded or defended.

    There is also a problem of method.

    In the examples given by Irvine and Dury, the obliteration of data about line-endings, and the choice of which of several variant forms to follow, is justified by invoking the easy but readily challenged assumptions of uniformity and mediocrity.

    When a variant usage occurs due to the breaking of a word by hyphenation at the end of a printed line, an editorial search is immediately begun – within the work, in all of Stevenson’s works, in other versions of the same work, in the works of his contemporaries, and in authorities such as the OED or even (perish the thought) Google – to see whether instances without the hyphenation (or other variation) occur, and how often.

    If such instances can be found, it is now assumed – about Stevenson’s text – that the end-of-line hyphenation (or other variation) is due solely to the word’s spilling over from one printed line to the next.

    In the edition, the version without hyphenation is then used everywhere, and silently – as if only one version were present in the copy text. When hyphenation does appear elsewhere, for example in a hyphenated compound word, the same approach yields an edited text in which hyphenation is used throughout.

    The trouble with all this is the assumption that the more common, more familiar versions, whether or not by Stevenson, are the versions that Stevenson wanted his printers to use, either because that is how he wrote them or because, if not, he wanted the more familar versions used instead.

    In most instances, this assumption actually rests on no evidence more substantial than that Stevenson passed a text containing the more common and familiar version(s) for the press. We have, in most instances, no manuscripts with which to compare the printed results, and this is particularly true of early works such as essays and book reviews.

    But it is well known that Stevenson had ideas of his own about many details of what we now call accidentals – hyphenation, compounding, even spelling – and that he expressed these ideas, in general terms, in several much-quoted letters, and (as Barry Menikoff has argued for Kidnapped and The Beach of Falesa) actually followed his own ideas in at least some of his manuscripts.

    He was also less than a perfect speller, not always consistent in details, even within a given work and still less over the course of his career, and his handwriting is often difficult to read.

    Therefore, tidying up the copy text, especially when, as in the approaches described here, the familar and ordinary is preferred over the eccentric (Stevenson sometimes spelled this word ex-centric) removes all evidence of what may be the result of compositors following what Stevenson actually wrote – details that he may have considered important in the literary effects that he wanted to achieve.

    Emendation is perfectly permissible in a scholarly edition, and most of the changes in Prince Otto and the essays mentioned in the discussion seem to me plausible and defensible. But this must not be done silently, in the name of (merely) tidying up the copy text from which all else derives.

    Readers more insightful than we are may well see in these seemingly inconsequential differences evidence as to intended literary effects of which we ourselves are unaware. Historians of printing practices may well want to know whether the word tomorrow was ever hyphenated in order to appear on more than one line.

    Pretending that decisions to suppress information altogether (Irvine) or to hide in a special category of emendations (Dury’s metadata) are not editorial or are not worth recording imposes an arrogant omniscience of our own that may not be welcomed by others.

    Surely a better method can be found for handling occurrences like these than the ones currently under discussion.

    Please think again.

    Roger G. Swearingen
    Santa Rosa, California USA

    15 July 2011

    e-mail = rogers99@prodigy.net

    • The ’emendations made while proofing’ are then transcribed into the Emendation List, so this information is not lost. The other two headnote categories are for things ‘noted while proofing’ and are the the equivalent of a personal note made in the margin for future reference.


      15/03/2016 at 2:40 pm

  5. As a general rule, and I emphasize general, Stevenson did not like hyphens, and his tendency was toward a more modern mode of composition — hence to-day, to-morrow, good-bye, &c, while a 19th c. practice, was not his preferred practice. On the other hand, he had his own peculiar or distinctive habits, so he would often hyphenate words beginning with dis (alas! i haven’t got a word in front of me, but it would be dis-) and we would not write or print that word this way. But Stevenson wrote for the ear, and the hyphen after dis was a way of accenting or pronouncing the word. Also, and this has already been noted, he was not robotic or consistent, so he could write “red coats,” red-coats,” and “redcoats,” all in the same text. What to do? Make consistent? Keep them as he wrote them? Editorial decisions all. Another thought, hyphenation comes up in words that are compounds (again, I’m not searching for examples, as I’m writing just in response to the discourse here). The 19th c. practice was toward hyphenation of compounds, the 20th c. is toward the elimination of the hyphen. Here again, Stevenson tended toward the modern. But it is not always clear, as Roger reminds us, when you read his hand if the word is written as two words or as one. And usually the printer would supply the hyphen and Stevenson would not remove it before publication, if he ever saw or noticed it. Thus if you stand back and look at a printed page of his text it will look much more Victorian than modern, even though I believe a fairer representation of that page would look more modern than Victorian. Finally, a small observe on the OED: great as it is, it is not infallible, and is best viewed as a repository of the printed versions of Stevenson’s language than as an indicator of his linguistic intentions.

    Barry Menikoff

    Barry Menikoff

    17/07/2011 at 9:16 pm

    • To the lay person, these small details of hyphenation will seem of little interest, but as the man himself says ‘A technicality is always welcome to the expert, whether in athletics, art or law; I have heard the best kind of talk on technicalities from such rare and happy persons as both know and love their business’ (Talk and Talkers).

      Concerning Barry’s point about the hyphenation of the prefix, I actually believe that what looks like a hyphen is just an isolated linking-line between letters. You find it typically between ‘s’ and a following left-facing small-bowl letter like ‘o’, ‘a’, ‘g’ or ‘d’. RLS lifts the pen from the bottom of the ‘s’, makes a short horizontal linking line which sometimes leaves a space, lifts his pen again to start the following small-bowl letter, sometimes linking with it, sometimes leaving a space. When a space is left one or both sides, it looks like a hyphen.

      School instruction was to link every letter (and even supply the lead-in link-line at the beginning of the word). Incidentally, Italians typically start a word beginning with small-bowl ‘a’, ‘o’ or ‘d’ with a similar lead-in horizontal line, frequently separated from the bowl (I’m not talking about present-day youth, who just write in capitals… but let’s not start on that).

      You’ll find this line between ‘mis-‘, ‘dis-‘ and and one of the small-bowl letters (sometimes a space is used instead). The lack of any historical precedent for a hyphen here and its use only in the context of a following small-bowl letter (never, for example, with ‘dislike’) is enough to convince me it’s a link-line. (Obviously I haven’t inspected every relevant word in the hand of RLS, so if anyone can provide an example of a “black swan”–a case of ‘dis-like’–then I’m willing to consider the hypothesis invalid.)

      In any case, hyphen or not, it doesn’t make much difference, as no-one is ever going to print them (an exception being Veeder in his transcription of the Dr Jekyll draft MS).

      While reading Barry’s sensible comments, it occurred to me that it might be a good idea for the edition to create a corpus of RLS MS transcriptions as they arrive, so that dubious or otherwise-undecidable cases can be checked against his general (or exclusive) practice.


      18/07/2011 at 8:44 am

  6. Thanks are due to Roger for keeping the Edition team on its toes, but I feel that there might have been a slight misunderstanding surrounding our policy.

    Both Bob Irvine and Richard Dury were discussing the process of transcribing the physical copy text itself into a digital file, prior to any kind of textual emendation or adjustment. Although the latter would be ideally identical to the former (in orthography, punctuation and paragraphing), the process will inevitably involve making some changes: that is in eliminating some end-of-line hyphens in cases of obvious typographical splitting (e.g. ‘giv-en’) and retaining others in cases of ambiguous lexis (e.g. ‘to-morrow’).


    1. All possibly problematic cases are recorded during this stage (e.g. by Richard in the metadata header for his transcription document; by Bob in a separate document that lists these instances);

    2. All decisions regarding ambiguous end-of-line hyphens will be recorded in an end-of-line hyphenation list that will form part of each printed volume (as has been a standard practice among numerous scholarly editions) and our online archive that will support the Edition;

    3. Our Edition will take a conservative policy for emendation, avoiding blanket standardization governed by any presupposed house rules: variant spellings and hyphenations will be retained, and decisions for one case will not set the precedent for all cases in the same text, let alone across the entire edition.

    Indeed, we are in complete agreement with Roger on these matters, and hope that my note provides sufficient clarification regarding our emendation policy regarding hyphenated and compound words.

    Anthony Mandal

    17/07/2011 at 10:31 pm

  7. […] often said that RLS puts a hyphen after the prefixes ‘mis-’ and ‘dis-’ (see earlier discussion). I think this is merely a linking line between ‘s’ and c/d/e/g/o/q and no attempt to […]

  8. […] There are also forms such as ‘dis-cretion’ (p. 115), which shows that the handwritten line between ‘s’ and a letter with left-facing bowl (c, d, g, o or q) has been interpreted as a hyphen. [For EdRLS, these marks have been interpreted as a non-significant link line; see this post in the blog and this one for a discussion. Barry defends his view in one of the comments to another post]. […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: