This is a draft, and I welcome comments.
Seventeen words appear in early printings of Uncle Tom’s Cabin both as a contraction and as a possessive form, including possessive nouns and the possessive case for gerunds, and they are listed in Table 1 below. The National Era serial (1851–1852) and the Jewett Illustrated edition (1853) have no typographical distinction between possessives and contractions, but two Jewett editions, the 2-volume First (1852) and the “Million” (1852/1853), and the Houghton Osgood “New Edition” (1879) have a typographical distinction: in contractions, a thin space precedes the apostrophe; but in singular possessives, the apostrophe abuts the preceding letter (with no thin space).
Table 1: ’s Contractions and Possessive Forms in the Era Serial and Four Editions of Uncle Tom’s Cabin (PDF)
All “ ’s” contractions with a corresponding possessive form in the same text together constitute a significant sample size of over 280 instances in each of five typographical settings of the text. In the three editions in which a distinction is observed, there are only two clear “errors” of execution: both are cases in which an “ ’s” contraction lacks the expected thin space before the apostrophe, a rate of consistency above 99.5 percent. For the purpose of these counts, I have excluded instances of possession in chapter titles (i.e., “Clare’s” or “Tom’s”). Though these spacing conventions are representative of the typographical texture of the early printings of Stowe’s text, they are not authorial because in the fragmentary extant manuscript pages the distinction is not present in Stowe’s hand. Also, I should add, spacing of contractions is not a distinction that concerns dialect (NB: To be the subject of another post). The setting of “I ’m” and “you ’re” and “does n’t” in the three editions with spaced contractions, for example, mirrors the practice of dialect forms like “I ’s” and “you ’s.”
When determining these counts, some conventions for regularization have been imposed. Typographically speaking, a “thin space” is a designated width of type: 5 thin spaces are the width of an em (i.e., 5 to the em), and the “em” is the standard unit of type measurement. In 19th century printing, a quad the width of an em is typically used to separate sentences. Whether a space before an apostrophe is made by a thin-space type–with an approximate width of 1.0 mm in the Jewett First Edition, 0.8 mm in the “Million” edition, or 1.0 mm in the Houghton Osgood “New Edition”–is a matter of judgment, one which can benefit from tools like a magnifying loupe scored with a tenth-millimeter gauge.
No, that’s not everyday literary scholarship, but I have measured several hundred thin spaces with a loupe and suppose that I may as well admit it. As a sanity check for assigning space width, I assigned spaces for contractions and dialect word forms during two keyboardings. If the two transcriptions differed, I measured the distance with the magnifying loupe and/or made a judgment call about the relative width of the space in this instance as compared to typical spacing in the immediate context of the same line. That is, spaces were double-keyed and checked whenever the two keyboardings disagreed.
So, the actual procedure requires a bit more judgment. When encoding spaces during transcriptions, I have marked some spaces as thin spaces outright and some spaces as thinner, relatively speaking, than the surrounding spaces on the same line. When words are widely spaced, with frequent thick spaces (3 to the em) between words, a medium space is thinner, relatively speaking, than the surrounding thick spaces. Also the opposite: when words are tightly spaced with frequent thin spaces between words, a hair space (7 to the em) is thinner, relatively speaking, than surrounding thin (5 to the em) spaces. I consider both of these as typographically equivalent to “thin” spaces. I disregard the distinction between “thin space” and “relatively thin space in context” in Table 1, but the transcriptions with the distinction recorded shall be made available. Because varying space width between words is typical of prose typography–sometimes so a paragraph may fit in a certain number of lines, sometimes so a chapter will fit on a certain number of pages with enough space to permit an illustration, etc.–the “relative” width of a space against the background of its surrounding content is a better way to assess the practice. Another complication is that the width of a space is indeterminate or an approximate space, because, for example, the space is preceded by a letter such as “f” with an overhanging terminal into the space. Again, judgment.
One consequence of identifying this conventional distinction between contractions and possessive forms is reasonably clear for modern reprint editions that aim to be scholarly. Mid-century compositors and proofreaders of the two Jewett editions must have recognized a spacing distinction between the two word forms as characteristic of quality typesetting. And, furthermore, the compositors for the 1878 Houghton-Osgood edition, over twenty-five years later, recognized and observed the same distinction. Therefore, for scholarly reprints, an editor who designates a reprint of the Jewett First or “Million” edition as “unmodernized” should reproduce type space in contractions, at minimum to distinguish contractions from possessive form. If you acknowledge that distinction, I believe the same principle should apply for all contractions that have the “ ’s” form without a corresponding possessive and for other contraction forms that close with “ ’ll,” or “ ’re,” etc., which are far more frequent. In reprints of the Era serial (1851-1852) or the Jewett Illustrated edition (1853), by contrast, no type space should be inserted into contractions: the convention is not present in the original texts.
As the practice of spacing in contractions has died out in modern printing (I am certain that it has, though I am not sure when.), and was not observed even in another edition by the same publisher (the Jewett Illustrated Edition ), no historical “flavoring” is achieved by retroactively applying a texture of spacing to contractions in a reprint of that text or of the National Era serial. Though I must leave the following comment as a side note, which cannot be duplicated except by re-transcribing the three editions that observe the distinction–imagine yourself spending 4 years double-keying all three texts with thin spaces, confirming that each thin space agrees, measuring with a loupe when they do not, and making a judgment call–I did not discover this difference between possessive and contraction forms until after both the Jewett first edition and the “Million” edition had been transcribed. I set out to record type spacing within words, but I only noticed the distinction when devising regular expression routines to normalize spacing for the purpose of reducing the number of variants that are noted during computer collation. The transcription of the Houghton Osgood edition along the same principles, recording typographical spacing in contractions, confirmed that the pattern was present in that edition also.
A second consequence must also remain speculative, and it can only be tested by preparing another edition. For editors who wish to establish an authoritative text that conflates the authority of multiple editions, the matter of contraction spacing is complicated. If the Era form is designated the copy-text for the entire reprint edition, in the sense that Sir Walter W. Greg suggested in “The Rationale of Copy-Text” (and Fredson Bowers and the early CEAA advocated) because it is held in most cases to be closest to authorial word forms and the authorial texture of punctuation, then no typographical distinction between possessive and contraction forms is present and should not be observed in a reprint edition. However, if the Jewett First edition is designated the Greg-Bowers copy-text for a reprint edition, its conventions should govern for the typographical reproduction. To do otherwise would be to modernize. One could modernize everything “for the convenience of readers,” but where’s the fun in that?
My preliminary analysis of variants suggests, however, that the manuscript served as the printer’s copy-text (setting copy) for the Era serial text, that the Era serial text served as the printer’s copy-text for most chapters in the Jewett edition, that two separate manuscripts served as printer’s copy-text for several chapters when the composition of the manuscript ran even both with serial publication and with First Edition typesetting, and that the typeset book or galley pages served as printer’s copy-text for the final installments in the Era serial. If my analysis is correct, an editor of a conflated text would need to explain the rationale for choosing a copy-text (in the Greg-Bowers sense) for each installment and either observe the typographical texture of the designated copy-text for each section or to regularize and normalize. Various arguments are defensible: one could appeal to authorial intention based on the practice of manuscript fragments, appeal to the convenience of readers who may be unsettled by alternate typographic textures in the same edition, or appeal to the prominence of the Jewett first edition as American literature’s textus receptus. Each of these options is speculative because no one has prepared a conflated text, and my editorial choice of the Era as a base text on which to hang variants and revision narratives (not the copy-text in the Greg-Bowers sense) will not resolve the matter either: it permits me in my editorial introduction to sidestep the fuller elaboration of principles that a conflating editor would be obligated to provide.
Be that as it may, these details are likely only to concern scholars of print history or editorial theory: they will not be reported in my printed scholarly edition except as a class of normalizations in the lists of variants. The practice perhaps should elicit some concern for deciding what a “word” is in quantitative analysis or so-called distant reading, as the common computing definition of a word, a series of letters separated by a space or punctuation, which is frequently invoked in word counts, requires a decision about whether a thin space, or a relatively thin space given the surrounding context, is actually a marker to separate words. Finally, even the literary critic who wishes to quote accurately for a study that concerns the exact wording of a historical version of the text would be best served by reviewing a facsimile and learning to read type space or by checking my archived transcriptions of other editions (though those are not yet posted).
In a culture in which whether one or two spaces should follow a period is a subject of popular though ill-informed discussion, I think it’s reasonable to maintain that such details as how to record and how to print spacing of contractions in historical reprints does matter. I hope at least one of the 25 eventual readers of this post find it useful.
- See notes to chart for errors. A caveat: The form “mass’rs,” which functions as a frequent possessive in the Era and is reprinted in chapter 10 of later editions, is arguably an error in form but is not treated as an error in execution for this analysis because it does not involve contraction spacing.
- I am simplifying here to describe the typographical conventions as practiced in these editions. Typographical conventions for conventional space widths differed over print history and within printing traditions. See Philip Gaskell, A New Introduction to Bibliography.