As explained above, there were in effect five more or less contemporaneous printings of the novel: the Australian serialisations in the Age, the Observer and the Evening Journal, the three-volume edition published by Bentley in London, and the one-volume paperback published by Harper in New York. Of these five printings, two need not be taken into separate consideration here. Collation has revealed that the New York edition was set from the London edition. It incorporates all of that edition's variations from the newspaper versions, and its own substantive variants are all explicable in terms of compositorial or, in some instances, editorial involvement.note The Harper and Brothers firm is known to have imposed its own system for punctuation and spelling on the "foreign" texts it reprinted, and there is no evidence that Martin herself had anything to do with these alterations or, indeed, with any other aspect of this edition.note

The other case involves the two Adelaide newspaper printings. As the weekly organ of the South Australian Register, the Adelaide Observer was printed on Saturday mornings (with an early country and inter-colonial edition bearing the Saturday dateline but printed on Friday mornings). The Evening Journal was the Register's companion afternoon daily, and printed "The Silent Sea" in its Saturday edition.note Apart from the instalment headings, the serialisation ofthe novel in the Evening Journal and the Observer was printed from the same typesetting (such elements as the coincidence of broken types make this clear), although it was shifted and re-formatted in galleys to accommodate the difference in column length between the two papers.note There are, therefore, three concurrent and authoritative type-settings and printings of the novel: newspaper versions in Adelaide and Melbourne, and the Bentley three-decker in London. The relationship between these three, described

  ― xxxi ―
below, has determined the selection of copy-text for the present edition.

It is uncertain how much time Martin actually had for revision of the Australia-bound proofs of the first volume: the transit time for London-Melbourne mails had been reduced to thirty-five and a half days, with weekly sailings alternating between the Peninsular and Oriental line and the Orient line.note Yet, despite her stated 22 January dispatch deadline, the serialisation did not begin until 2 April. Whatever the time constraints she was under, collation of the Age (hereafter, Mlb),note Evening Journal/Observer (Adl) and Bentley (E1) texts indicates that Martin made many more proof revisions in Volume I (approximately 400) than were subsequently made in Volumes II and III (approximately 250 in each), although other revisions to the later volumes may, as noted above, have already been made on the manuscript returned to her by Bentley, and would thus have made their way into all printings via the Bentley proofs.

Collation of Mlb, Adl and E1 also suggests that she undertook little or no revision of the set of proofs from which the Mlb version was printed. Evidence of this is provided, for example, by the instances (recorded in the List of Editor's Emendations) in which patently erroneous readings in Mlb evidently copy readings in the proofs which are corrected in Adl and E1.note The variant readings (detailed in the foot-of-page entries in this edition) which are shared by Adl and E1 and which improve upon Mlb also suggest strongly that Mlb was essentially unrevised.note

The foot-of-page apparatus shows that Martin significantly revised the proofs for the novel's serialisation in her native Adelaide. While there are a few instances in which it cannot be determined with certainty whether variants in Adl are the result of authorial revision or editorial or sub-editorial caution,note many of the substantive variations introduced in Adl, including its experimental changes in vocabulary and phrasing,note clearly show an author reworking her text, and Adls expansion of the list of the duties of a mine purser is patently authorial.note

The occurrence of shared readings between Adl and E1 suggests that in a number of instances Martin may simply have transcribed alterations already made to the proofs for Adelaide when she revised a third set of proofs for Bentley.note However,

  ― xxxii ―
foot-of-page evidence of variation between Adl and E1 makes it clear that this was not the case invariably and that the emendations made by Martin for Adelaide were not always reproduced on the third set of proofs. The nature of the variations between Adl and E1–including an occasional pattern of progressive change, in which an altered Adl reading is further emended in E1 by the addition of intensifiers or fulsome descriptive terms'snote–suggests that the revisions sent to Bentley entailed not only transcription but also a separate reading of the text, perhaps at a date when the pressure of deadlines for the Australian newspapers was no longer a factor.note

The marked extent to which the wording in E1 often varies from that in M1b and Adl is evident in the foot-of-page entries and clearly indicates authorial revision. However, apart from alterations to the novel's time scheme (creating, not correcting, an error in one instance),note the changes in E1 do not appear to meet Bentley's concern about "blemishes" in the novel's internal coherence. Inconsistencies in the names of several minor characters remain uncorrected,note and the arrangement of chapters is unchanged. However, in the first chapter E1 introduces a coyly sentimental exchange concerning Helen Paget's possible answer to Victor Fitz-Gibbon's proposal, in place of the remark in Mlb and Adl in which Helen had drawn a parallel between herself and a French hotel chambermaid.note This apparent concession to prudery may reflect Martin's concern to cater for her English readership, and, in particular, to address any possible objections (perhaps flagged by Bentley on the proofs) that might be raised by Mudie's circulating library, given Mudie's self-appointed role as the guardian of literary moral standards.note

On the whole, the variants in E1 reflect a move towards a smoother, more formal and perhaps more self-consciously "literary" style,note which deprives E1 of some of the vitality, immediacy and individuality of Mlb. It leads repeatedly to conventional and clichéd prose, and, in one instance, to the rejection of a particularly vivid simile in "his whole body was like a branch of shaking leaves".note Additionally, and perhaps further indicating Martin's sense ofher English readership in revising the proofset for Bentley, there is a subtle but significant lessening of the Australian flavour that is intrinsic to the earlier versions of the text.note

  ― xxxiii ―

Collation of Mlb, Adl and E1 also shows considerable revision by Martin of dialogue and of dialect speech. Some aspects of the dialogue are consistently "corrected" from Mlb to E1 across the three volumes. For example, the informal phrase "you/ I/ she/ they better" occurs a number of times in the speech of various characters in Mlb but is sometimes emended in Adl, and almost always emended in E1, either to "you'd better" or to "I/ she/ they had better".note There is also a considerable degree of difference between the three states in their representation of dialect and pronunciation by means of variations on conventional spelling. While it is probable that some alterations both in the dialogue and in the presentation of dialect are compositorial "corrections" or slips, the high number and consistency of these variants indicate that Martin was adjusting, and often accentuating, her representation of idiom and dialect as part of her revision of the proofs for Adl and E1.note Such variants have in general at least a semi-substantive force, and in some cases the variation carries a significant shift in meaning, as in the alteration of 'varmin' in Mlb and Adl to the less opprobrious "varmint" in E1.note

In the state in which they left Catherine Martin's hands, whether for the Adelaide and Melbourne newspapers or for return to Bentley, the proofs of The Silent Sea would all have reflected alike, in their accidentals, the manuscript from which they were set, as interpreted by Bentley's printer according to the printshop's house-style. As far as the set of proofs revised for Bentley is concerned, it seems likely that, having already been set in type, its accidentals would not have been systematically altered by publisher or printer. On this basis, therefore, it can reasonably be assumed that E1 reproduces, by and large, the accidentals of the proofs as sent to Martin and revised by her (although it is unlikely that she would have altered more than a small percentage of them).

In the case of the Australian versions, the picture is different. Collation shows that the accidentals of the proofs were altered to accord with the specific house-styling practices of the Melbourne and Adelaide newspapers. Mlb thus follows the normal practice of the Age in giving numbers in figures and of ending words with "-or" not "-our". Adl reflects Evening Journal/Observer practice in capitalising nouns such as Director, Company,

  ― xxxiv ―
Bank and Church. Compound words such as "salt-bush" (hyphenated in E1) were normally separated in Mlb but unified in Adl, while each paper favoured a different convention for prefacing direct speech. The resultant disparity in accidentals between the three versions is considerable, and ranges from minor differences in hyphenation to variations of textual significance. For example, a variation in accidentals between the Mlb/ E1 reading " 'Yes. I can.' " and the Adl version " 'Yes, I can. ' " in the first chapter affects our sense of the mood in which Helen agrees to a request of Victor's at this point.note Unique Adl readings also result from the frequent alteration of sentences beginning with "But" or "And" into clauses of the preceding sentence (through the substitution of a semicolon for the full stop), sometimes with consequent marring of sense or syntax. The evidence that the accidentals of the proof sets for Mlb and Adl were adapted to and thus obscured by the respective house-styles of the two newspapers means that, although Martin probably undertook some correction or revision of accidentalsnote in addition to revising the substantives in at least two of the three sets of proofs used for printing, her own revision of accidentals for the serial versions is, by and large, no longer recoverable. While E1 represents the last revised of the three states of the novel in substantives, its accidentals (even with the incorporation of whatever revisions of accidentals Martin made on the proofs returned to Bentley), paradoxically represent the least-altered state of the text.

In summary, The Silent Sea, as it existed in 1892, may be seen as a work in process–with its author prepared to allow publication in each of its phases of revision. In substantives, Mlb represents the earliest state, that is, the closest available state to the sets of unrevised proofs furnished by Bentley. Adl represents those proofs revised by Martin in some particulars, although not all its revisions were carried over to E1. In substantives, E1 represents the latest state, with Martin's fullest revisions, prepared for a different readership and a different form of publication, although these revisions do not always agree with or incorporate those she had made to Adl. The situation regarding the accidentals, however, is effectively reversed. E1 contains the earliest recoverable state of the accidentals, being initially set

  ― xxxv ―
from Martin's manuscript for the four sets of proofs and then corrected, probably only to a very slight degree in the case of its accidentals, from the revised proofs by Bentley's printers. Mlb and Adl, on the other hand, are both radically affected in their pointing and presentation by the house-styles of the respective newspapers.

In the present edition, E1 has been chosen, in all but a few instances, as the best available witness to the accidentals of the unrevised proofs.note It has proved possible to recreate the text of the missing proofs, or to approximate it as closely as possible, by basing the reading text of the present edition on the accidentals of E1 while incorporating the substantives of Mlb. Thus the reading text represents a state of the text which actually existed, and which is the earliest recoverable state and the one closest to the author's manuscript.note

Catherine Martin's revisions in Adl and E1 are given at the foot of the page in the Textual Apparatus, which thus constitutes a record of the competing authorial forms of the text. While the primary reading text in itself provides a reliable first acquaintance with the work, the entries in the Textual Apparatus are of particular value to readers wishing to understand the work as it actually was in 1892–in process, acted upon by authorial, editorial and audience influences. Accidental variants in Mlb and Adl are not recorded, except where clearly authorial (e.g. when associated with a substantive variant), given the probability that their variation from E1 is compositorial in origin. However, a full collation of a representative chapter in Mlb, Adl and E1 is given at pages 569-73, and the samples of the different printings reproduced at the front and back of the book show how the physical text appeared to readers of these versions.

Made up of the primary reading text and the foot-of-page apparatus, this Colonial Texts Series edition preserves for the reader a sense of The Silent Sea as an evolving and multi-state authorial work. It distinguishes between Martin's levels of involvement with her work–on the one hand, as an author in full compositional flow, and, on the other, as a sometimes judicious, but at other times self-censoring, formalising and occasionally inattentive reviser–and it respects Martin's own intention to offer her Australian newspaper readerships and her (mostly English) three-decker audience competing versions of her novel

  ― xxxvi ―
which differed in wording as well as in form, while at the same time it preserves the novel's original Australian quality.