Dates and ChronStructs:
Dynamic Chronology in the Orlando Project

Isobel Grundy, Patricia Clements, Susan Brown, Terry Butler, Rebecca Cameron,
Greg Coulombe, Susan Fisher, Jeanne Wood (members of the Orlando Project)

University of Alberta, Canada

(as printed in Literary and Linguistic Computing, 15:3 (2000). 265-89)

1 Why Electronic Chronology?

1.1 The Orlando Project

The Orlando Project is working to produce the first scholarly history of women's writing in the British Isles. In this work, literary history will be integrated with the historical events and processes - political, economic, scientific, cultural - that provide its material, shape the trajectory of literary history, and in turn are influenced by the activities of authors. Behind an analytic narrative history to be provided in printed volumes will stand an electronic textbase, which will open to users the massive accretion of detail on which the over-arching account will rely. Every aspect of this work is still in progress.

    The volumes will be available electronically as well as on paper, but they are designed to be read through in the order presented to the reader, whereas the textbase invites reading along multiple branching paths. New advances in humanities computing (some of them made by the project itself) are allowing the literary-historical quest for a more accurate and adequate account of women's writing.

    Both volumes and textbase are encoded in Standard Generalized Markup Language (SGML). By this means information about the texts under discussion, and about their authors, can be marked for complex retrieval by users, and particular threads of argument - concerning the influence of an individual writer, or the development of a genre, or the theatrical performance of works by women - can be followed through history in the detail of the textbase as well as in the more summary analysis in the volumes. SGML has been chosen because it is independent of particular hardware and software, and because it allows users to design document type definitions suited exactly to the needs of their research. For Orlando, SGML allows tagging of our documents for structure, content, and interpretation.

    Most projects using SGML for purposes of literary scholarship are using it to encode pre-existing, primary literary texts. Thus the Brown Women Writers Project offers texts by early modern women, encoded to serve the basic searching and navigation needs of readers who engage with these texts very much as they would with a printed book. We are doing something different. Instead of being faced with someone else's text and applying tags to it, we are encoding our own, original critical text even as we write it, designing our words to fit the tags as well as vice versa. The writing and encoding are two sides of the same process; words and tags function in symbiotic relationship.

    As well as bridging the fields of literary history and humanities computing, the Orlando Project is collaborative across international boundaries. It brings scholarly co-investigators together with post-doctoral fellows, specialist project employees, and graduate research assistants, besides a far-flung advisory panel. We have so far been funded by a Major Collaborative Research Initiatives grant of $1.6 million Canadian from the Social Sciences and Humanities Research Council of Canada, and by related contributions from the Universities of Alberta and Guelph.(1)

1.2 The scholarly issues

Scholars in the field of women's writing are still struggling free of the rampant misconceptions of the past. Mainstream literature can be studied in the light of an accumulation of accredited, relevant detail gathered by generations of scholarship. For women's writing such informative contexts have been lacking, and one result is that this writing is often dehistoricized and seen in essentialist gendered terms that impoverish response.

    This problem has been exacerbated by an academic environment in which until recently the pendulum of humanities investigation had swung away from specific, concretizing detail. Critical opinion has now shifted. In 1991 William Beatty Warner faulted Foucault's critical analysis of the novel on the grounds that Foucault 'rounds out the rough-edged historicity of happenings. [He] does not attend to the microlevel of actual behaviour in any time or place. Instead [he] abstracts and summarizes' (Warner, 1991, p. 195). At this juncture in women's literary history it is essential that scholars should revisit specific times and places to accumulate the evidence for new generalizing. In the words of Paula McDowell, historian of women's writing and of print culture, 'A new angle of vision reveals new questions that can be asked of familiar archival sources' (McDowell, 1999, p. 146).

    The Orlando Project starts out from a new angle of vision: to put writings by women at the centre of an account of the role of writers in the historical process. It has chosen to present its rounded account in volume form, and to use its electronic textbase as its means of capturing 'the rough-edged historicity of happenings.' Happenings that change the course of history are represented there at microlevel, along with large numbers of events whose significance is not particular but cumulative. The total of these dated statements in the textbase so far is 24,484. Not only happenings are being captured, but processes too, which have a more complex but no less rough-edged relationship to the passage of time.

1.3 The uses of chronology

In the context of studying history or literary history, chronology has a bad name, associated in the popular mind with rote learning, with lists of monarchs' reigns or of Great Books. Recent historiography has turned away from events to processes, and has dismantled familiar and facile concepts of periodization. There are almost no dates in Michel Foucault.

    But attitudes to dates, too, are historically constructed. To writers of the early modern period, dates were a kind of guarantee of the authenticity of experience. Seventeenth-century poets (Katherine Philips, Martha Moulsworth) specified the precise dates of the deaths of loved ones, the precise duration of marriages or the all-too-brief lives of babies: for them, to keep a careful account of time was to sanctify it. Writing for a more public sphere, Eleanor Douglas and Elinor James used dates (usually political anniversaries) in an analogous manner to enforce the importance of the message they had to deliver. In the early modern period dates had transcendental significance, and coincidences of dating were seen as part of the divine plan.

    Modernist women took the concept of the dated historical period, made fun of its arbitrariness, but seized and transformed it for important uses of their own. Ada Leverson's Reminiscences of Oscar Wilde offer a fantastical, Impressionist sketch of the contrasted 1880s and 1890s (the time of Wilde's flourishing), before narrowing to the precise year, the precise final evening of triumph before his fall. She uses dates, that is, to depict the collision of life with art, history with imagination. Virginia Woolf creates in Orlando an equally fantastical account of English literary history (the onset of the Victorian age represented by damp settling down, shrubberies and beards sprouting). She then goes on to head each section in The Years with a baldly stated year-date, as a tool by means of which to arrest the flow of time and to observe the historically fashioned interiority of the creatures living in it.

    Our history of women's writing in some sense emulates The Years. We have chosen chronology as perhaps the most vital tool for relating historical events and processes to each other, for trying to recreate simultaneously various aspects of history: on one hand the randomness of detail (on which chaos theory has recently shed new light), on the other the seemingly inexorable process of gradual change; on one hand a hegemonic narrative constructed of dates generally known, on the other various less authorized stories of writers' careers and women's issues. By this means we can present history as consisting of innumerable overlapping narratives. The entire textbase (including documents running to several thousand words) can be drawn on for multiple chronologies generated on the fly by readers according to their needs and interests.

    The choice of chronology as a central (though not the single) organizing principle is firmly in step with current practice. Chronologies are on offer everywhere as helps to students: rudimentary timelines of the author's life and works in paperback teaching texts have been succeeded by fuller and more nuanced ones. The Oxford English Novels series, which began in the late 1960s, provides early examples of such chronologies. They provided half a page each on Clara Reeve and Ann Radcliffe: for each author the birth, death, and one life event as well as the dates of their published works were given. Top-of-the-range chronologies of comparable novelists in the last few years (Elizabeth Griffith, whose The Delicate Distress was edited for the University Press of Kentucky by Cynthia Ricciarelli and Susan Staves (1997), or Eliza Haywood, whose History of Miss Betsy Thoughtless was edited for Broadview Press by Christine Blouch (1998)) run to seven or eight times that length, and make a real effort to explain the significance of the events and sequences that they record. Still, the usefulness of such single-author chronologies on their own is severely limited.

    Contemporary scholarly interest in chronology has also produced more complex ways of representing temporal relationships. Brief timelines in teaching and reference texts sometimes go beyond a single list, to separate various categories of event into separate columns or other groupings. The Oxford Illustrated History of English Literature (Rogers, 1987), for instance, provides one colume for 'Events' and another for 'Literary Works'. General or specialized reference books are proliferating that consist entirely of chronology. One example, Timelines of the Arts and Literature (Brownstone and Franck, 1994), provides an average of about one page per year, with lists in separate boxes for Literature, Visual Arts, Theatre and Variety, Music and Dance, and World Events. Detailed single-author chronologies (Edward Bishop (1989) on Virginia Woolf, Norman Page (1988) on Samuel Johnson) provide a kind of skeletal, time-dominated biography. Only with these last do we reach a sufficiently granular level to differentiate, on a regular basis, months and often days as well as years.

    In the life of an individual, or even on the world stage, it is hard to say anything meaningful about sequences of related events without differentiation by year and month. In a chronology of Mary Wollstonecraft the statement '1797, marries William Godwin, bears daughter (the future Mary Shelley), dies' would be almost wilfully uninformative. Clearly preferable is the form:

    29 March 1797, marries William Godwin.

    31 August 1797, bears daughter, the future Mary Shelley.

    2 September 1797, dies.

In Timelines of the Arts and Literature, this death and this birth are mentioned, in the Literature box; the marriage is not. Unsurprisingly in view of this work's overall scope, the Wollstonecraft narrative disappears.

    Even with (in some cases) the benefit of separate columns to categorize events, and (in other cases) the further benefit of specifying day and month, all these printed chronologies share a serious handicap in their static or passive character. They yield no information besides what has been pre-selected.

1.4 Electronic chronology

The electronic medium offers obvious advantages for flexibility and comprehensiveness. But even electronic chronologies are often not far removed from mere lists. They are usually abstracted from the research materials used to create them, and therefore have the same static quality as printed chronologies. They tend to be reductive; they summarize results rather than stimulating further enquiry.

    A critic of hypertext (a genre to which the Orlando Project textbase belongs, albeit as an atypical example) argues that the collection of historical material into a secondary archive flattens it out, diluting meaning and emphasis by making everything equally available and apparently undifferentiated. In such an electronic text 'there is no future and no history, only a timeless succession of instants'. 'History thus becomes timeless, above rather than submitted to time' (Johnston-Eilola, 1997, pp. 167, 95). It will become apparent from the account that follows that the Orlando Project has taken note of these pitfalls, and put in practice methods designed to avoid them.

    In this context it is interesting that the electronic Romantic Chronology composed by Laura Mandell and Alan Liu (2000) (http://english.ucsb.edu:591/rchrono/default.htm) quotes as its epigraph William Wordsworth addressing his sister: 'No joyless forms shall regulate / Our living calendar." (They have then succumbed to jokiness by continuing the quotation, 'We from to-day, my Friend, will date' - as if the verb 'to date' were an absolute, intransitive construction and dating, in the sense of making chronologies, were a way of life.) The Romantic Chronology (which runs at varying depths of coverage from 1785 to 1851, with preliminary subordinate chronologies of the seventeenth and the eighteenth centuries as well) is undoubtedly the nearest tool in existence to the Orlando chronology.

2 The Orlando Chronology

In contrast to this norm, the Orlando chronology has the following advantages: it is closely linked with the more organically structured discursive materials from which the chronological material emerges; and it emphatically fosters further enquiry. Furthermore, in its full form it encodes the record of its bibliographical sources, and calls attention to points of scholarly controversy over the interpretation of its material.

    To date, the Orlando textbase comprises two distinct types of documents. The first type consists of sometimes extensive bio-critical articles on individual writers (primarily British women writers but also a selection of male and international women writers) which are deeply tagged for structure (e.g. paragraphs, document divisions), content (e.g. names, organizations), and interpretive material (e.g. political affiliations, sexual identity, occupation; authorship issues, intertextuality, landmark texts). The second type consists of briefer records of related material, of the historical landmarks and minutiae that contextualize our view of literary history.

    The Orlando Project has taken advantage of the electronic medium by structuring both types of material in such a way that dated statements can be lifted from their original document or location, and produced in new chronologically ordered combinations. As the dated statements, encoded as <ChronStruct>, have been written by many different team members (from co-investigators to graduate research assistants) in the course of researching many different topics (in literature, biography, politics, law, history of medicine, etc., as well as on particular authors), they offer to scholars interacting with them a genuine unexpectedness of results - unlike the familiar hypertext use of hard links, which resemble printed material in having been pre-selected.

2.1 ChronStructs

The ChronStruct is the Orlando Project name for the tag enclosing a statement anchored to a particular date. Our massive accumulation of ChronStructs allows multiple pathways through the material available in the textbase, including pathways not foreseen by any of those who entered the discrete information. Just as a researcher in a library is free to let the results of past research dictate the direction next to be taken, so will users of the Orlando Project be free to roam through far more extensive material than could be contained within the covers even of a whole series of books, and to order the results through the basic relationship of temporal sequence or precedence.

    In the case of biocritical documents the basic unit for transferring into potential timelines is the ChronStruct, encoded as follows. All the fairly self-explanatory elements reproduced here (the date, the statement, the citation reference) are compulsory:

<CHRONSTRUCT>
    <DATE></DATE>
    <CHRONPROSE></CHRONPROSE>
    <BIBCIT></BIBCIT>
</CHRONSTRUCT>

Here is an actual example from the document on Clemence Dane (from which tags not relevant to the example have been excised):

<CHRONSTRUCT>
   <DATE>22 July 1956</DATE>
   <CHRONPROSE>Clemence Dane's play Ellen Terry in the Theatre was performed at the Barn Theatre on Christopher St John's property at Smallhythe, Kent, for the Ellen Terry anniversary celebration.</CHRONPROSE>
    <BIBCIT>Cockin 177</BIBCIT>
</CHRONSTRUCT>

2.2 ChronEvents

In the case of historical material not residing in biocritical documents (and therefore relating tangentially, not primarily, to women's writing), the basic unit to be transferred into an on-the-fly chronology consists of a ChronStruct enclosed in a ChronEvent, as follows:

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE></DATE>
    <CHRONPROSE></CHRONPROSE>
    </CHRONSTRUCT>
    <RESPONSIBILITY>
    <DATE></DATE>
    </RESPONSIBILITY>
    <BIBCIT></BIBCIT>
</CHRONEVENT>

Here is an example.

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>1656</DATE>
    <CHRONPROSE>J. Scultetus used pictures of instruments employed in mastectomies to illustrate his Armamentarium chirurgicum.</CHRONPROSE>
    </CHRONSTRUCT>
    <RESPONSIBILITY>
    <DATE></DATE>
    </RESPONSIBILITY>
    <BIBCIT>Green 36</BIBCIT>
</CHRONEVENT>

    Again all these elements are compulsory; irrelevant ones have been erased. The chronological unit named ChronEvent (designed to encode historical material which does not come from documents on authors) contains two further mandatory elements in addition to those in the ChronStruct (which, when it stands alone, comes from a biocritical document, and is designed to encode material on an individual).

    One extra element in the ChronEvent is the Responsibility tag, which records which member of the team has worked on a document, on what day, and which action they have performed, whether writing and tagging, revising, or checking various aspects of the data. Biocritical documents have a single site for Responsibility tags applying to the entire document.

    The other extra element is the ChronEvent tag, which encloses the whole. A ChronEvent (unlike a simple ChronStruct) may further include the non-compulsory elements ShortProse and P. Within these the original or core statement contained in the ChronProse can be elaborated, explained, or built on, thus:

<CHRONEVENT>
    <CHRONSTRUCT>
     <DATE></DATE>
    <CHRONPROSE></CHRONPROSE>
    </CHRONSTRUCT>
    <SHORTPROSE><P></P></SHORTPROSE>
    <RESPONSIBILITY>
    <DATE></DATE>
    </RESPONSIBILITY>
    <BIBCIT></BIBCIT>
</CHRONEVENT>

    There are reasons inherent in the material for differentiating the ChronStructs lifted out of biocritical documents from the ChronEvents which do not make part of any longer document. The ChronEvent tag ensures that any ShortProse will be lifted into potential chronologies together with its parent ChronStruct (as in the example immediately above). This is useful for ChronStructs that convey historical rather than directly literary material, as they often require the gloss or elaboration provided by a ShortProse. (Again, in the following example, tags marking, for instance, Title and Name have been removed, in order not to confuse the issue.)

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>September 1826</DATE>
    <CHRONPROSE>The conservative Quarterly Review, discussing Walter Scott's Lives of the Novelists, omitted all mention of any female writer.</CHRONPROSE>
    </CHRONSTRUCT>
    <SHORTPROSE><P>It also attacked the low class position, lack of talent, and dangerous politics of the radical novelist Robert Bage, and the moral damage done by Henry Fielding. It is a good example of the kind of ideological pressure to which, at this juncture, the novel was subject.</P></SHORTPROSE>
    <RESPONSIBILITY>
    <DATE></DATE>
    </RESPONSIBILITY>
    <BIBCIT></BIBCIT>
</CHRONEVENT>

    In biocritical documents, on the other hand, the contents of the ChronStruct tag are all that is lifted into the potential chronology. In this case there is no ChronEvent tag to hold accompanying or elaborating ShortProse together with the ChronProse. Such an elaborating ShortProse is to be seen only in the parent document, not in the freshly searched chronology. For example, let us consider the following ChronStruct from the writing document on Aphra Behn (from which, again, tags extraneous to the present argument have been removed).

<CHRONSTRUCT>
    <DATE>22 April 1689</DATE>
    <CHRONPROSE>An anonymous 'Young Lady' (most probably Delarivier Manley) published a pindaric elegy on Aphra Behn.</CHRONPROSE>
    <BIBCIT></BIBCIT>
</CHRONSTRUCT>

    This statement makes sense on its own: it is meaningful and comprehensible. Its implications, however, and the developments that relate to it, are far from simple. In its original document, a setting not dominated by chronology, this statement leads in to a lengthy ShortProse which does not belong in a timeline of any description. It is therefore better to provide for the user who wishes to move beyond a biocritical ChronStruct to do so in the original document. Here is what follows the ChronStruct above.

<SHORTPROSE><P>Behn's death, this elegy says, is a disaster for women's writing, for no other woman 'dares her Laurel wear'. For a while it remained possible for women writers like Jane Barker to claim descent from both Katherine Philips, the supposedly shy poet of retirement, and Aphra Behn, the woman of the court, the theatre, and the streets. Thereafter kinship with Behn was increasingly felt to preclude kinship with the respectable Philips. Critic Jane Spencer argues that if it was hard to be a 'daughter of Behn' it was even harder for those many male writers who felt her influence to own themselves her sons: the whole notion of poetic lineage, like property descent itself, was gendered male. Therefore Behn's fundamental and wide-reaching influence on Thomas Southerne, for instance, has gone unnoticed outside a couple of acknowledged theatrical adaptations.
    <BIBCIT></BIBCIT>
</P></SHORTPROSE>

    Orlando chronologies therefore stand without the ShortProses which may be associated with ChronProses within biocritical documents. But the delivery system will facilitate a user's movement by a click from the ChronStruct displayed in the chronology into the parent biography or writing-career document. This reflects the fact that temporal statements lifted from biocritical documents often need to be read in the context of those documents for adequate understanding. The biography and writing documents are longer and more complex texts than anything in the Events database, and in them a ShortProse is just as likely to raise new issues as to tie up those raised by the ChronStruct. Users will in effect be encouraged to pass freely from chronologies into biocritical documents for further investigation. The chronological function is important in the ordering of Orlando results, but in the presentation of literary history it is still only one conceptual method among many.

2.3 Attributes

Many of the elements already mentioned as constituent of ChronStructs have attributes (not yet mentioned) that encode more specialized discriminations. The BibCit tag, for instance, has an attribute bearing the overall project identification number for the work cited. The Responsibility tag has attributes that identify the team member responsible for this ChronEvent, the task performed and the day when it was performed. These attributes provide a vital tool for internal work management.

    Attributes contributing to the chronology's value to users are those on ChronStruct: the attributes of ChronColumn and Relevance. On these depend the sorting of ChronStructs into a timeline. Assigning these attributes involves complex judgement calls (they carry their own Resp attribute to signal who made the decision).

3 Sorting

3.1 ChronColumn

The ChronColumn attribute performs for Orlando timelines the function that is performed in print by sorting events into separate columns or boxes. ChronColumn has four values: BritishWomen'sWriting, WritingClimate, SocialClimate, and NationalInternational. Our central material on women's lives and writings has the ChronColumn value BritishWomensWriting. The value WritingClimate marks equivalent information about the lives and writings of male and non-British writers, but also information about printing technology, censorship, the book trade, and anything else bearing on the literary climate. (The example above, in which the Quarterly Review commented on male novelists and failed to mention female ones, bears the WritingClimate value.) The value SocialClimate marks information on a host of topics: education, marriage law, scientific and technical advance, fashion, etc. The value NationalInternational gathers events and processes having to do with the exercise of political power and the life of nations: governments, laws, wars, etc. To search on each particular column in turn provides four markedly different readings of the concerns of any period in history.

    For instance, in 1815, the year of the battle of Waterloo, items with the value BritishWomensWriting are divided among Romantics such as Ann Radcliffe and Mary Bryan, radicals such as Mary Hays, Augustans such as Anne Grant, popular-fiction professionals such as Elizabeth Gunning and Barbara Hofland, and future Victorians such as Jane Welsh Carlyle. The past is disinterred, in the form of extracts from Lady Mary Wroth's Urania. One woman writer renounces a passionate love-relationship of which her parents disapprove; another is active in founding a branch of the Society for Bettering the Conditions of the Poor; others publish abstruse theology and popular moralizing on the topic of British India. All these social concerns are grouped together with the BritishWomensWriting value, since they all bear on women's literary history. Their variety reflects the focus of the project on literature and its relationship to other areas.

    The WritingClimate value yields information from both sexes, from Britain and elsewhere: this year Byron is publishing, and Jean Lamarck, and Lydia Howard Sigourney. Anthony Trollope is born. With the SocialClimate value, the first so-called bazaar opens in London; the singer Madame Vestris makes her debut; Irish nuns strive to educate the poor; committees concern themselves with suppressing vice and uncovering abuses in private madhouses; Robert Owen issues a clarion call on behalf of factory children. And on the National/International front, Acts of Parliament dealing with Britain and India are overshadowed by affairs in France, as Napoleon makes his last march and fights his last battle, and the ancien règime is restored.

    To pick a single ChronColumn value and search on it over a span of years is to reveal processes and progression. Writing careers are built; literary movements develop and come to flower. SocialClimate weaves together its narratives about modes of commerce, entertainment, education, health and welfare. Both nationally and internationally, political events of one year are now revealed as developing from those of previous years.

3.2 Sorting on ChronColumn

Where several ChronStructs share the same Date value, they are sorted by ChronColumn in the following order: BritishWomensWriting, then WritingClimate, SocialClimate, and lastly NationalInternational. This reflects our sense of where the heart of our project lies, in an image which might be drawn from the practice of SGML encoding. The centre of the Orlando Project's research concerns the history of British women's writing. This is situated inside the mainstream of writing in general. Writing in turn is situated in a broader matrix of cultural practices. Culture in turn is shaped by being practised within the context of political systems.

    As an example of this sorting principle, two renderings of the same event sort into an order which places the event's effect on a woman writer before its effect on theatre history and women's employment. To put this another way, a ChronStruct from Mary Latter's writing document, with ChronColumn value BritishWomansWriting, sorts before a ChronEvent with SocialClimate value:

<CHRONSTRUCT ChronColumn='BritishWomensWriting'>
    <DATE>12 February 1762</DATE>
    <CHRONPROSE>The death of John Rich, licensee of Covent Garden Theatre, ended Mary Latter's hopes for the production of her blank-verse tragedy The Siege of Jerusalem, by Titus Vespasian.</CHRONPROSE>
</CHRONSTRUCT>

<CHRONEVENT>
    <CHRONSTRUCT ChronColumn='SocialClimate'>
    <DATE>12 February 1762</DATE>
    <CHRONPROSE>On the death of John Rich (holder of the licence for Covent Garden Theatre) his widow, Priscilla, took nominal control of the theatre.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

3.3 Relevance

    The Relevance attribute on ChronStruct signals the weight to be attached to this information in the broader historical scheme. It too has four values: Selective, Period, Decade, and Comprehensive, designed to produce increasing granularity of result, from the carefully-winnowed to the positively indiscriminate.

    The value Selective marks indispensable statements, which the Orlando team regards as necessary to preserve even in the most pared-down search results. The value Period marks events which might be valuable to a user interested in some time-frame like that between the two world wars, or the years that saw the struggle for women's suffrage, or the kind of period commonly covered in a university literature course. The value Decade brings in more material, such as might be wanted by a user researching more limited stretches of time: those of, say, the French Revolution, or the heyday of the New Woman, or the few years surrounding some particular event or literary text. Finally, the value Comprehensive calls up all the dated material recorded anywhere across the project, including many events (such as the beginnings and ends of schooling or employment, the births and deaths of children) that are vital in the chronology of an individual writer, but not to literary history as such.

    Birth and death dates of British women writers always have the Selective value, because of the way authors and other historical figures are traditionally labelled by their birth and death date. Each writer who receives biocritical treatment (even the more obscure) has a Selective value assigned to at least one publication or other writing event, as we are asserting these writers' worthiness for scholarly attention. The publication of well-known texts (like the successive novels of Jane Austen or George Eliot) are certain to be ranked Selective. But a prolific novelist like Iris Murdoch would be over-represented in a Selective chronology by a complete list of her publications, and some are ranked less generously.

    Other publication events are ranked according to a combination of factors: the respect accorded them by critics both past and present, the extent to which the general public is aware of them, the competitive pressure of events from that particular writer and that particular period, but most importantly the place they hold in the author's writing career. First and last publications normally receive a Selective value; so do first and last examples of a significantly-used genre. A novelist-playwright thus has her first and last play as well as her first and last novel ranked as Selective.

    To search on the name of Naomi Mitchison and use the Selective value is to generate a bird's-eye-view of the highlights or headlines of a career which spanned more than a century and produced almost a book a year. To do the same search on Period and then on Decade brings in progressively more and more of her publications (as well as her political activity). It takes a Comprehensive search to bring up every last pamphlet and children's book by Mitchison which the documents have mentioned in ChronStructs, and it is at this stage of granularity that purely personal events are also revealed. The user therefore has four choices as to how much chronological Mitchison he or she wishes to see, even without opening the actual biocritical documents.

3.4 Other encoding in ChronProse

Within ChronProse, as within ShortProse, other project encoding functions in the same way that it does anywhere else in our documents. This encoding embraces structural tags such as Quote, Socalled, Foreign etc. and core tags for Name, Place, OrgName, Title, etc. Each core tag indicates a fruitful category for searches to generate timelines. Any chronology on the screen can be transformed into a different one simply by clicking on a name, date, organization name, title, or place. In this way, our content tagging will facilitate chronology generation based on associations that arise as the user reads our material.

    More specialized Orlando tags particular to our biocritical documents will allow for subject- or theme-based searching. For example, a chronological search on literary <MOTIF> will show historical trends in the use of motifs. Similarly, a search for 'Quaker' in our religious <DENOMINATION> tag will provide list of writers whose membership in the Society of Friends has been noted (as opposed to the lavish results produced by a search for 'Quaker' as an open search term across the textbase, which would turn up much material about anti-slavery and anti-war activism).

3.5 Searching on date

For our users, searching on Date and specifying a year can produce a timeline of everything recorded by the project for a particular year, or a particular month or day within that year. We have already seen, in connection with the year 1815, the panorama or frozen snapshot that such searches produce, which create an appropriately dizzying sense of multifarious human activity.

    Searching on Date without specifying a year can produce a timeline of what has happened on a particular month or day (or group of days) throughout history. (See Orlando's webpage for some results of a search on each current week, changing every Monday. These results are chosen according to different principles from those of the Selective value on Relevance attribute, so as to produce a chronology which, even in brief space, is representative of both public and private events and processes.)

    Searching on a particular day or month Date across the lapse of years yields results which are largely dictated by chance. One kind of result among them is largely predictable for the month or month/day concerned, and clearly marks the superiority of the precise-date over year-only method of recording chronology. Years have their rhythm: the spring produces a regular micro baby boom. The publishing year, too, has its rhythms, which vary from one period to another, with August almost always a slack time. Political events may be purposely timed to fall on anniversaries like St Cecilia's Day or International Women's Day. Individual events too may be scheduled to fall on anniversaries: Brigid Brophy was married on her twenty-fifth birthday, Georgiana Duchess of Devonshire on her seventeenth; Elizabeth Barrett had her verses published by her proud father on her fourteenth. Lady Mary Wortley Montagu's most famous attack on Alexander Pope, Verses to the Imitator of Horace, appeared on the anniversary of Pope's first known printed attack on her: a symmetry that no scholar had noted before the Orlando chronology revealed it. (The Orlando documents are, however, silent about the question of cause: whether this was authorial intention, or mere coincidence, or a hostile practical joke.)

3.6 Searching chronologically by name, OrgName, or place

In the case of British women writers, a name search of the chronology will often give identical results with a request to see the biocritical documents on that writer in chronology-only form. But in the case of major figures - Jane Austen or Mary Wollstonecraft as well as Samuel Johnson or Karl Marx - results continue to register long after the subject's death, from sources that are not biocritical documents. In the case of Austen or Wollstonecraft, users are likely to know before they search something of the kind of results they will obtain. In the case of Johnson or Marx, and in the case of mentions of Austen and Wollstonecraft that fall outside the span of their lives and documents, the results will be more idiosyncratic and unpredictable.

    Use of a Standard attribute on Name tags ensures that no matter how an individual is named in the ChronProse, references to her or him will be sorted together. A chronology on 'Queen Caroline' will produce a timeline on the activities of the wife of George II (chiefly her patronage of women writers) followed by a timeline on the far less prepossessing wife of George IV, whose husband's unjust treatment of her nevertheless made her something of a national heroine. The two individuals can be separated by searching 'Caroline of Anspach, Queen of England' or 'Caroline of Brunswick-Wolfenbüttel, Queen of England'.

    Chronological searches on OrgName have similar potential. Pocket histories can be generated of gender-specific institutions like the Foundling Hospital or the Women's Social and Political Union, which are designed to meet the expectations of users rather in the same way as those on Austen or Wollstonecraft (although not so full, as they do not draw on documents devoted exclusively to the subject of search). In the case of, say, the East India Company, as with those of Johnson or Marx, it will be the special function of the Orlando Project chronology to bring together the general history of the organization with its intersection with women writers' lives and texts.

    Searches on titles are useful for anyone wanting to gather information on the incidence of Poems on Several Occasions during the eighteenth century (sixteen listed at last count), to obtain a quick fix on Gothic novels about castles, or to chart the shifting significance, in titles, of the word Tale, or Woman, or Women's, or Girl's. Authors' borrowing of titles from one another can be nicely revealed by such searches.

    Chronological searches on Place are eminently capable of revealing unexpected links or other relations. Country houses, like Garsington Manor, can be revealed as having touched and fertilized many lives and careers. At different moments in history, the town of Norwich, and the London districts of Chelsea and Bloomsbury, were each a centre for dynamic circles of writers, most or many of them women. The chronology includes evidence of these circles, with other kinds of information - on trade and manufactures, other residents, local politics, or architecture - not strained off but standing in historical relation to the more clearly literary-historical search results.

    Interesting results can also be obtained from the chronology by simple searching on words of every kind, from opera to smallpox to railway.

    Chronological searching, whether on Name, Place, OrgName or whatever, will yield less full but more clearly structured results than a search across the whole text archive for, for instance, the word Norwich inside a Settlement tag. When our delivery system is completed, users may wish to generate a timeline first, and then to use it as a guide or pointer for constructing further, non-chronological or less chronological, searches.

3.7 Searching by keyword

ChronStructs in the Orlando Project are also, in many cases according to significance, equipped with Keyword tags chosen in accordance with a project thesaurus or controlled vocabulary. (This work is still far from complete.) Keywords provide another method of searching, more flexible and discriminating than ChronColumn but much broader and more inclusive than searching on a Name or OrgName. The Keyword 'Women and Education' brings together material from writing documents on authors who discussed educational issues, from biographical documents of professional educators and those whose education held some special historical interest (a normal course of schooling does not receive a Keyword tag), and from historical ChronEvents with a ChronColumn value of SocialClimate. Among events with the ChronColumn value of NationalInternational, a Keyword search can bring together a timeline on the Corn Laws or the First World War.

    Keyword searching can be used to bring either fuller results, on the one hand, or more discriminated results, on the other, than a simple chronological wordsearch. The chronology generated by searching on the Keyword 'Suffrage' embeds the British votes-for-women struggle, conducted by the Pankhursts and others, in less detailed coverage of women's achievement of the vote in other countries around the globe, and also in a sketch of earlier and later events in which the issue of the vote is not inextricably bound up with the issue of exclusion by gender. The earliest and latest items in this timeline do not have the word 'suffrage' or any of its derivatives in their ChronProses.

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>1754</DATE>
    <CHRONPROSE>Women of Bristol were specifically canvassed in print at election time.</CHRONPROSE>
    <KEYWORD>Suffrage</KEYWORD>
    <KEYWORD>Elections</KEYWORD>
    </CHRONSTRUCT>
    <SHORTPROSE><P>This was because in Bristol (as at a few other boroughs) the status of freeman (which carried the right to vote) was passed from father-in-law to son-in-law: from man to man through woman. The author of a pamphlet, The Dundry Petition: or, The Countryman's humble Address to the Freewomen of the City of Bristol, clearly felt this gave women political influence.</P></SHORTPROSE>
</CHRONEVENT>

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>17 April 1969</DATE>
    <CHRONPROSE>The Representation of the People Act (1969) extended the vote to people in Britain aged eighteen and over.</CHRONPROSE>
    </CHRONSTRUCT>
    <KEYWORD>Suffrage</KEYWORD>
    <KEYWORD>Legislation</KEYWORD>
</CHRONEVENT>

    In this case to search the prose for 'suffrag' (taking in such words as 'suffragist') would produce a more limited and more specialized chronology, more completely dominated by the efforts of British suffragists, than the Keyword search on 'Suffrage' produces.

    On the other hand, a Keyword search on 'Witchcraft' brings up legislation and accounts of trials up to the mid-eighteenth century, and literary treatments of witches as they were believed to exist up to a similar date. A search on 'witch' in ChronProses, however, adds the following.

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>September 1947</DATE>
    <CHRONPROSE>US Senator Joseph McCarthy's UnAmerican Activities Committee began issuing subpoenas to leading figures in the entertainment industry; this marked the start of the infamous 'communist witch-hunt'.</CHRONPROSE>
    </CHRONSTRUCT>
    <SHORTPROSE><P>In 1954 communist 'witch-hunting' activities were shown on national television. McCarthy was subsequently censured and condemned by the Senate.</P></SHORTPROSE>
</CHRONEVENT>

    This particular wordsearch spreads a wider net than the Keyword search. Searching on words not infrequently exposes such shifts in the implication of language, thereby introducing an extra element in temporal change.

3.8 Sorting by date

We have mentioned the Date tag: in fact Orlando (borrowing from the Text Encoding Initiative) has three methods of encoding this information: Date, DateRange, and DateStruct. The first captures simple dates, the second captures time-span, and the third breaks down a date into its separate components so that attributes can differ across the components (day, month, year, season) that make up a date. These three elements have several attributes. Each has the attribute Value, for normalizing into figures a date which might be expressed in prose as <DATE value='1869-12-26'>Boxing Day 1869</DATE> or <DATE value='2000-04-20'>Passover 2000</DATE> or <DATE value='1930-01-29'>On her mother's birthday</DATE> or 

<DATESTRUCT value='1935-07-'>
    <SEASON>Summer</SEASON>
    <YEAR>1935</YEAR>
</DATESTRUCT>.

The DateRange tag alone has two Value attributes named From and To, for encoding both the beginning and ending of the range, thus: <DATERANGE FROM='1935-06-' TO='1935-08-'>June - August 1935</DATERANGE>.

    Another attribute, that entitled Certainty, offers the following choice: Cert (for certain), C (for circa), By, After, RoughlyDated (for an event which occurred at some specific but unidentified time within the span of a DateRange), and Unknown (for dates which are uncertain not because they are approximate, but because the source of specific information is suspect).

<DATE VALUE= '1855-03-29' CERTAINTY='C'>About 29 March 1855</DATE>

    The default Certainty value on dates is 'Cert'. Taggers need only fill in the attribute if it differs from the default. Similarly, dates tagged in predictable ways will have their attribute supplied by our systems, thus saving our taggers time in their work. For example, a person encoding text does not need to add the Value attribute to '1855', or 'March 1855', or '29 March 1855'. But if extra content is placed inside a Date tag ('About' or 'From' or 'Probably') then a Value attribute must be supplied.

    The DateRange tag (alone) has a further attribute allowing the encoder to flag the level of Certainty as applicable to either the To date or the From date, or both.

    Searches by date can be selected in two ways: by 'From date' or 'To date'. Each one brings up all ChronStructs including a Date or a DateStruct tag bearing the numerical value searched on; but the former kind of search brings up ChronStructs with DateRange tags by their From value, and the latter kind brings up those by their To value. We regard the historical processes that are often (although not always) the material of DateRange events as particularly important in a full and nuanced rendering of the workings of history. The Orlando Project is making particular efforts not to downgrade or overlook process in favour of simple events, since a tendency this way is a failing of the chronological medium.

    Here are some examples of the workings of DateRanges. As usual, tags not necessary to the point being demonstrated are omitted. A search on From dates would bring up, under 24 April 1916:

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATERANGE FROM='1916-04-24' TO= '1916-04-29'>24-29 April 1916</DATERANGE>
    <CHRONPROSE>In what became known as the Easter Rising, the Irish Volunteers and the Irish Citizen Army took control of Dublin.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

<CHRONSTRUCT>
    <DATE>24 April 1916</DATE>
    <CHRONPROSE>During the Easter Rising Constance Gore Booth fought at St Stephens Green, and was second-in-command to Michael Mallin.</CHRONPROSE>
</CHRONSTRUCT>

    A search on To dates under 11 November 1918 would bring up the following.

<CHRONSTRUCT>
    <DATERANGE FROM='1914-08-04' TO='1918-11-11'>During World War I</DATERANGE>
    <CHRONPROSE>Naomi Jacob joined the Women's Legion and worked in a munitions factory.</CHRONPROSE>
</CHRONSTRUCT>

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE>11 November 1918</DATE>
    <CHRONPROSE>At 11 a.m. (the eleventh hour of the eleventh day of the eleventh month) the Armistice, signed at Compiègne, went into effect, officially ending World War I.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

3.9 Sorting within sorting

Whether a search is conducted on To dates or From dates, the search produces a chronology in which the results are sorted in two successive ways. They are sorted first by the Value attribute on Date. This places year-only dates before year/month dates, and those in turn before year/month/day dates. Thus 
<DATE VALUE='1928--'>1928</DATE> 
comes before 
<DATE VALUE='1928-10-'>October 1928</DATE>, 
which in turn comes before 
<DATE VALUE='1928-10-11'>11 October 1928</DATE>, 
which in turn comes before 
<DATE VALUE='1928-11-'>November 1928</DATE>.

    Putting the year-only dates first facilitates a year-at-a-glance reading of results from a search on, say, 1621 (which has currently three year-only dates with Certainty value Cert, two year-only dates with Certainty value C, and ten year/month or year/month/day dates, two of them important events in women's publishing). Given a year that yields eight times more results, 1921, this ordering produces a sequence of 62 year/month or year/month/day dates, preceded by a sequence of 70 unspecified year-only dates. Some of these, but not all, will become more precise through further research. In such cases the placing of the year-only dates first provides a kind of snapshot of the concerns of the period, within which the further automatic sorting by ChronColumn creates a sweep from British women's writing outwards through its various contexts. After that the more precise dates follow up by providing a partial narrative of the year. Both the snapshot and the narrative provide rough-edged detail, of a granularity unthinkable in printed-page chronologies.

    The second process of sorting on Date uses the Certainty attribute, and reflects the literal meaning of the attribute values by the order into which it sorts them.

    Dates with Certainty value By are placed first - a rational ordering since <DATE CERTAINTY='By'>By 11 October 1928</DATE> really indicates an event that took place on an unspecified date before 11 October. Next comes Cert - <DATE CERTAINTY='Cert'>11 October 1928</DATE> - and after that C - <DATE CERTAINTY='C'>About 11 October 1928</DATE> - and RoughlyDated (in the cases of From and To dates in DateRanges). Last come dates bearing the Certainty value of After.

    For example:

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE CERTAINTY='By'>By 1900</DATE>
    <CHRONPROSE>Women accounted for twelve percent of the library staff in Britain whereas in America ninety-five percent of library staff were women.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

sorts before

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE CERTAINTY='Cert'>1900</DATE>
    <CHRONPROSE>Quantum Theory was introduced in Max Planck's Zur Theorie des Gesetzes der Energieverteilung und Normalspectrum (Leipzig).</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

which in turn sorts before

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATE CERTAINTY='After'>From 1900</DATE>
    <CHRONPROSE>The Birmingham School of Music developed under the supervision of composer Granville Bantock.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

Or in an example involving a DateRange:

<CHRONEVENT>
    <CHRONSTRUCT>
    <DATERANGE From='1916-04-24' To='1916-04-29'>24-29 April 1916</DATERANGE>
    <CHRONPROSE>In what became known as the Easter Rising, the Irish Volunteers and the Irish Citizen Army</ORGNAME> took control of Dublin.</CHRONPROSE>
    </CHRONSTRUCT>
</CHRONEVENT>

<CHRONSTRUCT>
    <DATE CERTAINTY='After' VALUE='1916-04-29'>After Easter 1916</DATE>
    <CHRONPROSE>Edith Somerville signed a petition on behalf of the condemned leaders of the Easter Rising.</CHRONPROSE>
    </CHRONSTRUCT>

    The first sorting parameter is degree of precision - year before year/month, year/month before year/month/day. The second is sorting by value of Certainty attribute - By, Cert, After. Last comes the sorting by ChronColumn already referred to.

3.10 Issues in sorting

Dates can be expressed in prose in so many imprecise ways that fairly elaborate sorting protocols are needed, and even these cannot forestall all problems. Where one source says 'Early January' another may say 'At the beginning of January' and another 'In the New Year'. Furthermore, 'Early January' might equally well be classified as <DATE VALUE='2000-01-01' CERTAINTY='After'>Early January</DATE>, or as <DATE VALUE='2000-01-10' CERTAINTY='By'>Early January</DATE>, or as <DATE VALUE='2000-01-05' CERTAINTY='C'>Early January</DATE>. Perfect consistency across a number of researchers (and different kinds of source expressing their dates in different ways) is a goal which can be approached but hardly attained.

    One helpful decision was not to introduce extra efforts at precision beyond what the source specifies. 'Early in 1910' is better represented by <DATE VALUE='1910-01-' CERTAINTY='C'> than by either <DATE VALUE='1910-01-10' CERTAINTY='By'> or <DATE VALUE='1910-01-01' CERTAINTY='After'>. Use of either of these latter forms causes awkward placing of these early year-dates among dates which have a clear link to specific January dates, rather than among those which are linked to the month alone.

    Several aspects of sorting problems can be controlled by devoting some time to searches through our existing material whenever new material is being added, to establish the correct relationship between the existing and the new. If a source says that Elizabeth Rigby and her husband left Venice for London in autumn 1852 to attend the funeral of the Duke of Wellington, then the DateStruct in the Orlando ChronStruct applies its Value and Certainty attributes to ensure that the event sorts before that in which Jane Welsh Carlyle is recorded as watching the funeral procession. This is a necessity although the two events belong in two separate biography documents.

3.11 Duplicates

Another recurrent problem for the makers of a collaborative chronology is that of duplicate events. Different team members reading different sources in pursuit of different lines of research can easily create three or more separate ChronEvents based on, for instance, the same Act of Parliament (if that Act covers several distinct issues), with different ChronColumn values. And depending on how attentive or inattentive the different sources are to the matter of dating, the three ChronEvents may be differently dated: by year-only, say, by year/month, and by year/month/day. It is essential for team members adding new ChronEvents to our events database to keep conducting rigorous searches, on Date, Name, OrgName, and Title, for instance, so that any duplicates can be caught and either deleted or merged. (Incorrect dates in sources produce all the same problems. When a source is found to be in error, this fact is noted in a ScholarNote tag which will be available for users to consult on deman.)

3.12 Comparing Orlando with the Romantic Chronology

A comparison of these two chronologies can usefully centre on their rendering of the year 1815 (already mentioned). Differences fall into three categories: coverage, emphasis, and structure. The Romantic Chronology is smaller in scope than Orlando, offering less per year: its twewnty-four events for 1815 are fewer than a third of Orlando's eighty-nine: its smaller coverage must be borne in mind when comparing emphasis.

    Differences in overall emphasis are nonetheless clearly marked. Whereas each chronology has the same number - five - of men's publications for 1815, Orlando has thirty-five women's publications and other writing events to the Romantic Chronology's three. The Romantic Chronology mentions the Drury Lane production of The Family Legend without explicitly revealing the author (Joanna Baillie) in the first screen it presents to the reader - though clicking on 'Details' identifies her by implication. Orlando, for its part, mentions this play's success in Edinburgh in 1810 but not that in London in 1815 - in this case it positively slights the centre for the margins. Both chronologies have the Corn Laws and the birth and death of Mary Shelley's first baby, as well as the end of Napoleon and restoration of Louis XVIII. The Romantic Chronology has more (and a far higher proportion of) events that in Orlando's system would have a ChronColumn value of NationalInternational. It has what Orlando lacks in the shape of attention to the career of Beethoven, but lacks what Orlando has in the shape of, for instance, the Parliamentary Select Committee on madhouses, the Critical Review's launching of a 'Works in the Press' feature, the latest Stamp Act, emigration to the USA, and the debut of Madame Vestris. Orlando's attention to social and cultural matters, broadly defined, is immediately visible.

    So much for comparing the whole content of each chronology. The gap in size between the two is diminished by comparing the Romantic Chronology's 1815 with Orlando's 1815 events with Relevance value Selective only. This, however, makes yet more apparent the difference in emphasis. Orlando has forty Selective events (still a significantly higher number than the Romantic total), and, although these include most of the NationalInternational events that it shares with the Romantic Chronology, almost half concern women's writings or publications. Of works ranked Selective, some have that attribute specifically because they are the first or last productions of their authors.

    Differences in structure are flagged by differences in usage. The Romantic Chronology uses telegraphic forms like "Robert Burns born" (not, of course, under 1815) and in most cases the present tense. Orlando uses complete sentences, and the past tense. This is a pointer towards the generic differences between database forms and SGML-encoded text. The Romantic Chronology presents its search results in database format, with sentences enclosed in boxes, and clicking on Details produces data in further boxed categories. Orlando's encoding offers an even flow of sentences, and its equivalent to those boxed categories is either a ShortProse (in the case of historical material) or an entire document (in the case of individual authors or their work).

    The Romantic Chronology offers searching on dates, words, or topics, as well as links to primary texts and elsewhere. Its word searches embrace the kinds of search which in Orlando might be conducted on Name, Place, OrgName, or Title (or even on Keyword, although this is logically parallel to the Romantic Chronology's topics). The Romantic Chronology lists 'France' and 'Scotland' as topics. These are too general to be Keywords in Orlando, where the choice is between 'French Revolution' or 'Jacobite Rebellion' as Keywords, or France or Scotland in Geog tags.

    The Romantic database offers among its mines of information nothing comparable to the juxtaposing of two different ChronColumn values, or the gradual broadening out from Selective via Period and Decade to Comprehensive, or the move out of chronological ordering into the more spacious field of a biocritical document (and back), to say nothing of more complex searches on marked-up text.

    These differences go far beyond merely style-sheet or design matters, and they have a powerful effect on reader's responses. More research is needed here. But all the indications are that the experience of browsing along the shelves of a library (where information can be found and used at different levels of specificity or breadth) is more like browsing in the Orlando Project's textbase than is browsing in the Romantic Chronology (excluding the latter's links). For straightforward rapid reference to generally available facts, databases and SGML-encoded documents are equally effective. But Orlando seeks to go beyond that in various ways: to track the less predictable backgrounds of literature, to cover women's writing to a depth that forestalls any easy construction of a female alternative canon, and to hybridize chronology with critical investigation.

4 Using the Chronology for Research

Specific single dates are generally of slender importance compared with dates woven together to create a bigger picture. The well-known eighteenth-century writing sisters Elizabeth Montagu and Sarah Scott were until a few years ago universally supposed to have been born a year or two later than in fact they were. When the poet Patricia Beer died in August 1999, her obituaries revealed that her birth-date, too, was wrongly given in every extant reference book. But these corrections to the record have local importance only. Also of limited value to the scholar are many issues of priority. Claims for literary firsts are often without substance, and the Orlando Project will probably prove to be of greater value in debunking such claims than in substantiating them.

    Dates are valuable chiefly as they establish relationships. Women's activity in writing, as in other fields, has historically depended on networks of other women. It has also been more frequently and more deeply involved than scholars have sometimes realized with politics and with social and cultural change. Authorship in itself, as Robert Darnton (1982) has argued, is not carried out alone but in circles of communication, with publishers, readers, and others. Already the Orlando Project chronology has generated a mass of minor new discoveries about the personal networks and the topical connections of women's texts. It turns out, for instance, that the racy fiction of Eliza Haywood more often made capital from current headline incidents than had been supposed.

    The Orlando Project will be instructive to researchers at every level of expertise. Let us suppose a student has to write an essay on a play by Timberlake Wertenbaker. An Orlando timeline for the author, similar to the one that might be contained in a printed text, might provide a series of dates (composition, production, publication). This, however, is still just preliminary. Let us suppose the play is a comedy with historical aspects. In that case the student might generate timelines of these genres for the previous decade or so. If the play was first produced at, say, the Royal Court Theatre, a search on the theatre will provide a different kind of context: an account of some recent productions there, with women's work predominating (as well as a residence there by Caryl Churchill). Then, of course, the year of the play's appearance can be searched at any level from Selective to Comprehensive, to obtain a snapshot and a narrative of the year from varying focal depths. And in the case of the historical drama Our Country's Good, the researcher could go on to locate the date and central facts about the arrival of the first transported convicts in Botany Bay, Australia (the subject of the play), and about George Farquhar's The Recruiting Officer, 1706 (which is performed by convicts as a play within the play).

    The project chronologies can thus be expected to produce, for those who enquire about them, answers about literary history and about broader history. More important than this, however, is its potential for exploration of a less directed kind, for research that is curiosity-driven rather than produced within the constraints of a specific demand for results. A project timeline for, say, William Shakespeare or the East India Company will mingle on the one hand events which any student of those topics would justifiably expect to find in it, with events thrown up by a number of individual project members' attention to the possible relationships of women and their writing to other topics. The Shakespeare timeline, having dealt with the poet's life and works, mentions Charlotte Charke's performing the role of Falstaff, as well as Mary Cowden Clarke's The Girlhood of Shakespeare's Heroines and Mathilde Blind's Shakespeare Sonnets. The East India Company timeline mentions the company's founding and access to political power (Relevance attribute Selective), as well as the employment or marriages of several company servants whose wives or mothers were writers. A timeline of this kind hardly comes under the heading of research results, as it is assembled in response to no specific problem-solving agenda. It is, instead, a provocation to research, a jumping-off point, a provision of multiple suggestions and opportunities for the increase of knowledge.

    Let us imagine a scholar who is almost the converse of the one pursuing Timberlake Wertenbaker: someone with the massive project of writing a book on women's attitudes to and representations of war in the twentieth century. For this scholar, initial keyword searches on World War I and World War II would generate chronologies that chart the course of the conflicts through events with ChronColumn value of NationalInternational, from trench warfare to the dropping of the atomic bomb. But with these predictable results the same searches would map out areas of women's experience: service in the various women's branches of the forces and as Land Girls and VADs (nurses belonging to the Voluntary Aid Detachment); munitions work, ambulance-driving, intelligence work; loss of fathers, husbands, and brothers; refugee experience; and challenges faced by the civilian population, such as air-raids, rationing of food and clothes, and the mass evacuation of children. This material would suggest a host of follow-up wordsearches on significant terms. It would also include the publication of novels, poems, and plays about the wars (often written long after the event) and suggest names of organizations for further searching, as well as authors whose biocritical documents should be explored.

    Searching on the Wars/Peace keyword would add lesser conflicts to the list of research areas: the Irish Civil War, the wars in Korea and Vietnam, etc. It would also add more organization names (like the Women's International League for Peace and Freedom, and the Campaign for Nuclear Disarmament) whose investigation would foreground the need for work on another group of authors.

    Searching on dates, rather than on direct references to conflict, would add information on books, films, and radio programmes which were particularly popular during wartime though their subject-matter had nothing to do with fighting.

    Questions would start up on every side: which writers were children in wartime, and which were adolescents? Who had parents killed in previous wars, or who was the daughter of war heroes or of profiteers? What kinds of distinctions can be made between first-hand re-creation and that based on research? When, after a war, does the peak period arrive for publication about it? Does class, or age, or nationality (English, Irish, Welsh, Scots), or locality (urban or rural) dictate broad differences in approach? How far do literary representations shift over time, reflecting generational attitudes?

    Work with the Orlando Chronology will not decrease by one iota the need for our scholar to invest her time in reading books and articles. But it will most excitingly stimulate and direct such reading. The project chronologies have less resemblance to a report on completed research than to suggestions for future research by new generations of users.

References

Bishop, E. (1989). A Virginia Woolf Chronology. Basingstoke: Macmillan.

Brownstone, D., and Franck, I. (eds.) (1994). Timelines of the Arts and Literature. New York: HarperCollins.

Darnton, R. (1982). "What is the History of Books?" Daedalus, 3: 65-83.

Foxon, D. F. (1975). English Verse, 1701-1750, a Catalogue of Separately Printed Poems . . . Cambridge: Cambridge University Press.

Griffith, E. (1997). The Delicate Distress. Ricciardi, C., and Staves, S. (eds), Lexington, KY: University Press of Kentucky.

Haywood, E. (1998). The History of Miss Betsy Thoughtless. Blouch, C. (ed.), Peterborough, Ont.: Broadview Press.

Johnston-Eilola, J. (1997). Nostalgic Angels: Rearticulating Hypertext Writing.  Norwood, NJ: Ablex.

Mandell, L., and Liu, A. (2000) Romantic Chronology. Available: (http://english.ucsb.edu:591/rchrono/default.htm)

McDowell, P. (2000). "Women and the business of print." In Jones, V. (ed), Women and Literature in Britain, 1700-1800. Cambridge: Cambridge University Press.

Page, N. (1988). A Dickens Chronology. Boston, MA: G. K. Hall.

Radcliffe, A. (1970). The Mysteries of Udolpho. Dobrée, B. (ed.), Oxford: Oxford University Press.

Reeve, C. (1967). The Old English Baron. Trainer, J. (ed.), Oxford: Oxford University Press.

Rogers, P. (ed.) (1987). The Oxford Illustrated History of English Literature. Oxford: Oxford University Press.

Warner, W. (1991). "Social Power and the Eighteenth-Century Novel: Foucault and Transparent Literary History". Eighteenth-Century Fiction, 3: 185-203.


1. Since this paper was printed we have been awarded funding by the Canadian Foundation for Innovation.