The Application of Form Data to Works of Fiction: Discussion Paper


by the British Library Fiction Indexing Group

[Andrew MacEwan; Sue Lambert; Lesley Firth; Peter Morriss]


cc: Tom Yee, Assistant Chief CPSO, Chair of LC's Form/Genre Working Group

Mary Charles Lasater, Chair of SAC Subcommittee on Form Headings/Subdivisions Implementation

Pat Thomas, Chair of SAC Subcommittee to Revise the Guidelines on Subject Access to Fiction

[Please circulate further as required]

1. Introduction

 This paper reviews the application of form and genre headings from the Library of Congress Subject Headings to works of fiction. The purpose of this study is to highlight the issues arising from the planned implementation of the new 655 field and $v subfield for form data on the LCSH file in the specific context of one major literary genre.

 There are two reasons for choosing fiction as an ideal case study for looking at implementation issues for literature:-

a) Use of the 655 field for individual works of fiction is already well-established through the OCLC/LC Fiction Project

b) At the British Library we have gained some experience of using the 655 field through our application of the Guidelines on Subject Access to Works of Fiction, Drama, etc. used by the OCLC/LC project. This experience has generated a number of unresolved questions concerning the relationship between LCSH and GSAFD which are the stimulus for writing this paper.

This second point also has the particular significance that the overall context for indexing fiction at the BL is similar to that of the Library of Congress: access needs to be provided to a large, multi-language collection of books.

 The paper is angled towards a series of practical, specific recommendations based upon the cataloguing experiences and access needs of a large national library. One of the key difficulties experienced at the British Library has been living with the differences between the GSAFD and the standard LCSH approach to indexing fiction. The aim has therefore been to review fiction indexing as a whole and to come up with a coherent and complete set of "best option" recommendations for how that "whole" can be reconciled to the possibilities inherent in introducing the new MARC codes into LCSH. The recommendations are meant to stand together as a holistic, coherent plan for 655/$v implementation for works of fiction, but each recommendation can, of course, be taken as an individual discussion point. The intention is to stimulate debate towards a co-operative consensus on how best to deal with the issues identified.

The paper is structured as follows:-

Section 2 identifies the principal background parameters on which the ensuing analysis is based.

Section 3 describes current LCSH practice for assigning form data to works of fiction

Section 4 identifies the principles for constructing and assigning form headings for fiction, and makes recommendations based upon these principles. This is the critical section comprising our attempt to re-think the nature of indexing by form in order to exploit the full potential of 655 and $v.

Section 5 makes recommendations which draw heavily on recent work of the GSAFD subcommittee on the theme of disentangling form data from subject data

Section 6 considers how other non-fiction form data fits into the application of fiction form headings, focusing in particular on pictorial works and readers.

Section 7 concludes with a summary of the recommendations made, divided into those affecting the LCSH file, and those affecting policies for application of form data.

 Appendix A contains examples of searches which illustrate the current status of access by form to works of fiction.


 Although we have consulted with LC on several specific interpretations of current LCSH policy in the course of compiling this paper, any errors in our account of current practice (hopefully few) are entirely our own.


2. Background parameters

 There is already a well-developed and ongoing debate on the issues involved in the distinct identification of form data. The following common assumptions have been for the most part gleaned from recent discussions, announcements and recommendations on form data. These assumptions form the key parameters for the ensuing discussion.

 i) The following recommendation made by the SAC Subcommittee on the Nature and Use of Form Data in July 1994 (1) is taken as a fundamental principle:-

"We recommend that form data applicable to individual items be treated in the same way as is form data applicable to collections.
e.g. Use 655 Science fiction
whether the item is a novel or an anthology."

The key principle involved here seems to be that the literary form data applied should be the same for all works of fiction.

The issue of whether or not individual items should be indexed at all with form data is a separate matter of policy for individual libraries to determine for themselves.

 ii) We take it to be a logical implication of the first assumption above that with the advent of 655 as a LCSH field, with corresponding 155 authority records, the need for a separate GSAFD list of headings for individual works of fiction will disappear. Since the same form data should be available for application to individual or collected works then it is only natural that this form data should all be held in a single file. Those GSAFD headings which already correspond to equivalent 150 form headings on the LCSH file, will in future correspond to 155 form headings on the LCSH file.(2) Others that currently only appear on GSAFD should be added to the LCSH file.

 iii) LC's form/genre implementation plan (3) makes the assumption that 655 and $v will be introduced over 1997-1999 without making any change to the way subject headings strings have been constructed to date. In sympathy with this plan this discussion makes the assumption that minimal change to LCSH practices is to be preferred.

However, some reconciliation of the GSAFD and LCSH approaches to indexing will be desirable in order to avoid introducing more confusion when the 655 field becomes common to individual and collected works.

 iv) The fundamental purpose of 655/$v implementation has been well-defined by David Miller as "extricating access to form and genre from subject access."(4) Purely topical access to fiction is not reviewed closely in this document. $v in headings of the type 650$a[Topic]$vFiction is considered uncontroversial. This document only examines the distinct organisation of access to fiction by form through the range of potential 655 access points for fiction which appear on LCSH (and potential $v's under this range).

 v) The retrospective conversion of files as part of the implementation of 655 and $v is not discussed in this paper. The principle of minimal change cited in (iii) above is the only point of consideration given to the problems of retrospective alignment of files to the new coding.


3. The range of headings expressing types of form/genre in fiction

This section provides a summary of the types of fiction form headings which currently exist on LCSH.

The complete genre of fiction comprises a range of form and genre headings that sub-categorise fiction into types. These types are the critical focus for implementation of 655 and $v, since collectively they define the LCSH treatment of fiction as a distinct form. It is this range of form data which needs to be considered as the basis for providing a consistent set of form data applicable to collections and individual works.

Other form data outside these types are not intrinsic to the fiction genre, but may be used in conjunction with it or merged with these types in complex phrase headings. These additional forms are discussed in Section 6 below.

LCSH applied according to the policies outlined in H1790 of the Subject Cataloging Manual allows for the full available range of form/genre headings to be assigned only to collected works by several authors. This range includes:

1) Primary genre: fiction itself, expressed by headings of the form, English fiction

2) Minor genre: there is only one minor genre type defined under fiction, e.g. Short stories, English

3) Specific form headings with topical aspects: e.g. Western stories

and Specific form headings without topical aspects: e.g. Radio stories, Children's stories, English


1) Primary genre headings are appropriately qualified by language/nationality (N.B. qualifiers may express either concept).

They may be further qualified by subdivisions to express:-

Author groups: e.g. American fiction--Women authors, English fiction--Scottish authors; (author groups can also be expressed post-coordinately in separate headings, e.g. Lesbians' writings)

Time period: French fiction--19th century

Translations: Portuguese fiction--20th century--Translations into English

Locality where written: American fiction--20th century--Southern States


2) Minor genre headings qualified by language/nationality qualifier.

May be further qualified by subdivisions to express:-

Translations: Short stories, American--Translations into Spanish

Author groups: when established, e.g. Short stories, American--Women authors

NOT allowed on a free-floating basis under minor genre, but are exceptionally established as heading- subdivision combinations on authority file

Locality where written: Short stories, American--Southern States


3) Specific form headings qualified by language/nationality qualifier.

Are not further qualified by subdivisions except:

Translations: e.g. Detective and mystery stories, English--Translations into German

Locality where written: Detective and mystery stories, American-- Southern States


4. Principles of fiction form headings

This section attempts to identify the principles underlying current LCSH treatment of fiction forms and makes recommendations for maintaining those principles in the new context of form access disentangled from subject access.

4.1 Principles underlying the structure of fiction form headings

4.1.1. Phrase headings

All fiction form or genre headings are established as phrase headings.

Primary genre is always expressed in the form:- [Language] fiction.(5)

Minor genre is always expressed in the form:- Short stories, [Language]

Specific genres are variously expressed according to usage or precedents:-

e.g. War stories, [Language]; Political fiction, [Language]; Satire, [Language]

N.B. Phrases expressing genres are often first established on LCSH for use as topical headings on works about those genres. This can make some phrases clumsy as form headings applied to works of fiction: e.g. Black humor (Literature) (used at LC only as a topical heading but on GSAFD the same phrase is used as a genre heading in field 655).

All levels of form heading are also established without language qualifiers in order to provide form access points for collected works which include more than one language or for topical access where language qualification would not be appropriate. These are also often used to provide the fullest reference structure for the genre, without reproducing that reference structure on every instance of the genre. (See below 4.1.3 Reference structures)


Recommendation 1

The principle of minimal change suggests that the phrase forms already established on LCSH should in general be retained when creating 655 equivalent headings, particularly where headings are already used both as topic and form in the 650 field. As a rule the same phrase should always be used for 650 or 655.

However, on a case by case basis headings could be treated exceptionally for one of three reasons:-

(a) to create a variant phrase equivalent for a topical heading where a difference between the genre as topic and the genre as form is judged appropriate. (e.g. Utopias/Utopian fiction) (6)

(b) to improve the phrase heading used. (The GSAFD Subcommittee have identified a number of improvements to specific phrase terms on LCSH, e.g. 655 Detective and mystery stories to be replaced by 655 Mystery fiction)

(c) some genre phrases on LCSH should be viewed as representing topics in fiction, rather than as true genres and these should not be copied as 655 form headings. [The GSAFD Subcommittee has identified a number of recommendations of this kind already. e.g. Prefer 650 Christmas--Fiction to 655 Christmas stories. See also Section 5 Form/Genre vs. Topical access below]

4.1.2 Language qualifiers


The standard for all genre/form headings are pre-coordinated phrases which express both the literary form and the language/place of origin of the work. e.g. English fiction, Short stories, English, Science fiction, English. Headings without qualifiers are used as form headings only for collections covering more than one language.


Language qualification is the major element of divergence between LCSH and GSAFD. The GSAFD allows for no genre/form headings to be qualified by language.

If the principle of applying the same form data to individual works and collections is to be followed then a standard phrase structure for genre headings needs to be established. Clearly this must either involve changing GSAFD practice to conform with LCSH or changing LCSH to conform with GSAFD with respect to the practice of language qualification.

Importance of access by language

Firstly, it is vital to be clear that access by language is necessary for a number of reasons:-

1) Access to large collections: the practice of language qualification divides up large catalogues and improves search options on such catalogues.

2) The notion of providing access by language as well as form is well-established across a range subject access systems, e.g. DEWEY, MLA coded descriptors (7)

3) LCSH represents a huge historical investment in the principle of associating access by form and access by language for collections of fiction.

If we assume that access by language must continue to be a part of access to forms of fiction then there are only two possible ways to achieve this using LC subject headings:-

EITHER the current pre-coordinate headings at all levels of fiction form should be assigned to both collections and individual works; e.g. 655 Science fiction, English

OR specific form headings assigned without language qualification should be assigned doubled with headings for the major genre to provide access by language post-coordinately. (8) e.g.,

655 English fiction
655 Science fiction

The latter would reconcile LCSH practice to current GSAFD practice, since GSAFD only includes specific genre/form headings. The former would involve reconciling GSAFD to current LCSH practice. Both options offer a range of advantages/disadvantages as outlined below.

Changing LCSH by losing language qualifiers on specific form headings


(a) access by specific genre headings would retrieve everything in that genre, irrespective of language or country of origin.

(b) a post-coordinate system of doubling the broader major genre heading would allow for clear Boolean searching by combining subsets of the catalogue. e.g.,

655 Science fiction
655 English fiction
NOT 655 Science fiction, English


(a) loss of pre-coordinate specificity

(b) involves major change to LCSH practice: i.e., major costs

Changing GSAFD by adding language qualifiers on all GSAFD range headings


(a) greater specificity in headings

(b) a consistent hierarchically structured approach combining language and literary form at all levels of specificity:- English fiction; Science fiction, English

(c) avoids post-coordination to express language

(d) involves minimal change to LCSH


(a) pre-coordinate system limits the number of search options that might be achieved through post-coordinate access (advantage (b) under first option above)


 Recommendation 2

The cost of changing LCSH is surely the critical factor.

The advantages of rationalisation and incorporation of GSAFD headings into the parent system LCSH seem to far outweigh the disadvantages of having to adopt standard LCSH headings. In addition to the costs LCSH's internal consistency in using language qualification at all levels of the form hierarchy is a feature worth retaining in its own right.

The GSAFD approach should be changed to conform to the LCSH style of language qualified form/genre headings.

4.1.3 Reference structures on 155 authority records

The creation of 155 authority records out of existing 150 authority records is further complicated by the fact that references on 150 records often connect form headings with topical headings and references. This happens particularly on the "parent" genre headings without language qualifiers. If we are in the business of disentangling form access from topical access then we should be considering how much of this reference structure it is appropriate to carry over from the matching topical headings. Since the USMARC authority format now allows for 7XX links between equivalent headings with different thesaural references, then the appropriate link from the main 155 heading to other topical headings can be made via the single 750 reference to the topical equivalent. A solution is suggested in the following analysis of the reference structure of two genre/topical headings on the LCSH authority file:-

 e.g., LCSH genre heading

     Science fiction
	UF	Science$xFiction
		Science stories

	BTs 	Adventure stories     (z)              	
		Fantastic fiction     (z)
		Literature and science     (x)

	RT	Future in literature     (x)

	NTs	Cataloging of science fiction     (x)
		Classification$xBooks$xScience fiction     (x)
		Interplanetary voyages     (x)
		Interstellar travel     (x)
		Traveller (Game)     (x)
		Time travel     (x)
		Star Trek fiction

The see also references that are followed by an "x" represent topical headings which would not be appropriate on the 155 form heading reference structure.

The BT references on this heading that are followed by a "z" reflect the fact that this has an unevaluated reference structure.(9) BT references should only refer one hierarchical step. Reference evaluation at LC would probably result in the removal of the references in bold typeface.

References of the kind 450 Science$xFiction merit close consideration in the light of recent work by the GSAFD Subcommittee. See the discussion on genre vs topic in Section 5 below. A clearer example of the problem is Detective and mystery stories with the UF Detectives$xFiction. The use of such [Topic]$xFiction references seems unduly restrictive on the flexibility needed for providing access by both genre and topic to works of fiction. A work may merit the class of persons access Detectives$xFiction, without belonging to the genre Detective and mystery stories.

Also with reference to the work of the GSAFD subcommittee the status of the heading Star Trek fiction is questionable. Is this too specific as a genre heading? Should it only be used as a topical heading for works about Star Trek fiction? See Recommendations under Section 5.

Creating 155 equivalents to all these parent headings on the LCSH file will require analysis of the reference structure to decide what can be copied.

By contrast the equivalent language qualified genre/topical heading for Science fiction has a simple reference structure that can be copied without revision:-

Science fiction, English
UF English science fiction

BT English fiction

 Recommendation 3

 Short term:

(i) copy reference structure on all language qualified genre headings into new 155 records;

(ii) model reference structure on unqualified parent headings on the simple structures found on the language qualified versions of those headings.

 Long term: review reference structure on the 150 parent headings and transfer any appropriate references to the 155 equivalents.

 A model short term reference structure:-

Science fiction

  UF Science stories

  BT Fiction

  750 Science fiction


4.2 Principles for the application of fiction form headings

This section attempts to identify the principles involved in assigning form headings to works of fiction. Having made recommendations to achieve a standard form of heading which can be applied to all types of work, the next question is "should the same level of access always be assigned to collected and individual works?"

The GSAFD approach is an attempt to extend a range of access points, only allowed on collected works under the LCSH system, to individual works. Whether a particular library wishes to index individual works to this level remains a matter of local policy, but the GSAFD provides the common standard by which to extend such access. Is there scope for improving this standard and better integrating it with the principles for application of form headings already well-established in LCSH practice for collected works?

As it stands the GSAFD does not simply take the range of form headings applicable to collected works and make it applicable to individual works: instead it selects a range of specific form headings only, and makes no provision for using subdivisions in conjunction with its range of 655 genre headings. Similarly, mainstream LCSH makes distinctions between form access that is appropriate to collections by several authors and form access that is appropriate to collections by one author. [H1790 (2)]

It is not clear whether the subcommittee recommendation "that form data applicable to individual items be treated in the same way as is form data applicable to collections" is intended to address the question of what data is applicable, but the question is surely appropriate. If we are in the business of disentangling form access from subject access in general, then we should be asking what principles are involved in applying form data to collected and individual works. Are there good reasons for applying some form headings only to collections and not to individual works? Or are there good reasons for applying the full range of fiction form headings both to collections and to individual works?

There seem to be two principles involved: Specificity and depth of analysis.


The specificity of form headings is governed by a different relationship to the item indexed than topical headings. Because a form heading does not attempt to indicate what a work is about, but what kind of a work it is, the degree of specificity leans towards the identification of the broad categories under which the work falls. As with subject analysis the rule is to apply the most specific form heading available working up to broader form headings when no specific heading can be found.

So, in current LCSH practice, for a collection of science fiction it is appropriate to assign a heading Science fiction, English. That is the correct form analysed to the most specific level possible.

For a more general literary collection a heading English fiction is the most specific form appropriate.

Considered as a heading that categorises or classifies a work a form heading only needs to be as specific as the work allows. However, the specificity principle does not provide a reason for distinguishing between collections and individual works. Analysis to the most specific level of form data is the same for an individual work as it is for a collection. If the individual work is general in nature it is surely as appropriate to categorise it with a general form heading as it is for a collection. This is, after all, analogous to the same level of data expressed in a classification number.(10)

There is an argument that this will make an access point which produces impossibly large hit lists. But this is simply what form headings do. As access points all (or at least most) form headings are going to involve large hit lists. Access by form is not the same as access by subject. Having made the distinction between form and subject clearer on LCSH it is important to realise that each provides a different type of access. Few people are going to want to search for English fiction. When they see the size of the hit list not many more will want to search for Detective and mystery stories, American. But both levels of heading add a depth of access to the work catalogued which can be used in more sophisticated search combinations. What matters is that the form heading performs a collocating function which can be exploited properly only if access by form is applied consistently to all works, individual and collected.

An example from the LC file demonstrates this point. Out of 13 of Colin Dexter's Inspector Morse books only 3 (collections) have been assigned the access point Detective and mystery stories, English; the rest as individual works receive no form access point (setting GSAFD aside). This must be viewed as a failure to provide appropriate collocation. All the works share this form so all should be assigned the appropriate access point. There is no reason why the same cannot be said of form access at any level of categorisation, even the most general - the level of the primary genre. The critical issue is not numbers and high recall, but relevant recall and precise search options/combinations.

 Depth of analysis

There may be no reason to distinguish between collections and individual works on the grounds of specificity, but there is reason to question the depth of subject analysis that is appropriate to each. The Inspector Morse example above shows the importance of form as a collocating principle that should be applied in common to individual and collected works. The success of the GSAFD and the OCLC/LC project demonstrate clearly that specific form/genres can as easily and appropriately be identified for individual works as for collected works. However, the GSAFD does not allow for the subdivision of its 655 headings, while LCSH permits the use of subdivisions to express locality where written, time period, author groups, and translations, as described in section 3.

 It is rightly argued that the act of compilation and presentation involved in a collected work provides additional fodder for subject and form analysis which is not there in an individual work. A collection may bring together works not only (or not even) associated by specific genre, but associated by author groups, specific locality of author groups, time period when the works were all written. All of these qualifications to the form of a work are expressed in LCSH assigned to collections. Attempting to identify any of these for an individual work involves a degree of subjective judgement about what is an appropriate collocating access point for the work in hand. An individual work does not usually present itself in terms of author group, time period or specific locality where written, so there is no reason to analyse it in these terms.


Recommendation 4

Collections and individual works should be assigned the same range of fiction form headings as appropriate, working through the hierarchy from the most specific to the most general:

assign: specific form/genre (e.g. Love stories, American)
if none identified then

assign: minor form/genre (e.g. Short stories, American)
if none identified then

assign: major form/genre (e.g. American fiction)

N.B. This recommendation advocates extending current practice for collections to individual works. However, current practice for collected works on the LC bibliographic file reflects a greater degree of flexibility in applying this hierarchy (See Appendix for example). Often the forms Love stories and Short stories are doubled, presumably because they are viewed by the cataloguer as different form aspects and which are not hierarchically related in the reference structure on the specific form heading. Official LC policy is that these form headings should not be doubled and the hierarchy above should be followed strictly [at H1790 2(a) "These headings are used to designate both the form and topic of collections and no additional heading is usually required"]. So associated with this recommendation is a need for clearer guidance in the SCM on when it is appropriate to double form headings and when not.

This is a question for form headings applied without subdivisions only and does not affect the common practice where headings such as American fiction are doubled with a specific genre to allow a period subdivision or an author group to be added (see Recommendation 5 below).


Recommendation 5

The general principle that underlies the depth of analysis currently applied to collected works is equally applicable to individual works.

It can be stated briefly as "when qualifying form headings only assign access points warranted clearly and unambiguously in the presentation of the work in chief bibliographical sources of information (inc. title page, preface, covers)". If this principle is applied then for the most part the distinctions between GSAFD and LCSH practice for collections will remain much as they are now, with only a few exceptions.


The following sub-recommendations are suggested to standardize subdivision practice under fiction form headings according to this single principle:-

$vTranslations; $vTranslations into [name of language]

Current practice

Current LCSH practice allows this subdivision under all levels of fiction form heading: primary, minor and specific genre headings.

Recommendation 5 (a)

Since translation is a form aspect which may equally be found unambiguously presented in an individual or collected work then this subdivision should be applied whenever appropriate, based on the information found in the work.

$xAuthor groups under primary genre heading; 655 [class of persons] writings, [language]

Current practice

The Author groups subdivisions are only applied under the primary genre type headings, e.g. English fiction$xCatholic authors. These are doubled with any specific form headings assigned. Author groups which do not qualify as free-floating subdivisions because they are not national, ethnic or religious groups (SCM H1156, p.3) are expressed as headings such as Children's writings, English; Prisoner's writings, French, etc. and are assigned in addition to other applicable form headings.

In most cases individual works do not identify authors with a particular class of persons. Even where a class of persons or ethnic group is easily identified it could be considered arbitrary and potentially offensive to "index the author" in this way. Then the general rule should be that author group subdivisions and class of persons' writings' headings should only be assigned to collections.

An alternative view might be to consider that guidelines should be given for making a decision on a case by case basis. Some individual works do clearly present the author in terms of their significance as a member of a group, e.g. Scottish authors, Gay or lesbian authors, in a way which is no different to the way in which a collection might be assembled. In the case of the heading English fiction$xScottish authors there is a further justification for a scope note in that the subdivision is bound up wi th the headings assigned to express language/country of origin, so this might have to be allowed as an exception on individual works. The correct array of headings for a collection of Scottish fiction written in English is:

655 Scottish fiction

655 English fiction$xScottish authors

Recommendation 5 (b)

As a general rule author group subdivisions and class of persons' writings headings should only be assigned to collected works. Exceptions should be established only by explicit scope notes under particular headings.

Geographic subdivision

Current practice

Geographic subdivisions can be assigned to all levels of fiction form heading on collected works, to indicate the more specific locality where written, e.g. 655 Detective and mystery stories, American$zNew York (State)$zNew York for a collection of stories written by American authors living and working in New York.

This is akin to the identification of author groups and although such biographical author information may appear incidentally on an individual work, it would be rare that this could be said to be given significance.

Recommendation 5 (c)

To avoid subjectivity on the question it is recommended that geographic subdivision is used only under form headings for collections.

Chronological subdivision

Current practice

Chronological subdivisions are only assigned under primary genre headings, e.g. 655 English fiction$y19th century. These are doubled with any specific form headings assigned.

All individual works are clearly dated, but the time of writing is naturally not presented in the work as significant. At present LCSH allows for cataloguers to assign access for period as appropriate on collected works. In practice this means that most collections are dated by century, in addition to other specific form access assigned, i.e. it is not clear that the principle suggested above is being applied. Rather it appears that century subdivisions are assigned as a matter of course to all or most collections, whether or not they are presented as a significant gathering device in the title or preface of the collected work being catalogued. If ease of identification of date were taken as justification for period subdivision the century subdivisions under primary genre headings would be assigned to all individual works.

Recommendation 5 (d)

It is recommended that only collections are indexed to include period subdivisions according to the principle stated above: that the period covered by the collection is presented in the work as a significant gathering device.

Collections vs. individual works

Having established that in general access to forms of fiction should be the same for individual and for collected works the de facto distinction provided by the current LCSH practice of only indexing collections with form headings will be lost. It might be appropriate to introduce a new $v subfield, e.g.$v Collections, in association with the recommendations made above. This would be used under 655 headings assigned to collected works. Since it is not currently possible to distinguish collections by one author or several it may not be considered appropriate to have a further subdivision, $vCollections by one author. There is also a question whether a construction such as 655$aShort stories $v Collections is a redundant use of the subdivision.

Recommendation 5 (e)

To answer the question above in terms appropriate to all the preceding recommendations it is recommended that a new free-floating subdivision, such as $v Collections, is established for use under form headings of all levels assigned to collected works by several authors, with the exception of Short stories where this would seem an unnecessary qualification. Collections by individual authors should not be distinguished by any form subdivision, but are better associated with individual works by the absence of the qualification. This means that the more distinct form of collections of several authors is explicitly added as an access point.


5. Form/genre vs. Topical access

This section examines the relationship between access by form and access by topic to works of fiction where the two modes of access appear to be conflated in "topical form headings" on the LCSH file.

Pursuing the idea that form and genre access should be extricated from subject access leads to a problem that has been the focus of much discussion on the GSAFD Subcommittee. The SCM at H1790 2(a) defines certain phrase headings as "used to designate both the form and the topic of collections". The emphasis on the topical aspect of these headings is underlined at H1775 3(a) by the assertion that "themes or topics may be expressed as [Topic]--Subdivision ... or as a phrase ([Topic] stories)". In terms of specificity such phrase headings are measured against the hierarchy of topical headings rather than a hierarchy of form headings, which gives rise to the rule at H1790 2(a) which forbids combinations of the kind:
War stories, American
Jungle warfare--Fiction

in favour of:
American fiction
Jungle warfare--Fiction

The former combination would be appropriate under the terms of the GSAFD where War stories is coded separately as a form/genre heading which does not impinge on the specificity of the 650s assigned. An approach that is justified on the grounds that the specific genre is still the appropriate point of collocation as a form heading.

The problem identified by the GSAFD Subcommittee is that many of the topical phrase headings seem better suited for topical access rather than genre access.(11) A topical theme does not always make a new genre of fiction. If the phrase headings are considered as topical theme headings the SCM rules seem correct, but if they are considered to be genre headings they seem wrong. The idea of proliferating genre headings in the revised GSAFD was rejected by the Subcommittee on the grounds that the notion of genre would become debased into a pseudo-topical access. The point was well expressed by Joseph Miller in a GSAFD listserv discussion: "To make every topic that can appear in fiction into a genre is only to obscure the concept of genre altogether."(12)

In order to extricate genre from topic the British Library has picked up this issue by distinguishing two general groups of specific form headings within the fiction genre.(13) These have been labelled "genre headings" to express topical form headings, and "form headings" for all other types. Working definitions of these headings are given below.

5.1 Form headings

Form headings are defined as indicating various aspects of a work :

(a) the special presentation of the narrative, e.g., as letters or diary entries

(b) the provenance of the novel, e.g., adapted from a film or television programme

(c) the intended audience, e.g., children, young people, etc.

(d) its publication in a certain form, e.g., as a dime novel, etc.

These include specific form headings with nontopical aspects, e.g., Humorous stories, Radio stories, Childrenís stories, etc.

5.2 Genre headings

The definition provided in Arlene G. Taylor's paper written for the SAC Subcommittee on Form Data encapsulates British Library policy on genre headings:
"Genre is used rather narrowly to encompass any recognised category of fictional works which is characterised by recognisable conventions, i.e., a group of works all of which tend to explore the same themes and use the same plot formulae, character-types and icons."(14)

The definition in the Fiction Policy Document of the British Library Fiction group is:
"In fiction genre represents a type of novel, so that novels belonging to a certain genre share certain characteristics - of setting (e.g. westerns), theme (e.g. detective and mystery stories), plot, etc. Genre terms are determined by convention, as set by the bibliographic community of publishers, booksellers, librarians and readers. The genre or genres to which a novel belongs should be easily identifiable."(15)

In general, novels belong to one "primary" genre. As Joseph Miller says, "They are like slots that books fall into."(16) The assignment of a single genre heading ensures that a work is categorised according to its primary characteristics, rather than being inappropriately labelled according to secondary or incidental characteristics.

Genre headings are more than topical phrase headings. The GSAFD Subcommittee decided against the use of multiple sub-genres following the view that genres should be easily recognisable and not invented wholesale as a substitute for topical access. Genre headings must be useful and meaningful. A proliferation of genre or sub-genre headings mean that works could become "lost" to readers unaware of the range of genre headings available, and their use would require a sophisticated knowledge of sub-genres by both the reader and the cataloguer.

When considering the issue of extricating genre from topical access, genre headings must be scrutinised to decide whether they do fall within the criteria of genres and provide useful access points as such, or whether they would be more appropriately expressed as [Topic]--Fiction. For example, Detective and mystery stories is a genre heading within the criteria, while Tennis stories is not a conventional or easily identifiable type of writing, and therefore cannot be regarded as a genre heading. Tennis stories as a topical phrase heading is appropriate as a 650 heading for works about tennis stories, but as a subject in a work of fiction should be expressed as Tennis--Fiction.

5.3 Topical phrase headings

At the moment there are topical phrase headings which are being used in two ways - as subject headings and as form headings, for example, Tennis stories. With the implementation of the 655 field it would only be appropriate to convert these headings to 655 headings if they satisfy the criteria for genre headings. However, it may be necessary to retain these as 650 headings for use in critical works. If retained, their use as topical headings need not be retrospectively amended, therefore allowing minimal intervention on the LCSH file. In future, the subject within works of fiction should be assigned the heading in the form [Topic]--Fiction.

5.4 Separating form from subject access

Implicit in the definition of a genre is a certain degree of topical focus. This should be recognised when assigning them in conjunction with topical headings. It is not necessary to duplicate the same level of subject access with additional topical headings. For example, it is not appropriate when assigning the genre heading War stories to also assign War--Fiction. However, it is appropriate to bring out a specific aspect of the work by assigning topical headings such as Jungle warfare--Fiction.

Recommendation 6

Following the work of the GSAFD Subcommittee identifying appropriate headings, it is recommended that topical phrase headings on the LC file which do not fit the criteria for genre headings should not be converted to 655 headings. It is recommended that the see references in these topical phrase headings be removed. This would allow their use both as 650 topical headings, and also in the form [Topic]--Fiction, e.g., retain 650 Christmas stories, and also allow 650$aChristmas$vFiction. The GSAFD Subcommittee should be given the charge of identifying and recommending headings to be converted, and 450s to be deleted, since they are already well-advanced on this theme.

Recommendation 7

It is recommended that the specificity of form/genre access should be kept separate from the specificity of topical access. Therefore, the specificity of form/genre headings which is applied to individual works should also be applied to collections. For example, 655 War stories used in conjunction with Jungle warfare--Fiction.(17) The rule on this at SCM H1790 2(a) should be deleted.


6. Other form data

Other types of form data are also assigned to works of fiction which do not represent sub-genres of the primary genre "fiction". Here the question of extricating access by form from access by subject is given a different slant because the form data transcends the boundaries of any one literary genre. Because of this wider scope this section makes more tentative recommendations.

Where headings represent categories and genres of fiction it has been argued above that form access should always be assigned "sui generis", i.e. along lines that completely disentangle forms from topics. This ensures that access by literary form can be assigned according to the principle of specificity following the hierarchy of form data only. In the case of other form data, indicating pictorial content of a work, or labelling a work as a reader, this principle is less clear. Since this form data does not fit into the hierarchy of a literary form it is not clear that it needs to be disentangled from topical data to the same degree. However, determining how far the data does need to be disentangled is further complicated by the fact that it is currently "entangled" with both topical data and fictional form data.

6.1 Pictorial works

The pictorial or graphic nature of a work can be expressed through form headings or form subdivisions following topical headings.


Main headings can be assigned simply to give post-coordinate access to the pictorial form:-

655 Comic books, strips, etc.

655 Toy and moveable books

Or main headings may combine pictorial form with fiction form in a single phrase heading:-

655 Graphic novels

655 Stories without words

655 Science fiction comic books, strips, etc.

Some apparent form headings are only assigned as topical headings by LC as a matter of policy (unwritten):-

650 Picture books for children


Subdivisions are assigned under topics to give specific access in a pre-coordinate string:-

650 Dogs$vPictorial works$vJuvenile fiction

650 Dogs$vComic books, strips, etc.

650 Dogs$vFiction

 [N.B. In practice the construction: 650 Dogs$vComic books, strips, etc.$vFiction is never used by LC although no rule seems to disallow it.]


 1) Hierarchy of pictorial form headings

 (a) Because not all the main headings which express a pictorial form are assigned as form headings there is a need to review reference structures carefully when establishing 155 authority records for those headings which are to be assigned as form headings, e.g.,

 Picture books for children : is used as a topical heading only

Toy and movable books : is used as topical and form heading

But both are have references to the BT : Illustrated children's books 

(b) The hierarchical relationship of the heading Graphic novels to specific forms of graphic novel, e.g. Science fiction comic books, strips, etc. is puzzling. Currently both are treated equally as NT's of Comic books. strips, etc. Graphic novels is surely a BT of Science fiction comic books... and a NT of Comic books...?

2) Subdivisions vs. main headings

In general subdivisions such as --Pictorial works and --Comic books, strips, etc. are only used after topical headings. The introduction of 655 and $v will allow the possibility of using $v form data unambiguously as qualifications of form headings. This allows for various combinations to express precisely the form of a work:- e.g.,

655 Childrenís stories$vPictorial works
655 Toy and moveable books

the first 655 gives a specific access point to categorise the work as a picture book for children which is also a work of fiction. The second is added postcoordinately to define more precisely the physical form of the book, but clearly the addition of the subdivision to the second heading would be redundant.

655 Young adult fiction
655 Young adult fiction$vPictorial works

this combination expresses the level and the significance of the pictorial content in at least 50% of the work.

3) Disentangling pictorial and fictional form

Headings such as 655 Science fiction comic books, strips, etc. already combine the genre and form data into a single phrase heading. Expressing the language of the genre would currently require doubling:-

655 Science fiction comic books, strips, etc.
655 Science fiction, American

But use of a free-floating subdivision would allow a single pre-coordinated heading to express the same data.

655 Love stories, American$vComic books, strips, etc.

This would provide greater flexibility for the application of specific genres of graphic novel. This assumes, however, that all graphic novel genres are derivative from standard fiction genres. It may be argued that a graphic novel genre should always be established in its own right.(18)


Recommendation 8

It will be necessary to review the hierarchical references when creating 155 headings from 150 headings for pictorial forms. In particular, the hierarchical relationship of the heading Graphic novels should be reviewed.


Recommendation 9

The free-floating subdivisions$vPictorial works and $vComic books, strips etc. should be authorised for use under form headings as a means of further qualifying form or genre data.


6.2 Readers

LCSH practice for expressing the form Reader follows two patterns:-

Readers can be combined with topical and form aspects in pre-coordinate strings. E.g.,

650 French language$vReaders$vDetective and mystery stories

655 Readers$vScience fiction

655 Readers$xAgriculture

Specific types of Reader, however, are only assigned post-coordinately as individual headings. E.g.,

655 Readers (Adult)

655 Readers (Elementary)

655 Readers for new literates

 The practice of constructing pre-coordinate strings requires doubling of headings for post-coordinate access. E.g.,

655 Readers (Adult)

655 Readers$vDetective and mystery stories

655 Detective and mystery stories

The application of the Readers--[Topic] pattern to topics in works of fiction is not envisaged in the SCM H1975, although it is not explicitly excluded, e.g.

655 Readers$xUnited States$yCivil War, 1861-1865

seems wrong because the fictional nature of the work is not apparent. Or:

655 Readers$xUnited States$yCivil War, 1861-1865$vFiction

is confusing because it is not clear whether Readers is itself being qualified by the form subdivision.



The general application of form headings and subdivisions for Readers merits further consideration on the theme of comparing post-coordination with pre-coordination. It would be worth considering whether Readers and the more specific Readers (Elementary) should be applied in the same way: i.e. either Readers should only be assigned without topic or form subdivision or Readers (Elementary) should be capable of being assigned with the same subdivisions.

However, in the narrow context of fiction indexing (which may be a pattern for other forms of literature) specific options might apply.

(1) The heading 655 Readers might be applied as a post-coordinate access point only on works of fiction. This would allow a consistent pattern for bringing out the form Reader with specificity only expressed in terms of the level of audience. Specificity of the form Reader would be disentangled from topical or literary qualification, which would make for broader headings, but complete clarity. In the case of English language Readers this would mean that it would no longer be necessary to double 655 Readers (Adult) with 655 Readers .

Non-English language readers which are also works of fiction would follow the same restriction and the subdivision $v Readers would not be further subdivided on any works of fiction.

(2) The current pattern of subdividing 655 Readers or $v Readers could be maintained as now, but it should be stipulated that for fictional works only the literary form headings should be assigned as subdivisions. Any other topical access on the work should be assigned separately with the standard pattern: 650 [Topic] $v Fiction, and not associated pre-coordinately with the form heading 655 Readers.

This would basically leave LCSH practice unchanged, but ensure that precoordinate strings conform to a consistent pattern.


Recommendation 10

Neither option is specifically recommended. What is recommended is the need for some clarification on the way in which the form Readers can be assigned together with access to fiction forms and topics on fiction. The two options are suggested, but wider considerations on the treatment of readers in general may generate other options for fiction readers.


7. Summary of recommendations

Recommendations have been grouped into proposed changes to the LCSH file and proposed changes to GSAFD and LCSH indexing policy.


Recommendations involving file changes

1) The phrase form of 655 headings should be identical to existing 650 headings, with exceptions treated on a case by case basis (Recommendation 1 at 4.1.1.)

2) The GSAFD Subcommittee should identify present LCSH form/genre headings which are inappropriate for conversion to 655 headings. See references of the form [Topic]-Fiction should be deleted from these headings (Recommendation 6 at 5.4.)

3) The reference structure of the 155 records should follow the present structure of the language qualified form headings as a short term measure. In the long term, the reference structure of the 150 headings should be reviewed and the structure of 155 records transferred from this (Recommendation 3 at 4.1.3.)

4) A new free-floating subdivision $vCollections should be established for use under form headings assigned to collected works by several authors (Recommendation 5 (e) at 4.2.)

5) For pictorial form headings, it will be necessary to review the hierarchical references when creating 155 headings from 150 headings, especially with regard to the hierarchical relationship of the heading Graphic novels (Recommendation 8 at 6.1.)

6) The free-floating subdivisions $vPictorial works and $vComic books, strips etc. should be authorised for use under form headings as a means of further qualifying form or genre data (Recommendation 9 at 6.1.)


Recommendations involving policy changes

1) GSAFD practice should adopt the current LCSH style of language qualified form/genre headings (Recommendation 2 at 4.1.2).

2) Collections and individual works should be assigned the same range of fiction form headings as appropriate, working through the hierarchy from the most specific to the most general (Recommendation 4 at 4.2.).

3) The analysis of individual and collected works by form should follow the common principle that "when qualifying form headings only assign access points warranted clearly and unambiguously in the presentation of the work in chief bibliographical sources of information". Sub-recommendations are suggested to standardize subdivision practice under fiction (Recommendation 5 at 4.2.)

4) The rule at SCM H1790 2(a) prohibiting the combination of a topical phrase heading and a more specific [topic]-Fiction heading should be deleted (Recommendation 7 at 5.4.)

5) The policy for assigning the form Readers to works of fiction needs to be clarified (Recommendation 10 at 6.2.)

Appendix A

back to the top



 (1) Available at
back to text

 (2) This seems to fit with the trend of recent decisions and discussions of the GSAFD subcommittee, which is currently focusing on aligning GSAFD forms of heading with LCSH (and vice-versa).
back to text

 (3) In: Report to ALA ALCTS CCS SAC Subcommittee on Form Headings/Subdivisions Implementation, by Tom Yee, Feb. 1997.
back to text

(4) In "Response to the question "Is all of this worth it?" by David Miller, posted at
back to text

(5) For language read language OR nationality. Language qualifiers in LCSH literature headings are often conflated with the nationality of the literature, e.g. Scottish fiction is assigned to works in English by Scottish authors.
back to text

(6) Utopias was identified as a doubtful genre heading in the form established for topics in a discussion on the GSAFD listserv. See archives at (under February 1997)
back to text

(7) Access provided by MLA coded descriptors are used as a benchmark to assess LCSH access to French fiction in French fiction: LCSH applications, by Florence E de Hart and Karen Matthews, Cataloging & Classification Quarterly vol 9 (2) 1988. The article identifies the lack of language access to individual works of fiction as a limitation of current LCSH practice.
back to text

(8) This position was suggested on the GSAFD listserv in response to Joseph Miller's objection that specific genres of fiction, e.g. Science fiction, should be regarded as transcending language distinctions.
back to text

(9) With reference to the ongoing LC project to improve reference structures on topical headings in the LCSH file: evaluated/unevaluated status of headings is indicated by the values a/b in 008/29.
back to text

(10) This begs the question of whether this level of access should be left to the classification schemes. However, LCSH already provides broad form access that duplicates access through the class mark on collected works. Including this level of access in the indexing language as well is simply to rationalise this particular means of access by form.
back to text

(11) See discussion between Ruth Bogan and Joseph Miller posted to GSAFD listserv November 1996 subject "Fishing stories" at:
back to text

(12) Joseph Miller posted at:
back to text

(13) See BLís Fiction Policy Document at: and
back to text

(14) In "Definition of "Form", "Genre", "Physical characteristics" : paper written for the purpose of discussion by the SAC Subcommittee on Form Data, January 1992 by Arlene Taylor posted at:
back to text

(15) BLís Fiction Policy Document posted at: and
back to text

(16) Message posted to GSAFD listserv July 1997 subject "Genres and genre indexing"
back to text

(17) This is a change of practice to the Subject cataloging manual H1790 2(a)
back to text

(18) Whatís in a Name? Form and Genre Terms for Cartoon Art. A paper presented at the Annual Meeting of the Popular Culture Association, Popular Culture and Libraries, April 9, 1994 by Lucy Shelton Caswell.
back to text



Appendix A

back to the top