DATE: May 14, 1999

NAME: Coding non-Gregorian dates in the MARC 21 Bibliographic Format

SOURCE: Research Libraries Group

SUMMARY: This paper considers the need to code publication dates from non-Gregorian calendars when they appear on items and various coding requirements and options.

KEYWORDS: Field 008/06 (BD); Field 008/07-14 (BD); Field 046 (BD); Type of publication date (BD); Non-Gregorian dates (BD); Special coded dates (BD)

RELATED: DP106 (Jan. 1998); 98-07 (June 1998)


5/14/99 - Forwarded to the MARC Advisory Committee for discussion at the June 1999 MARBI meetings.

6/27/99 - Results of MARC Advisory Committee discussion: There was not a lot of experience with non-Gregorian dates and the need for them in records. It was suggested that there might also be a need for natural history dates as part of a future proposal. Follow-up on this paper by those requesting this will include querying USMARC subscribers and specialized groups inside and outside ALA about special needs for these types of dates.

DISCUSSION PAPER NO. 117: Coding non-Gregorian dates


The publication date may be expressed in various ways on an item. The MARC 21 format defines the 260 subfield $c as the place to describe the publication date for current cataloging. MARC 21 also defines the 008/07-14 and 046 field to hold a coded form of an item's publication date. The coded form is consistently structured to facilitate machine indexing and other processing of the data.

In the experience of most catalogers from the Americas and Europe, nearly all items have only publication dates from the Gregorian calendar. This calendar was created in 1582 as a reform of the Julian calendar then in use in much of Europe. The use of the calendar has spread in the last few centuries so that it is in widespread use throughout the world today. Yet, many other calendars are in use throughout the world, and many items are published with dates from those calendars. For simplicity in this paper, those calendars will collectively be called "non-Gregorian" calendars.

The MARC 21 format currently only accommodates coding the Gregorian calendar publication date of an item. If an item's source of information includes a non-Gregorian calendar publication date, the cataloger may express it in a descriptive field, such as the 260. However, there is no MARC element in which to record a coded form of the non-Gregorian publication date.

The lack of a standard coded form for non-Gregorian publication dates also makes it difficult to create indexes of those dates, to qualify search results by such dates, and to customize displays of search results involving such dates.

The lack of accommodation for non-Gregorian dates in MARC 21 makes retrieval more difficult. A year according to a non-Gregorian calendar may correspond to parts of successive Gregorian years. Therefore, retrieval by Gregorian year is less precise in such cases. Furthermore, when the source of information has multiple non-Gregorian dates, the corresponding Gregorian year ranges cannot be completely expressed in 008/07-14.

The lack of correspondence of years among different calendars is due to different definitions of the length of a year, what constitutes the first day of a year (i.e., when a year starts), or the start of the calendar's epoch (i.e., what is the first year of the calendar). The length of a year is often based on astronomical cycles such as those of the moon and sun, but different calendars relate lunar and solar cycles to a year in a variety of ways. The first day of a year may be based on the observation of a particular astronomical event such as an equinox or solstice, on an arithmetic calculation which may itself be based on an astronomical event, or on some other system. The basis of a calendrical epoch may be any event, such as the Prophet Mohammed's emigration (hijra) to Medina, the founding of the French Republic, or the start of the fourth and last stage (lasting 432,000 years) of the 4,320,000 year era that began with the last re-creation of the world (Hindu calendar's current Kali Yuga epoch).

The following table indicates the length of a year and first day of the year in a few calendars in current use:

Calendar Length in days* First day of year
Baha'i 365 (366)Vernal equinox
Chinese365 (366)12th or 13th new moon after previous winter solstice
Islamic (civil)354 (355)1st day of month Muharram
Jewish353-355 (383-385)New moon of month Tishri

*Days in parentheses are for leap years.

Many calendars in current use have variants to suit some civil, ecclesiastical, local, national, or regional purpose. Many calendars in current use have historical variants as well.

When dates from other calendars are converted to the Gregorian calendar for the 008/07-14 or the 046 field, the difference in the length of the year and/or its first day often result in a single non-Gregorian calendar year corresponding to 2 or 3 Gregorian calendar years. The usual Jewish or Islamic year is shorter than the usual Gregorian year, so those Jewish and Islamic years often span two Gregorian years. The leap year in the Jewish calendar is longer than the Gregorian leap year, so it is possible for a Jewish leap year to span three Gregorian years. Even calendars that define a year as the same length as the Gregorian calendar often span two Gregorian years because those calendars define the first day of the year differently than does the Gregorian.

Users of catalog records that have Gregorian dates in place of non-Gregorian dates must account for the translation to Gregorian numeration, using a span of 2 or 3 years in some cases to retrieve the desired record. More precise retrieval will be possible if MARC records contain the non-Gregorian calendar publication date in a standard coded form. Catalog users will not be burdened with converting known dates in another calendar to the Gregorian calendar.


2.1. Requirements for coded non-Gregorian dates in MARC

1) Identify that the coded data is in a non-Gregorian calendar.

2) Identify the calendar of the date to allow a system to distinguish the calendar and/or to create an index for dates from a particular calendar, qualify searches, or customize displays of search results.

3) Allow any date type code defined in the 008/06 to be used to identify a non-Gregorian publication date or dates.

4) Allow coding of more than one non-Gregorian publication date when a particular item has more than one.

5) Allow identification of an incorrect publication date in a non-Gregorian calendar.

2.2. Options

The 008 field has bytes 07-14 defined for coded publication dates. The dates are expressed according to the Gregorian calendar. For consistency of retrieval and indexing, this restriction to a single calendar should be maintained. Dates from other calendars should not be entered in the 008.

The only other field that contains coded publication dates is the 046 field. In 1998, its coverage was expanded to add incorrect dates, so it is an obvious choice for further expansion to accommodate non-Gregorian publication dates.

The option exists to add a new field for coded publication dates in non-Gregorian calendars, but that would create a third MARC element that systems would have to process to deal with coded publication date information. If no problems are identified in expansion of the 046 field, there appears to be no advantage to creating a new field for this information.

2.3. Expansion of the 046 for non-Gregorian publication dates

The 046 field is currently defined as a supplement to the 008 for coding of publication dates in the Gregorian calendar when a B.C. date is involved. In 1998, MARBI approved a change to expand the field to cover incorrect dates with Proposal No. 98-07 (Recording Incorrect Dates in Field 008/06-14 in the USMARC Bibliographic Format). The 046 is the obvious choice for coded publication dates in other calendars.

In the revision of field 046 to be published in MARC 21, the following was changed:

- Definitions/names of data elements no longer restrict the field/subfields to B.C. dates
- The name of the field was changed in Proposal No. 98-07 to: Special coded dates
- Subfield $6 (Linkage) has been included
- Subfields $c (Date 1 (C.E. date)) and $e (Date 2 (C.E. date)) were expanded to include incorrect dates.

See attachment A for the revised definition of field 046 to be included in MARC 21 Format for Bibliographic Data.

Based on the coding requirements stated above, the 046 would need the following changes:

2.3.1. Define a means of identifying the calendar. The use of subfields specifically defined for a particular calendar is not an option because there are so many calendars. The use of indicator codes to identify calendars must be rejected for the same reason.

The use of subfield $2 is an option. It would require definition of an authorized list of calendar codes.

Is there an existing standard that defines a code or mnemonic of some sort for calendars? The use of subfield $2 to identify the calendar would require extra system processing to relate calendar and coded dates for indexing, display and retrieval purposes. Is there a better option?

2.3.2. Define two additional subfields for non-Gregorian calendar dates. Two new subfields could be defined to include only non-Gregorian calendar dates. This would allow for systems to look in one place for such dates. It is suggested that subfield $f be defined for Date 1 (Non-Gregorian calendar) and subfield $g for Date 2 (Non-Gregorian calendar).

2.3.3. Define additional codes (c, d, e, u) in subfield $a to correspond with those defined in 008/06. The 046 subfield $a currently only defines a subset of the publication date type codes defined in 008/06. Non-Gregorian calendar publication dates could be of any type defined in 008/06, except for code b. All other 008/06 codes must be defined in 046 subfield $a.

2.3.4. Make the field repeatable. In a case where publication dates in more than one non-Gregorian calendar and/or an incorrect date occurs for an item, it will be necessary to repeat the 046 field.

2.3.5. Clarify use of codes in subfield $a. The existing definition of 046 subfield $a does not limit the subfield to a single code. Limiting to a single code seems appropriate to avoid difficulties in machine processing. The limitation to a single code should be made explicit.

Code x was defined for 046 subfield $a for incorrect publication dates. It is possible in some cases to determine that an incorrect date is also a single date or multiple date. Is there any advantage to making such distinctions of incorrect dates?

It is also possible for an incorrect da te to be in a non-Gregorian calendar. Incorrect non-Gregorian dates should be sufficiently treated with the "x" date type code, the incorrect date(s) in the appropriate subfield(s) and the code to identify the calendar (such as using $2 as proposed above).

2.4 Impact of 046 expansion on 008

If the 046 field is chosen for expansion to include non-Gregorian publication dates, then several considerations arise.

The 008/07-14 does not explicitly restrict dates to the Gregorian calendar. It is likely that MARC records have both Julian and Gregorian dates in the 008. Should MARC 21 restrict the 008/07-14 to Gregorian dates for the sake of consistency? Such a restriction might be particularly beneficial to allow systems to distinguish detailed dates in the Gregorian and Julian calendars. Where the Gregorian and Julian date information is the same, however, the recording of Julian dates in the 046 in addition to Gregorian dates in the 008 should be strictly optional (for those cases where some advantage is found for that practice).

The definition of 008/06 code b must be evaluated. The code is currently defined to mean that at least one B.C. date is involved, so the publication dates are found in the 046 field. If the purpose of the code is to send users to field 046, then it could be redefined as "No dates given, use field 046". Since field 046 has already been expanded to include incorrect dates, systems should no longer expect to find only publication dates involving at least one B.C. date in the 046.

2.5 Rules for recording Gregorian equivalents in 008

When a non-Gregorian year equates to a range of Gregorian calendar years, the MARC 21 Bibliographic format provides no explicit instruction to determine whether to put the earliest or latest Gregorian date(s) in the 008/07-14. There is an example under code "s" (Single date) that shows putting the earliest Gregorian date in 008/07-10. Is there a need to explicitly state which date to choose for each type of date? For example, if a multipart item's non-Gregorian publication dates equate to the Gregorian dates 1963 or 1964 to 1968 or 1969, should the definition of code "m" give instructions on which Gregorian dates to choose?


Assuming that the 046 is expanded as indicated above, a few coding examples will be useful to illustrate the discussion. In the examples $f represents the first non-Gregorian date subfield and $g represents the second such date.

3.1. Item has publication date 1377 in the civil calendar of Iran:

     008    990401s1998    ...
     046    $as$f1377$2<code for calendar>
     260    ... $c1377 [1998 or 1999]

3.2. Item has publication date Germinal, an 4 of the French Revolutionary calendar:

     008    880818s1796    ...
     046    $as$f0004$2<code for calendar>
     260    ... $cGerminal, an 4 [1796]

3.3. Item has publication date 1605/51 from the English Old Style calendar (the Old Style calendar observed March 25 as the first day of the year, so a date range indicated that a work was published between January 1 and March 25), but the1605 is a typographic error (should have been 1650):

     008    921016s1651    ...
     046    $ax$f1605$2<code for calendar>
     046    $as$f1650$2<code for calendar>
     260    ... $c1605/51 [i.e. 1650/51]

3.4. Item has publication date min kuo 59 from Chinese calendar in use in Taiwan, and the item is a reprint of 1926 edition (min kuo 15):

     008    841107r19701926...
     046    $ar$f0059$g0015$2<code for calendar>
     260    ... $cMin kuo 59 [1970]
     500    Reprint, with new introduction. Originally published:  ... min kuo
              15 [1926]


1. Are there calendars that have specific notation features that require additional changes in the 046?

2. How will detailed dates be represented in various calendars? Are there calendars that commonly express a date using more than 4 digits for a year, 2 digits for a month, and 2 digits for a day? Are there calendars that commonly express a date using more than three elements for year, month and day?

3. Are there publication practices in other countries that require additional date type codes beyond those already defined in 008/06?

4. Are there calendars in use that have no standard means of representing dates using numerals (i.e., in coded form)? If so, what authority would be used to create such a standard for those calendars to permit MARC coding?

[The key point for issues 2 and 4 is that the recording of a coded non-Gregorian date must be according to a uniform method for that calendar. The uniform method for a particular calendar could be considered like a code list and maintained by LC. Appropriate organizations should be invited to contribute to this work, including:
- for the Jewish calendar, the Association of Jewish Libraries (AJL);
- for the Islamic calendar and any other calendars used in the Middle East, the Middle East Librarians Association (MELA);
- for East Asian calendars, the Council on East Asian Libraries (CEAL);
- for South Asian calendars, the Committee on Research Materials on Southeast Asia (CORMOSEA).
Are there others?

Go to:

Library of Congress
Library of Congress Help Desk (07/14/99)