DATE: May 14, 2001
NAME: Spans of enumeration and chronology in expressing publication patterns in the MARC 21 Holdings Format
SOURCE: CONSER Task Force on Publication Patterns and Holdings
SUMMARY: This paper considers alternatives for expressing patterns for enumeration and chronology that span issues or years. Alternatives considered are the addition of a new subfield $p (Span interval) or enhancement of subfield $y (Regularity pattern).
KEYWORDS: Subfield $y in field 853-855 (HD); Subfield $p in field 853-855; Span interval; Regularity pattern
RELATED: 2001-DP10 (June 2001)
5/14/2001 - Made available to the MARC community for discussion.
6/16/01 - Results of the MARC Advisory Committee discussion - Although participants were generally interested in providing a method for expressing patterns for enumeration and chronology that span issues or years, some members expressed concern over the complexity of doing so. Overall, the group favored both using a technique like in subfield $y, and defining a new code for spanned years. Participants also favored adding a code in subfield $y, as indicated in the examples under section 2.2. A proposal may be written for the midwinter meeting.
Currently, the prediction patterns that can be coded in field 853-855 predictions do not offer sufficient opportunities to express spanned enumeration and chronology for serial issues.
The following are inadequacies of the current coding:
Currently, limited opportunities to express spanned chronology are covered in $y, Regularity Pattern. In the current model, if a title combines its July and August issues, one can issue the following 85X prediction:
853 02 $a v. $b no. $u 12$v r $w 12 $ypm 01,02,03,04,05,06,07/08,09,10,11,12
to yield a prediction like this:
|v.1 no. 1||2001, Jan||v.1 no. 07/08||2001, Jul/Aug|
|v.1 no. 2||2001, Feb||v.1 no. 09||2001, Sep|
|v.1 no. 3||2001, Mar||v.1 no. 10||2001, Oct.|
|v.1 no. 4||2001, Apr||v.1 no. 11||2001, Nov|
|v.1 no. 5||2001, May||v.1 no. 12||2001, Dec|
|v.1 no. 6||2001, June|
In cases where the combined July/August issue has a combined chronology attributed to single enumerative value (as shown below), one has no means in the 853-5 fields to accurately express the result:
|v.1 no. 1||2001, Jan||v.1 no. 07||2001, Jul/Aug|
|v.1 no. 2||2001, Feb||v.1 no. 08||2001, Sep|
|v.1 no. 3||2001, Mar||v.1 no. 09||2001, Oct.|
|v.1 no. 4||2001, APR||v.1 no. 10||2001, Nov|
|v.1 no. 5||2001, May||v.1 no. 11||2001, DEC|
|v.1 no. 6||2001, June|
In the first example, it is assumed that if the chronology combines, the enumeration also combines. In the second example, we have the chronology in combination, but attributed to a single enumerative value. Some possible solutions to these problems are examined below.2.1. Define Subfield $p to allow for specification of spanned issues at each level
It may be desirable to embed the intervals of spanned issues within the specification of the enumeration levels themselves. With this method it may be possible to define spanned enumeration that embraces both continuous and restarting numbering continuity ($v).One alternative is to introduce a new subfield for ranges within the enumeration levels. This new subfield, suggested here as subfield $p (Span interval), could be introduced into any level of enumeration. It could be applied in cases where a level of enumeration was consistently receiving a span of enumeration values in every instance of publication. This formulation would permit us to predict the following:
anno 50, no. 21-23 (1999, Jan)
anno 50, no. 24-26 (1999, Feb)
anno 50, no. 27-29 (1999, Mar)
Thus, if a title has a span of 2 for its enumeration, we could express the span interval as follows:
853 02 $a anno $b n. $u 12 $p 2 $v c $w 12
in the case where $p is the span interval.
In this case the enumeration picks up the range value of 2 and is able to calculate the sequence 21-23, 24-26, and 27-29. It is clear that titles such as these could be subjected to predictive calculation.
Even the intervention of a subfield $p to control continuous numbering ranges would only work in those instances where the intervals were consistently maintained across all issue enumeration at a given level. This method does not provide for the situation where combinations occur at irregular intervals, such as the case of "Astrophysics and Space Science." This Kluwer title has a pattern in 2000 that emerges like this:
|v. 271, Issue 1, 2000||v.272, Issue 4, 2000|
|v.271, Issue 2, 2000||v. 273, Issue 1/4 (2000,Sep.)|
|v.271, Issue 3, 2000||v. 274, Issue 1/2 (2000)|
|v.271, Issue 4, 2000||v. 274, Issue 3, 2000|
|v.272, Issue 1/3, 2000||v. 274, Issue 4, 2000|
We can tell that there are four numbered volumes in a given calendar year. Beyond this the combinations at the second level of enumeration vary from one volume to the next within a given year. Titles like this go beyond conventional prediction formulas and are the logical candidates for heavily edited inventory requiring extensive staff intervention.
If subfield $p were defined as span interval and subfield $s as intervening interval, as suggested in Discussion Paper 2001-DP10, there would still be three subfields left in the field (subfields $n, $q, and $r).
2.2. Expand Chronology codes in Subfield $y to include spanned years.
Various options may be available to adapt $y to expand the scope of its coverage. One method includes the modification of the current Subfield $y chronology codes to allow for spanned years. The current chronology code definition specifies d (day), m (month), s (season) and w (week). The new code, "y", could take the following form:
|Chronology Type Code||Possible Chronology Code Pattern||
where the last digit of the four digit year field would indicate the range of the span. For example, yyy1/yyy3 would take an initial input year of 1999 and add two years to the initial year to derive the ending value, 2001. From a programming standpoint, the system would compute an interval for the year range by subtracting the value in the first year range (in this case "1") from the value in the second year range (in this case "3") to yield an interval of "2". Then "2" would be added to the value in the starting range to derive the ending range. This alternative illustrates the need for additional values for frequency in subfield $w. It may be worth considering additional alpha codes for triennial, quinquennial and other multi-year frequencies.
If the definition is expanded to embrace "y" (year) we could potentially output a field such that:
853 02 $a v $i (year) $w g $ypy yyy1/yyy2
863 02 $a 10 $i 1999/2000
This example expresses the idea of an issue whose chronological dates span 1999 to 2000.
2.3. Expand Subfield $y to allow for spanned enumeration
An alternative method for spanned intervals to that contained in section 2.1 expands the use of the $yo (omitted) and $yp (published) character codes in referring to publication or non- publication of an item to include enumeration values.
For enumeration only titles the level of enumeration affecting the combinations could be expressed as the following for the example in section 2.0 above:
853 02 $a v $b no. $u 12 $v r $w 11 $ype1 no.1, no.2, no.3, no.4, no.5, no.6, no.7/8, no.9, no.10, no.11,no.12
Here the code "el" stands for enumeration level 1. This approach to coding subfield $y is detailed in Discussion Paper 2001-DP10. Such methods could only prove effective where numbering continuity ($v) restarts. This is due to the fact that the enumeration, when its continuity is continuous, does not assume the form of a repeating pattern but has different values every year. It should also be noted that high frequency titles like biweekly would result in very long Subfield $y strings of data.
In extending the coding to a case with both enumeration and chronology, consideration might be given to combining the spanned chronology with the unspanned enumeration in brackets.
One could explicitly express the published items paired with their enumeration partners as values in Subfield $y as follows:
853 02 $a v $b no. $u 12 $v r $w 11
$ypme2 01(no.1),02(no.2),03(no.3),04(no.4),05(no.5),06(no.6), 07/08 (no.7),09(no.8), 10(no.9),11(no.10),12(no.11)
In this example, we augment the $y chronology code, "m" with a new enumeration code, "e2", that acknowledges the presence of months in combination with their equivalent levels of enumeration. A problem with this approach is that the codes defining the regularity values would require four positions if it is necessary to express both chronology and enumeration aspects, while currently in subfield $y only three positions are expected.
An alternative approach would allow subfield $y to be repeated:
853 02 $a v $b no. $u 12 $v r $w 11 $ypm01,02,03,04,05,06,07/08,09,10,11,12 $ype2no.1,no.2,no.3,no.4,no.5,no.6,no.7,no.8,no.9,no.10,no.11
The first subfield expresses regularity in terms of chronology, while the second in terms of enumeration.
3. QUESTIONS FOR FURTHER DISCUSSION
3.1. What are the practical limits of prediction in the communication format? Is it reasonable to expand $y definitions to account for combinations that can be expressed in the kind of iteration of repeating strings of enumeration and chronology seen in the examples from this paper?
3.2. Which approach listed above is most flexible for encoding these types of patterns?
3.3. How does the approach considered desirable in 2001-DP10 affect our decisions about the alternatives described in this paper?
3.4. If a new subfield is defined, is it reasonable to define $p at all levels of enumeration? If so, should we also consider defining it for levels of chronology to cover such things as multi-year combinations?
3.5. If additional values for frequency are needed for prediction purposes, must they also be defined for serials in fixed field 008/18 (frequency) of the bibliographic format?