Library of Congress

Program for Cooperative Cataloging

The Library of Congress > Cataloging, Acquisitions > PCC > CONSER > Publication Patterns Initiative > Appendix C. The Harvard Load

On February 27, 2001, the OCLC database was "seeded" with 40,000 Harvard holdings records reformatted into 891 fields,using a matching algorithm on the OCLC number. These records were loaded under the following conditions:

  • All were CONSER records with data current as of late January 2001.
  • The Harvard data did not replace any existing 891 fields in the CONSER records.
  • Itwas not added in cases where the allowable record length would be exceeded.
  • The 891/853 and 891/863 pair(s) show only the first issue to which each pattern applies; a hyphen was placed at the end of the statement to show an open range.
  • Because there is no way to verify automatically that the issue was actually the first in the pattern, all holdings data subfields were enclosed in angle brackets < > .
  • Supplement and index data was not included.
  • Indicators for the 891/853 were left as they were in Harvard's Hollis database.
  • Since the statements are open-ended, all 891/863 indicators were set to 40.
  • Since relatively few of the statements contained patterns, the 891/863 field carries a subfield $x "provisional."
  • Hollis lacked sequence numbers in its 863 $8 fields; these are also missing from the 891/863 data.
  • Multiple 891 pairs were loaded if they were present in the source database and fit in the target records.In such cases the linking numbers incremented in the prescribed manner.

Editing these records may involve replacing the existing fields with more complete data.In particular, pattern data may have to be added.As time goes on, some of the patterns cease and should be closed; at that time subsequent patterns may be added.Editing ceased patterns, though optional, is useful and therefore should be done if the situation is fairly apparent so that  fields can be completed without difficulty.  This may not always be the case!  In the usual closing, the editor may have to adjust indicators, remove angle brackets and hyphens from 891/863, and place a sequence number (always 1) after the linking number in 891/863 $8.The subfield $x "provisional" may be removed if the field is edited.You may accept the data already in the record (frequency, 362, numbering peculiarities) as evidence in order to update the 891, even if you do not have the issue on which the description is based.If the data seems contradictory, you may leave the Harvard data unedited and input one or more subsequent patterns based on your own issue.

Example:You have v. 25- of a journal whose frequency is q, whose 362 field says Vol. 1, no. 1 (Jan./Mar. 1965)- and it has Harvard 891 fields as follows:

891 03$9 853 $8 1 $a v. $b no. $i (year) $j (month)
891 40$9 863 $8 1 $a <1>- $i <1965>- $x provisional

You verify that it is quarterly from your own issues.You can edit as follows:

891/853:

Change the indicators to 20, meaning that the pattern subfields will be complete and the issues are correctly captioned.
Insert $u and $v after $b, with values 4 and r.
Add subfields $w q and $x 03.
Add subfield $y pm01/03,04/06,07/09,10/12 (you can also use the new code "cm" in place of "pm").

891/863:

Change the indicators to 41; you will be stripping the hyphen and filling out the second level of first-issue data, changing the nature of the statement to uncompressed;
Add a dot and sequence number 1 after the linking number in subfield 8;
Add a subfield $b containing 1 for the first issue numbering;
Add a subfield $j containing 01/03 for the first issue date;
Remove angle brackets and hyphens from subfield $a and $i;
Remove subfield $x.

Resulting field:

891 20$9 853 $8 1 $a v. $b no. $u 4 $v r $i (year) $j (month) $w q $x 03 $y pm01/03,04/06,07/09,10/12
891 41$9 863 $8 1.1 $a 1 $b 1 $i 1965 $j 01/03

Variation:

If your present issues are not quarterly, and you need to close the Harvard pattern and add your subsequent pattern but don't know when the first pattern ceased, you can use this strategy: 

Verify all of the first pattern except the closing data from the existing data in the record;
Leave the pattern open, but remove the angle brackets and "provisional";
Do not change 891/863 indicators, since "40" is correct for a range of  holdings;
Explain the situation in 891/853 subfield $3:  "closed before 1980" if you know that, or just, "closed"
Add your new pattern:

891 20$9 853 $8 1 $a v. $b no. $u 4 $v r $i (year) $j (month) $w q $x 03 $y pm01/03,04/06,07/09,10/12 $3 closed
891 40$9 863 $8 1.1 $a 1- $b 1- $i 1965- $j 01/03-
891 20  $9 853 $8 2 $a v. $b no. $u 6 $v r $i (year) $j (month) $w b $x 02 $y pm 01/02,03/04,05/06,07/08,09/10,11/12
891 41  $9 863 $8 2.1 $a <25>  $b <1> $i  <1989> $j <01/02>

In your statistics, this counts as a "subsequent" pattern.

Questions? Contact the Project at: CONSERHOLD-L@cornell.edu

Back to Top