I appreciate the paper’s motivation to improve structures for controlled provenance data, especially in the holdings format, but I have concerns about the methods proposed. 

6.1: Yes, it is unquestionably valuable to accommodate control numbers for agents associated by provenance and for controlled terms associated with provenance. Control numbers for specific markings (e.g. bookplates) may be a use-case unique to (or infrequently found beyond) the D-A-CH community, but I don’t dispute the value there too.

6.2: Of the four options presented, I’d prefer nos. 1 and 2 (neither requiring a change to MARC).  For option 3, I’m not sure if it’s appropriate for a control number to be associated only with an agent when the full accompanying custodial history note (subfield $a) may reference the agent and markings, dates, places, etc. That seems a mismatch. 

I have some appreciation for the two novelties introduced by option 4 (field 361): [1] an explicit connection between marking evidence and agents (in controlled subfields), which would enable the display of bundled provenance data somewhat in the manner of CERL’s Material Evidence in Incunabula (for example, please see: https://data.cerl.org/mei/02000344); and [2] a mechanism to record provenance agents and provenance terms in controlled form within a holdings record, which is currently impossible. But, as noted in the CCM and BIC responses, a single 361 field appears inadequate to record all of the information proposed (or needed), especially all of the control numbers and sources for the agents, terms, and markings. 

I’m also concerned that the semantics of subfields proposed in field 361 would overlap with existing fields defined for provenance data (541, 561, 562, 563, 655, 7xx), especially in the context of a Bib record. If our goal is to improve or maintain standardization, introducing more variation seems undesirable.

I strongly endorse this CCM comment: “CCM would welcome further exploring the use of the holdings format for provenance information. The examples in section 4.4 show how much clearer this is than attempting to code provenance of multiple copies in the same bibliographic record, particularly in a shared environment.” And this National Library of Spain comment: “Maybe devising a proper access point in the Holdings format could be the shortest way to implement Agent-Item relationships from LRM/RDA.” I would like to see fields 655 and 7xx mirrored in the Holdings format to fully express structured data for provenance terms and agents. 

6.3: No opinion. 

6.4: No opinion.

6.5: See my comments to 6.2.

6.6: See my comments to 6.2.

6.7: No opinion. 

6.8: The Art and Rare Materials (ARM) extension to BIBFRAME defines separate classes for Custodial history and Marking. Examples are given on pages 34-36 of the ARM 1.0 implementation guidelines (PDF—accessible through Git Hub). Though both are associated with an Item entity, I don’t see a method in the modeling for asserting connections between a specific marking and related custodial history, as proposed in the current discussion paper.  

I would support the addition of a new subfield 246 $s to contain an unparsed PPDM or "provision" statement. This would appear to be well-suited to accommodate various RDA manifestation statement elements: manifestation production statement, manifestation publication statement, manifestation distribution statement, and manifestation manufacture statement. Per RDA, manifestation PPDM statements could include any or all of place, date, or name of the providing agent, instead of or in addition to the separate elements for those data. 

Especially in a rare materials context, fuller transcription of imprint statements from manifestations plays an important role in both representation and differentiation between similar manifestations. In many cases, statements on manuscript or early printed texts, maps, scores, graphics, etc. may vary widely from modern Western conventions of title page-like presentation of data on the manifestation. Statement data may be spread out across different sources on the manifestation, they may be grammatically inseparable from data belonging in other elements or presented in an order and style that doesn't correspond to ISBD order or punctuation. Depending on the standard being applied, the cataloger may need to transpose data, omit linking words or addresses, and/or interpolate punctuation and bracketed data in the 260/264 field, in ways that increasingly distort what is actually found on the manifestation, to assemble a "complete" statement in the 264 field. Then, DCRM instructions may also require separate notes to explain these transpositions or interpolations. After all that, display and discovery layers are increasingly no longer presenting the data in the ISBD syntax and order we went through so much work to formulate in the first place.

So, the ability to record the statement(s) instead (or also) in the order and form presented on the manifestation could lessen the need for catalogers to parse the data at the expense of representational transcription.

Whether the 264 field data is recorded in parsed or unparsed form (or both), I agree with the DP that it is often valuable to supplement the transcribed statements with controlled forms of places, agent names, and dates in 7XX, 008, etc., to support more consistent searching, compensating for the wide variation in whether or how the provision data may be presented on manifestations.

All of the examples provided in the DP show statements in 264 $s that are still "parsed" by ISBD syntax/punctuation and supplied data ("[place of publication not identified]" etc.) for "incomplete" statements. If ISBD parsing is still expected within a single subfield $s, then I don't see as much benefit to the new subfield over current practice of separate subfields that more or less duplicate the ISBD parsing. Rather, I would consider the new $s more useful for statements as transcribed, without being reconfigured into ISBD syntax/order or interpolated with supplied data. For example:

$s Emprynted the yere of oure Lorde a. MCCCCC & xiij by Richard Pynson, prynter vnto the kyng[es] noble grace.

$s Verlegt zu Jena von Joh. Jacob Ehrdten und Gedruckt zu Mühlhausen von Tobias David Brücknern, anno 1698 

$s Printed in the year MDCCI

Additionally, when a statement is not parsed, it may not be possible or practical to select a single appropriate value from the existing second indicator values. This may happen when the type of provision is ambiguous, or when the statement represents multiple provision types (either explicitly, as with "Printed and sold by" or implicitly, as early European printers often functioned also as publishers). Or, this may be the case when no attempt was made to determine or code the type of provision (e.g. when the cataloger is not fluent in the language, if the data is generated by an automated process or converted from a card), or other reasons. To account for such cases, I would suggest either or both options:

The definition of 260 $s (as suggested in question 6.5) for unparsed statements when it is not possible or desirable to associate a single provision type with the statement.

and/or

The definition of new 264 second indicator value(s) to indicate cases when the provision type or function is unknown or mixed/multiple, such as:

# - No information provided

and/or

5 - Mixed function or Multiple functions

Lastly, with 264_4, I wonder if it could be appropriate for $s to contain transcribed copyright statements beyond dates, corresponding roughly to RDA element: manifestation copyright statement, e.g.:

264_4 $c ©1866 $s Entered, according to Act of Congress, in the year 1866 in the clerk's office of the Dist. Court of the U.S., for the Southern District of New York

Though I think that might require a corresponding RDA revision proposal as the manifestation copyright statement element seems more restricted to dates. 

As with 2022-DP10, I would support the definition of 490 $s, which could correspond to the RDA element: manifestation series statement and consider it more useful for transcribed series statements that may not have been manipulated into ISBD syntax. For example:

490_0 $s Tract no. I of the American Peace Society

490 _0 $s No. XII, Modern standard drama, edited by Epes Sargent

A more general comment, getting beyond the scope of these two discussion papers, but for the reasons described above, I can see a case for defining similar subfields for other unparsed manifestation statements in the MARC Bibliographic Format:

A new subfield in 245 for unparsed Title/SOR statements, corresponding to RDA element: manifestation title and statement of responsibility statement (though 245 $s is already defined for version so another subfield code would need to be used)

250 $s for unparsed edition statements corresponding to RDA element: manifestation edition statement.  

An entirely new field for unparsed manifestation statements, corresponding to RDA element: manifestation statement, for statements that cannot be more specifically typed or combines title, edition, series, PPDMC, etc., into a single statement, possibly in combination with other source data that doesn't fall into any of the "types" (such as prices, pious invocations, privilege statements, limitation statements, etc.), that we might otherwise omit or give as quoted notes in the 500 field. If 264_4 $s as proposed above is not appropriate for copyright statements, those could go in this field instead.