When is a part not a part?

Or, to be more precise, under what circumstances can a work be a part of more than one work?

I have come across several instances where a work has been listed as a ‘part’ of another work where the relationship hadn’t struck me as being “proper”. The one that struck me most recently was the soundtrack Neon Genesis Evangelion which listed several original Bach works among its parts (in one case duplicated by an arrangement). I deleted these relationships, but haven’t fixed it with alternatives (at least, not yet…) as the matter got me thinking about the whole issue of claiming an original work to be part of some other later ‘work’ just because the ‘composer’ (compiler?) chose to include it. This strikes me as wrong and potentially chaotic, with many famous pieces being listed as “part of” lots of derivative works as well as the the original work.

I searched in vain for something in the style guidelines - perhaps I missed it?

One obvious approach is to use an “arrangement” relationship, but what if the soundtrack or whatever just used the original composition?

On further investigation, there are more subtle instances. For example, Holst’s ‘Planet Suite’ has two MB works: The Planets, op. 32 (original version) and The Planets Suite for Orchestra, op. 32 (with “Pluto” by Colin Matthews). At least the latter has the “version of” relationship, but surely any of the original ‘planets’ should only be shown as a part of the original?

Perhaps there is something missing in the relationship structure that leads to this state of affairs? In other words, should there not be a “used in” relationship (reverse = “uses”) where the other work is not the composition of which it is a part but simply quotes it in full or part?

Any thoughts before I go mindlessly hacking more relationships :wink:?

EDIT: Looking into Neon Genesis Evangelion a bit more, I have no regrets about removing the relationships. It seems that the list of “parts” is a general mish-mash of stuff that appears on various media. The classical works seem to be simply recordings, not part of separate works, from what I can see. The more general question remains, however. Not wishing to denigrate Colin Matthews’ addition, but what’s to stop me adding another random rock to the suite and listing a new work, calling it “The Planets Suite, op. 32 (with two tings on a triangle for Astraea by MetaTunes)”?


It would be really annoying if the more popular classical works were cluttered with dozens of links to ‘soundtrack’ compositions, so I’d rather not have these links in any form.

There is often a difference between the parts in the ‘suite’ and the corresponding parts of the larger, original work the suite is based on, so that would require new part works and links between the old and the new (also between the parts). If the parts are exactly the same, I would reuse them, but they often have a slightly different name anyway. If the name is different I would also create new works.

Perhaps the most obvious work with multiple versions is Requiem in D minor, K. 626. Here I note that none of the versions claims the parts of other versions as part of themselves - they either just list the revised parts or have distinct works with disambiguation.
In my view The Planets Suite for Orchestra, op. 32 (with “Pluto” by Colin Matthews) is not a work. There is the original Planets Suite and then there is Pluto (and possibly a revision of Neptune). I would be in favour of removing the work entirely and just leaving Pluto as a stand-alone work. Unfortunately I don’t have a recording of Pluto or the apparently revised version of Neptune to go with it.

EDIT: An alternative would be to rename the parent work to be something like “Addition to the Planets Suite” and delete all the duplicated parts. Votes?

In the absence of objections, I propose to rename the extended work to "The Planets Suite extension (with “Pluto” by Colin Matthews) and delete all but Pluto from it. If someone wants to add a revision of Neptune (ending changed) then feel free (EDIT - actually I’ve done that too). I will also delete all the Bach work relationships from the above-mentioned soundtrack.

IMO while the soundtrack case is one where it makes sense to remove the links (they can be in a soundtrack album but they’re not really part of the score), the Planets case is a very different one, and the parts should remain linked there. The Colin Matthews version of The Planets is not just two movements, but all of them - if you only played Pluto, then you didn’t play the work. Similarly, a revision of an opera adding, say, a balleto for the French market shouldn’t only have that new bit, but all the parts of the opera needed to have an actual performance of the opera.


I really don’t get that. He had absolutely no creative input to the other parts. To my mind, what he has done is added an extension. As I said earlier:

Also your suggestion is inconsistent with the way Mozart’s requiem has been dealt with - only revised parts included in the various versions, or at least a new (disambiguated) part included. An option would be to add new works like “…Mars, the bringer of war (Colin Matthews version)” but that implies that he did something to it (although I suspect the Requiem may occasionally suffer from this issue too).

An additional downside of including parts in “works” where they are not part of the created content of those “works” is that a part now has multiple “parents”, only one of which is the true created work. This causes problems for my Classical Extras plugin as it does not know which is the true parent. If there were a relationship attribute like “quoted in” or some such, then it would not be such an issue. As it is, it would be quite messy to work out (via the “version of” relationship of the two parents), which is the true parent.

This is still an issue in some cases even if you discount these ones. If Holst himself had made two versions, changing only one movement or two for the second version (that certainly exists for some works) the shared movements should be linked as parts of both parents, and none of them would be “the true created work” in any way :slight_smile:

Quote from style guidelines:

Any original classical works can be added to MusicBrainz

To my mind, that means that you can’t add, as a parts of a work, things that are not an original part of the work.

Fair comment (as in this case the composer did create the work) - if you can link the works (I can’t bring them to mind just now) then that would help :wink:

To return to the planets (per last.fm):

In 2000, the Hallé Orchestra commissioned the English composer Colin Matthews, an authority on Holst, to write a new eighth movement, which Matthews entitled, “Pluto, the Renewer”. Dedicated to the late Imogen Holst, Gustav Holst’s daughter, it was first performed in Manchester on 11 May 2000, with Kent Nagano conducting the Hallé Orchestra. Matthews also changed the ending of “Neptune” slightly so that movement would lead directly into “Pluto”.

He was not commisioned to write, neither did he write, a new work “The Planets Suite for Orchestra, op. 32 (with Pluto…)”

what’s to stop me adding another random rock to the suite and listing a new work, calling it “The Planets Suite, op. 32 (with two tings on a triangle for Astraea by MetaTunes)”?

Unlike yours, Colin Matthews’ work has several recorded performances by different artists, and thus satisfies the criteria for a work.

Still, when you look at the ways those are presented, I do agree that they are not attempting to present a new composite work, but a performance of Holst’s The Planets followed by a performance of Matthews’ Pluto. As such, I agree that the composite work we have at present of The Planets plus Pluto is a mistake.

Pluto has been downgraded now anyway. It isn’t even a planet any more.

Working on it :laughing:
But seriously, I’m glad someone understands the issue with this particular “work”.

On the other matter which @reosarevok raises regarding work revisions by composers where only some parts were revised. I’m still trying to track these down. One instance that comes to mind (now) is Stravinsky’s Firebird suites (of which there are 3 - 1911, 1919 and 1945), but I think that the orchestration is different throughout, so all the parts are subtly different. In MB the revised versions have separate work names, so I’m still looking for an example of a part which is genuinely an identical part in two separate parent works.

Assuming such an example exists, I think it is a MB database issue, not just an issue for my plugin: i.e. for any particular recording, how are we supposed to know what work it is actually a recording of? It may be possible to see this from the context (either by the other parts in a release, or from a track title in a release), but for the recording per se, this is not possible without either (a) a disambiguation of the related work or (b) a specific relationship attribute. Otherwise it would appear to be a quantum superposition of two works.

One example that comes to mind is the parts of “The Four Seasons”, e.g.:

(This is more out of pragmatism than being a separate Work per se, due to how often and well know the Four Seasons is outside of “Il cimento dell’armonia e dell’inventione”.)

The subtle difference there is that the relationship with its parents is “part of a collection”. I see no reason why works cannot be parts of several collections, depending on how they were published. I’m still looking for a work which is a part of two or more created works.

With some uncertainty:

Beethoven: Str4 13 Bb Op130: Original finale Große Fuge later replaced by a daintier allegro.

Mahler: Symphony 1, in its earlier forms (1889, 1893—4) having its Blumine movement.

Milhaud: Str4 14 5 , ad lib performed simultaneously as octet ; all 3 works sharing opus number 291.
(From thread Levels in the structure of works, @derobert looking “for works that are part-of more than one work”.)

Atterberg: Symphony 2 (originally in 2 movements and as such expressly content; later added a finale III. Allegro con fuoco)

Ben Johnston: Sonata for Microtonal Piano / Grindlemusic

This “Janus-faced work” comes from reördering the same 3 of 4 movements:

Piano Sonata:

  1. “sonata-allegro”
  2. “scherzo”
  3. “slow movement”
  4. “finale”


  1. Premises [“finale”]
  2. Questions [“sonata-allegro”]
  3. Soul Music [“slow movement”]
  4. Mood Music [“scherzo”]

In opera also numerous cases of number revision (Fidelio/Leonore), arias and else going in and out of score, sung recitative replacing originally spoken dialogue (Medea, Carmen), notably differing versions (Don Carlos), …


Excellent! Thanks @Griomo. I knew about Op130 but had forgotten. The others add to my education :smiley:.

In terms of how they play into the current discussion (regarding works with multiple parent works):

  • Große Fuge: This is shown as a stand-alone work, separated from its original parent, which contains the late allegro as well as all the other parts. I think this is correct since it was subsequently published as a separate work (Op.133)
  • As regards the Mahler, all the parts of the different versions are separate records in MB. I am assuming that there are, at least, subtle orchestration differences throughout so that none of the parts are the same as the corresponding parts of previous versions.
  • Milhaud: I’m not familiar with this work, but as far as the MB database structure is concerned, none of the parts are shown as “part of” more than one parent work. The linked thread took me to Work movement reused in another work - Relationship type?. which I realise is almost identical to the question I am now trying to resolve. Unfortunately, it did not get answered. In that post, @alex_s7 asks what approach to use where a composer re-uses one of (their own) work movements:
  • I think there is a fourth option: Place only the new/revised works in the later “container” work. That said, I think (in an ideal world) option 2 would be best, but it would require an additional “(re)used in” attribute. Failing that, I would go for option 3 (with disambiguation to make it clear that the movement is just re-used).
  • Atterburg: Symphony 2 (another new one to me!). Here the original work in MB has no parts - the parts are just movements of the later revision.
  • The Ben Johnston work would be an excellent example (when the movements of the sonata are played in an alternate order, the piece is titled Grindlemusic) . Unfortunately, Grindlemusic is not in MB as a separate work.

So, to sum up, although the specific quoted works are all excellent examples of re-used movements, none of them have been implemented in MB as a part belonging to more than one work. I haven’t investigated all the suggested operas, but I looked at Fidelio and, again, that is implemented in MB by using different work records for the parts of the revisions - i.e. no parts with more than one parent.

As far as I remember, when asking that question, I was editing works of Luigi Boccherini. He was a very prolific composer (nearly 600 works in the catalogue), but in his late years he sometimes took two movements from two earlier works, added a new movement and published all three as a new work. Sometimes he transposed re-used movements into another key, sometimes just keep as they were published earlier. Lazy bastard ! :slight_smile:

1 Like

Can you remember how you handled this - added a new MB work with a related name or linked the movement to two parent works?

After 1 year I do not remember 100%, but I believe I have added new MB works since I got no answer which relationship to use. To really clean up Boccherini, one has to go work-by-work through the catalogue of his works. I was satisfied with more-or-less improving meta-info for a CD box I own, without attempting to attain perfection :slight_smile:

1 Like

I just remembered one slightly different case where I indeed assigned a part to 2 different parents.

Handel’s oratorio “Belshazzar, HWV 61” was originally written in English, and later translated to German. There are recordings of “Belshazzar” in both languages. For a German translation I have created a separate MB work linked to the original work as “translated version of”. For most parts (“numbers”) I have also created separate MB works linked to original works as “translated version of”. But there were a couple of parts without lyrics. For such parts creating a “translated version of” would be wrong: there is no translation, it is exactly the same part. But such instrumental parts belongs to both the original work in English and to the German translation. As a result, the Overture and the Sinfony belong to both parents.

1 Like

Well remembered. Now I need to put my thinking cap on.

As does A Martial Syphony [sic]. The last 2 of those 3 are actually not correctly named (in addition to the typo) - they have been given names which are consistent with the other parts in the German translation, but not with the other parts in the original. If they were to have names consistent with the other parts in both cases, then the works would need to be duplicated. Otherwise, I think the name should be from the original. Alternatively, the names of the translated parts should be exactly as per the original plus the disambiguation, then the identical parts will fit into both schemes.

Ironically, none of those 3 parts cause my Classical Extras plugin any problems as the parent works have the same name, disambiguation aside. However all the other parts do cause problems, for reasons that I have not yet been able to fathom. Thanks for finding a good test case :sweat_smile:.