Cataloguing Classical Music – Part 1

The Cataloguing Game

I want you to imagine that you have just been made Chief Librarian for a public library that has no computers or other electronics. You have to devise a way of making your book collection accessible to people that involves nothing fancier than ye ancient index cards. You are allowed to add one or more cards to your index per book -but if you choose to use more than one card for one book, you have to use more than one card for all books. Similarly, if you decide to use just one index card for one book, you can only use one index card each for all future acquisitions.

Right: those are the rules of the game. Now tell me what sort of index card (or cards) you’d create for this book:

Now, obviously opinions might be divided on the matter, but I think we would generally agree to decide to have an index card filed under ‘Tolkien, J.R.R’. On that card you’d record ‘Two Towers, The’. You might even mention “2nd part of The Lord of the Rings”, since it is, and you might need to know that in the future.

Anyone coming into your library and asking, ‘have you got a copy of ‘The Two Towers’ by Tolkien?’ would then very easily be able to find out that your library does indeed have such a tome available. They’d search through ‘T’, find ‘Tolkien’, then ‘Two Towers’… bingo: correct index card found without difficulty! Such efficiency!

But what if someone came and asked, ‘Do you have the second volume of the trilogy by Tolkien?’. Well: you could answer that question less efficiently, but still reasonably quickly, by finding all index cards labelled ‘Tolkien’ and searching through them one by one. Not knowing the name of the book means you can’t pull out the exact card required first time of asking, but you’d be able to do a swift scan of them all to pick up the pertinent information. So, it would be a possible search, not one ruled out by the way you decided to catalogue the book.

But now what happens if someone asks, ‘There’s a book by someone I can’t remember the name of. It’s got a chapter called ‘The King of the Golden Hall’. Do you have that book, please?’. Well, you’re stuffed at this point, aren’t you? You simply can’t answer that question -not because your library visitor can’t remember such things as author or book title, but because your chosen index system was “one card per book”, and the card didn’t record chapter titles (not unreasonably, since many books don’t have individual chapter titles). Had you catalogued this book with 19 separate index cards, each one headed up with the chapter title, you’d be able to locate the entire volume just by searching for one of them. But you didn’t choose to catalogue things like that -and who could blame you!- so your cataloguing system cannot answer the new question satisfactorily.

The point I’m getting at here is that the way we catalogue things -by author, by book title, by chapter title, by publishing house or by price of purchase- really depends on what questions we expect our users to ask when they come to our information store. Do most people look for books by price or by author, do you think? Hence most library card catalogues are arranged in author alphabetical index, not by price. Do most people come in looking for the plays of William, or the plays by Shakespeare, do you think? Hence most library catalogues will be arranged by author surname in alphabetical order, rather than by author first name!

Generalising from this, we can say that the structure of the catalogue is governed and directed by the questions we expect people to ask of it -and, in turn, the structure of the catalogue we decide on restricts the questions that can usefully be asked of it in the future. Once we’ve decided to catalogue by author surname, we cannot answer questions about ‘where’s the book published by OUP in 1977’ or ‘have you got the one about Leonardo da Vinci and Mary Magdalene’.

The Music Cataloguing Game

Cataloguing music is no different: we need a catalogue because if you own 50,000 different pieces of music, you’ll never find a specific one without some organising principle being applied. But which organising principle to use depends entirely on the questions we expect to ask of our catalogue and the uses to which we will therefore put it.

Take this piece I found on the Internet after about 14 seconds of searching one morning, for example:

Probably the most important tag is the name or title of the music track. This is usually the existing name but it often gets sullied by other information. […] Knowing who produced or made the track can help you find tracks by the same person or group.

Before I say anything else, let me say that the article I’ve ripped those two sentences out is perfectly well-written and I’m criticising neither it nor its author in anything I now go on to write! I’m using it as an indication of how ‘the other side’ does this sort of thing, and learning from it, not dismissing it or rubbishing it in any way.

Notice how, for the author, the track title of a piece of music is his ‘organising principle’ because that’s what he searches for things by. As a DJ, I expect he gets asked by event attendees things like “have you got, ‘Living on my own’?” or “can you play, ‘Under Pressure’ next, please?” Thus, for him, title is ‘king’. So much so that knowing it’s ‘Under Pressure by Queen’ is not that important and thus knowing the name of the group who ‘made’ the track is, for that author at least, secondary in importance.

That article is also interesting because it then goes on to suggest different possible ‘organising principles’ for non-classical music, such as ‘BPM’ or ‘beats per minute’. DJs (I’m not an expert on the subject!) presumably can become quite specialised in the type of music they play and thus need to identify ‘all music with 120bpm’ quickly, so they can play multiple tracks that seamlessly blend into each other without an unexpected ‘bump’ in tempo. He also mentions knowing the key of the music can be important -presumably so that you don’t mix two tracks together whose respective keys are so far ‘off’ as to make the mix sound extremely jarring.

The real thing to get from that article is that the organising principle of your non-classical music can vary, depending on what sort of DJ you are (or want to be), which is another way of saying: it comes down to knowing what questions you’re going to want to ask of your music library.

This brings us neatly on to the point of this particular article! What questions do you think you’d ask of a classical music library?

Let me begin by saying that I don’t know rational, reasonable people that would genuinely ask “what pieces do you have in E flat major, please?”. I mean, classical music people are perhaps more likely than others to ask such a specific question, but I doubt anyone really wakes up and says, “It feels like a B flat minor day today!” So I don’t think a classical music catalogue needs to be organised by key, though it might contain a note of the key of a piece on its index card, just in case! Similarly, I think beats per minute isn’t going to be a question our idealised classical music catalogue needs to answer! I mean, yes, I get that some days you might fancy an up-tempo waltz over a funeral march, but I don’t think anyone writes in to Radio 3 or Classic FM requesting “a piece of music with more than 80 crotchets per minute, please”!

What people might ask for is, “The Kyrie from Verdi’s Requiem”. That seems to me to be a perfectly reasonable thing to want to find in a hurry. So you might think “Organise by the name of the movement”… that way, all the Kyries will be stored together, then the Glorias, then the Sanctuses and so on. Finding any one Kyrie will then be really simple!” Well… you might think that, but I have to say I would respectfully and gently submit at this point, however, that you’ve lost your mind! The question actually asked for Verdi, then his Requiem, then the Kyrie from that Requiem: the correct ‘ordering principle’ here is therefore composer -> composition.

Likewise, most people, I suggest, would ask for ‘the slow movement of Beethoven’s 6th symphony’, not “can I listen to Szene am Bach, please”: and the more likely version of the question again makes Beethoven of primary importance, the overall composition name of secondary importance and the individual movement name being of tertiary importance.

So, when we build our classical music library catalogue, we should organise it by Composer, then by Composition, and within each composition, by movement. Details such as key and tempo might be worth recording if there’s space to do it on the index card, but they won’t affect the way the index card as a whole is placed within the catalogue.

Composer -> Composition is King… So what’s the catch?

Having therefore established (at least to my satisfaction!) that ‘composer-> composition’ is king for classical music, the inevitable exceptions and questions arise. How, for example, do you deal with this CD in your catalogue:

It’s a “Maria Callas” greatest hits CD, consisting of various excerpts from a number of operas by either Puccini or Bellini. How on earth does this fit into the ‘composer -> composition’ organising principle, when we have a mix of composers …and, er… we don’t have any real compositions (in the sense that Puccini wrote Madama Butterfly as a coherent, whole opera; Un bel di vedramo is merely a small piece of a composition, not a composition in its own right, surely)?

Well, I try and avoid buying CDs like this for a reason! Remember, you can’t suddenly change the basis of your index card organisation: if you picked composer->composition for one, it has to be composer->composition for them all. If you had earlier decided to classify ‘Beethoven -> Symphony No. 6’ or ‘Verdi -> Requiem’, you cannot now suddenly decide to catalogue this new CD as (for example) ‘Maria Callas -> Puccini and Bellini Arias’. That would make Maria Callas a composer -and whilst she was many things, she was never one of those!

Take this CD as another example of the problem:

You could almost catalogue that as 18 separate compositions: at least ‘The Bell Carol’ or ‘The Lamb’ are genuinely standalone compositions by their respective composers, for example, in a way that a Bellini aria can’t sensibly be described as a standalone composition by Bellini! So you could catalogue this as ‘William Mathias -> Sir Christèmas’, ‘Hector Berlioz -> The Shepherds’ Farewell’ and so on. That would properly observe the composer -> composition organisational principle that works so well for Verdi’s Requiem or Beethoven’s Symphonies.

Except… 18 index cards for 18 different compositions… when most of those compositions only last for 2 or 3 minutes at a time?! It seems a little excessive!

Surely, too, you must take account of the fact that whilst “The Bell Carol” is a nice piece, albeit a short one, it gets its significance and impact in this case at least by being part of a coherent ‘programme’ of Christmas music. Catalogue it as a standalone piece, metaphorically with its own index card, and you strip it of that context, and thus of much of its meaning.

There aren’t any perfect answers as to how you should catalogue the sorts of CDs I’ve just thrown at you here! As I say, my general principle is to avoid buying this sort of CD altogether if I possibly can… but if I am forced to catalogue them anyway, I would say: stick to composer->composition, but declare the composer to be “Compilation” and say that the composition name is the name of the compilation CD.

In other words, I would catalogue the Maria Callas CD as Compilation -> Maria Callas sings Puccini and Bellini Arias and the Love Came Down at Christmas CD as Compilation -> Carols from Wells Cathedral. In this scheme of things “Compilation” becomes the catch-all equivalent of ‘anonymous’ or ‘miscellaneous’. It certainly occupies the ‘organisation space’ of ‘composer’… but it is so readily distinguishable from a ‘real’ composer’s name that there’s no great danger of confusing roles or function.

Let me finish by throwing another possible exception at you:

That’s my copy of Wagner’s Götterdämmerung. If you know anything of Wagner (or of Solti’s recording of that particular opera), you’ll immediately know that it was a 4-CD set, since there’s about 5 hours of opera to sit through from beginning to end!

So does this get 4 catalogue index cards, one per CD?

Surely not. On the basis that the organising principle is Composer -> Composition, surely this gets indexed simply as Wagner -> Götterdammerüng?

Well, that sounds sensible …except that I think you could at least plausibly argue that the composition name isn’t Götterdammerüng at all, but ‘The Ring’ or possibly Der Ring des Nibelungen. That is, Götterdammerüng is just one part of the Ring Cycle tetralogy, and it’s the Ring as a whole that should get the index card, not the specific parts of it.

I’d buy that argument except for one thing: remember the questions that users would ask of your music index. Would anyone come to you asking, ‘Do you have Wagner’s Ring?’ or are they more likely to ask, ‘Do you have Wagner’s Siegfried?’ and, (much!) later on, come back and ask ‘Do you now have Wagner’s Götterdämmerung?’ My suspicion is that they’d be more likely to ask the latter questions than the former: few people would check out the entire 19 hours of the Ring for a single sitting, after all! They’ll nibble away at it (pun intended!) part by part as time and mood allows.

We can conclude by making two observations at this point, then. Firstly, when Composer-> Composition is your organising principle, you must ignore the number of CDs the music is supplied on: four CDs in this case gets just one metaphorical index card. In a different case, you might find a single CD containing recordings of Beethoven’s symphony numbers 1 and 2.: that would be one CD but two compositions. The media count is therefore irrelevant to determining the number of compositions (and hence the number of metaphotical index cards required to catalogue it properly).

Secondly, you sometimes have to make a call for what counts as a ‘composition’. Mostly, it’s obvious: this symphony, that requiem, those operas. But sometimes, declaring something to be a composition in its own right elevates something to an importance it doesn’t really have (and which would be of no functional relevance anyway, since no-one’s ever going to ask, ‘Have you got a copy of X’ when “X” lasts 38 seconds and is the 15th track on a 38-track CD! If a piece of music loses context or meaning when catalogued as a standalone composition, then don’t regard it as such at all but see it merely as a part of a bigger ‘compilation’.

Some Further Thoughts: What’s a Composer?!

So, at this point, we’ve gotten to the point where we can agree to organise our music library by Composer first, and by the name of the compositions within each Composer -albeit, making an allowance for ‘compilation CDs’ when they happen. But this raises an interesting further question. How, precisely, do we specify the composer?

By that I mean, do we say a piece is by ‘Mozart’, or by ‘Bach’ or by ‘Strauss’? Or would such a cataloguing approach be deficient in some way? Obviously, I have deliberately set that question up to prove the point: that would be a terrible way of specifying the composer, since you wouldn’t know if a piece was by Wolfgang Mozart or his father, Leopold; or by Johann Sebastian Bach or his son Carl Philipp Emanuel; or whether you meant Richard Strauss or Johann Strauss the Younger (or the Elder, come to that!).

No: when you mention a composer, you need the first names, not just the surnames. That’s an absolute given, I think!

Now, in a traditional card index, you’d probably index composers by “surname, firstname(s)”. You would therefore end up with index cards labelled “Bach, Johann Sebastian” and “Bach, Carl Philipp Emanuel”. That’s probably because people have generally gone to a library and asked about ‘Books on Bach’: in that case, it makes sense to keep all the Bachs together, so they can all be found at once, and then filtered at leisure.

You could choose to catalogue your digital classical music in the same way if you like: it wouldn’t be “wrong” as such. However, I personally can’t see any real reason for doing so.

In a music collection, what’s the point of keeping all the Bachs together?! They didn’t all live at the same time; their music is quite different from one another’s. There’s a family relationship, obviously -but even that wouldn’t be true if you filed all the Strausses together, for example (Richard being no relative of Johann). Practically, what does sticking all the Bachs together achieve, apart from making it easy to know that whatever Bach you’re interested in, he’ll be in <points vaguely in the direction of an old-style card index draw—> there, somewhere!

Meanwhile, we have moved into the digital age and every digital contraption out there has this thing called a search engine which is perfectly capable of finding the relevant data, no matter what order you typed it in. For example:

That’s me searching my actual music collection for anything containing the word ‘bach’. Note that it’s found Johann Sebastian Bach, Carl Philipp Emanuel Bach, Johann Michael Bach and Johann Christoph Friedrich Bach… despite me cataloguing them in firstname,surname order. The computer is sufficiently clever, indeed, that it has found a piece by Samuel Barber that is essentially a set of variations on a theme of J.S. Bach’s, as well as something by Mozart (Wolfgang!) that’s a variation of a J.S. Bach original composition. On a computer, it doesn’t matter where you stick your surname, the machine will find it wherever!

I suggest, therefore, that muddling the ‘natural’ word order of a composer’s name is no longer appropriate for a digital music index or catalogue. We poor analogue mortals might have needed encyclopaedias and card indexes to group all Bachs together by surname in the old days, because otherwise we’d never know where to find the entries for whichever obscure member of the family we were interested in on a particular day. But we’ve moved on since then and computers can find information for us even when it’s not the ‘up-front’ piece of information we’re searching for.

Leave composer names as their parents intended them to be, in other words. It’s Johann Sebastian Bach and Wolfgang Amadeus Mozart and Benjamin Britten, not ‘Bach, Johann Sebastian’, ‘Mozart, Wolfgang Amadeus’ or ‘Britten, Benjamin’. The computer (and you!) will cope regardless!

It will cope regardless the other way round too, of course: even if you decided to catalogue in surname, firstname order, the computer would be able to find things written by ‘wolfgang’ but catalogued under ‘Mozart, Wolfgang’ perfectly well! In which case, I can’t compel you to type things in firstname/lastname order, since lastname/firstname would function just as effectively. But I would nevertheless strongly recommend using the firstname/lastname approach simply because, having done so for the past twenty years, I definitely feel that I have come to know my various composers a little better by knowing (and being forced to remember!) their first names. I feel the days of calling people by their surnames should be left to schoolmasters of a certain class and generation

Whilst I’m talking about composer’s ‘proper’ names, can I also recommend not mentioning any titles, honours or post-nominals they might have gained in their lives? That’s to say, stick to cataloguing things as ‘Edward Elgar’, not ‘Sir Edward Elgar’: you want to find him in the E section of your catalogue, not the S. Similarly, it’s Benjamin Britten, not Lord Britten of Aldeburgh: he’s a B, not an L. It is true that Ralph Vaughan Williams was inordinately fond of his doctorate and liked being called Dr. Vaughan Williams by all and sundry… but if you file his music under D, no-one will ever find it again!

I have put together a list of about 450 composer names, each of whom have a place in my own music collection. It’s not intended to provide an exhaustive list of who counts as a ‘proper’ composer. It’s simply there to show you how I catalogue my composers -which names I use, in what order, and what I do and don’t include by way of titles and so on.

…and what’s a composition?

Just as there can be debate as to the best way of specifying a composer’s name, so there are many ways of referring to their compositions.

Do you like listening to BWV 140 or to Wachet auf, ruft uns die Stimme, for example? They are two perfectly accurate ways of describing the same 1731 composition by J. S. Bach after all. Or would you prefer to listen to Mozart’s K.550 or his Symphony No. 40? Again, two different ways of referring to the same thing.

Obviously, not all composers have had proper experts cataloguing their works like the Bach-Werke-Verzeichnis or Köchel did for Bach and Mozart’s works respectively. More commonly, you’ll see people referring to things like Beethoven’s Symphony No. 6, Op. 68 -where “Op.” is short for “Opus” which is, in turn, latin for “[published] work”. His 6th symphony was, in other words, his 68th published composition …except that it probably wasn’t, because composers notoriously get forgetful and error-prone about their publishing history. Benjamin Britten’s last opus number was ’93’, for example: yet the Britten Thematic Catalogue lists over 1184 actual compositions by the man. Opus numbers are usually a poor way of itemising things about any composer, in short!

I should mention, too, that some people would prefer to refer to Beethoven’s Symphony No. 6 in F major, Op. 68 …the inclusion of things like catalogue numbers and keys in a composition’s ‘title’ starts making them rather long and cumbersome, but full of information that some (slightly odd!) people might want to ask of their music index.

What the composition name actually should be recorded as is, therefore, not as easy a matter as it might first appear. Should you include catalogue or opus numbers? Keys? Date of publication? Anything else?!

There’s no definite answer to that. I would suggest that your composition names should be enough to make one recording easily distinguishable from the next, but not overloaded with extraneous data that you won’t find functionally very useful later on. Remember always that the art of cataloguing is to know what questions your users will ask of the catalogue. If you are the kind of person who says, ‘Have you got the piece Bach wrote in 1744’, then maybe including the data in the composition name is going to be important (but given that so few of his works can definitively be attributed to a specific year, I’d argue this is a piece of functionality you would do well to avoid in the first place!)

One important feature of such an approach to naming compositions is that you will need a way of distinguishing multiple recordings of the same ‘work’ by different conductors, orchestras and soloists. I have, for example, 6 different recordings of Beethoven’s fifth symphony. I can tell them apart because the conductor’s surname is included in the composition name at the end and in brackets:

Sometimes, though, even that’s not enough to truly distinguish between different recordings. Adrian Boult, for example, had a habit of recording the same Vaughan Williams Symphony more than once:

As you can see, I get around that difficulty by adding the year of recording within the end brackets. I can thus distinguish between Boult’s 1958 and 1969 recording cycles simply by visual inspection.

With this inclusion of the conductor (or other principle artist) in the composition name, you neatly make it possible to deal with user queries of the form ‘have you got Solti’s Das Rheingold’, ‘have you got Karajan’s recording of Beethoven’s ninth symphony?’ or ‘have you got Callas’ Traviata rather than Sutherland’s?’. People do sometimes want to interrogate a music index by a key performer or conductor: by consistently including the conductor or other primary artist in the composition name, you give your music index the ability to assist with such queries.

And again, choose your ‘distinguishing artist’ based on context. To distinguish between different recordings of the same symphonies, I’d use the conductor’s surname. But to distinguish between two recordings of the same Beethoven violin concerto, I would probably use the violinist’s surname rather than the conductor’s. It all depends on what makes most sensible, functional use to you when you later come to interrogate the catalogue.

Otherwise, you’ll note that I don’t stuff my composition names with opus or catalogue numbers, still less with keys, tempo markings or other information that isn’t pertinent to making a composition name unique and functional for searching. I don’t ever say, “I need to listen to Mozart’s G minor symphony”, so I don’t need the composition name to mention that Symphony No. 40 is in G minor.

Your mileage might vary on that sort of thing: I have, after all, known some classical music connoisseurs to refer to works by their key in casual conversation, though I always thought it was pretty pretentious and merely a way of showing off their immense cleverness, usually with the intentional side-effect of putting you in your ignorant place! But as I say: your mileage might vary.

There are a couple of exceptions to that ‘no technical extras’ rule that I make, however. There are so many Bach cantatas, for example, that I always like to see them listed by their catalogue number:

This has the beneficial side of effect of making all the cantatas begin with the letters ‘BWV’, which means they all get grouped together. You could achieve the same thing by naming them ‘Cantata 022’, ‘Cantata 023’ and so on. I preferred not to do that, though, because seeing the same word (‘cantata’) appear time and time again got me a bit annoyed at the redundancy of it all! If you understand how the BWV catalogue was put together in the first place, you know that all the cantatas got given BWV numbers 1 to around 220ish, so by using a BWV number less than 220 or so you are already, if implicitly, saying ‘it’s a cantata’… in which case, no need to spell that word out in full.

The worst thing you could do, though, is to catalogue them by their first name: Ach wie flüchtig… would get catalogued before Brandenburg Concerto… whilst Du wahrer Gott… would appear afterwards: different types of music by Bach would thus all get jumbled up and you’d never find a thing!

You’ll notice, too, that I pad my BWV numbers out to a consistent three-digit length: it’s BWV 020, not BWV 20, for example. If you don’t do this, computers will tend to sort 1 and 10 together, along with 2 and 20. You’d end up with BWV 1 listed next to BWV 10, or BWV 2 followed by BWV 20 and so on. A ‘natural’ numerical sorting order of composition titles requires the numeric component to be forced to sort properly -and that will require padding. Beethoven obviously saw this problem coming his way: it’s the little-known reason he decided to not write a 10th symphony. If only Shostakovich had been so foresighted! 🙂

So you see that in these cases, my composition names are directly affected by the things I know, the order in which I want to see things, the order in which I want things grouped: your catalogue needs to be organised according to your needs, and specifically according to the questions you will ask of it.

The same trick gets applied to any large-ish collection of similar things. Thus Haydn’s symphonies:

But, still: whatever the subtle variations, you won’t see me (usually!) declaring things to be ‘opus this’ or ‘in C sharp minor’, unless doing so is the only way to distinguish between two otherwise identically-named compositions:

I couldn’t usefully think of a way of telling apart the two Caresana sonatas without mentioning both their opus number and their number within the opus. But that sort of thing is quite rare in my catalogue, which places an emphasis on concision, readability and functionality over and above showing off how much musical minutiae I can squeeze onto my virtual ‘index cards’!

In Conclusion

I’m going to leave it there for now. I hope I’ve shown you that the way you organise your music depends on a number of things:

  • The questions or searches are you going to ask of your catalogue
  • Knowing whether your chosen organizational principle will prevent you performing some searches -and whether any such searches are truly important or merely frivolous. Frivolous edge-cases can be ignored!
  • What information makes one item in your catalogue truly unique from another
  • Whether something is of sufficient, standalone significance to be worth cataloguing on its own or whether it derives its ‘significance’ only from its context
  • The need to be consistent in the way you organise your music, so that you don’t start treating or classifying ‘mere’ Prima Dona singers or Cathedral choirs as if they were composers

Taking these things all together, then, we arrive at the conclusion that our classical music organisational principle will most usefully be, for the vast majority of CDs, the Composer -> Composition principle.

But we also learn from all this that:

  • Sometimes you’ll need to be a bit flexible about what a ‘composer’ is: sometimes it will be ‘Compilation’!
  • You need also to be a bit forgiving about what a ‘composition’ is: whilst short and sweet works best, always including a conductor or significant artist in its name will be needed to make it unique within your collection

Finally, I hope I’ve shown you that:

  • The physical media a piece of music comes on is irrelevant. If a symphony comes on half, 1 or more CDs, it doesn’t matter: it’s still a single symphony
  • Computers make searching for text trivially easy, so there’s no need to enter text in a special, artificial word ordering. Natural word orders make most sense, most of the time

The fact that I called this ‘How to Catalogue Classical Music – Part 1’ should give you a clue that there’s more to be said about the art of classifying and cataloguing classical music… but I think those few points are more than enough to get you started for now. Watch this space for part 2 soon enough!