We all know the problem. You hear a tune, but you can't place it. What's it called? Who wrote it? Who recorded it? It may be so familiar that it drives you half-crazy trying to pin it down. Or it may be completely new to you, but you just love it and want to get the record.

If you are musical enough you might be able to write it down in music notation and show it to friends, or hum, whistle or even play it to them. But it isn't everyone who can do that in a way that makes it recognisable, even if your friends would just happen to know the details of the tune you are looking for (assuming they could recognise it from your tuneless whistling).

Of course, if it was a word and you wanted to find out what it means or how to spell it, you could look it up in a dictionary. Even though a dictionary contains many thousands of entries, you can find what you want because the entries are arranged according to that seemingly illogical jumble we call alphabetical order. Fortunately, the ABC was drummed into most of us in early childhood, and with a bit of luck we can still use it to navigate our way through a dictionary. But a dictionary of tunes?

Written Music

It is quite possible for a body of knowledge to accumulate over centuries without any of it being written down. Folk tales, myths and legends, poetry, plays, dances, historical chronicles, religious beliefs and even codes of law have been handed down the generations by oral tradition. So it can be with tunes, especially folk music and religious chants.

Our present methods of representing sounds on paper go back over 1,000 years, when early music notation schemes were being worked out. Then, the main priority was to record music on paper as a reminder of the general shape of the tune - its ups and downs and general rhythm. Then the system of notation developed to allow for the pitches, note lengths, associated words and so on to be read by anyone who had learned how to do so. The system has been further refined in various ways over the generations, but the musical notation we use today uses basically the same ideas. Parallel lines (and the spaces between them) indicate pitches; different symbols indicate notes of varying lengths; and there are marks to indicate other elements that go into a performance, such as rhythm, tempo, phrasing, dynamics and expression.

Today in the English-speaking countries and elsewhere we use letters of the alphabet (A - G, plus H in some systems) to refer to musical pitches, though some countries, including France and Italy, prefer naming the notes: do (or ut), re, mi etc. The ancient Greeks and also the Romans had alphabetical systems relating to music, too. In fact 'gamut', the old word for a scale, is derived from the Greek letter gamma (the derivation is actually gamma ut).

So the score, the written record of the music, came to ensure that music, once composed, could be kept for posterity as the composer originally wrote it, and read and performed at any time by anyone with sufficient musical ability. But this did nothing to solve the problem of how to identify a particular tune.

Barlow and Morgenstern

In 1948, in an attempt to do for tunes something of what a dictionary does for words, Harold Barlow and Sam Morgenstern published 'A Dictionary of Musical Themes', containing the first notes of more than 10,000 musical themes or tunes, and a newly-developed notation index. Since musical notes cannot easily be indexed whereas letters of the alphabet can, the idea was to convert notes into letters.

But there is a fundamental problem with this. If you hear a tune but don't know which key it is in, you do not know what the notes are. But unless you have 'perfect pitch' - a fairly rare phenomenon - you are unlikely to be able to tell which key a tune is in just by hearing it. To deal with this problem Barlow and Morgenstern first transposed every theme in the dictionary into the same key, the key of C (major or minor, as appropriate). In other words, whichever key a theme may have been composed in, the notes were given as though every theme in the book was in the key of C, this being the major key in which you are likely to encounter the least number of sharps and flats. The dictionary gave the first six to ten notes of each theme listed.

So if you were able to hear a tune and write down the first notes as they would be if the tune was 'in C', you would have a string of letters - for example CCDBCD EEFEDC (for 'God Save The Queen', known in the USA as 'My Country, 'Tis Of Thee'). You could then look up CCDBCD EEFEDC in the alphabetical index and find out the name of the piece and maybe get other information about it too.

Barlow & Morgenstern, as the book is usually called, became quite well known and used among musicologists. But for various reasons it is not always successful (some reviewers reported an early success rate of only about 20%), and of course it is not much use to anyone who does not possess the required skills.

The Parsons Code

In the early 1970s a British scientist, the late Denys Parsons (who also played the cello), started to produce a dictionary of tunes based on a different principle to that of Barlow & Morgenstern. The idea was to index the tunes, not according to the names of their notes on the musical scale, but according to the relationship of each note to the note which immediately precedes it in the tune.

This is very much simpler than it might sound. Parsons' method was simply to write down, for each note of a tune, whether it was higher than the note before, or lower, or the same. Any tune could be represented using combinations of only three letters: D (for Down), U (for Up), and R (for Repeat).

The first note of the tune was represented by an asterisk - you have to start from somewhere - then the rest of the tune would be a combination of D, U and R. So, to take our earlier example, 'God Save The Queen' would appear as

*RUDUU URUDD DUDDU

The spaces after the sixth and eleventh notes are added merely as an aid to readability, and have nothing to do with the rhythm or phrasing of the tune. The size of the musical interval, up or down, is irrelevant and so are the note-lengths and the key. The method is simple and ingenious, and requires no technical knowledge of matters such as key, rhythm, or even the names of the notes. Anyone over the past thousand years might have invented such a method, but as far as is known Denys Parsons was the first to do so.

Parsons initially tested the method on a relatively small sample of tunes. Once he had satisfied himself that it was sound and workable, he wrote letters to various journals and elsewhere, seeking help with the massive task of putting themes into what has now become known as the Parsons Code.

So it was that in the spring of 1975 Parsons' The Directory of Tunes and Musical Themes was published, containing some 13,000 themes. There had been all sorts of problems, such as how to get access to all the scores required, what to include and what to leave out, what happens when more than one tune has the same profile¹, what to do about such things as trills and other musical 'ornaments', and so on.

The book was divided into two sections: 'classical' themes (the major part) and 'popular' themes, which Parsons worked on almost single-handedly. In each section the themes are listed alphabetically, starting with

*DDDDD DDDDD DDDDD

and ending with

*UUUUU UUUUU UUUUU

so you use it like any dictionary or telephone directory. The people who actually worked on the book called it the DRUM, standing for Down, Repeat, Up Method. It was rapidly acknowledged as being a very useful complement to Barlow & Morgenstern, and is sometimes referred to as The Up-Down Book.

Further Research

Parsons' further work had some unexpected results. He began to find certain patterns in the Parsons Code. For example, he discovered that in a sample of more than 7,000 classical themes, the most common initial profile was *UU. In other words, the pattern that occurred most frequently has the second note of the theme higher than the first, and the third higher than the second. The next most frequent was *UD. The pattern least likely to occur was *DR.

Parsons then enlarged his sample to include a further 3,005 classical melodies and 3,763 popular songs. For each sample he ranked the nine possible pitch profiles, for the first three notes, in order of frequency. With the opening note marked with an asterisk,
the order was:

*UU, *UD, *DU, *DD, *RR, *RU, *UR, *RD, *DR

Parsons announced his discovery in the New Scientist journal in 1977, but it was several years before anyone took much notice of it. Further research, however, showed a definite tendency for the second note of a theme to be higher than the first, and while the next note might be Up or Down it was far less likely to be Repeat. These results seem to hold good for works by Handel, Mozart and Beethoven.

Parsons did begin to explore the possibility of using his system to analyse the characteristics of a composer's thematic output. And others, using this system, have begun to explore the similarities and differences between music of different cultures, and between 'art' music, 'folk' music, and 'popular' music.

This research has demonstrated that ancient Greek music, medieval monks and baroque composers all show the same preference for starting their tunes with the *UU pattern. It is believed that this is no accident or coincidence, but that it reveals a fundamental tendency of European music.

Research has also been carried out on traditional songs collected in the early part of the 20th Century from what was left of four North American tribes, and on a collection of songs stemming from the Peyote religion of North America. Evidence suggests that such songs retain the melodic profiles of their tribal ancestors. Analysis shows that the 'European' pattern of *UU is one of the least common in these songs. The preferred pattern is *RR, which is only fifth in the list of European preferences. And songs of Australian Aborigines show an even stronger preference for the *RR pattern.

The full significance of such differences in preferred song patterns is the subject of continuing research. It might even be possible to use such data to help trace, for example, patterns of migration and perhaps of religious or cultural affiliation among peoples, by using Parsons' simple Up-Down-Repeat method to analyse their folk music.

And some people are interested in using this method to analyse bird-song...

Identifying a Tune Online

• Parsons' Directory has long been out of print, but there is now an On-line Parsons Code Search facility.

• Mobile phone owners can now use a service that recognises a snatch of a tune and identifies it from a database of 1.6 million songs, using a technology known as 'hashing'. See
  Mobiles guess that song.