Organized Mapping: Documenting a Complex Musical System

James David Mason

Abstract

Before the rise on industrialized manufacturing in the early nineteenth century, the most complex product of human technology was the pipe organ. Even today, with a proliferation of other industrial products, a large instrument may be comparable in complexity to an airliner. Although the complexity of the pipe organ arises in large part through the replication of comparatively simple structures, how an entire instrument, with thousands of pipes and tens of thousands of moving parts, fits together is difficult to document. Most documentation projects in the past have consisted largely of stoplists, with no attempt to visualize how all the parts become a whole organ. Topic maps, designed to deal with complex relationships, would seem ideal for a project to document these instruments.

Keywords: Topic Maps

James David Mason

James D. Mason, originally trained as a mediaevalist and linguist, has been a writer, systems developer, and manufacturing engineer at U.S. Department of Energy facilities in Oak Ridge since the late 1970s. In 1981, he joined the ISO’s work on standards for document management and interchange. He has chaired ISO/IEC JTC1/SC34, which is responsible for SGML, DSSSL, topic maps, and related standards, since 1985. Dr. Mason has been a frequent writer and speaker on standards and their applications. For his work on SGML, Dr. Mason has received the Gutenberg Award from Printing Industries of America and the Tekkie Award from GCA. Dr. Mason was Chairman of the Knowledge Technologies 2002 conference sponsored by IDEAlliance. He is currently working on information systems to support the classification community at DOE's Y-12 National Security Complex ( Y-12) in Oak Ridge, Tennessee.

Long an enthusiast of the pipe organ and its music, Dr. Mason studied organ design in graduate school and has visited and heard many major organs in Europe and North America. He is a member of the Knoxville Chapter of the American Guild of Organists.

Organized Mapping: Documenting a Complex Musical System

James David Mason

Extreme Markup Languages 2007® (Montréal, Québec)

Copyright © 2007 James David Mason. Reproduced with permission.

What is a pipe organ?

More than two thousand years ago, an engineer in Alexandria named Ctesibius added a playing mechanism and wind supply to a set of panpipes, thereby becoming one of the earliest organ builders. The oldest instrument for which we have physical remains is a Roman one from the third century A.D., found in Budapest. Although the art of organ building was lost in the West, it continued in Byzantium, whence it was reintroduced to northern and western Europe in the latter Middle Ages. By the Renaissance the organ, as it was known down to the Industrial Revolution, was already developing regular patterns and even regional styles of construction. The great age of organ building was the early eighteenth century, when figures like Arp Schnitger and the Silbermann brothers worked in Germany and the Cliquot family flourished in France.

The panpipe, which uses a separate tuned pipe for each note in its gamut, shows the first principle of organs: unlike the player of a stringed instrument like a violin or guitar, the organist cannot change the dimensions of the resonating body to change pitch. A mechanism must thus exist to allow selection from a set of pipes so that individual pitches can be played. An organ, then, uses some sort of keyboard to open valves under pipes, allowing wind to flow into their sound-producing parts. From the Middle Ages on, all but the smallest instruments had more than one set of pipes, and the windchest, the box holding the valves under the pipes, works as an x-y selection matrix. The x axis corresponds to the keys; each can open a valve into a wind channel in the chest, extending under all the pipes, for example, that play middle C. The y axis is the stop action: pipes do not sit directly above the note channels but rather on a sandwich whose middle layer can slide to open or stop the connections between all the pipes of one set and their respective key channels. As organs grew in size (some at the beginning of the Renaissance might have as many as 50 pipes over each keychannel), so did the need for complex mechanisms to control them. Keyboards proliferated for both the hands and the feet, allowing contrasts of sound or allowing one sound to accompany another. And many different types of pipes were developed.

The most essential kind of organ pipe, called a “principal” (also known as a diapason, praestant, or montre) is a cylindrical metal tube with mouth parts at one end not unlike those of a recorder or penny whistle. A Roman organ builder would easily recognize a modern principal pipe. The characteristic sound of the organ is from a chorus of principals whose members reflect the natural harmonic series: unison, octave, octave fifth, superoctave, superoctave third, superoctave fifth, supersuperoctave, etc. Because it creates varieties of sounds from components of the harmonic series, the organ is the oldest form of synthesizer. Besides principals there may be flutes, usually with relatively larger cross sections than principals, but also often assembled into simple choruses, and strings, with narrow bodies, that attempt to emulate the sounds of bowed string instruments. Besides these flue pipes, there may also be reed pipes, whose sound-producing mechanisms look like clarinet mouthpieces made in brass. Reeds come in a great variety of body shapes, corresponding to orchestral instruments like trumpets and oboes but also sometimes taking on truly bizarre shapes to produce exotic sounds.

A set of pipes producing a single kind of sound, running throughout the compass of pitches they produce, is called a rank. A rank associated with a manual keyboard generally has from about 50 to 61 pipes, depending on the historical period of the organ. Pedal ranks typically have from 20 to 32 pipes. Sometimes ranks contain more pipes than the notes that appear on their corresponding keyboards; this enables the ranks to be used at more than their basic pitches.

Because it opens or stops the wind passages under the ranks of pipes, the selection mechanism has come to be called stop controls. In the earliest organs, there was a one-to-one mapping between ranks and stops. But by the late Renaissance, groups of pipes from the upper end of a harmonic series were combined into compound stops such as mixtures; while three to five ranks are most common, mixtures of up to a dozen ranks are known in the largest organs. Until the late nineteenth century, an organ almost always had more ranks than stops. Electrical control systems made it possible to make a rank play at more than one apparent pitch and location on the keyboards, leading to organs with far more stops than ranks (certainly a subject for a topic map).

Organ ranks and stops are also identified with their base pitches. A rank speaking the same pitch as a piano is known as an 8' rank because the longest pipe (speaking two octaves below middle C) is approximately 8 feet long. The length of pipes is halved as the pitch goes up an octave. Accordingly, the suboctave to an 8' rank is known as a 16' rank, the octave as a 4' and the superoctave as 2'. Pipes that speak odd harmonics have fractional lengths, such as 2 2/3' for the octave fifth and 1 3/5' for the superoctave third. The longest pipes ever made are 64' long, speaking two octaves below the pedal unison of 16' and producing a sound at about 8 Hz, which can be felt, if not really heard. (Only two such sets of pipes have been installed, one in Sydney, Australia and the other in Atlantic City.) The top pipes of a typical organ have a speaking body only a fraction of an inch long. A typical mixture based on octaves and fifths for the 8' harmonic series on a manual keyboard might bring on pipes for 4', 2 2/3', 2', 1 1/3', and 1' at middle C, in addition to any of those pitches that might be available from separate ranks and stops.

Although the defining characteristic of a pipe organ is the production of sound by wind blown through pipes, that may not be the only source of sound. J. S. Bach was fond of tuned bells that appeared as a special stop; the theater organ might add to that a marimba or xylophone. Percussion instruments, such as drums and cymbals, have been known for centuries. What the theater-organ fancier calls a "Toy Counter" with bird whistles and car horns, also has its origins in the Baroque period. (Of the four bird whistles in Knoxville, two are on a Wurlitzer of 1926, and two are on church organs built on eighteenth-century models.)

All but the smallest organs are arranged into divisions. Before the nineteenth century, each division has its own keyboard. The largest organs of the time had five manuals and one pedal keyboard. With the development of electrical switching technology late in the century, it became possible to create floating divisions that could be assigned to whatever keyboard was convenient. So the largest playable instrument in the world, the John Wanamaker Grand Court Organ in Philadelphia (http://www.wanamakerorgan.com/), has 9 divisions containing 466 ranks and about 29,000 pipes, playable from six manual keyboards and pedal. Several divisions have no fixed keyboards of their own, and one keyboard is used only to be a home for floating divisions and a few solo stops.

In most styles of organ design before the nineteenth century, divisions played on contrasts of pitch of their principal choruses. Thus a typical organ of three manuals and pedal would have a pedal based on the 16' series, the main manual on the 8' series, and secondary manuals on the 4' and 2' series. A manual/division with emphasis on higher harmonics usually had ranks beginning at 8', but from flute or accompanimental timbres. Later organs from the Romantic period might have divisions to emphasize tonal colors. Thus the Wanamaker has two string divisions, one division dominated by colorful orchestral reeds, one division for soft echo effects, and one division of powerful solo stops, both reed and flue.

Describing Complexity

There is a long tradition of documenting organs through their stoplists. The Organ Historical Society has a large database of stoplists (http://organsociety.bsc.edu/), and there is another in the Osiris Archive at the University of Vienna (http://www.wu-wien.ac.at/earlym-l/organs/local.html). The stoplist for the Wanamaker Organ occupies a small book. But a stoplist is barely scratching the surface for documenting an organ. No mere list of stops will begin to convey the complexity, much less the sound, of the 118 ranks (combined manual and pedal) of the Main String division of the Wanamaker.

What a rank sounds like is the product of many factors. Scaling is the ratio of width to length; thus an 8' string may be only an inch or two in diameter, while a flute of the same pitch may be three times as wide (and not nearly so long because cross section interacts with length in determining pitch). The proportions of the mouth parts of a pipe affect both how it attacks a note and what its steady state sounds like. Materials, whether wood or metal, have a profound affect on sound. (Some of the best metallurgists of the eighteenth and earlier centuries were makers of pipe alloys.) Few published stoplists begin to document these essential properties of pipes, leaving it for the reader to surmise from the names how the pipes are made. That works well for principals and other well-established families of tone. But how is one to know from a name like "Choral Bass 4'" that in one particular organ it is a rank of metal pipes, much less that they are tapered rather than straight?

Organs have long had a kind of internal hyperlink, called a coupler. Since the Renaissance there have been mechanisms that allowed one manual to be played from another or a manual from the pedalboard to supplement its division's native resources. Little more than that could be done before the development in the nineteenth century of mechanical and then electrical supplements to the organist's muscles. Supplementing the organist's fingers with electricity meant that a manual could be coupled to itself at the octave or suboctave, and so couplers proliferated. But there is another kind of hyperlink that appears in various forms as borrowing, duplexing, and unification: applied to an individual rank, these allow one set of pipes to serve more than one keyboard or to appear at various pitches on one or more keyboards.

Several years ago, the creator of the Encyclopedia of Organ Stops (http://www.organstops.org/) attempted to create an XML DTD for describing organs. The most serious limitation of the DTD design, I felt, was that it left too many options, so that it was not clear where a particular feature should be described. Should an accessory be connected to a particular keyboard or to a division? If something affects the whole instrument (such as a blower for wind), how can it be attached to a particular division? What information should be attached to ranks and what to stops when the two are not equivalent?

In short, the author needed a Topic Map. With Topic Maps, the user doesn't have to select one location for a description over another: descriptions can be made self supporting and attached to as many places as they are needed. Since a topic map is a hypertext structure, the concept of being attached to one place in a linear document is irrelevant. Thus an organ specification is transformed from a simple table to a polydimensional artifact that can be navigated in many dimensions.

A Topic Map for an Organ

The basic structure of a pipe organ lends itself well to developing an ontology. After all, an organ has many components, and much of the technology that has been developed over the millennia has concerned itself with allowing some of those components, such as keyboards, to control others, such as pipes. Topics to name the components of an organ fall naturally into associations that represent the organizational and control structures.

The history of the organ can be traced through the proliferation of types of pipes, as control structures allowed the performer to bring more and more ranks under his control, and the growing complexity of the control structures. Down through the middle of the nineteenth century, all control systems, particularly those that connect the keys to the pipe valves, were entirely mechanical and powered solely by the player's muscle power. The size of organs was limited by human strength: the player could stand to perform complex music only up to a certain weight of keyboard force. Furthermore, all the wind to supply the pipes needed to be supplied by the muscles (and weight) of organ blowers. The very largest organs, such as the 1735 Christian Müller instrument at St. Bavo's in Haarlem, had about a hundred and twenty ranks, in that case controlled by three manual keyboards and a pedalboard. There were a few organs with four manuals in Germany, such as Arp Schnitger's masterpiece at St. Jakobi in Hamburg, with about a hundred ranks (http://www.jacobus.de/kirche/default_englisch.htm, http://www.arpschnitger.nl/schnit.html), and even a few with five manuals, like the Henri Cliquot instrument at St. Sulpice in Paris.

The nineteenth century introduced industrial techniques to organ building as well as other businesses. Throughout the century, mechanisms to supplement finger power for connecting keys to valves proliferated and drove their rapid development of large organs with ever-increasing varieties of ranks, paralleling the rapid growth of the tonal palette of the symphony orchestra. Thus Aristide Cavaillé-Coll's expansion of the organ at St. Sulpice in 1862 was the first instrument with over a hundred stops, made possible by the Barker Lever, a pneumatic assist to finger power (http://www.culture.gouv.fr:80/culture/cavaille-coll/en/sommaire_general.html, http://www.stsulpice.com/, http://www.cavaille-coll.com/). (The organ was powered by eight men, treading on sixteen bellows. Cavaillé-Coll's other innovations also included, it has been said, the invention of the circular saw.) Electricity led to even greater complexity of controls. Robert Hope-Jones, an English telephone engineer, pioneered intricate switching systems at the end of the nineteenth century, culminatingin the the Wurlitzer theater organ, where a few ranks of pipes could be reused and reused to supply dozens of stops. But explaining how a "Hope-Jones Unit Orchestra" with more than 80 stops made from only 14 ranks does its work is one of the reasons why traditional specifications like stop lists need to evolve into topic maps.

Topic Types

A topic map to represent information about an organ should begin with the basic elements of an organ, essential concepts such as stop, rank, and pipe. The major organizational structures, such as division and chamber, must be present. Control structures, such as keyboards and the stop mechanism, as well as the elements of the wind supply and, for organs with electric controls, the electrical system, are also needed. The console, which aggregates the manual and pedal keyboards, the stop controls, and perhaps other control structures, must be accommodated. Some organs may have more than one console. The Wanamaker, in addition to the immense main console that is in public view, has many small consoles scattered throughout its chambers, for use by the organ tuners.

The typology for pipes, to document an organ adequately, must allow for more than just the names that appear on the stop knobs on the console. Analysis of organ specifications generally proceeds by recognition of the families of pipe tone, such as principals, flutes, strings, trumpets, and the like, that need typing topics. The pitch series needs topics, not only for analyzing the components of choruses and harmonic series, but for comparing functions across divisions.

Because the timbre of a pipe depends on many factors, those must be recognized. The basic proportions of length and width that are the basis for scaling are essential. Topics are needed for the elements of mouth parts for flue pipes: width of the mouth, proportion of height to width ("cut-up"), width of the windway between the lower lip at the front of the pipe and the languid (the "tongue" that separates the foot of the pipe from its body or resonator), presence of nicking (small cuts on the languid or lip), shape of the upper lip of the pipe mouth. Materials of which pipes are constructed need their own typology, beginning with a division into wood and metal, and the latter may need subdivision into the different alloys used. Reed pipes need their own typology. Not only are there the shapes, dimensions, and materials of the resonating pipe bodies, there are a whole set of shape and dimension characteristics of the sound-generating mechanisms, the reeds and the shallots against which they beat. Pipe construction may also include special features, such as the mitering (folding) of large pipes to fit into cramped spaces.

Association Types

Associations naturally begin with those structural connections that link pipes to their ranks and ranks to stops. Keyboards need to be linked to divisions, and stop controls to the stops in the organ. Divisions need also to be linked to the physical structure of the organ, such as the design of wind chests and the placement of the chests in the case or chambers of the organ. Chests need to be tied to their wind supply and to the control systems that link the valves and stop controls to their controlling structures in the console.

The study of control structures just begins with key and stop actions. Couplers, borrowing, duplexing, and unification open a world that can be adequately documented only through tools like topic maps. And then there is the whole field of combination actions. These systems allow the organist to push a button and change the combination of stops and couplers engaged, either for a division or for the whole organ. Although these systems arose before the introduction of electrical controls, it was telephone switching technology that made them practical. Large English or American consoles might have a hundred or more buttons, called pistons, between the manuals or above the pedalboard. The 1928 mechanical version of the Wanamaker combination action was said to have a memory unit roughly the size of a railroad boxcar. Computer technology has enabled the development of even more flexible controls, so that the new system being installed on the Wanamaker will allow an organist to carry on a USB thumb drive several thousand settings for the hundred or so pistons that affect over 700 stop controls. This needs a topic map, indeed!

The traditional organ specification has for centuries provided a framework for listing the stops in the various divisions of an organ. What it does not do is provide a framework for looking at the relationships among the stops, for turning the organ inside-out and seeing how the components fit together in other ways. The topic map allows comparisons, for example, among the representatives of a family of stops. How does the 8' principal in one division compare to that in another division? How do ranks of a given family but different base pitches compare? What is the relationship of pipe scaling, say, between principals of two feet in length, which speak middle C for an 8' rank, but tenor C for a 4' or the bottom note on the keyboard for a 2'? The traditional specification makes such comparisons cumbersome, but a topic map, being polydimensional, makes it easy to build the associations.

Another issue with associations is arity: how many members does an association need to work. Or, put another way, how many members can it tolerate and still communicate the right information? Some associations seem at first to be natural candidates for binary relations. For example a rank of pipes sits on a chest: this appears to be binary, with one rank, one chest, the rank playing the role of being controlled by the chest and the chest playing the role of the controlling mechanism. But it is not always so simple. Pedal pipes at the low end are frequently too large to sit on chests with pipes of middle-of-the rank sizes. So they are either connected to the chest by tubes or are set on special chests designed for large pipes. So a rank may be associated with multiple chests. (The two largest ranks in Richards, Fowkes &Co., Opus 15 are associated with no fewer than four chests each.) For many ranks, there is a one-to-one mapping of ranks to stops, as would typically be the case for the unison principal in the main division of an organ. But with compound stops like mixtures, there may be many ranks to one stop. Or when stops are duplexed, borrowed, or unified, there may be many stops to one rank. In the theater organ, for example, a single tibia may appear at 16', 8', 4', 2 2/3,' 2', 1 3/5' and 1' pitches: a whole chorus, of sorts, from one rank! Or in many organs the bottom pipes of a 16' rank in a manual division may be shared with a rank in the pedal. So associations may not be simply binary.

Collections across an organ also raise the issue of arity. How should one accumulate all the members of a family of stops, such as principals? Does associating each rank individually with the principal family have the same effect as building a single association with many members? Each way will produce the result of listing all the ranks if one selects the principal family. But looking at the association from the rank end rather than the family end shows something different. When each rank is associated individually with the family, one sees only the family. But when a higher-order association is built, with all the ranks as members of one association, the result is that from any one rank one sees the connection not only to the family but also to the list of all the other ranks in the family. Which association one uses may depend on whether one wishes to emphasize the attributes of the individual rank or the participation of the rank with its kindred in the family. Because I am accumulating properties of ranks, I tend to use the collections of binary associations rather than the large associations, though those are easier to build when a topic map is being assembled by hand rather than by software. Besides, the effect of the large association is still achievable through a query language.

There are some associations that are naturally of higher order. Couplers, I feel, are best dealt with as ternary associations. Thus the association for a coupler has three members: the named control from the console, the keyboard coupled from, and the keyboard coupled to. In the case of octave and suboctave couplers, even higher arity may be needed: not only are there source and target keyboards (which may be the same), there is the pitch relationship of source and target.

Growing complexity

Topic map technology also enables comparisons among instruments. Just looking at the stoplists of two Knoxville organs by Richards, Fowkes & Co. (http://www.richardsfowkes.com/) led me to see similarities between the Great division of their Opus 7 and the Positive of a much larger instrument, their Opus 15. When I mentioned this to Bruce Fowkes, he confirmed that not only were the names the same, the scaling of the flue stops was identical. So what was the main manual division on one organ became a secondary manual division on the larger instrument. I could guess that, and the builder knew that, but the topic map can make it clear.

Associations, then, are needed between ranks and stops, not just on one instrument but across instruments. And associations are needed between ranks and their physical (and thus aural) characteristics.

What is the dividing line between topics and occurrences in a topic map of organs? The traditional specification makes, in effect, most things to be occurrences. Thus all principals are occurrences of a family. But to make relationships visible in the map, each rank needs its own topic. Are the pipes then occurrences in the rank? I think not. To make comparisons such as that among principals of two-foot length from different ranks, one needs each pipe in a rank to have its own topic. But then one also needs each pitch in the gamut and each note on the keyboard to have its individual topic. (Remember that a given pitch can speak from more than one key, and any individual key in the synthesizer can play more than one pitch!)

Dimensional data, such as that associated with scaling, would at first seem to be appropriate for occurrences associated with individual pipes. (In previous topic maps, I have made such things as the weights of manufactured parts into occurrences.) But then Bruce Fowkes asked me whether the topic map could locate all the principals of 150-mm scale. So at least some dimensional landmarks need to be reified so that more comparisons can be made in the map.

And then there is the matter of names. Pipes of the principal family may be called "principal," "diapason," "montre," or "octave," perhaps even in a single organ. Having multiple names for the same kind of rank helps disambiguate them. But what if there are multiple ranks or stops with the same name? Even a relatively small organ of two manuals and pedal might easily have three stops called "Octave 4'", one per division. In larger organs, naming gets more complex: the Wanamaker has no fewer than six stops called "Diapason 8'" in a single division (and four in another); the second diapason is not scaled or voiced like the first, and so produces a different sound. Each rank, and each stop, obviously needs its own topic. Assigning unique names to ranks (as opposed to stops, which are named on their controls), such as "Great Octave 4'" and "Swell Octave 4'" helps some, but goes only part way. When the topic map covers more than one organ, the reduplication of common names, like "Trumpet 8'" becomes a major problem. Scope is probably the best answer, applied in multiple layers. Then it becomes easier to disambiguate the "Gamba 8'" of the Wanamaker Solo from that on Richards, Fowkes Opus 7 Swell, which is built quite differently and so shares with it only the general characteristics of narrow scale and string tone.

Other types of associations can be useful for study of organs. Organs, particularly those that have survived for many years, gather much metadata. Organs pass through the hands of restorers, rebuilders, and recreators. Ranks and even whole divisions are added, removed, reassigned, repurposed, revoiced, and rescaled over the centuries. Adding Barker Levers to the Cliquot organ of St. Sulpice was but one of Cavaillé-Coll's changes. He took the original Positif from its position on the balcony rail, behind the organist's back, and made it the core of the Recit, fifty feet up at the top of the organ! And that was just part of converting a Baroque organ to a Romantic-Symphonic one. Sometimes such processes are reversed, as was the case of the five-manual organ in the Cathedral of St. Cecilia in Albi, where most of the accretions of the nineteenth and early twentieth centuries were ejected, removing electric action to restore mechanical and even shortening the compass of some keyboards that were intended to play only in the treble in their original eighteenth-century form. Backdating in Albi made sense, because the earlier form of the organ was musically more interesting than the later, as well as being better constructed. The organ at the Wenzelskirche (St. Wenceslas) in Naumburg, near Leipzig, was similarly backdated because J. S. Bach was involved in its original design (and he thought so much of it that he got one of his sons-in-law the organist's job there). But backdating St. Sulpice to its Cliquot form would be unthinkable because Cavaillé-Coll's instrument is a major landmark of musical history, associated with many major composers and organists. Nonetheless, the topic map could show how Cavaillé-Coll rethought and reshaped the earlier instrument.

Mapping an Organ

The organ I know best is Richards, Fowkes & Co., Opus 7 (http://www.discoveret.org/wpc/organ.htm, http://www.richardsfowkes.com/pages/opus/07/07a.html). I saw it under construction, helped install its case, and have heard and recorded it on a regular basis. It is largely designed in the style of Arp Schnitger and serves the music of Baroque masters like Bach and Buxtehude well. It also provides excellent support for congregational singing.

Figure 1
[Link to open this graphic in a separate page]

The organ of Westminster Presbyterian Church, Knoxville, Tennessee

Table 1: Richards, Fowkes & Co. Opus 7
Great   Swell   Pedal
1. Quintadena 16   10. Viol d'Gamba 8   20. Subbaß 16
2. Principal 8   11. Viol Celeste (t.c.) 8   21. Octave 8
3. Rohrflöte 8   12. Gedackt 8   22. Gedackt (from No. 20) 8
4. Octave 4   13. Octave 4   23. Octave 4
5. Spitzflöte 4   14. Rohrflöte 4   24. Posaune 16
6. Nazard 3   15. Gemshorn 2   25. Trompet 8
7. Octave 2   16. Quint/Sesquialtera 3/II    
8. Mixture V   17. Scharff IV    
9. Trompet 8   18. Dulcian 16    
        19. Schalmey 8    
Accessories
Couplers Great to Pedal,
  Swell to Pedal
  Swell to Great
Tremulant to whole organ
Vogelgesang
Zymbelstern
 
Mechanical stop and key action
Flexible winding through two wedge bellows
Unequal temperament by Kellner, 1979
Wind pressure: 69 mm

Although this instrument is particularly refined in its execution and the tonal result has been described as "elegant" by the reviewer of a recording, it is in many ways typical of a moderate-sized two-manual organ with mechanical action. The Great has a principal chorus (items 2, 4, 7, and 8), a secondary flute chorus (3, 5, and 6), suboctave support for the choruses (1) that can also serve as a solo stop (played up an octave), and a reed (9) that can serve either as added weight to the principal chorus or a solo stop. The Swell is a smaller echo of the great, an octave higher. It has a small principal chorus based on 4' (13, 17), with suboctave support (12) from a flute that also supports its own chorus (14, 15). The compound stop (16), together with the 8', 4' and 2' flutes, forms what is called a Cornet, frequently used as a synthetic solo reed in Baroque music. For other solo possibilities, the string (10) can be used by itself or with its celeste (11, a similar rank tuned slightly sharp to produce an undulating heterodyne that simulates vibrato), and there are two reeds (18, 19) that can also serve as a chorus with the principals. The foundation tone of the Pedal begins with the 16' flute (20), which provides a base for principals at 8' and 4' (21, 23). The only example of extension in this organ is the 8' flute (22), which is derived from the 16'. The 16' reed (24) is not only the largest rank in the organ, it is designed to support the entire ensemble above it. The 8' reed (25) can supplement the 16' or serve, in French Baroque style, to provide a solo line in the pedal.

Such an organ, built largely in the North German tradition represented by Arp Schnitger, is primarily an ensemble instrument. That is to say, its characteristic sound is of principal choruses, and the choruses are distinguished by contrasts in pitch emphasis. So the Great is based on 8', the Swell on 4', and the Pedal on 16'. Although many of the stops, notably the flutes and reeds, as well as the 8' and 4' principals, serve well for solo tone or in small combinations, such an organ does not depend on the highly individual solo stops, particularly powerful reeds, that characterize French and English organs of the Romantic period. The designers of this instrument balanced the tone of the individual ranks so that almost any stop can be played by itself or in combination with almost any other stops so that ensembles of many sizes and tonal characters are possible. It is the antithesis of an instrument such as a Wurlitzer theater organ in which a few stops are primarily for accompaniment of a great variety of solo stops but there is little attempt an an ensemble built of choruses.

Organ specifications have traditionally been presented as lists of stops, frequently lists in the form of tables such as that above. Learning to read such a table is a basic skill for organ students. The interpretation of the specification above can largely be interpreted from the list, though the actual tonal character of an instrument can be determined only from hearing it.

What, then, does a topic map offer beyond what is provided by a list?

Figure 2
[Link to open this graphic in a separate page]

Opening page for Opus 7 topic map

A map, for a start, allows extracting groupings of stops from the table. Identifying the elements of choruses of principals, flutes, or reeds is not difficult when there are only 25 items in the list. but as specifications get longer, it becomes increasingly difficult to extract particular perspectives from the list. (Even after nearly half a century of reading specifications, I frequently find myself baffled by large German organs of the late nineteenth century.) Because topic maps are nonlinear, they can bring together items from different parts of the list, allowing, for example, comparisons among various choruses from different divisions.

A topic map also allows extracting combinations of information from across the list, such as finding all the 4' stops, all the ranks of principals, or all the orchestral reeds in a large specification. The Wanamaker organ has the bulk of its strings in one immense division, but there are dozens of other strings scattered throughout the rest of the organ; a topic map could easily bring them all together.

Figure 3
[Link to open this graphic in a separate page]

Four-foot stops in Opus 7

Figure 4
[Link to open this graphic in a separate page]

Opus. 7 principals

The relationship between stops and ranks can also be elucidated by a topic map. The Westminster organ has only one duplexed rank, the 16' flute in the Pedal, but it is represented in two stops. The topic map can show clearly, if it provides identities for the 42 individual pipes in this rank, how they are mapped two different ways onto the 30 keys of the pedalboard. Because pedal playing is frequently one note at a time, slow, and only in the bass, it is common to borrow the bottom notes of manual ranks to make pedal stops. Such duplexing as appears in the one case at Westminster is much more common when it can be brought about through electrical switching. The first large concert organ I heard, the Kotzschmar Memorial in the City Hall of Portland, Maine (given by the publisher Cyrus Curtis in honor of his first music teacher, http://www.foko.org/), is an instrument of five manuals and 102 ranks. It appears to have a very substantial Pedal division of 25 stops, yet there are only two partial ranks dedicated to this division, the rest being drawn from the 100 ranks in the manual divisions. There, indeed, is fertile ground for a topic map to explore.

In an organ with direct mechanical connection between keys and pipe valves, there is usually, as at Westminster, a one-to-one mapping between keyboards and divisions of the organ. Thus, the first manual at Westminster is in effect synonymous with the Great division. But larger organs are not so straightforward. Even at Westminster there is not a simple mapping between divisions and windchests. Part of the Pedal is on a chest behind the organ, and the rest is tucked in with the Great on a chest above the console. In Richards, Fowkes' Op. 15, a much larger organ, most of the Pedal is divided between two chests, one on either side of the console, but the lowest pipes of a 16' flute and a 32' reed are located behind the organ on offset chests. At Westminster, the 42 pipes of the Pedal flute are all controlled by the chest behind the organ, but only the middle 18 sit directly on the chest, while the top and bottom octaves are at other locations behind the organ, connected to the chest by tubing. Here, again, the topic map can lead the way from the chests to the pipes.

Figure 5
[Link to open this graphic in a separate page]

The Great division of Opus 7

Figure 6
[Link to open this graphic in a separate page]

The Great chest of Opus 7

When it comes to the pipes themselves, the topic map can make other connections. Most organ pipes are made of alloys of tin and lead, but some are wood (at Westminster, the Pedal flute and part of the 8' flute on Swell). But even in such a relatively small organ, three different alloys are used. The strings and the lower pipes of the Great 8' principal are made of 90% tin; the rest are more lead than tine. Much of the remaining pipework is 28% tin, but the 8' flutes on Great and Swell and the big 16' Pedal reed are mostly lead, with just enough tin for hardness. Here, again, the map can show the route from materials to pipes. (Metal pipes are not always the rule; the 1610 Compenius organ in Frederiksborg palace near Copenhagen, which is similar in size to the Westminster organ, is unusual in that all the pipes are wood.)

Figure 7
[Link to open this graphic in a separate page]

The Great Principal 8'

Pipe shapes can be quite varied. Most are cylinders (the wooden ones are generally rectangular prisms). But some are not. The Spitzflöte, as its name indicates, is conical. But there is nothing about the name Nazard to show that it, too, is tapered. The topic map, however, can map pipes to shapes. The Great and Pedal reeds and the Swell 8' reed have resonators of inverted conical shape but different proportions. When it gets to proportions, the topic map becomes even more useful. To begin with, one needs to be able to pull out from the ranks individual pipes of a given pitch to see how scaling works. At Westminster, the 8' pipe of the Great principal is 148 mm in diameter. The corresponding pipe of the Swell string is only 118 mm in diameter. Both pipes are made of the same alloy, so the quite audible difference in timbre between the two is largely attributable to their proportions. The map can pull the pipes out of their separate positions and show them together to illustrate the relationship. Principals and strings belong to different families, so the difference in proportions is perhaps expected. However, there are subtle differences even within a family, as the map can bring out. Looking through the map not at pipes that play the same note on the keyboard but rather at pipes that play the same pitch, one can see something of how the designers use scaling to shape the overall sound of their instruments. Principals that play the note middle C are approximately 2' long. Selecting from the 8', 4', and 2' principals on the Great, one finds diameters of 48, 47, and 45 mm. So the tone tends to be ever so slightly more like strings as the pitch of the rank goes up. The map shows that the corresponding pipes from the Pedal 8' and 4' ranks are 50 and 49 mm in diameter, slightly more foundational and less string-like than their Great counterparts.

Length is the primary determinant of pitch for flue pipes. However, there are other factors that affect the pitch. The term 8' means that a principal, which is an open pipe, is about 8' long at the bottom of the rank. The same pitch from a pipe closed at the top, like the 8' flute on Swell, needs a pipe only 4' long. And then there is the Swell string. Strings, like principals, are open pipes. There isn't room in the swell box for a pipe 8' tall, so the lower pipes use a subterfuge of having a small closed tube inserted in the middle of a shorter open pipe to produce the lowest harmonic. One of the things the map can show is which pipes have such Haskell tubes. The map can be even more illuminating when it comes to the reeds. Reed pitch is determined more by the length and stiffness of the reed tongue in the pipe motor than by the length of the resonator, which just needs to be the right length for some harmonic of the fundamental pitch. Both 16' reeds at Westminster have fractional-length resonators for the lowest notes. (Westminster also has an older organ with a 16' pedal reed, but its longest resonators are quarter length, only 4' at 16' C.) The map can show the complex breaking of lengths for the pipes.

The information that can be shown in the map doesn't stop with the characteristics of the pipes. Organs, with their hundreds of pipes and the supporting mechanisms, occupy space. The extreme case is the Wanamaker instrument, which occupies parts of five floors of one end of a seven-storey atrium, with some extensions to two other sides. It is so huge inside that the staff have their offices, and part of their shops, with hot and cold running water, tucked in among the pipes. The Westminster organ, much smaller, is nonetheless a freestanding piece of furniture nearly 30' tall (with earthquake braces to the walls of the building). Windchests and pipes are located on three levels. How things are distributed is best shown by a map.

The layout of the pipes is particularly complex because pipes don't simply sit above their keys. After all, the largest pipes are far wider than even the pedal keys, and even for the smallest pipes, the valves and chest structure are wider than manual keys. Furthermore, pipes are not on the chests in keyboard order. For reasons of weight distribution and aesthetics of display, pipes are generally arranged symmetrically, so that the lowest C is on the left, C-sharp on the right, D back on the left, D-sharp on the right, and so on up the gamut. On the Westminster organ, the order is further subdivided to avoid tuning interactions among nearby pipes.

Connecting the keys to the valves in the chests on a mechanical-action organ requires complex linkages. The force from the keystroke is transmitted to the valves through hundreds of feet of thin wood strips, called "trackers", from which the overall mechanism is often called "tracker action." The movement is spread from the width of the keyboards to that of the chests and routed back and forth from side to side through a mechanism called "roller boards". Westminster's organ has four of these, one each for Great and Swell, and two for Pedal because it sits on two chests. Finding a path from a key to a pipe, across the width and height of the instrument, and from the keys at the front to the valves at the back surely needs a map!

Figure 8
[Link to open this graphic in a separate page]

Association type: Division to Keyboard

Figure 9
[Link to open this graphic in a separate page]

Association type: Division to Chest

Location of ranks within an instrument also needs a map to explain. As I indicated above, the 16' flute in the Pedal has pipes on three different levels in the organ, even though all are controlled by only the lower Pedal chest. Figure 7, which shows information collected about the Great Principal 8', also shows another special feature, that part of this rank is shown in the facade of the organ. These are the larger pipes that appear in Fig. 1. The presence of these pipes in a position of high visibility is another reason why they are made of a high-tin alloy, as indicated above: the alloy takes a high polish.

Complexity upon Complexity

Although building a topic map of an individual instrument can elucidate its specification in ways not readily evident in the traditional presentation, it is in comparing instruments that the mapping technique becomes most valuable. As I mentioned above, the Great division of Opus 7 is the starting point for the Positive in the same builders' Opus 15 (http://www.richardsfowkes.com/pages/opus/15/15b.html). But that realization is only the beginning of understanding the relationships that could be exposed through a topic map. The map would show that, for example, the Nazard on the Opus 7 Great swapped places with the Sesquialtera on the Swell, so that Opus 15 Positive has the Sesquialtera and its Swell has the Nazard and gains a Tierce (the Sesquialtera is equivalent to the combining of these two stops). Opus 15 has a new Great, based on a Principal 16'. The design of this new division reflects the need to produce a sound bold enough to fill a large auditorium with rather dead acoustics. Those acoustics also call for a 32' reed in the Opus 15 Pedal.

But that is not the end of what a topic map of the two organs would show. Despite the differences in their specifications, the two Great divisions have a functional similarity, to provide the main chorus of principals plus ensemble reeds that establishes the core of the organ's tone. The "works principle" that was established in North Germany and the Low Countries by the seventeenth century still applies to both organs: Their divisions are distinguished by both physical position in the organ and pitch emphasis. So the Opus 7 Great is just above the console and the Swell above it as an Oberwerk. The Opus 15 Great is likewise above the console (except for the longest pipes, which are at floor level) with the Positive in Oberwerk position. The topic map would thus show a web of relationships in which the Opus 7 Great is sometimes the counterpart of the Opus 15 Great and sometimes of the Opus 15 Positive. (Sometimes such relationships appear within one instrument: at St. Sulpice, Cliquot's originial Positif, mounted on the gallery rail, was moved toOberwerk position by Cavaillé-Coll to become the core of the Recit, or Swell.)

Similarities between two organs by the same builders are perhaps not unexpected. Richards, Fowkes built almost the same Pedal for their Opus 12, in New Brunswick, New Jersey, as they did for Opus 7. Ernest M. Skinner, the great American organbuilder of the early twentieth century, had certain favorite stops, such as the Erzähler (a narrow-scale tapered flute of light but singularly bright tone) that could be mapped across many of his instruments, from the small to the immense. Similarities can be traced across builders. Richards, Fowkes has made extensive studies of Schnitger's instruments, and it would be possible to map their scaling tables, as well as their metallurgy, onto his. (The Opus 7 Vogelgesang, for example, is a near copy of one Schnitger made at Norden.) Functional mappings work well across instruments of all sizes, styles, and builders. The smaller organ at Westminster, built nearly forty years ago by Alfred Lunsford, has a Great, based on 4', whose stops are near counterparts of the Opus 7 Swell.

Functional mapping can lead into many places. One I would like to pursue would be mapping the ensembles of Ernest Skinner's instruments to those of English builders he knew, such as Henry Willis. Most students of organ building have a general sense of what that would show, but it would be interesting to see what details would turn up in the execution. Functional mapping can extend even further afield. In the theater-organ tradition, a chorus of principals is rarely well developed. Instead, the ecological niches of the harmonic series are occupied through unification of tibias, which are large-scaled wooden flutes. The sound of a slim, high-tin Schnitger principal has little in common with that of a fat wooden Wurlitzer tibia, but the ensemble roles can nonetheless be mapped.

I have mentioned that organs accumulate history. Kal Ahmed showed several years ago a pattern for mapping historical events with his topic map of Samuel Pepys' Diary (http://www.techquila.com/blog/archives/000037.html). The same techniques could be applied to organs. The evolution (and occasional devolution) of instruments through the centuries often reflects trends in musical style and taste, such as the adding of swell boxes to organs from earlier centuries in the nineteenth century. Understanding such patterns is sometimes critical to reconstructions such as Jürgen Ahrend's rebuilding of Schnitger's instrument at St. Jakobi in Hamburg. There war had destroyed many important portions of the instrument, from mechanical components to the large 32' principals that framed the case. In rebuilding the Hildebrandt organ of the Wenzelskirche, Herman Eule saved even worm-eaten topboards from windchests: the boards would no longer support pipes, but it was important to preserve their information content. The boards clearly showed how later builders, such as Walcker and Sauer, had modified Hildebrandt's original work. One might not initially think of boards having information content, but they clearly do for organ restorers. When pipes have been lost, as in Hamburg, having even parts of the supporting structure allows reconstruction of pipe spacing, which leads to pipe diameters and, since the necessary lengths can be calculated, thus to scaling.

One of the great areas of contention in the history of technology is whether it is better to preserve an artifact as a static monument or as a working device. John Harrison, the inventor of the marine chronometer that revolutionized navigation in the eighteenth century, left behind several of his clocks. Some are running to this day, but the great watch that confirmed his achievement sits silent in the Maritime Museum at Greenwich because putting it back in operating condition would require lubricating it, and that might damage some of its components. When the Smithsonian restored the John Bull, their oldest intact locomotive, to run for its 150th birthday celebrations, even rusty bolts that were no longer safe to use were preserved when replaced by modern stainless steel. An organ is such a large and expensive device that one can hardly just let it sit around silent. Sometimes, only parts of a case or a few pipes are left of an ancient organ, so that all that can be done with it is place the remains in a museum, like the late-Renaissance instrument hanging high on a wall over a stairway in the Rijksmuseum in Amsterdam. If it is in a church, it needs to support worship. But restoring an organ may confuse, if not destroy, parts of its history. Taylor and Boody Organbuilders have documented the care with which they reconstructed the David Tannenburg organ of 1800, now in a museum in Winston-Salem (http://www.taylorandboody.com/opuses/opus_40.htm). In some cases, they had to cut strangely shaped pieces of wood to create a functional component while still retaning as much of Tannenburg's work as possible (http://www.taylorandboody.com/opuses/opus_40_restoration_pages/dovetail_repair1.htm). Flattened pipes needed to be returned to playing shape (http://www.taylorandboody.com/opuses/opus_40_restoration_pages/flat_facade2.htm). A topic map is obviously not going to restore the wood that works have eaten or the pipes that have been melted down into bullets, but it can help bring order into understanding the historical process.

Conclusion

Building a topic map of pipe organs is for me very much an experiment. I have been thinking about it for several years, but thinking about it is different from executing it. This topic map is still very much a work in progress. At the point when final papers had to be submitted, I had not yet extracted scaling data for Opus 7 from the builders' spreadsheets. In XTM notation, which is notoriously verbose, the map is already something over 15,000 lines, all hand created, at this time. In that state, it covers two Richards, Fowkes organs from the perspective of both ranks and stops, without beginning to cover the complexity of control structures, such as tracing the action from keys to pipes.

Nonetheless, in its rudimentary state, the map shows many things not possible with traditional documentation, such as the derivation of one organ from another. But even applied to a single instrument, it allows shifting of perspective, so that one can look at pipe families rather than divisions or pull out the stops available at a given pitch level. For me, one of the best features of topic maps, in addition their ability to link clumps of information, is their ability to keep absorbing more and more information. In that way, they are like the Wanamaker organ, still growing after more than a century. I shall keep adding to this map, and who knows what new combinations I shall find!


Acknowledgments

I am grateful to Ralph Richards and Bruce Fowkes, of Richards, Fowkes & Co., of Ooltewah, Tennessee, for sharing design details of their Opus 7 for use in this paper. John Brock, Professor of Organ at the University of Tennessee and a leader in their acquiring Opus 15, was my principal teacher in organ design. Peter Van Eenam, the organist at Westminster Presbyterian, is a trusted adviser on matters organistic, as well as a close friend to my wife and me.



Organized Mapping: Documenting a Complex Musical System

James David Mason