Aspects of Sound Structure in Historic Organs of Europe

Abstract

This article sketches the development of pipe organs in Europe from Roman times to the Baroque era in order to shed light on concepts of sound structure that apparently guided the design and construction of certain types of organs. For a better understanding of empirical measurements presented in Parts 4 and 5 of this study, Part 2 provides some basic organology. In Part 3, the development from the so-called ‘Blockwerk’ to organs comprising several wind chests, manuals, and various types of pipes and stops is outlined. In part 5, relations of sound structure to tunings and temperaments are discussed, including actual measurements.

1 Introduction

One of the most ancient and widespread musical instruments in Europe is the organ, as it is used in churches and concert halls as well as in private settings. A few years ago, UNESCO included the organ and organ music in its list of world cultural heritage. The historical record for organs in Europe from the Middle Ages to the Renaissance and Baroque era and also in modern times is very broad and has been studied in great detail, with a focus on archival source material as well as the development of organ building and organ music. Thereby, a number of historical and regional organ types have been identified, and major stages in the development of pipe organs and organ music have been outlined (for comprehensive surveys, see Williams 1966, Klotz 1975, William & Owens 1984, Eberlein 2011). Works on organ building provide information on technical aspects such as the mensuration of pipes and their peculiar geometry in relation to sound generation and the organisation of various pipe ranks within the overall structure of organs (e.g., Adelung 1976). Pipe organs are known for their complex mechanical construction as well as for the amazing variety of ‘sound colours’ they can produce from pipe ranks of different designs and manufacture. Beginning in the 1930s, characteristics of sound recorded from historical organs have been investigated (e.g., Trendelenburg, Thienhaus & Franz 1936, 1938, Lottermoser 1940, 1983a/b). However, early recordings made for documentation and empirical research on sound properties included only a small number of extant instruments while organs built in the 17^th and 18^th centuries, respectively, have been used quite frequently for recordings of music (mainly of the Renaissance and Baroque era) within the context of historical performance practice, which involves ‘original instruments’ of a given period. A milestone, in this respect, was the recordings the organist Helmut Walcha made of Johann Sebastian Bach’s complete works for organ, where he played instruments of the famous organ builders Friedrich Stellwagen, Arp Schnitger, and Andreas Silbermann. These recordings began in 1947 at Lübeck, where the organ of Friedrich Stellwagen from 1637 was reinstalled in the St. Jakobi church after WW II (see Wölfel 1980, 53ff.). Walcha’s recordings of Bach were issued by the ‘Archiv Produktion’, a special label of the Deutsche Grammophon devoted to ‘scientifically documented’ recordings of works from various periods of music history. A re-issue of Walcha’s recordings of Bach’s works for organ appeared in 2000 (Helmut Walcha: J.S. Bach. The Organ works, 12 CDs. Archiv Produktion 2000). Such recordings honoured a number of historical organs in Europe that were esteemed for their excellent craftsmanship and sound quality, which saved them from destruction and removal. Still, quite many historical organs from the Baroque era were badly damaged, to the point of losing original wind chests, part of the mechanical action as well as a significant number of pipe ranks, in the second half of the 19^th and the first decades of the 20^th century; from various instruments only the organ cases remained (into which a new organ manufactured around the period 1850–1910 was inserted). The reason to abandon old organs was not so much their need for maintenance or repair but a change of concepts when organ builders and organists alike opted for a more ‘modern’ sound, whereby pipe stops in organs were often designed to emulate orchestral timbres (in particular, strings). Also, devices suited to vary the dynamics of sound were installed in many new organs. This included the ‘swell case,’ which can be opened or closed in quasi-continuous motion so as to decrease or increase the SPL [dB] of sound radiated into the ambience. Another example is the so-called ‘crescendo roller,’ a device suited to activate stops in successive order, thereby increasing SPL and spectral density stepwise. Against such ‘progressive’ inventions, organs from the Baroque era were often viewed as old-fashioned and unworthy of proper maintenance. It was more or less by chance (and due to the fact that not all villages and small towns could afford to order new instruments) that some of the most outstanding organs, as, for instance, the Schnitger organ of Cappel (1679–80; originally built for the St. Johannis monastery church of Hamburg; see Fock 1974, 33f., Vogel, Lade & Keweloh 1997, Edskes & Vogel 2009) survived nearly untouched.

In the 1920s, a group of organ builders, organ experts, musicians, and musicologists unsatisfied with industrial organ manufacture, pneumatic instead of mechanical action, and contemptuous of the gadgetry in contemporary organs (such as “high pressure” stops), discussed the merits of ‘classical’ organs (the period from ca. 1600–1770) and proposed a program calling for the restoration of historic organs, possibly to their original state. This movement, which in Germany is known as ‘historische Orgelbewegung’ (historic organ revival, with similar organisations in the Netherlands, France, Italy, and other countries), had practical relevance in that, in the first stage, a survey of surviving organs and their present state was initiated. From inspection of individual instruments as well as from archival studies, their history with all the previous repairs and modifications became evident. Notwithstanding regrettable losses suffered over two or three centuries, there was still a substantial mass of original parts (organ cases, wind chests, more or less complete pipe ranks, ducts, bellows, keyboards, etc.) in place, in various organs, to allow for restoration and/or reconstruction. Working from a comparative basis (one part missing in a certain organ, fortunately, was preserved in another of the same master or one of his contemporaries and could serve as a model for reconstruction), the process of restoration, especially since ca. 1950, has been a continuous and very effective one. Of course, there were severe problems as organ builders had to understand the manufacture of pipes based on pre-industrial techniques of casting organ alloys (from tin, lead, copper), as well as how intonation of flue and reed pipes was facilitated hundreds of years ago. Along with the reconstruction of original pipe ranks and revoicing of pipes, in many historic organs, the tuning has been changed from equal temperament (ET12) back to one of the systems used around 1700 (like meantone or Werckmeister, see below). Though there is no doubt that all the work invested into the restoration of historic organs has brought back an impressive range of instruments diverse in design and ‘sound colours,’ the question remains of how close the actual sound might come to sound properties an instrument had when it was first installed centuries ago.

What can be said, with some confidence, is that a restoration (of a painting or other work of art as well as of a monument etc.) is always an attempt at finding a convincing solution on the basis of available evidence. The goal is to preserve as much as possible of the original substance and to reconstruct what is missing using appropriate materials and techniques. In regard to organs, the corpus of data gained from restoration projects over several decades is extensive, and so is the experience of organ builders who specialise in restoration. Since their knowledge and craftmanship have been proven in the technical reconstruction of wind chests and actions, re-adjustment of wind supply and revoicing of pipes etc., their efforts likely revived also sound properties as assumed for the original instrument. Of course, this is a process of approximation since almost all historic organs have undergone modifications, usually in the period ca. 1780–1870, which changed their pitch and tuning. The original pitch, referring to some practical usance like the ‘Chorton’, was generally considerably higher than our standard a¹ = 440 Hz. The tuning system applied to organs for more than 200 years had been one of the variants of meantone temperament (see Lindley 1987, Ratte 1991, Schneider & Beurmann 2017). When equal temperament (ET12) came into use ca. 1780–1820, most organs then were tuned to this system (especially in cities wealthy enough to afford such a process that often included an exchange of pipe ranks since certain old stops were not compatible with ET12, see below). Retuning to ET12 and lowering the pitch level were often combined. All these modifications of the past had to be corrected in a process of careful restoration so that a state close to the original construction and voicing was achieved. The expectation of such proceedings is that also the sound of each pipe rank, and of the organ as a complex unit, will come close to what must have been the ‘original sound’ (‘Originalklang’) of a Renaissance or Baroque organ. Since we have no recordings from 1600 or 1700, attempts at finding the ‘original sound’ are demanding and may, to some degree, remain conjectural (that is, they are based on factual evidence yet include inferences). The task to approximate the ‘original sound’ is by no means restricted to historic organs but exists, in similar ways, for almost all instruments from past centuries. For example, violins and other string instruments of famous makers such as Stradivari, Guarneri, or Stainer were not left untouched over several centuries but have been subjected to repair and re-adjustment, including replacement of strings as well as of bridges and even necks. For appropriate repair of historical violins, specialists had to study in depth the principles of design of those masters and had to become familiar with the materials (wood, glue, lacquer) they had used. Thus, it was possible to restore such instruments in detail, thereby regaining superior playability and excellent sound quality, which, as a huge number of recordings made with historic instruments amply demonstrates, cannot be too far from the original. From all the evidence available, we may conclude that, by skilful and well-informed restoration, a close approximation to the ‘original sound’ of a historic instrument such as a violin, bass viol, flute, oboe, harpsichord, or organ is possible. In this respect, it seems justified to regard sounds recorded from a Baroque organ fully restored to its original state as authentic. Though we cannot relive the past, we can revive its instruments and concepts of sound.

Research directed to characteristics of the sound properties of historic organs gained new momentum when digital recording and signal processing tools became available on a greater scale in the 1990s. As data for such research, it is mandatory to record sounds from each pipe rank on site since the voicing and intonation of pipes receive a final adjustment in the room into which an organ is placed. Also, it is important to record instruments before and after restoration in order to document the previous sound characteristics and to assess the changes that result from the restoration process (cf. Schneider et al. 2006, Ahrens, Braasch & Schmidt 2006). Sound recorded from pipes mounted on their wind chest can be subjected to signal analysis whereby temporal and spectral features suited to describe sound generation in pipes and timbral quality of peculiar pipe ranks can be studied and documented objectively (see Beurmann, Schneider & Lauer 1998, Schneider, von Busch & Schmidt 2001). In this chapter, we continue and expand previous research, which includes actual organ sound as produced with combinations of pipe stops viewed in relation to tuning and temperament.

In the following section, I shall first address some basics of organology, including terminology, as certain concepts and terms will be needed, in Sect. 4, in conjunction with sound analyses of pipe ranks. In Sect. 3, the development and history of some organ types are briefly reviewed since organs of the Baroque era found in parts of Northern Germany and adjacent regions of the Netherlands, preserved certain features known from older types of organs. As in many cultural phenomena, one can observe the interplay of continuity and change also in organ building.

2 Some Basic Organology

A pipe organ is a wind instrument (aerophone) that consists of a system supplying wind to a chest on which one or several ranks of pipes are mounted (for technical aspects, see Adelung 1976 and Williams & Owen 1984). Pipes are distinguished by their mode of operation into flue and reed pipes. In a pipe organ, pressing a certain key on the keyboard will open a valve whereby air streams from the wind chest into a flue pipe or reed pipe, where the airflow will activate a pulse generator coupled to a resonator. Regular sequences of pulses from the generator elicit periodic vibrations in the air column enclosed in each pipe, which acts as the resonator part of the coupled system. Standing waves will be formed in a cylindrical or conical tube of a given length l if the resonance condition ω_e = ω_r is met (ω_e = exciting frequency, ω_r = resonance frequency, for ω = 2πf). Standing waves and resonance, in turn, is the condition necessary for the production of harmonic sound that is radiated from the open end of a tube (e.g., a diapason pipe). In aerophones such as flutes and reed instruments, the generator can be described as a nonlinear oscillator, whereas the tube resonator reacts to excitation in a linear response (within certain operation limits).

The oscillator/generator typically interrupts a continuous stream of air by a valve-like mechanism which, in reeds and horns, opens and shuts in a basically periodic motion controlled by, first of all, the pressure and the speed of the air fed into the oscillator. A valve-operated pulse generator can be formed, in real instruments, for instance, by the two lips of a musician pressed into a mouthpiece (as in trumpets and horns). In wind instruments, a single reed (as in a shawm or clarinet) and double reeds beating against each other (as in the oboe and bassoon) can serve as a valve. Instead of a valve, an edge-tone generator can produce a pulse train in a complex cyclic process (of 360°, see Meyer & Bork 1987, 20ff.) controlled by velocity and pressure parameters (cf. Fletcher & Rossing 1991, ch. 16).

In a block-and-duct flute, air passing the duct forms a laminar jet which streams against the edge opposite the duct where the jet bends inwardly into the pipe and outwardly away from the pipe while forming vortices and, consequently, eddies. The periodic change of direction the jet undergoes is brought about by the interplay of velocity and pressure differences. The pulses transmitted to the air column inside the resonator excite vibrations which result in standing waves when resonance is achieved. Since air molecules inside a tube do not undergo shear stress, only longitudinal motion is observed. Eigenmodes and resonance frequencies in an ideal tube open at both ends are in harmonic ratio. In regard to modes of vibration, there are nodes (minima) and antinodes (maxima) for the displacement and pressure amplitude, respectively; in an ideal tube open at both ends, the (alternating) pressure p_~ at each open end must be minimum while displacement x and velocity v of particles must be maximum. Hence, the open end viewed as a boundary condition (see Kalähne 1913, 76ff.) has a pressure node and a displacement antinode while pressure reaches a maximum at l/2 for the first mode, and displacement has a node there. The modes of vibration in the air column inside the open tube correspond to frequencies whose ratio is harmonic, that is f_n = nf₁ (n = natural number 1, 2, 3,…), where f₁ = $\frac{c}{2l}$, with l = length of the tube, and c = speed of sound in air (~340 m/s at 15 ℃, sea level).

A standing wave fits into a tube if its length l equals 1/2 of the wavelength, λ, or an integer multiple of λ/2, thus: l = n $\frac{\lambda }{2}$ and λ = $\frac{2l}{n}$ with λ = $\frac{c}{f}$.

Since only half of such a standing wave fits into a tube of length l, it is a λ/2-resonator. For a cylindrical tube closed at one end, it must have a displacement node and a pressure maximum at the rigid wall, while the open end has a pressure node and a displacement antinode. The distance between node and antinode, in this case, is l; a standing wave in the open tube closed at one end requires that l must be 1/4 of the wavelength λ or an odd multiple of λ/4. Hence l = $\frac{\left(2n+1\right)\lambda }{4}$ for n = 0, 1, 2, 3, … and λ = $\frac{4l}{\left(2n+1\right)}$ where λ₁ = 4l; for the lowest mode of vibration in a λ/4-resonator, the corresponding fundamental frequency is f₁ = $\frac{c}{4l}$ and frequencies of the next higher modes that have a pressure antinode and a displacement node at the closed end are 3f₁, 5f₁, 7f₁ etc. Thus, the cylindrical tube closed at one end yields only odd harmonics. In principle, this is also the case with organ flue pipes closed (stopped) at one end. However, in a real vibrating system, one has to take more parameters into account, such as particle velocity (v) and input impedance (Z_in), as well as energy losses (D) due to friction and damping (cf. Gough 2014, 635ff.). Though the input impedance to a tube filled with air is small, a certain force from wind pressure is needed to overcome the resistance.

Measurements of impedance in a cylindrical tube as well as in wind instruments yield curves where the maxima approximate harmonic ratios. In organ flue pipes, the actual resonances can deviate somewhat from the exact harmonic ratio so that higher partials increase in their frequency above the f_n = nf₁ ratio. A more general factor is that the pressure nodes of an air column vibrating in a tube do not match its end plane but lie somewhat outside (otherwise, sound radiation, which requires energy transport into open space, would be impossible). Thus, the effective length of a tube is l + Δl, where Δl is the end-correction which relates the distance a of the pressure node from the end of the tube to its radius (r) or diameter (d). The term a depends on frequency (cf. Meyer 1960) and diminishes with rising frequency; since the effective length of the air column under vibration decreases with frequency, actual resonance frequencies rise accordingly. The quantity a can be calculated (cf. Kalähne 1913, T. II, 221) like a = $\frac{\pi r}{4}$  = 0.7854 r, where r is the radius of the tube’s opening. Average values given for the end-correction Δl of open pipes usually are between 0.6r and 0.85r.

In addition to the end correction Δl, another term Δm is required to account for the fact that pressure is not zero at the labium (l = 0) but has a positive value (cf. Mühle 1966/1979 for measurements taken from the block-and-duct flute which is comparable to a small flue pipe; see also Fletcher & Rossing 1991, ch. 17.3). Thus, Δm increases the effective length of the air column in the direction of the labium. Slight deviations of resonance frequencies in a tube from harmonic ratios are also caused by friction of particles on the walls of the tube. Friction effects are more marked in tubes or pipes of small diameter (where the wall plane is relatively large as compared to the dimensions of the air column). Further, it should be noted that flue pipes and the resonators of reed pipes develop structural vibrations of their walls (their geometry corresponding to a cylinder, a cone, or to a compound of elements). Wall vibration is relevant in long pipes with a thin wall, as in pipes made of an alloy with a high percentage of tin. Though these effects are measurable (cf. Runnemalm, Zipser & Franke 1999), structural vibrations are very small in amplitude (unless structural eigenmodes and modes of the air column coincide in frequency so that resonance is achieved).

A pipe organ consists of at least one wind chest equipped with one rank of pipes (for details, see Adelung 1976 and Williams & Owen 1984). Most instruments, however, comprise several such ranks of flue and reed pipes, and many organs installed in churches or concert halls combine several so-called ‘works’ (German: Werke) or ‘divisions,’ which can be viewed as separate units or even as separate organs. A ‘work’ generally has its own wind supply (in the era under consideration here, based on bellows), which feeds one or several wind chests via ducts. On each wind chest, there are several rows of pipes representing various organ stops which differ by the type of sound generation (flue and reed pipes; certain pipe models make use of overblowing into the 2^nd or 3^rd harmonic, see Mahrenholz 1942/1968) as well as by their size and geometry. Figure 1 shows a range of different flue and reed pipes standing on their wind chest from the Oberwerk (OW, upper organ) of the large organ built by Arp Schnitger 1689–93 for the St. Jacobi church of Hamburg. The picture was taken after the restoration of the organ in 1993/94, see Ahrend (1995).

Fig. 1.

Pipe ranks of the OW of the Schnitger organ at St. Jacobi, Hamburg. The pipe ranks on the wind chest are (from left to right): Prinzipal 8’, Rohrflöht 8’, Holtzflöht 8’, Spitzflöht 4’, Octava 4’, Nasat 3’, Octava 2’, Gemshorn 2’, Scharf IV-VI, Cimbel III, Trommet 8’, Vox humana 8’, Trommet 4’ (from Ahrend 1995).

From the picture, it should be clear that pipes vary in regard to their length, diameter and shape. Flue pipes can be cylindrical or conical or may show a combination of cylindrical and conical segments in the resonator. Also, flue pipes can be open at the upper end or stopped, being either completely closed (a λ/4-resonator, meaning the pitch is about one octave below that of an open pipe of equal length) or partly closed as in the Rohrflöte shown in the picture, where a tube of small diameter is inserted into a disc on top of the pipe. The disc, in turn, is part of a ‘hat’ which covers the top end of each pipe. The hat is air-tight and can be moved up and down, which changes the effective length of the pipe and, thus, its pitch. A quite simple construction for stopped pipes usually was to close the upper end by soldering a plate on top; for fine-tuning, there may be holes drilled into such a top plate. Wooden flue pipes usually have a quadratic cross-section (as does the ‘Holtzflöht’ in the picture). In regard to the pitch, timbre, and intensity of the sound emitted from flue pipes, there are several relevant parameters related, most of all, to the geometry of the mouth, which incorporates the lower lip and the upper lip, whose edge acts as a pulse generator. In the process of voicing, an organ builder may make minute changes to the width of the windway (the flue), which alters the thickness of the laminar jet passing through the flue, and at the same time, alters the pressure and the speed of the jet. Among the variable parameters are also the height of the ‘cut-up’ (German: Aufschnitt) between the lower and upper lip as well as the width of the mouth (for technical details, see graphics in Adelung 1976, Williams & Owen 1984). The result of voicing should be a stable pitch and a harmonic timbre at a sound level as desired. Since the generator is coupled to a resonator, the actual system behaviour also depends on the geometry of the resonator, and in particular, on the relation of the effective length to the diameter of the pipe. With conical or double-conical pipes (e.g., Spitzflöte, see Fig. 1) or with even more complex shapes, numerical calculation of pitch can be quite demanding while actual mensuration is based very much on rule-of-thumb estimates, and even more so on practical experience as it did grow over centuries of organ building.

From medieval treatises on the mensuration of pipes (mensura fistularum; see Sachs 1980), it is evident that theorists considered first the length of different pipes in terms of small integer proportions (analogous to string sections on a monochord). Apparently, just a few theorists recognised that the analogy of strings and pipes did not hold as such, and was insufficient to determine dimensions and pitches for real pipes. There were some considerations where fractions of the diameter of a pipe were added to its length (to account for the factor later understood as the end-correction of the pipe; see Sachs 1980, 65ff.), however, the approach was theoretical rather than empirical. A general aspect inherent in these mensuration problems is that appropriate scaling of organ pipes involves more than one parameter of pipe length since the design of pipes must consider not only their pitch but also the specific timbral quality of a rank. The task to model a row of pipes thus is threefold. First, the sounds emitted from the pipes must realise the steps of a musical scale defined, for flue pipes, in the main by their fundamental frequency. Second, the sequence of sounds from such a row of pipes must bring about an increase in brightness proportional to the increase in pitch per scale step since brightness is a component of pitch and at the same time, a timbral factor (see Schneider 2017). Third, while spectral centroid and sensation of brightness change along the steps of a rising or falling musical scale, spectral energy distribution and spectral envelope, as well as temporal characteristics of sounds for one pipe rank, should follow a certain pattern so as to maintain the timbral quality (by which a rank is identified, by musicians and listeners). This was already a problem in late medieval times when the compass of an organ was restricted to 2–3 octaves, and more so in modern instruments where four or even five octaves in manual keyboards are standard. One of the facts probably experienced in medieval organ building was that continuity in timbral quality cannot be achieved if only the pipe length is varied, with the diameter (and all other parameters) kept constant. With such a design, pipes low in pitch will have a timbre that is too bright, whereas pipes high in pitch will sound dull (see Adelung 1976, 80ff). The lesson learned early from scaling was that several parameters in regard to the geometry and also voicing of pipes must be taken into account, and in doing so, pipes of a certain type (e.g., a cylindrical diapason, an open conical flute, or a trumpet) can be built in different size so as to match pitch levels for a certain octave (32’, 16’, 8’, 4’, 2’, 1’). In this respect, scaling demands that actual measures must be altered in proportion to each other along relevant dimensions (pipe length, diameter, height and width of cut-up, etc.). Assuming such proportionality, pipes and pipe ranks can be described, first of all, by their pitch level as defined by the pipe length expressed in ‘foot.’ In modern standard tuning (a¹ = 440 Hz), an open flue pipe of approximately 262 cm effective length will produce a sound with the fundamental of 65.4 Hz when the key for C₂ is pressed. The wavelength, in this case, is ca. 525 cm while the pipe of 262 cm approximates ‘eight times a foot’ (in olden times, spatial extensions were often measured in ‘foot’, one foot is ca. 32 cm). Thus, the pipe in question will be labelled 8’ (eight-foot). To produce a fundamental at 32.7 Hz for the key C₂, the pipe length must be doubled to 16’. Conversely, a 4’ pipe at the same key will have a fundamental at 130.8 Hz, a 2’ pipe at 261.6 Hz etc. In historical organs of Northern Germany and the Netherlands, some stops with flue and with reed pipes are found in the 32’ register, while in most organs, pipe ranks from 16’ to 2’ are implemented (spanning four octaves); some organs have or had 1’ ranks (cf. Praetorius 1619, 162ff., Edskes & Vogel 2009, 167, 169, 173, 198). Mixture stops, as well as special pipe ranks, can incorporate very small pipes (<1’), which add high harmonics and increase spectral brightness (see below).

The diameter of open flue pipes changes in proportion to their pitch. However, while pipe lengths approximate a ratio of 2:1 per octave, the diameters of pipes do not correspond to this ratio and, in fact, can vary considerably to adjust the timbre (number and strength of partials) in a rank of pipes. For example, diameters for the Principal 16’ in the HW of St. Jacobi at Hamburg have been measured (cf. Ahrend 1995, 255) as shown in Table 1.

Table 1. Pipe diameters and ratios, Principal 16’, HW, St. Jacobi, Hamburg

Viewed from a historical and geographical perspective, the variety of organ stops in Europe is immense notwithstanding certain standards had been established over time (for detailed accounts, see Mahrenholz 1942/1968, Klotz 1975, Williams 1984, Eberlein 2009). Organ builders experimented a lot to find optimal designs for pipes as well as for the resonators coupled to reed generators. Since the labelling of organ stops was not uniform, one has to take historical and regional traditions of organ building into account. To complicate things further, one and the same name, like the German ‘Nachthorn’ or the French ‘cornet de nuit,’ can stand for stops of quite different designs and musical functions.

Most pipe organs from the 15^th to the 18^th centuries were primarily designed to be used in church services as well as in music events related to religious practice and recreation; this does by no means exclude recitals and concerts where the organ was used as a continuo instrument in an ensemble of strings, woodwinds or brass. Depending on factors such as the size of churches or other rooms chosen to house an organ, financial means available to a community that would order and pay for an instrument as well as the strength of musical activities pursued in certain regions, organs of different sizes and complexity were build that ranged from small instruments (in general, one manual, one wind chest, number of stops < 10, no separate pedal work, or pedal completely missing) to middle-sized and large organs. Middle-sized, in this respect, typically would be a two-manual organ with separate pedal work and an overall number of stops from about 18 to 25 distributed to the three works. Large historical organs of the late Renaissance and Baroque era generally comprise three manuals (in some organs, even four) plus a fully equipped pedal. Thus, there are four or five separate ‘works’ (divisions) which typically can be distinguished by their location within the overall spatial structure of a pipe organ as well as by different musical functions and sound designs.

Large organs from places like Hamburg, Lübeck, or Stralsund by the time of ca. 1600–1700 typically had three or even four manual works like Hauptwerk (HW, great or main organ), Oberwerk (OW, placed above HW), Brustwerk (BW, right in front of the organist sitting with his face in the direction of the HW and OW), and a Rückpositiv (RP, chair organ in the back of the organist) plus a pedal work (Ped) that had gained special importance by then. One of the peculiarities of northern Germany was that in a number of large organs in use around 1650–1730, four manual works were played from three keyboards; that is, one could either couple two manual works (like OW and BW) to the respective keyboard or use them alternately. In a schematic drawing, the spatial arrangement of a large organ with four manual works plus a pedal work split into two towers flanking the organ to the left and right is shown in Fig. 2.

Fig. 2.

Scheme of a North German Baroque organ with four manual divisions (HW, OW, BW, RP), three keyboards, and a pedal whose pipes are mounted in two flanking towers.

A real instrument is shown in Fig. 3, a photo taken of the three-manual organ built for St. Nikolai at Altenbruch (near Cuxhaven, close to the North Sea) by Johann H. Klapmeyer in 1727–1730. This instrument (III, Ped, 35 voices) incorporates a considerable number of stops from older instruments that had been built in the 16^th century; the RP very likely dates back to 1577 (see Vogel et al. 1997, 218ff.). In the years 1647–1649, Hans Christoph Fritzsche from Hamburg renewed the HW and added stops to the RP. Since organs were expensive in regard to the materials needed for construction (different kinds of wood and metal, such as tin and lead hard to get hold of in those days), it was customary to repair existing stops and wind chests and to keep them in use as part of a new instrument. Thanks to this cost-saving attitude, the organ at Altenbruch and many others of the Baroque era preserved pipe ranks from the 16^th and 17^th centuries.

Fig. 3.

Klapmeyer organ at St. Nikolai, Altenbruch, with the HW in the back and the Renaissance RP and the two pedal towers integrated into the balustrade of the gallery. The BW is masked by the RP and not visible in this picture.

Splitting the pedal pipes into two towers with a division of C and C#, D and D#, etc., has the advantage that pipes that are just a semitone apart will not interfere with each other in chromatic bass lines (as are found in organ music of the late Renaissance and Baroque era). Flue pipes in an organ can interact in several ways. One factor is that pipes mounted on the same wind chest share the supply of air delivered from the bellows (which was the traditional wind supply before electric ventilators came into use). If the pressure in the wind supply is not strong enough or unsteady, simultaneous use of several large flue pipes can cause a slight but sometimes audible pitch shift relative to the pitch of each single pipe as tuned. Flue pipes standing close to each other moreover influence each other in pitch, an effect known as acoustical coupling (‘Mitnahme’; cf. Lottermoster 1983a, 56) that can be explained as a synchronisation of vibration regimes (Abel 2008, Fischer, Bader & Abel 2016).

3 A Few Notes on the Development of Pipe Organs

Though any detailed account of organ history is far beyond the scope of this article (see Williams 1966, Klotz 1975, Williams 1993, Eberlein 2011; for an overview, Williams & Owen 1984), a few remarks concerning some major stages in organ development seem apt to understand aspects of continuity and change in organ design. The very beginnings of the pipe organ lead us back to Roman times. In addition to written and iconographic sources, we have a small number of pipes and other parts from archaeological excavations. Perhaps the most significant find brought remnants of a small pipe organ from the camp at Aquincum (today, a part of Budapest, Hungary) to daylight. This organ, parts of which lay in the rubble after the camp was destroyed by fire, dates to 228 CE; it might have been a hydraulis, a type known for its mechanism of hydraulic pressure used to provide wind to the wind chest and further on into the pipes. While the exact mode of wind supply in this instrument is not quite clear, the wind chest survived in relatively good condition and allowed, together with a number of pipes and other parts, a tentative reconstruction of the organ, including inferences as to the dimensions and original tuning of the pipes. These were ordered into four rows, one with open pipes and three with stopped pipes, which yields 13 × 4 = 52 pipes (see Kaba 1976 and the article ‘Orgel von Aquincum’ in the German Wikipedia for pictures). There have been some suggestions in regard to the compass and the scale to which the organ was tuned. Apparently, the four pipe ranks could be activated individually—as in a modern organ—by some mechanism. If used together, pressing a single key (or, rather, pulling a lever) would join the sound from three stopped pipes (of different lengths) plus an open pipe into some sonority that possibly could have included musical intervals (the pitches of a stopped pipe and an open pipe of equal length are an octave apart; see below, Sect. 4).

This aspect is of interest since an important late medieval organ type, the so-called ‘Blockwerk,’ apparently was designed to produce complex harmonic sounds. In its most basic form, a Blockwerk assembled several or even many rows of pipes ordered according to their length on a single wind chest (see Williams & Owen 1984, Klotz 1975, 10f.). The idea behind this arrangement was that, by pressing a single key (which was broader and longer in size than in a modern keyboard; see Praetorius 1619 and Bormann 1966), a number of pipes responded, which were in harmonic pitch ratios and formed chord-like sonorities. In certain respects, a Blockwerk thus can be viewed as an early large mixture stop which, however, is different from later mixtures in that a Blockwerk typically had only a few pipes for the low keys and a larger number of pipes for high keys, that is, the number of pipes per key increased from the bass to the discant register (see Preaetorius 1619, 94ff; Bormann 1966, Klotz 1975). The next step in organ building was to add ranks of diapason pipes which either were rigidly coupled to the compound of pipes in the Blockwerk or could be switched on and off as desired; thus, an elementary registration was possible (diapason pipes with and without Blockwerk, or the latter alone). Musically, one could use such ranks of diapason pipes to carry a hymn or other melody while the Blockwerk, as a sonorous unit, could support the diapason as well as a group of singers. Since the ranks of diapason pipes were placed in the front of such organs, they often were labelled ‘Praestant’; the Blockwerk mixture set behind these diapason pipes was labelled ‘Hintersatz’ (both terms remained in use over centuries).

The structure of a late medieval organ that was built at Halberstadt around 1360, and revised in 1495, was as follows: this organ had three manuals and a pedal, with 22, 22, 12, and 12 keys, respectively. The compass for the two upper manuals apparently was H–a¹ (B₂–A₄), with g#¹ (G#₄) missing, and the key for H (B₂) probably sounding the tone B (Bb₂). The uppermost manual was connected to a compound of pipes of various lengths that made up the discant Blockwerk (see Table 2).

Table 2. The distribution of pipes relative to the keys likely was as follows (cf. Bormann 1966, 44).

From the middle keyboard, one could play a double row of 16’ diapason pipes (suited to carry a melodic line); however, the keys of this keyboard also activated the pipes of the discant Hintersatz. The third manual activated the pipes of another Hintersatz, which was an octave lower in pitch than the discant Blockwerk unit, and coupled with the pedal.

The structure of the discant Blockwerk unit reveals that the number of pipes increased towards high pitches and small pipes. The sound pressure level for each of the 16’ and 8’ pipes will be higher, at a given wind pressure, than that of a single 2’ or 1’ pipe. Still, the sheer number of the 2 $\tiny{\raise0.7ex\hbox{$2\,\,$} \!\mathord{\left/ {\vphantom {1 3}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$3$}}$’, 2’ and 1’ pipes will reinforce the sound level considerably and, moreover, will shift the spectral centroid upwards in the treble range. Praetorius (1619, p. 100) noted that a Blockwerk such as the old organ of Halberstadt must have produced “ein uberaus starcken schall und laut und gewaltiges geschrey” (an immensely strong sound and enormous screaming).

Praetorius understood the ‘Hintersatz’ as a forerunner to the more modern mixture stop. The difference, however, is that the Blockwerk of late medieval and Renaissance masters, in general, did not use the concept of repetition, which is characteristic of a mixture stop. For example, in the organ built by Berendt Huß and Arp Schnitger for the church St. Cosmae & Damiani at Stade (Altes Land, see Edskes & Vogel 2009), the OW from 1675 has a compass (with a so-called short octave in the bass) C, D, E, F – c’’’ (44 keys spanning four octaves). Among the 11 stops in the OW is a mixture VI (Vogel 1982), as shown in Table 3.

Table 3. Composition of mixture stop (sixfold), Huß-Schnitger organ, Stade

Such a mixture stop is called ‘sixfold’ since there are six pipes per key who have interval relations of octaves and fifths. From the structure of the pipes in regard to their length and pitch, it is clear that the mixture serves to ‘brighten up’ sounds of the basic diapason stops (like Principal 16’ or Quintadena 16’ available in this OW). In particular, for tones low in fundamental frequency like C (in Helmholtz designation), which, in this organ (historically tuned to g’ ~442 Hz, a so-called ‘choir tone pitch’) is at ca. 74 Hz. Conversely, tones much higher in fundamental frequency, like c’ (295.4 Hz in this organ), shall have less treble sound added from high-pitched small pipes so as to avoid a sound quality sensed as ‘sharpness’. Instead, for these tones in the upper octaves of the keyboard, the pipes from the mixture stop should reinforce, to some extent, midrange frequencies. Since the pipes of mixture stops internally are tuned to just intonation, that is, harmonic ratios, the acoustical function of such a mixture stop thus is to provide additional partials on top of the sound of pipes from other stops (e.g., diapason- or flute-like stops). This effect can be labelled ‘harmonic spectral enhancement.’ The overall perceptual effect of such a mixture stop is to supplement the sound of diapason- or flute-like stops with a certain amount of spectral brightness that, approximately, should be constant over the whole compass of the keyboard. Therefore, the spectral centroid of sounds from a mixture stop should not change very much over several octaves so that the sensation of spectral brightness from a sequence of sounds played with, for example, Principal 16’ + Octave 8’ + Octave 4’ + Mixtur VI does not surpass a certain range.

The original Blockwerk concept implied that several, if not many, pipes attached to each key would produce a rich, chord-like sonority based on octaves and fifths (see above). The Blockwerk, with its peculiar sound structure, has been linked with the medieval organum as a musical form (cf. Klotz 1975, 9). However, during the 14^th and, more so the 15^th century, the structure of organs was modified and expanded in line with musical developments. The Halberstadt organ had three manuals and a pedal suited to play some elementary two-part polyphony, perhaps including long-held pedal notes or simple ostinato patterns. In the 15^th century, a number of organs already showed separate ‘works’ (like HW, RP, Ped), each equipped with one or several pipe ranks, which could be activated individually and combined at will with one or with several mixture stops. From the treatises on organs by Henri Arnaut de Zwolle from 1447 (facs. ed. 1972; Latin text with commentaries in Bormann 1966, 157ff.), the structure of an organ built by Jehan du Mexe for the cathedral Notre Dame at Dijon (Bormann 1966, 163f., 169f., Klotz 1975, 35, 40ff.) has been inferred like shown in Table 4.

Table 4. Organ of Notre Dame, Dijon (15^th century), stop list (reconstructed)

According to this stop list, the diapason in the HW, RP, and Pedal each consisted of several, i.e., two to seven rows of pipes (marked with Roman capitals). The HW had a substantial mixture with up to fourteen pipes per key (thereby continuing the tradition of the Blockwerk), and the Pedal also had its mixture (fivefold). The important point here is that the HW included a special mixture stop generally known as a threefold cymbal (German: Zimbel, fr. cymbale). Arnaut de Zwolle (fol. 133 v° and 134 r°) gave more information in regard to this stop which offered high-tuned major chords; for instance, pressing the key f² would produce sound from a total of 18 pipes, the c³ key would activate 20 pipes (cf. Bormann 1966, 163) shown in Table 5.

Table 5. Organ of Notre Dame, Dijon, compound of pipes attached to single keys

The major-third cymbal (German: Terzzimbel) is of particular interest since, in the course of the 15^th century, the major third was accepted as a consonant interval in both music theory and composition, a fact that led to significant changes also in the tuning of organs and other keyboards. By about 1500, the so-called meantone temperament based on just major thirds (see Schneider & Beurmann 2017) had become the predominant tuning system in Europe which was in use, in a number of variants, well into the second half of the 18^th century or even later (it was laborious and costly to change the tuning of organs with their multiple pipes).

Splitting the former Blockwerk into separate stops (like mixture and cymbal) as well as dissolving the compound of diapason pipe rows into several independent stops were notable developments in organ building in the 15^th and 16^th centuries. The dissolution into individual stops is especially clear in Italian organs where the order of diapason stops often was as reported for S. Pietro at Modena (Giovanni B. Facchetti, of Brescia, 1519) and for Santa Maria Rotonda at Brescia, built by Gian Giacomo Antegnati in 1536; see Klotz 1975, 71, 133). The stop list for the Facchetti organ is shown in Table 6.

Table 6. Organ of S. Pietro, Modena, early 16^th century, stop list

The Italian terms and the Roman capitals relate to the claves naturales of the diatonic scale; foot marks are relative and indicate interval relations between stops, not the absolute length of pipes. The organ built by Antegnati had an additional flute stop (XXII, 1’) and an independent pedal (F G A-d¹, 20 keys) with a single stop labelled Contrabassi 16’. The feature that is of interest here (as different from a fixed mixture) is the possibility to select and combine pipe ranks that form harmonic interval ratios; activating those diapason stops one after another means expansion of an additive harmonic synthesis whereby the sound gets brighter with every pipe rank added on top. The concept of additive synthesis of diapason stops, being apparent in many Italian organs, has a modern follow-up in electronic organs of the 20^th century, such as the Hammond B3 and the Vox Continental, where the player can mix partials from generators with ‘drawbars’ like pipes of different foot length (16’, 8,’ 5 1/3’, 4’, etc.).

From historical sources, it seems the range of ‘sound colours’ available from those Italian organs was rather small (with a dominance of diapason and flute stops). However, from iconography and written sources, it is well known that late medieval and particularly Renaissance musical practice included many wind instruments (flutes, horns, reeds).. At the beginning of the 17^th century, yet much in retrospective, Michael Praetorius, himself a skilled musician and composer, in his ‘Organographia’ gave a detailed account of musical instruments and put special emphasis on trombones, trumpets, the Zinck, several long and cross flutes, and various types of reed instruments such as the Pommer (alto and tenor shawm, Bombart), Schalmey (treble shawm), Dulzian (an early bassoon-like reed), the Krumbhorn (crumhorn), Rankett, etc. (1619, 31–43). In the same work, the chapter on the ‘Historia veterum Organorum (81ff.) elucidates the concepts behind organs of previous centuries, in particular instruments of the Blockwerk style. The chapter on the Historia novorum Organorum (119ff.) discusses the types of organ pipes and the different pipe ranks that came into use, mostly in the 16^th century, and which Praetorius knew from first-hand experience. He describes the more common diapason and flute-like ranks, followed by flue pipes with more complex geometry (like the Gemshorn), the stopped pipes and the various reed pipes. He also adds chapters on the tuning of reed pipes and on the suitable design of organs and presents a comprehensive survey of stop lists (Dispositionen) from various organs that had been recently built for churches at Danzig, Lübeck, Hamburg, and other places. More information is condensed into a catalogue of pipe ranks. Finally, his ‘Sciagraphia oder Theatrum Instrumentorum’ offers many figures that illustrate the instrument types addressed in the text.

From Praetorius and other sources, we understand how diversified pipe ranks had become between roughly 1450–1600. One of the reasons already mentioned was the dissolution of huge compounds of pipes (Blockwerk and Hintersatz) into separate ranks, another was that organ builders strived to emulate the broad range of flutes, horns and reed instruments that played an important role in Renaissance music. These instruments all had a peculiar sound quality (some came close to the human voice, some reeds had a nasal sound, etc.), which made them distinct and identifiable in an ensemble. Organ builders must have recognised the benefit they could have for organs devised as multi-timbral instruments. If several distinct ‘sound colours’ were available, the organist could play a cantus firmus or characteristic melodic line with a reed stop against other voices, for which a soft sound (from stopped pipes like a Gedackt) might be appropriate. Such a concept would work easily in a two-manual plus pedal organ with different pipe ranks available in each department. In small instruments (one manual, no separate pedal), parallel usage of two ‘sound colours’ was possible if some stops could be assigned to either the bass or the discant half of the manual (which was thus divided into two registers).

From a historical perspective, the diversity of pipe ranks and an increase in the number of stops is evident from many organs of the 16^th century that were built in France as well as in a large region comprising the Low Countries (understood as a geographical term) and parts of Germany (for a detailed account, see Klotz 1975, ch. X-XVI). The division of pipe stops into distinct groups according to sound properties and musical function, in general, followed a scheme like:

• A. Diapason stops (flue pipes from 16’ to 2’ with relatively narrow diameters like Prinzipal 16’ or Praestant 8’, Oktave 4’, Oktave 2’); mixtures (usually III to IV) and related aliquot stops (like Zimbel or Sesquialter) composed of rather small and narrow flue pipes;

• B. Open and stopped flue pipes with a wider diameter (like Hohlpfeife, Quintaden, Nachthorn); the sound quality in theses pipe ranks is more mellow or flute-like, in stopped flue pipes it can be hollow (like a voiced syllable ‘hu’) due to the prevalence of low odd partials;

• C. Reed pipe stops like Posaune 16’, Trompete 8’, Krummhorn 8’, Schalmei 4’.

In a well-designed, middle-sized or even large organ, one could expect a selection of stops from all three groups in every department (HW and/or OW, RP and/or BW, Ped). Perhaps the largest organ in use before 1600 was built 1583–1585 by Julius Antonii (from Bergues-Saint-Vinoque; Flemish: Sint Winoksbergen) for St. Marien at Danzig (56 voices on HW, RP, BW, Ped plus 3 tremolo units and a wind-operated drum). This instrument offered an enormous range of pipe ranks (as listed by Praetorius 1619, 162f.), including various diapason and mixture stops, many different flutes (from 16’ to 1’), and no less than 11 reed stops such as a trombone 16’ (Ped), two trumpet 8’ (RP, Ped), two Krummhorn 8’ (RP, Ped), two Schalmei 4’ (RP, Ped), two Zink 4’ (RP, BW), a Regal 8’ (BW), and a Kornett 2’ in the pedal. Praetorius (1619) presented the stoplists of some other large organs as were built at, for example, Lübeck (St. Petri, Gottschalk Johannsen, 1587–91, 45 stops on OW, BW, RP, Ped), Stralsund (St. Marien, Nicolaus Maaß ca. 1592, 43 Stops on OW, RP, BW, Ped), Hamburg (St. Jacobi, 53 stops on OW, BW, RP, Ped). In 1680–87, Arp Schnitger succeeded in building an even more complex organ for the St. Nikolai church of Hamburg (67 stops on HW, OW, RP, BW, Ped; four manuals with ‘short octave’ in the bass register, C to c³, 47 keys; see Fock 1974, 46ff.). This marvellous instrument, the result of a long tradition of organ building first developed in the Low Countries and continued in Northern Germany, unfortunately was destroyed, in a disastrous fire, on May 5^th 1842. Another famous organ that fell victim to this fire was the organ built by Henrick Niehoff for St. Petri of Hamburg (ca. 1550, 42 stops, see Praetorius 1619, 169f. and Fock 1939, 298ff.). Niehoff, who worked for many years from s’Hertogenbosch in the province of Brabant, is understood as a foremost organ builder of his era as he pursued a concept of contrasting sound colours produced from various flue and reed stops. In particular, he recommended the Terzzimbel (also labelled ‘klingende Zimbel’ and ‘rauschende Zimbel’), a special type of mixture which, due to its composition, added high harmonics including major thirds to the sound of other stops, thereby amplifying both spectral fusion and brilliance. In the historic organ of Altenbruch (see above), there is an original Zimbel in the HW (probably built by Hans Christoph Fritzsche in 1649) that produces significant spectral energy in high-frequency bands (up to and even beyond 10 kHz; see Schneider et al. 2006).

4 Sound Generation: Empirical Observations

One obvious feature in the speech of flue pipes is the noisy transient in the onset of sounds which has been studied extensively (cf. Fletcher 1976, Nolle & Finch 1992, Castellengo 1998). The main reason for this phenomenon is that alternating pressure p_~ needs time to build up from the pulse train passing from the edge tone generator into the resonator tube and that, after reflection at the open end, a stable regime of periodic vibration needs to be established resulting in standing waves. The pipe viewed as a cylinder filled with a mass of air has a certain input impedance Z_in which is quite small, but so is the wind pressure in most historic organs as measured in a duct or chest (usually 50–80 mm water column depending on the size of the organ and the room in which it stands). As a mass of air enclosed in a large flue pipe has some inertia, the onset in 16’ and 8’ pipes can last quite long (t_on > 50 ms, for some pipes even t_on > 100 ms; see examples in Beurmann et al. 1998, Schneider et al. 2001, 2006). Quite often, the second mode of vibration (the octave in an open flue pipe, the twelfth in a stopped pipe) is activated before the fundamental sets in. The higher partial kind of ‘signals’ the onset of such a tone to the listener ‘(see Fig. 4)’:

Fig. 4.

Stade, St. Cosmae & Damiani, the organ built by Berendt Huß and Arp Schnitger 1668–1673; RP, Oktave 4’, key/tone C; onset begins with the 2^nd harmonic (Oscillogram).

Another characteristic feature of the onset of many flue pipes is the ‘spitting’ noise (“Chiff”) preceding periodic vibration. The noisy transient, together with the attack of the partials, has a sound quality of its own that helps listeners sense the onset of single tones. While the length of the pipe that produced the transient shown in Fig. 4 was ca. 1.20 m, transients appear even in very small flue pipes. In Fig. 5, the onset for the tone/key c in the Nachthorn 1’ of the pedal in the organ of St. Cosmae at Stade is shown. The pipe length for this tone is actually 1/2 foot; the fundamental frequency consequently is high at ca. 1177 Hz.

Fig. 5.

Nachthorn 1’, tone/key c, Onset with noisy transient, periodic vibration.

In flute-like aerophones, one can observe, in a series of overlapping short-time spectra, the process whereby relatively broad and flat components carrying energy turn into harmonic partials with marked peaks (see Schneider 1998). The physical process thus covered is the transition from relatively broad resonance zones to definite resonance frequencies in the tube as the standing wave regime becomes stable. The time needed to establish standing waves in the tube is dependent, among other parameters, on the length and width of the tube. In a small pipe like the Nachthorn 1’, a periodic vibration pattern appears after ca. 10 ms (Fig. 5). In the spectrum of this sound taken shortly after onset, four partials are prominent.

In general, reed pipes differ from flue pipes in that there is a fast and hard attack in their sound with less noise involved in the transient. The periodic regime of vibration is often established almost instantly, even in large generator plus resonator systems, as is demonstrated in Fig. 6, which shows the onset of sound radiated from the c pipe of the Dulzian 16’ in the RP of the Huß/Schnitger organ at Stade.

Fig. 6.

Dulzian 16’, tone/key c (f₁ ~ 77 Hz). The section to the left marks the time from the start of wind supply to this pipe and the transient (ca. 34 ms) which is immediately followed by full periods of vibration (T ~ 13 ms).

Sounds from reed pipes typically have a rich spectrum with dozens (in some cases more than 100) harmonics. The spectra, in general, show a cyclic structure where spectral amplitudes and the spectral envelope are similar to the envelope of a Sin[x]/x function, and certain harmonic partials are more or less suppressed due to the duty cycle of the valve defined by τ/T (τ = pulse width, T = length of the period in ms; for examples from the large Schnitger organ of St. Jacobi at Hamburg see Beurmann et al. 1999, 159ff.). Numerous reed pipe spectra show considerable energy in frequency bands known from phonetics as ‘formant zones’. Such a concentration of spectral energy lends sounds a vocal quality. The spectrum of a sound of the Regal 8’ in the RP of the organ at St. Jakobi of Lüdingworth (see Edskes & Vogel 2009) illustrates this peculiar aspect (Figs. 7 and 11). The Regal 8’ was built by Antonius Wilde in 1598/99 for an organ expanded by Arp Schnitger in 1692. In this stop, the tongue, the shallot, the tuning wire and the resonator of each pipe are made from brass which gives the pipes a distinct ‘sound colour.’ As is evident from Fig. 7, one of the energy concentrations is around 3.4 kHz (the region of the so-called ‘singing formant’ for male opera singers is at ca. 3 kHz; see Sundberg 1987).

Fig. 7.

Regal 8’, tone/key C, formant-like spectral energy concentrations at 1.3 and 3.4 kHz.

A closer inspection of the same sound with the Burg algorithm (see Marple 1987, ch. 8) reveals that, in fact, two concentrations of spectral energy can be identified as formants, with centres at about 1.3 and 3.4 kHz, respectively.

The temporal and spectral composition of such sounds from reed stops is of perceptual and musical significance. First of all, their pitch is clearly defined both from the fundamental f₁ and the periodicity pitch f₀ = 1/T resulting from the joint effect of numerous harmonics. Second, the presence of formant-like energy concentrations in the spectrum gives such sounds a vowel-like quality (which is also observed in the tone of Italian master violins, see Mores 2017). In effect, reed stop sounds, such as those produced by the Regal 8’ of Lüdingworth provide the listener with ample information in regard to pitch structure and timbre. With reed stops available in each division of the organ, one could emphasise prominent voices in a polyphonic setting or could give consecutive sections of a musical work alternating ‘sound colours.’

5 Sound Structure and Tuning

The design of pipe organs from the medieval Blockwerk and the Italian instruments of the 16^th and 17^th century up to a range of organs with multiple stops built in the Low Countries and Northern Germany (with some extensions also to Denmark and Sweden) between ca. 1600 and 1750 converges in one fundamental aspect which can be described as ‘massive additive sound synthesis’. To understand this concept, one has to remember that each pipe in an organ is a sound generator of its own which produces a more or less complex sound with a periodic time function and a harmonic spectrum. In a large organ of the Baroque era, such as the (extant, fully restored) Schnitger organ of St. Jacobi at Hamburg, there are more than 4,000 pipes (see Ahrend 1995), and even in smaller instruments, there are several hundred sound generators each tuned to a certain pitch. Thus, any combination of pipes will produce a complex harmonic sound where many spectral components carry energy and reinforce the tonal structure (e.g., in a major chord played with organo pleno registration).

As has been pointed out, the early Blockwerk organ (without any facility for registration) must have had a complex sonority for every key (in historic treatises, the expression indeed is “die Orgel schlagen”). From medieval treatises on music theory and organology, we may assume that the early Blockwerk was tuned to a chain of pure fifths. Thus, the tuning was ‘Pythagorean’ based on ratios such as 3/2, 4/3, 9/8, 81/64, etc. To be sure, these ratios concern the horizontal dimension of tuning (i.e., the distances between fundamental frequencies of the tones within a scale which, in modern times, can be expressed in Hz or in cents calculated therefrom). The vertical dimension of tuning, in a Blockwerk as well as in various mixture stops and special pipe ranks consisting of several rows of pipes (e.g., the Rauschquinte or the Sesquialter) concerns the relations of fundamental frequencies of the pipes within that pipe rank and in particular, the interval and frequency relations of pipes activated by each key of the manual. The pitch intervals in the vertical direction were always (and still are) tuned to small integer ratios, that is, in just intonation, in order to produce a high degree of spectral fusion and harmonicity.

As we know from Arnaut de Zwolle (see above), the Terzzimbel, which comprises pipes tuned to sound as just major thirds 5/4 and major chords, was introduced quite early into organ building. This did not cause problems as long as the music played on a Blockwerk may have been restricted to hymns or other melodic formations. In this case, every note played on the keyboard would produce a rich harmonic sound from the group of pipes assigned to a particular key (similar to a modern synthesiser with several harmonic oscillators where a complex sound can be activated from a single key). The problems began when separate ranks of diapason pipes and extra manuals were added to the Blockwerk and when musical settings or improvisations played on those organs included polyphony (already beginning in the 14^th century and clearly so in works of the 15^th century; see Klotz 1975). Though simultaneous intervals according to music-theoretical rules of that era were restricted to perfect consonances (the octave 2/1, the fifth 3/2, the fourth 4/3), in the 15^th century, major thirds appeared in various sources (like the well-known Buxheimer Orgelbuch). Given that the horizontal tuning in the keyboard was still Pythagorean, most of the major thirds in a twelve-note scale would be of the size of a ‘ditonus,’ comprising two whole tones 9/8, which results in the ‘Pythagorean major third’ of 81/64 (of 408 cents). The Terzzimbel, however, had just major thirds 5/4 (of 386 cents). Playing the interval of a major third on the keyboard and at the same time activating the pipes of a Terzzimbel would inevitably bring about a controversy of two major thirds differing in interval size by a so-called comma of ca. 22 cents. The sound of two groups of pipes which differ in their tuning by ca. 22 cents, results in severe amplitude modulation, which listeners sense as roughness.

The problem was solved, in one essential aspect, when the so-called meantone temperament was introduced into keyboard tuning ca. 1500 (or shortly after). The term ‘meantone’ (which was coined in the 19^th century) refers to the fact that, in this tuning, an interval of a just major third like c–e (386 cents) is divided, by the tone d, into two intervals of equal size (193 cents). However, the so-called quarter-comma meantone temperament (see Schneider & Beurmann 2017) was derived from practical tuning based on four pairs of major thirds like b_b–d–f^#, f–a–c^#, c–e–g^# and e_b–g–b as is evident from the tone lattice A shown in Table 7.

Table 7. Tone lattice for A: Meantone tuning and B: Just intonation pitches

In the left tone lattice, the sign│denotes a just major third 5/4, and the sign ↔ denotes a fifth narrowed by one-quarter of a comma, that is by ca. 5.5 cents (the ‘meantone fifth’ thereby is ca. 696.5 cents). In the right lattice, the│also denotes a major third 5/4, and ─ denotes a pure fifth 3/2. Tones in the −1 row are flat by one comma (22 cents) relative to the pitch of a tone of the same name in the row marked 0 (which represents the basic chain of just fifths). Tones in the −2 row are two commas flat, and tones in the +1 row are one comma sharp relative to their equivalents in the 0 row (the lattice of just intonation tones can be extended in the horizontal and in the vertical as needed but is restricted here to the section shown for demonstration). From the two schemes, it is evident that they share the formative interval of the just major third in the vertical and that they differ in the size of the fifths. The main reason is that tuning an instrument to just intonation ratios requires more than 12 pitches and tones per octave since it distinguishes between sharps and flats; for example, an E-major chord needs the g^# as major third, a f-minor chord needs an a_b as minor third, etc. With but twelve tones and pitches to be tuned per octave and the decision for implementing as many just major thirds as possible (which, in addition, would bring about just minor sixths 8/5 of 814 cents), so-called ‘quarter-comma meantone temperament’ with no less than eight just major thirds was the best possible solution. This tuning soon became widespread and was reflected also in works of Renaissance and Baroque keyboard music which indeed feature the ‘sweetness’ of just major thirds and minor sixths. A piece that clearly demonstrates such features is John Dowland’s ‘Lachrimae Pavan’ (originally for lute) which was set, with variations, for organ and other keyboard instruments by composers like William Byrd, Jan Pieterszoon Sweelinck, Peter Philips, Melchior Schildt, and Heinrich Scheidemann. These variations sound great when played on an organ or other keyboard instrument tuned to 1/4-comma meantone temperament because of the high degree of fusion in simultaneous major thirds and minor sixths. However, while there are eight just major thirds in 1/4-comma meantone tuning, the other four are far too wide (at 428 cents), and some of the minor thirds are far too narrow. Besides the slightly narrowed ‘meantone fifths’ of 696.5 cents, there is one very poor fifth at g^# – e_b of 738.5 cents (blamed as the ‘howling wolf’). In effect, 1/4-comma meantone offers a number of highly consonant major chords (B_b, F, C, D, A, E, E_b) as well as a number of harmonious minor chords (c, d, a, a, e, f^#, c^#). In contrast, some major and minor chords sound rather harsh, particularly those where the wide major thirds c^# – f, f^# – b_b, g^# – c, b – e_b of 428 cents are involved. To overcome these deficiencies, one had to avoid the poor intervals and chords (or could use them, in certain settings, as an expression of grief and pain as was appropriate in the context of the musical ‘Affektenlehre’).

A technical and musical remedy suited to overcome the limits of 1/4-comma meantone was to increase the number of pitches per octave beyond twelve. A practical solution for harpsichords and organs was to provide extra strings and pipes so as to split g^# and a_b, e_b and d^#, thereby eliminating not only the ‘wolf’ but also improving the compass of chords that can be played in acceptable quality (that is, with a sufficient degree of harmonicity and the absence of unbearable roughness). With 1/4-comma meantone tuning as background, the compass of keys and chords used in musical works, in general, was from A-Major to B_b-minor. Inserting the two tones/pitches a_b and d^# into lattice A above shows that the B-major chord and the A_b-major chord, the f-minor chord and the g^#-minor chord are now at hand. The development of keyboard instruments with more than twelve keys/pitches per octave seems to have started in Italy, in the 16^th century, in attempts at reviving classical Greek chromatic and enharmonic scale models to be used in contemporary music. In 1548, Zarlino had a harpsichord with 19 keys/pitches per octave, and Vicentino expanded the number of keys and pitches per octave to 31 (see Schneider & Beurmann 2017, 415ff.). Two such enharmonic instruments with 31 keys, the ‘Clavemusicum omnitonum’ built by Vitus de Transuntino in 1606 (see Barbieri 2008, 25f.), and a Hammerklavier from the late 18^th century (Johann Jakob Könnicke, Vienna; see Barbieri 2005, 463ff.) have survived.

Though instruments with 17, 19, or even 31 keys per octave were rare since their construction was far from easy, the concept of adding two extra keys and pitches per octave into the keyboard of harpsichords as well as of organs to improve on the meantone tuning must have been more common. Werckmeister (1698, 79, 81) complained that one finds keyboards with three or more subsemitonia implemented; in his opinion, this was an obstacle to musical performance. At Hamburg, Gottfried Fritzsche (also Frietsch, he originally came from Meissen in Saxonia) built a new organ for St. Maria Magdalena in 1629, which, according to Mattheson (1721, 180f.), had several subsemitonia in each octave. In 1633/34, Fritzsche expanded the organ of the St. Petri church, where he supplied the HW with a new chest and added several stops; he implemented subsemitonia for d^#/e_b, g^#/a_b and a^#/b_b in the HW as well as in the newly built BW (see Schröder 2006, 32). In 1635, he also must have implemented split keys for d^#/e_b, g^#/a_b and a^#/b_b in the RP of the organ in St. Jacobi (see Fock 1939, 350). Organs with 14 or even 16 keys per octave were still found in England in the 19^th century (and even new instruments with split keys were built there; see Williams 1968, 62f.). When, in the 1990s, a new organ was planned for Örgryte Nya Kyrka at Gotenburg that should emulate the large North German Baroque organs as were built at Hamburg by the Scherer family, Gottfried Fritzsche, and Arp Schnitger (as well as by Friedrich Stellwagen at Lübeck and Stralsund), a decision was made to incorporate split keys in this four-manual organ analogous to the extra keys Fritzsche had provided (see Speerstra 2003). The Gotenburg instrument also has split keys for e_b/d^# and g^#/a_b in the pedal.

Since split keys in organs meant extra costs for additional pipes and mechanics, and as organists perhaps found it difficult to master keyboards with 14 or even 15 keys, a less arduous way to deal with tuning problems was to try various temperaments (the Latin word ‘temperari’ means to balance) such as proposed by Werckmeister, Neidhardt, and other theorists and practitioners of music around 1700 (see Lindley 1987, Ratte 1991). A general tendency of most such proposals is to enlarge the just major thirds, which made up the core of 1/4-comma meantone tuning, and to widen the narrowed fifths of this system so as to approximate the 3/2 ratio. In the tuning scheme known as Werckmeister III (from 1691, see Rasch 1983), there are four narrowed fifths (similar to 1/4-comma meantone) while all other fifths are pure. Major thirds in this system vary from 390.2 to 407.8 cents, and minor thirds from 294.1 to 311.7 cents. Werckmeister III thus was a step back from a tuning based on just major thirds to a tuning based on pure fifths (like the Pythagorean system). While Werckmeister still maintained some grading within intervals and chords in regard to harmonicity vs. roughness, ET12 later levelled such differences. The advantage of a tuning like Werckmeister III understood as a ‘Wohltemperierung’ (making all major and minor chords acceptable though by no means equal like ET12) was that it allowed using most keys around the circle of fifths. As some impressive works of German organ music of the Baroque era are in E-major (a Praeludium and fugue by Buxtehude, BuxW 141, and a similarly complex work by Vincent Lübeck, LübWV 7), their rendition in 1/4-comma meantone is problematic since the B-dominant chord needed in E major suffers, with the pitches actually available in this tuning (see tone lattice A), from the false major third b–e_b (of 428 cents) and the narrowed fifth. It has, therefore, been suggested that those works (as well as works by other composers of the Baroque era, including J. S. Bach) would require a ‘Wohltemperierung’ like Werckmeister’s, as an adequate tuning system.

To assess the quality of a certain tuning or temperament like 1/4-comma or 1/5-comma meantone, Werckmeister III, ET12 etc. objectively, sound analysis directed to temporal and spectral parameters seems appropriate (see Schneider et al. 2004, Schneider & Beurmann 2017). In particular, measurements of all major and minor chords as played on an organ or harpsichord provide data for a comparative evaluation. The harmonic-to-noise ratio (HNR, see Boersma 1993), calculated from the periodicity of a signal as measured by autocorrelation or cross-correlation, shows differences between the twelve major and twelve minor chords of a chromatic scale for a given tuning as well as differences between several tunings (e.g., variants of meantone temperament, Werckmeister, ET12; see Schneider & von Busch 2015, Schneider, von Busch & Adam 2017, Schneider & Beurmann 2017). Empirical data from such measurements thus allow us to classify major and minor chords for each tuning in regard to spectral harmonicity where low HNR readings (quantified in dB) indicate a poor degree of harmonicity (the respective sounds are likely to give rise to a sensation of roughness). Higher HNR readings indicate that spectral components of the three complex harmonic sounds making up a chord are less divergent in frequency and amplitude, thereby enhancing the periodicity of the signal and that such chords are sensed as more consonant by listeners.

A comparison of tunings in use on pipe organs poses a problem, in one respect, since it is not possible, under realistic conditions, to change the tuning of a certain historic organ so that one could record sounds from pipes in meantone tuning on one day, and repeat the process a short time later after retuning the same organ to some other system. Thus, one has to compare sound data recorded from different organs. This, however, seems justified if the recordings can be done under nearly identical conditions from instruments of the same period, which have been restored recently according to the same criteria. For an actual comparison, we made recordings of the organ Arp Schnitger had built in 1688–90 for St. Mauritius at Hollern (Altes Land, close to Hamburg; see Edskes & Vogel 2009) and of the organ at St. Wilhadi in Stade (Altes Land), built by Erasmus Bielfeldt 1732–36 (see Vogel, Lade & Keweloh 1997). The Schnitger organ at Hollern is tuned to 1/4-comma meantone, the Bielfeldt organ at Stade to Werckmeister III. A comparison of the HNR data demonstrates that, for 12 major chords, meantone yields a number of relatively high readings (for C, D, E_b, E, F, G, A, B_b) in contrast to some poor (C^#, F^#, G^#, B). In Werckmeister, differences between the 12 major chords are still present but not as large as in meantone. For the 12 minor chords, meantone again shows a clear pattern of higher vs. low HNR readings, and also Werckmeister exhibits an uneven pattern. However, the differences (expressed in dB) between individual minor chords are not as big as in meantone. In conclusion, a comparative evaluation of HNR data suggests that Werckmeister, on average, is more balanced than meantone in regard to major chords and, to a lesser degree, also minor. HNR data thus confirm the concept of Werckmeister III as a tuning suited to compose and perform organ music within a wider compass of keys. This advantage, however, is not without problems.

One has to remember that the pipe ranks in a Blockwerk and then in mixture stops were tuned to just intonation, typically in octaves, pure fifths and just major thirds. Stops like a Terzzimbel, Sesquialter or Terzian (see Mahrenholz 1942/1968, 228ff.) work very well in 1/4-comma meantone for those chords which incorporate just major thirds, but must be avoided in remote keys where the poor major thirds of 428 cents of the horizontal keyboard tuning would create roughness against the just major thirds of 386 cents from the Zimbel. As the contrast between good and poor chords in Werckmeister is less marked, adding a Zimbel to a standard registration like Prinzipal 8’, Oktave 4’ + 2’ perhaps would yield tolerable or even fair results in regard to HNR readings (which can be related to the psychoacoustic parameters of harmonicity vs. roughness). In Fig. 8, HNR readings for twelve major chords played on the HW of the Bielfeldt-organ at Stade with three stops (Prinzipal 8’ + Oktave 4’ + 2’) are shown in Fig. 8.

Fig. 8.

Major chords, Werckmeister III, St. Wilhadi, Stade, HW, Prinzipal 8’, Oktave 4’ and 2’. The very short and high HNR at the onset of the A-major chord appears because the sound at the onset starts with a single harmonic partial corresponding to the 2^nd mode of vibration in one of the pipes.

If the Cimbel (threefold) available in the same HW is added, the effect is significant, as Fig. 9 demonstrates (all sounds had been normalised to −3 dB before analysis). For all twelve chords, HNR readings are markedly lower, indicating the overall level of periodicity and harmonicity is reduced. Moreover, the flux in HNR over time, which indicates modulation effects such as AM and roughness, is already visible in Fig. 8 and increases significantly with the Cimbel added (Fig. 9). Though Werckmeister III tuning allows for a greater compass of usable keys, on the one hand, it does not work well with mixture stops which incorporate just major thirds, on the other.

Fig. 9.

HNR readings, Major chords, Werckmeister III, St. Wilhadi, Stade, HW, Prinzipal 8’, Oktave 4’and 2’ plus Cimbel (threefold).

The discrepancy between horizontal and vertical tuning encountered in Werckmeister III is of a more general nature. In a fundamental way, tunings based on pure or slightly tempered fifths (such as Werckmeister and ET12) differ from tunings based on just major thirds, such as 1/4-comma meantone and its expansions on keyboard instruments with 17 or more keys (see Barbieri 2008, Schneider & Beurmann 2017). The reason is that powers of one prime number do not equal powers of another prime number (e.g., 3ⁿ ≠ 5^m), to the consequence that a Pythagorean major third 81/64, derived from four pure fifths 3/2 like c – g – d – a – e, differs from a just major third 5/4 by 21.5 cents (the so-called syntonic comma). This discrepancy, well-known from Greek musical theory, must have become a problem for organ builders in the 15^th century when horizontal keyboard tunings most likely were still based on chains of pure fifths, while musical works demanded just major thirds as consonant intervals. Moreover, the Terzzimbel was invented as an organ stop complementing the usual mixture (based on pure fifths, see above), and thus actual sounds produced from a combination of horizontal and vertical tuning could have employed two different major thirds at the same time. Since the just major third 5/4 was accepted as a consonant interval of structural importance (most clearly by Zarlino 1558, 1573), keyboard tuning had to adapt to this situation and did so by inventing 1/4-comma meantone (which was outlined, in a practical way, by the organist and organ expert, Arnold Schlick in 1511). This tuning system saw a number of variants (like 1/5-comma) and extensions to more than 12 tones/pitches per octave in order to expand the compass of usable keys and chords but, in general, was implemented in its original form (see above). For example, it was found in the course of restoration that long flue pipes of the Schnitger organ at St. Jacobi of Hamburg (1689–93) had been left untouched in their 1/4-comma meantone tuning implemented by Schnitger (see Ahrend 1995).

Demands on tuning and temperament began to change around 1680–1720, when organists, many of them active also as composers (like Dietrich Buxtehude, Nikolaus Bruhns, Vincent Lübeck, Georg Böhm, and of course J. S. Bach) ventured into more remote keys (in the circle of fifths) and also used hitherto unknown chord progressions and modulations. As a parallel process, one has to note the transition from a predominantly modal organisation of music in the 16^th and still in the 17^th century to modern concepts of major and minor tonalities. While organ music preserved modal structures even in advanced compositions like Fischer’s Ariadne musica (1702, 1710) and J. S. Bach’s organ chorales as well as the Duetto BWV 802 from the third part of the Clavier-Übung (published 1739), elements of major/minor tonality are also ingredients of those works. It has been suggested that a performance of Bach’s Fantasia in g-minor (BWV 542, coupled with a fugue) would need an organ “to be well-tempered, though not necessarily equal-tempered” (Williams 1980, Vol. 1, p. 120). In fact, the modulations found in this work are far-reaching (from D-major to D_b-major in bars 31ff.) and involve no less than 25 different pitches if one would intend to play this section in just intonation. If played on a ‘well-tempered’ organ tuned to Werckmeister III like St. Wilhadi at Stade, with a conventional registration (HW: Prinzipal 8, Oktave 8’, 4’, 2’; Pedal: Subbass 16’, Oktave 8’, 4’), HNR measurements yield the pattern shown in Fig. 10.

Fig. 10.

J.S. Bach, Fantasia g-minor (BWV 542), bars 31–35, Werckmeister III tuning.

The modulation in this section proceeds around the circle of fifths from sharps to flats; the relevant chords in bars 31–36 are D – g – G – c – C – f – F – b_b – B_b – e_b – E_b – a_b – A_b – d_b – D_b – dim – e (Uppercase: Major, lowercase: minor). Figure 14 shows this modulation up to the first diminished chord (dim) in bar 35; in the Fantasia, it is followed by two more sonorities which resolve to an e-minor chord in bar 36. From Fig. 14, it is obvious that the Werckmeister tuning yields quite good HNR readings for a number of major (D, G, C, F, Des [D_b]) and minor chords (g, c, f) while some others are acceptable (B [B_b], Es [E_b], As [A_b], as [a_b], des [d_b]), and the remaining b [b_b] and es (e_b) are rather poor. Thus, there is a grading with respect to spectral fusion versus roughness, as one would expect from a temperament such as Werckmeister III. The problem that becomes obvious is that with advanced compositions employing far more than 12 tones (as identified from the notation), a tuning system suited to realise harmonic and melodic structures with precise intonation also would need more than 12 tones and pitches per octave. In this respect, common temperaments like Werckmeister III or ET12 fall short of providing adequate acoustical means for the performance of music that makes use of advanced harmony. After various temperaments had been explored, from ca. 1500–1800, in tuning organs and other keyboards (see Lindley 1987, Ratte 1991), ET12 finally was accepted as a standard mainly for practical reasons. In this process, the newly invented pianoforte played a “key” role because the vast number of instruments manufactured in Europe per year required certain conventions in regard to compass, tuning, and standard pitch (of a¹ = 435 Hz, which came as late as 1858 in France and 1885 as international agreement).

As a tuning and pitch system, ET12 incorporates slightly narrowed fifths and markedly enlarged major thirds as well as narrowed minor thirds. In effect, ET12 is closer to Pythagorean tuning than to just intonation. Since the octave is divided into 12 steps of equal size (100 cents), ET12 allows modulation from arbitrary starting points to whatever target (key/chord) is chosen. However, the difference between sharps and flats is levelled, and none of the intervals besides the octave is just. In regard to historic organs, their mixtures and stops including just major thirds like the Sesquialter and especially the Terzzimbel did not fit, in their sound structure of harmonic partials, to the horizontal ET12 tuning, which, unlike Werckmeister III or a similar ‘Wohltemperierung’, does not provide for some keys and chords with major thirds closer to the 5/4 than the 81/64 interval. As a matter of fact, when ET12 was implemented in historic organs all over Germany and adjacent regions, ca. 1780–1870, in particular, mixture stops with pipes tuned to major thirds were altered or completely removed. Quite many old reed-pipe stops (like Trechterregal, Bärpfeife, Dulzian) met the same fate and were dismissed for their rich harmonic sounds (that is, for the quality that once had made reed pipes so attractive for organ builders and musicians alike). Instead of such stops, pipe ranks with a more mellow sound emulating bowed strings and other ‘orchestral colours’ were installed on wind chests to accommodate a much different concept of organ music inspired by predominantly homophonic genres.

6 Concluding Remarks

The design and construction of pipe organs from medieval times to the Baroque era show remarkable achievements in regard to mensuration and technical manufacture of pipes, formation of pipe ranks and stops as well as setting up a disposition for each organ where stops combine into an overall sonic unit. At the same time, they maintain a characteristic timbral sound quality. Such a concept of ‘diversity within unity’ became evident in particular in the 16^th century when organs were built, perhaps first in the Rhineland and the Low Countries (see Klotz 1975, 93ff.), with a growing number and diversity of stops. It is from this era that the typical tripartite organisation of stops results, that is, there are (a) diapason pipes of different foot lengths with relatively narrow diameters as well as mixtures and the occasional Zimbel; there are (b) flute-like stops with pipes of a wider diameter; and there are (c) various reed stops emulating reed instruments and horns of the Renaissance. In the course of the 16^th and further, in the 17^th century, the divisions of larger organs became well-equipped with stops from these three groups, whereby in particular, the pedal chest gained in volume and gravity. This was a condition prerequisite to assigning voices to the pedal for the performance of polyphonic settings such as bicinia, canons, or fugues. One can see a clear interdependency between developments in organ design and construction, on the one hand, and compositional practice, on the other. Organ building reached a zenith already around 1600, with a number of large three-manual organs (plus pedal) as listed by Praetorius (1619). It seems that Gottfried Fritzsche was the first to build a fourth division equipped with its own clavier as he expanded the BW in St. Jacobi, Hamburg (1635/36; Fock 1974, 55f.). This organ was enlarged and improved by Arp Schnitger (1689–93) and is fully restored (except that 1/5-comma meantone tuning has been substituted for the original 1/4-comma to allow for a wider range of usable keys). Another organ built by Schnitger with four divisions and four manuals for St. Ludgeri at Norden (1686–88, 1692, IV, Ped, 46 voices; see Vogel et al. 1997, Edskes & Vogel 2009) is of interest since it had to be fitted into a church of unusual architecture, where the nave is much lower in height than the transept and the choir. Schnitger chose to place the organ on a balcony on a side wall of the choir, which extends ‘round the corner’ into the transept. While HW, OW, BW and RP radiate their sound into the choir, most of the pipes in a single huge bass tower (which contains all pedal stops) ‘speak’ into the transept. The organ at St. Ludgeri proves masters like Schnitger could solve complex mechanical and even acoustical problems.

Significant developments in organ building between ca. 1500 and 1700 gave organists, many of them composers as well, ample opportunity to create a wealth of works written to be performed on ‘the queen of all instruments’, the pipe organ. As Praetorius (1619, 85) remarked, the organ should incorporate all other instruments by emulating their peculiar sound characteristics. The great variety of organ stops found in late Renaissance and Baroque organs and the diversity of sounds they produce must be regarded as an important part of our sonic and musical heritage.